Merge branch 'RED-10752-main' into 'main'

RED-10752: Enabled prometheus

See merge request fforesight/layout-parser!267

.gitattributes

@@ -0,0 +1 @@
+*.pdf filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -42,3 +42,5 @@ gradlew.bat
 gradlew
 gradle.properties
 gradle/
+.DS_Store
+.DS_Store/

.gitlab-ci.yml

@@ -1,3 +1,7 @@
+variables:
+  #  SONAR_PROJECT_KEY: 'fforesight_layout-parser_AYd5quv2mRkBOCG22hvF'
+  GIT_SUBMODULE_STRATEGY: recursive
+  GIT_SUBMODULE_FORCE_HTTPS: 'true'
 include:
   - project: 'gitlab/gitlab'
     ref: 'main'
@@ -17,5 +21,6 @@ deploy:
       dotenv: version.env
   rules:
     - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH 
+    - if: $CI_COMMIT_BRANCH =~ /^feature/ && $CI_COMMIT_TAG == ""
     - if: $CI_COMMIT_BRANCH =~ /^release/
     - if: $CI_COMMIT_TAG

.gitmodules

@@ -0,0 +1,8 @@
+[submodule "layoutparser-service/layoutparser-service-server/src/test/resources/files/basf"]
+	path = layoutparser-service/layoutparser-service-server/src/test/resources/files/basf
+	url = ssh://git@git.knecon.com:22222/fforesight/documents/basf.git
+	update = merge
+[submodule "layoutparser-service/layoutparser-service-server/src/test/resources/files/syngenta"]
+	path = layoutparser-service/layoutparser-service-server/src/test/resources/files/syngenta
+	url = ssh://git@git.knecon.com:22222/fforesight/documents/syngenta.git
+	update = merge

buildSrc/src/main/kotlin/com.knecon.fforesight.java-conventions.gradle.kts

@@ -8,6 +8,8 @@ plugins {
 
 group = "com.knecon.fforesight"
 
+val documentVersion by rootProject.extra { "4.433.0" }
+
 java.sourceCompatibility = JavaVersion.VERSION_17
 java.targetCompatibility = JavaVersion.VERSION_17
 
@@ -42,6 +44,19 @@ tasks.jacocoTestReport {
 }
 
 allprojects {
+
+    tasks.withType<Javadoc> {
+        options {
+            this as StandardJavadocDocletOptions
+            addBooleanOption("Xdoclint:none", true)
+            addStringOption("Xmaxwarns", "1")
+        }
+    }
+
+    pmd {
+        setConsoleOutput(true)
+    }
+
     publishing {
         publications {
             create<MavenPublication>(name) {
@@ -64,6 +79,7 @@ java {
     withJavadocJar()
 }
 
+
 repositories {
     mavenLocal()
     mavenCentral()

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/DocumentData.java

@@ -1,28 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.internal.api.data.redaction;
-
-import java.io.Serializable;
-
-import io.swagger.v3.oas.annotations.media.Schema;
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
-@Schema(description = "Object containing the complete document layout parsing information. It is split into 4 categories, structure, text, positions and pages: " + "The document tree structure of SemanticNodes such as Section, Paragraph, Headline, etc. " + "The text, which is stored as separate blocks of data. " + "The text positions, which are also stored as separate blocks. The Blocks are equal to the text blocks in length and order. " + "The page information.")
-public class DocumentData implements Serializable {
-
-    @Schema(description = "Contains information about the document's pages.")
-    DocumentPage[] documentPages;
-    @Schema(description = "Contains information about the document's text.")
-    DocumentTextData[] documentTextData;
-    @Schema(description = "Contains information about the document's text positions.")
-    DocumentPositionData[] documentPositions;
-    @Schema(description = "Contains information about the document's semantic structure.")
-    DocumentStructure documentStructure;
-
-}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/DocumentPage.java

@@ -1,30 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.internal.api.data.redaction;
-
-import java.io.Serializable;
-
-import io.swagger.v3.oas.annotations.media.Schema;
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-@Schema(description = "Object containing information about the document's pages.")
-public class DocumentPage implements Serializable {
-
-    @Schema(description = "The page number, starting with 1.")
-    int number;
-    @Schema(description = "The page height in PDF user units.", example = "792")
-    int height;
-    @Schema(description = "The page width in PDF user units.", example = "694")
-    int width;
-    @Schema(description = "The page rotation as specified by the PDF.", example = "90", allowableValues = {"0", "90", "180", "270"})
-    int rotation;
-
-}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/DocumentPositionData.java

@@ -1,28 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.internal.api.data.redaction;
-
-import java.io.Serializable;
-
-import io.swagger.v3.oas.annotations.media.Schema;
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-@Schema(description = "Object containing text positional information of a specific text block. A document is split into multiple text blocks, which are supposed to be read in order. Every text block can only occur on a single page.")
-public class DocumentPositionData implements Serializable {
-
-    @Schema(description = "Identifier of the text block.")
-    Long id;
-    @Schema(description = "For each string coordinate in the search text of the text block, the array contains an entry relating the string coordinate to the position coordinate. This is required due to the text and position coordinates not being equal.")
-    int[] stringIdxToPositionIdx;
-    @Schema(description = "The bounding box for each glyph as a rectangle. This matrix is of size (n,4), where n is the number of glyphs in the text block. The second dimension specifies the rectangle with the value x, y, width, height, with x, y specifying the lower left corner. In order to access this information, the stringIdxToPositionIdx array must be used to transform the coordinates.")
-    float[][] positions;
-
-}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/DocumentStructure.java

@@ -1,144 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.internal.api.data.redaction;
-
-import java.awt.geom.Rectangle2D;
-import java.io.Serializable;
-import java.util.Arrays;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-import java.util.stream.Stream;
-
-import io.swagger.v3.oas.annotations.media.Schema;
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-@Schema(description = "Object containing information about the parsed tree structure of the SemanticNodes, such as Section, Paragraph, Headline etc inside of the document.")
-public class DocumentStructure implements Serializable {
-
-    @Schema(description = "The root EntryData represents the Document.")
-    EntryData root;
-
-    @Schema(description = "Object containing the extra field names, a table has in its properties field.")
-    public static class TableProperties implements Serializable {
-
-        public static final String NUMBER_OF_ROWS = "numberOfRows";
-        public static final String NUMBER_OF_COLS = "numberOfCols";
-
-    }
-
-    @Schema(description = "Object containing the extra field names, an Image has in its properties field.")
-    public static class ImageProperties implements Serializable {
-
-        public static final String TRANSPARENT = "transparent";
-        public static final String IMAGE_TYPE = "imageType";
-        public static final String POSITION = "position";
-        public static final String ID = "id";
-
-    }
-
-    @Schema(description = "Object containing the extra field names, a table cell has in its properties field.")
-    public static class TableCellProperties implements Serializable {
-
-        public static final String B_BOX = "bBox";
-        public static final String ROW = "row";
-        public static final String COL = "col";
-        public static final String HEADER = "header";
-
-    }
-
-    public static final String RECTANGLE_DELIMITER = ";";
-
-
-    public static Rectangle2D parseRectangle2D(String bBox) {
-
-        List<Float> floats = Arrays.stream(bBox.split(RECTANGLE_DELIMITER)).map(Float::parseFloat).toList();
-        return new Rectangle2D.Float(floats.get(0), floats.get(1), floats.get(2), floats.get(3));
-    }
-
-
-    public EntryData get(List<Integer> tocId) {
-
-        if (tocId.isEmpty()) {
-            return root;
-        }
-        EntryData entry = root.children.get(tocId.get(0));
-        for (int id : tocId.subList(1, tocId.size())) {
-            entry = entry.children.get(id);
-        }
-        return entry;
-    }
-
-
-    public Stream<EntryData> streamAllEntries() {
-
-        return Stream.concat(Stream.of(root), root.children.stream()).flatMap(DocumentStructure::flatten);
-    }
-
-
-    public String toString() {
-
-        return String.join("\n", streamAllEntries().map(EntryData::toString).toList());
-    }
-
-
-    private static Stream<EntryData> flatten(EntryData entry) {
-
-        return Stream.concat(Stream.of(entry), entry.children.stream().flatMap(DocumentStructure::flatten));
-    }
-
-
-    @Data
-    @Builder
-    @NoArgsConstructor
-    @AllArgsConstructor
-    @FieldDefaults(level = AccessLevel.PRIVATE)
-    @Schema(description = "Object containing information of a SemanticNode and also structuring the layout with children.")
-    public static class EntryData implements Serializable {
-
-        @Schema(description = "Type of the semantic node.", allowableValues = {"DOCUMENT", "SECTION", "PARAGRAPH", "HEADLINE", "TABLE", "TABLE_CELL", "HEADER", "FOOTER", "IMAGE"})
-        NodeType type;
-        @Schema(description = "Specifies the position in the parsed tree structure.", example = "[1, 0, 2]")
-        int[] treeId;
-        @Schema(description = "Specifies the text block IDs associated with this semantic node. The value should be joined with the DocumentTextData/DocumentPositionData. Is empty, if no text block is directly associated with this semantic node. Only Paragraph, Headline, Header or Footer is directly associated with a text block.", example = "[1]")
-        Long[] atomicBlockIds;
-        @Schema(description = "Specifies the pages this semantic node appears on. The value should be joined with the PageData.", example = "[1, 2, 3]")
-        Long[] pageNumbers;
-        @Schema(description = "Some semantic nodes have additional information, this information is stored in this Map. The extra fields are specified by the Properties subclasses.", example = "For a Table: {\"numberOfRows\": 3, \"numberOfCols\": 4}")
-        Map<String, String> properties;
-        @Schema(description = "All child Entries of this Entry.", example = "[1, 2, 3]")
-        List<EntryData> children;
-        @Schema(description = "Describes the origin of the semantic node", example = "[ALGORITHM]")
-        Set<LayoutEngine> engines;
-
-
-        @Override
-        public String toString() {
-
-            StringBuilder sb = new StringBuilder();
-            sb.append("[");
-            for (int i : treeId) {
-                sb.append(i);
-                sb.append(",");
-            }
-            sb.delete(sb.length() - 1, sb.length());
-            sb.append("]: ");
-
-            sb.append(type);
-            sb.append(" atbs = ");
-            sb.append(atomicBlockIds.length);
-
-            return sb.toString();
-        }
-
-    }
-
-}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/DocumentTextData.java

@@ -1,36 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.internal.api.data.redaction;
-
-import java.io.Serializable;
-
-import io.swagger.v3.oas.annotations.media.Schema;
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-@Schema(description = "Object containing text information of a specific text block. A document is split into multiple text blocks, which are supposed to be read in order. Every text block can only occur on a single page.")
-public class DocumentTextData implements Serializable {
-
-    @Schema(description = "Identifier of the text block.")
-    Long id;
-    @Schema(description = "The page the text block occurs on.")
-    Long page;
-    @Schema(description = "The text the text block.")
-    String searchText;
-    @Schema(description = "Each text block is assigned a number on a page, starting from 0.")
-    int numberOnPage;
-    @Schema(description = "The text blocks are ordered, this number represents the start of the text block as a string offset.")
-    int start;
-    @Schema(description = "The text blocks are ordered, this number represents the end of the text block as a string offset.")
-    int end;
-    @Schema(description = "The line breaks in the text of this semantic node in string offsets. They are exclusive end. At the end of each semantic node there is an implicit linebreak.", example = "[5, 10]")
-    int[] lineBreaks;
-
-}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/LayoutEngine.java

@@ -1,6 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.internal.api.data.redaction;
-
-public enum LayoutEngine {
-    ALGORITHM,
-    AI
-}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/NodeType.java

@@ -1,22 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.internal.api.data.redaction;
-
-import java.io.Serializable;
-import java.util.Locale;
-
-public enum NodeType implements Serializable {
-    DOCUMENT,
-    SECTION,
-    HEADLINE,
-    PARAGRAPH,
-    TABLE,
-    TABLE_CELL,
-    IMAGE,
-    HEADER,
-    FOOTER;
-
-
-    public String toString() {
-
-        return this.name().charAt(0) + this.name().substring(1).toLowerCase(Locale.ROOT);
-    }
-}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/SimplifiedSectionText.java

@@ -14,7 +14,7 @@ import lombok.NoArgsConstructor;
 public class SimplifiedSectionText {
 
     @Schema(description = "The number of this Section. This is used to map the simplified section text back to the original Section.")
-    private int sectionNumber;
+    private String sectionNumber;
     @Schema(description = "The text in this Section.")
     private String text;
 

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/SimplifiedText.java

@@ -19,6 +19,16 @@ public class SimplifiedText {
     @Schema(description = "Number of pages in the entire document.")
     private int numberOfPages;
     @Schema(description = "A List of simplified Sections, which contains almost exclusively text.")
+    @Builder.Default
     private List<SimplifiedSectionText> sectionTexts = new ArrayList<>();
+    @Schema(description = "A list of the main section numbers ")
+    @Builder.Default
+    private List<String> mainSectionNumbers = new ArrayList<>();
+    @Schema(description = "A list of the header section numbers ")
+    @Builder.Default
+    private List<String> headerSectionNumbers = new ArrayList<>();
+    @Schema(description = "A list of the footer section numbers ")
+    @Builder.Default
+    private List<String> footerSectionNumbers = new ArrayList<>();
 
 }

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/taas/StructureObject.java

@@ -1,5 +1,7 @@
 package com.knecon.fforesight.service.layoutparser.internal.api.data.taas;
 
+import java.util.List;
+
 import io.swagger.v3.oas.annotations.media.Schema;
 import lombok.AllArgsConstructor;
 import lombok.Builder;
@@ -13,6 +15,8 @@ public class StructureObject {
 
     @Schema(description = "The ID of this StructureObject.")
     Integer structureObjectNumber;
+    @Schema(description = "The Tree ID of this StructureObject.")
+    List<Integer> treeId;
     @Schema(description = "This value indicates the start of the string offsets in this Object, with respect to the reading order.")
     int page;
     @Schema(description = "This stringOffset indicates the start of the string offsets in this Object, with respect to the reading order of the entire document. It is equal to the previous' StructureObject stringOffset + its length.")

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/queue/LayoutParsingFinishedEvent.java

@@ -8,13 +8,20 @@ import lombok.Builder;
 @Builder
 @Schema(description = "Object containing information about the layout parsing.")
 public record LayoutParsingFinishedEvent(
-        @Schema(description = "General purpose identifier. It is returned exactly the same way it is inserted with the LayoutParsingRequest.")
-        Map<String, String> identifier,//
-        @Schema(description = "The duration of a single layout parsing in ms.")
-        long duration,//
-        @Schema(description = "The number of pages of the parsed document.")
-        int numberOfPages,//
-        @Schema(description = "A general message. It contains some information useful for a developer, like the paths where the files are stored. Not meant to be machine readable.")
-        String message) {
+        @Schema(description = "General purpose identifier. It is returned exactly the same way it is inserted with the LayoutParsingRequest.") //
+        Map<String, String> identifier,
+
+        @Schema(description = "The duration of a single layout parsing in ms.") //
+        long duration,
+
+        @Schema(description = "The number of pages of the parsed document.") //
+        int numberOfPages,
+
+        @Schema(description = "A general message. It contains some information useful for a developer, like the paths where the files are stored. Not meant to be machine readable.") //
+        String message,
+
+        @Schema(description = "The app version of the layout parser.") //
+        String layoutParserVersion
+) {
 
 }

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/queue/LayoutParsingQueueNames.java

@@ -2,7 +2,9 @@ package com.knecon.fforesight.service.layoutparser.internal.api.queue;
 
 public class LayoutParsingQueueNames {
 
-    public static final String LAYOUT_PARSING_REQUEST_QUEUE = "layout_parsing_request_queue";
-    public static final String LAYOUT_PARSING_DLQ = "layout_parsing_dead_letter_queue";
-    public static final String LAYOUT_PARSING_FINISHED_EVENT_QUEUE = "layout_parsing_response_queue";
+    public static final String LAYOUT_PARSING_REQUEST_QUEUE_PREFIX = "layout_parsing_request";
+    public static final String LAYOUT_PARSING_REQUEST_EXCHANGE = "layout_parsing_request_exchange";
+    public static final String LAYOUT_PARSING_RESPONSE_QUEUE_PREFIX = "layout_parsing_response";
+    public static final String LAYOUT_PARSING_RESPONSE_EXCHANGE = "layout_parsing_response_exchange";
+    public static final String LAYOUT_PARSING_DLQ = "layout_parsing_error";
 }

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/queue/LayoutParsingRequest.java

@@ -19,7 +19,6 @@ public record LayoutParsingRequest(
         @Schema(description = "Path to the original PDF file.")//
         @NonNull String originFileStorageId,//
 
-
         @Schema(description = "Optional Path to the table extraction file.")//
         Optional<String> tablesFileStorageId,//
         @Schema(description = "Optional Path to the image classification file.")//
@@ -37,9 +36,12 @@ public record LayoutParsingRequest(
         @NonNull String positionBlockFileStorageId,//
         @Schema(description = "Path where the Document Pages File will be stored.")//
         @NonNull String pageFileStorageId,//
+        @Schema(description = "Path where the Document Markdown File will be stored.")//
+        Optional<String> documentMarkdownFileStorageId,//
         @Schema(description = "Path where the Simplified Text File will be stored.")//
         @NonNull String simplifiedTextStorageId,//
         @Schema(description = "Path where the Viewer Document PDF will be stored.")//
-        @NonNull String viewerDocumentStorageId) {
+        @NonNull String viewerDocumentStorageId
+) {
 
 }

layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/queue/LayoutParsingType.java

@@ -2,6 +2,11 @@ package com.knecon.fforesight.service.layoutparser.internal.api.queue;
 
 public enum LayoutParsingType {
     REDACT_MANAGER,
-    TAAS,
-    DOCUMINE
+    REDACT_MANAGER_OLD,
+    REDACT_MANAGER_PARAGRAPH_DEBUG,
+    REDACT_MANAGER_WITHOUT_DUPLICATE_PARAGRAPH,
+    DOCUMINE,
+    DOCUMINE_OLD,
+    CLARIFYND,
+    CLARIFYND_PARAGRAPH_DEBUG
 }

layoutparser-service/layoutparser-service-processor/build.gradle.kts

@@ -8,20 +8,33 @@ description = "layoutparser-service-processor"
 val jacksonVersion = "2.15.2"
 val pdfBoxVersion = "3.0.0"
 
+
 dependencies {
     implementation(project(":layoutparser-service-internal-api"))
     implementation(project(":viewer-doc-processor"))
 
-    implementation("com.iqser.red.service:persistence-service-shared-api-v1:2.144.0") {
+    implementation("com.knecon.fforesight:document:${rootProject.extra.get("documentVersion")}")
+    implementation("com.iqser.red.service:persistence-service-shared-api-v1:2.564.0-RED9010.0") {
         exclude("org.springframework.boot", "spring-boot-starter-security")
         exclude("org.springframework.boot", "spring-boot-starter-validation")
     }
-    implementation("com.knecon.fforesight:tenant-commons:0.21.0")
-    implementation("com.iqser.red.commons:storage-commons:2.45.0")
+    implementation("com.knecon.fforesight:tenant-commons:0.30.0") {
+        exclude("com.iqser.red.commons", "storage-commons")
+    }
+    implementation("com.iqser.red.commons:storage-commons:2.50.0")
 
     implementation("org.apache.pdfbox:pdfbox:${pdfBoxVersion}")
     implementation("org.apache.pdfbox:pdfbox-tools:${pdfBoxVersion}")
     implementation("com.fasterxml.jackson.module:jackson-module-afterburner:${jacksonVersion}")
     implementation("com.fasterxml.jackson.datatype:jackson-datatype-jsr310:${jacksonVersion}")
     implementation("org.springframework.boot:spring-boot-starter-web:3.1.3")
+    implementation("org.jgrapht:jgrapht-core:1.5.2")
+    implementation("org.apache.pdfbox:jbig2-imageio:3.0.4")
+    implementation("com.github.jai-imageio:jai-imageio-core:1.4.0")
+    implementation("com.github.jai-imageio:jai-imageio-jpeg2000:1.4.0")
+    implementation("org.tinspin:tinspin-indexes:2.1.3")
+    implementation("org.commonmark:commonmark:0.22.0")
+    implementation("org.commonmark:commonmark-ext-gfm-tables:0.22.0")
+    implementation("com.pdftron:PDFNet:10.11.0")
+    implementation("org.apache.commons:commons-text:1.12.0")
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/LayoutParserSettings.java

@@ -0,0 +1,20 @@
+package com.knecon.fforesight.service.layoutparser.processor;
+
+import org.springframework.boot.context.properties.ConfigurationProperties;
+import org.springframework.context.annotation.Configuration;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingType;
+
+import lombok.AccessLevel;
+import lombok.Data;
+import lombok.experimental.FieldDefaults;
+
+@Data
+@Configuration
+@ConfigurationProperties("layoutparser")
+@FieldDefaults(level = AccessLevel.PRIVATE)
+public class LayoutParserSettings {
+
+    boolean debug;
+    LayoutParsingType layoutParsingTypeOverride;
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/LayoutParsingPipeline.java

@@ -2,12 +2,15 @@ package com.knecon.fforesight.service.layoutparser.processor;
 
 import static java.lang.String.format;
 
+import java.awt.geom.AffineTransform;
 import java.awt.geom.Rectangle2D;
 import java.io.File;
 import java.io.IOException;
 import java.nio.file.Files;
 import java.util.ArrayList;
+import java.util.Collection;
 import java.util.HashMap;
+import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
 import java.util.concurrent.atomic.AtomicReference;
@@ -17,20 +20,35 @@ import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDPage;
 import org.apache.pdfbox.pdmodel.common.PDRectangle;
 import org.apache.pdfbox.pdmodel.documentinterchange.markedcontent.PDMarkedContent;
+import org.springframework.beans.factory.annotation.Value;
 import org.springframework.stereotype.Service;
 
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
+import com.iqser.red.service.redaction.v1.server.mapper.DocumentDataMapper;
+import com.iqser.red.service.redaction.v1.server.model.document.nodes.ImageType;
+import com.iqser.red.service.redaction.v1.server.model.document.nodes.NodeType;
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingFinishedEvent;
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingRequest;
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingType;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Character;
+import com.knecon.fforesight.service.layoutparser.processor.model.DocumentWithVisualization;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.Ruling;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextDirection;
+import com.knecon.fforesight.service.layoutparser.processor.services.classification.ClassificationService;
+import com.knecon.fforesight.service.layoutparser.processor.services.mapper.MarkdownMapper;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationDocument;
 import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
+
 import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.model.outline.OutlineExtractorService;
+import com.knecon.fforesight.service.layoutparser.processor.model.outline.SectionTreeBuilderService;
+import com.knecon.fforesight.service.layoutparser.processor.model.outline.SectionTreeEnhancementService;
+import com.knecon.fforesight.service.layoutparser.processor.model.outline.SectionTree;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.Cell;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.Word;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.adapter.CvTableParsingAdapter;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.adapter.ImageServiceResponseAdapter;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.adapter.VisualLayoutParsingAdapter;
@@ -38,23 +56,24 @@ import com.knecon.fforesight.service.layoutparser.processor.python_api.model.ima
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableCells;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
-import com.knecon.fforesight.service.layoutparser.processor.services.BodyTextFrameService;
 import com.knecon.fforesight.service.layoutparser.processor.services.RulingCleaningService;
 import com.knecon.fforesight.service.layoutparser.processor.services.SectionsBuilderService;
 import com.knecon.fforesight.service.layoutparser.processor.services.SimplifiedSectionTextService;
 import com.knecon.fforesight.service.layoutparser.processor.services.TableExtractionService;
+import com.knecon.fforesight.service.layoutparser.processor.services.TextRulingsClassifier;
+import com.knecon.fforesight.service.layoutparser.processor.services.blockification.BlockificationPostprocessingService;
+import com.knecon.fforesight.service.layoutparser.processor.services.blockification.DocstrumBlockificationService;
 import com.knecon.fforesight.service.layoutparser.processor.services.blockification.DocuMineBlockificationService;
 import com.knecon.fforesight.service.layoutparser.processor.services.blockification.RedactManagerBlockificationService;
-import com.knecon.fforesight.service.layoutparser.processor.services.blockification.TaasBlockificationService;
-import com.knecon.fforesight.service.layoutparser.processor.services.classification.DocuMineClassificationService;
-import com.knecon.fforesight.service.layoutparser.processor.services.classification.RedactManagerClassificationService;
-import com.knecon.fforesight.service.layoutparser.processor.services.classification.TaasClassificationService;
 import com.knecon.fforesight.service.layoutparser.processor.services.factory.DocumentGraphFactory;
-import com.knecon.fforesight.service.layoutparser.processor.services.mapper.DocumentDataMapper;
+import com.knecon.fforesight.service.layoutparser.processor.services.graphics.Box;
+import com.knecon.fforesight.service.layoutparser.processor.services.graphics.GraphicExtractorService;
 import com.knecon.fforesight.service.layoutparser.processor.services.mapper.TaasDocumentDataMapper;
 import com.knecon.fforesight.service.layoutparser.processor.services.parsing.PDFLinesTextStripper;
 import com.knecon.fforesight.service.layoutparser.processor.services.visualization.LayoutGridService;
 import com.knecon.fforesight.service.layoutparser.processor.utils.MarkedContentUtils;
+import com.knecon.fforesight.service.layoutparser.processor.utils.PageInformation;
+import com.knecon.fforesight.service.layoutparser.processor.utils.TextPositionOperations;
 
 import io.micrometer.observation.Observation;
 import io.micrometer.observation.ObservationRegistry;
@@ -69,26 +88,32 @@ import lombok.extern.slf4j.Slf4j;
 @Slf4j
 @Service
 @RequiredArgsConstructor
-@FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
+@FieldDefaults(level = AccessLevel.PRIVATE)
 public class LayoutParsingPipeline {
 
-    ImageServiceResponseAdapter imageServiceResponseAdapter;
-    CvTableParsingAdapter cvTableParsingAdapter;
-    LayoutParsingStorageService layoutParsingStorageService;
-    SectionsBuilderService sectionsBuilderService;
-    TaasClassificationService taasClassificationService;
-    RedactManagerClassificationService redactManagerClassificationService;
-    DocuMineClassificationService docuMineClassificationService;
-    SimplifiedSectionTextService simplifiedSectionTextService;
-    BodyTextFrameService bodyTextFrameService;
-    RulingCleaningService rulingCleaningService;
-    TableExtractionService tableExtractionService;
-    TaasBlockificationService taasBlockificationService;
-    DocuMineBlockificationService docuMineBlockificationService;
-    RedactManagerBlockificationService redactManagerBlockificationService;
-    LayoutGridService layoutGridService;
-    ObservationRegistry observationRegistry;
-    VisualLayoutParsingAdapter visualLayoutParsingAdapter;
+    final ImageServiceResponseAdapter imageServiceResponseAdapter;
+    final CvTableParsingAdapter cvTableParsingAdapter;
+    final LayoutParsingStorageService layoutParsingStorageService;
+    final SectionsBuilderService sectionsBuilderService;
+    final SimplifiedSectionTextService simplifiedSectionTextService;
+    final RulingCleaningService rulingCleaningService;
+    final TableExtractionService tableExtractionService;
+    final DocuMineBlockificationService docuMineBlockificationService;
+    final RedactManagerBlockificationService redactManagerBlockificationService;
+    final BlockificationPostprocessingService blockificationPostprocessingService;
+    final DocstrumBlockificationService docstrumBlockificationService;
+    final LayoutGridService layoutGridService;
+    final ObservationRegistry observationRegistry;
+    final VisualLayoutParsingAdapter visualLayoutParsingAdapter;
+    final GraphicExtractorService graphicExtractorService;
+    final OutlineExtractorService outlineExtractorService;
+    final SectionTreeBuilderService sectionTreeBuilderService;
+    final SectionTreeEnhancementService sectionTreeEnhancementService;
+    final LayoutParserSettings settings;
+    final ClassificationService classificationService;
+
+    @Value("${LAYOUT_PARSER_VERSION:}")
+    private String layoutParserVersion;
 
 
     public LayoutParsingFinishedEvent parseLayoutAndSaveFilesToStorage(LayoutParsingRequest layoutParsingRequest) throws IOException {
@@ -100,62 +125,59 @@ public class LayoutParsingPipeline {
         File viewerDocumentFile = layoutParsingStorageService.getViewerDocFile(layoutParsingRequest.viewerDocumentStorageId())
                 .orElse(originFile);
 
-        VisualLayoutParsingResponse visualLayoutParsingResponse = new VisualLayoutParsingResponse();
-        if (layoutParsingRequest.visualLayoutParsingFileId()
-                .isPresent()) {
-            visualLayoutParsingResponse = layoutParsingStorageService.getVisualLayoutParsingFile(layoutParsingRequest.visualLayoutParsingFileId()
-                                                                                                         .get());
-        }
+        VisualLayoutParsingResponse visualLayoutParsingResponse = layoutParsingRequest.visualLayoutParsingFileId()
+                .map(layoutParsingStorageService::getVisualLayoutParsingFile)
+                .orElse(new VisualLayoutParsingResponse());
+        ImageServiceResponse imageServiceResponse = layoutParsingRequest.imagesFileStorageId()
+                .map(layoutParsingStorageService::getImagesFile)
+                .orElse(new ImageServiceResponse());
+        TableServiceResponse tableServiceResponse = layoutParsingRequest.tablesFileStorageId()
+                .map(layoutParsingStorageService::getTablesFile)
+                .orElse(new TableServiceResponse());
 
-        ImageServiceResponse imageServiceResponse = new ImageServiceResponse();
-        if (layoutParsingRequest.imagesFileStorageId()
-                .isPresent()) {
-            imageServiceResponse = layoutParsingStorageService.getImagesFile(layoutParsingRequest.imagesFileStorageId()
-                                                                                     .get());
-        }
+        LayoutParsingType layoutParsingType = settings.getLayoutParsingTypeOverride() == null //
+                ? layoutParsingRequest.layoutParsingType() : settings.getLayoutParsingTypeOverride();
 
-        TableServiceResponse tableServiceResponse = new TableServiceResponse();
-        if (layoutParsingRequest.tablesFileStorageId()
-                .isPresent()) {
-            tableServiceResponse = layoutParsingStorageService.getTablesFile(layoutParsingRequest.tablesFileStorageId()
-                                                                                     .get());
-        }
-
-        ClassificationDocument classificationDocument = parseLayout(layoutParsingRequest.layoutParsingType(),
+        ClassificationDocument classificationDocument = parseLayout(layoutParsingType,
                                                                     originFile,
                                                                     imageServiceResponse,
                                                                     tableServiceResponse,
                                                                     visualLayoutParsingResponse,
-                                                                    layoutParsingRequest.identifier().toString());
+                                                                    layoutParsingRequest.identifier());
 
         log.info("Building document graph for {}", layoutParsingRequest.identifier());
 
-        Document documentGraph = observeBuildDocumentGraph(classificationDocument);
+        DocumentWithVisualization documentWithVisualization = observeBuildDocumentGraph(layoutParsingType, classificationDocument);
 
         log.info("Creating viewer document for {}", layoutParsingRequest.identifier());
 
-        layoutGridService.addLayoutGrid(viewerDocumentFile, documentGraph, viewerDocumentFile, false);
+        layoutGridService.addLayoutGrid(viewerDocumentFile, documentWithVisualization, viewerDocumentFile, layoutParsingType, layoutParserVersion, false);
 
         log.info("Storing resulting files for {}", layoutParsingRequest.identifier());
 
-        layoutParsingStorageService.storeDocumentData(layoutParsingRequest, DocumentDataMapper.toDocumentData(documentGraph));
-        layoutParsingStorageService.storeSimplifiedText(layoutParsingRequest, simplifiedSectionTextService.toSimplifiedText(documentGraph));
+        layoutParsingStorageService.storeDocumentData(layoutParsingRequest, DocumentDataMapper.toDocumentData(documentWithVisualization.document()));
+        if (layoutParsingRequest.documentMarkdownFileStorageId()
+                .isPresent()) {
+            layoutParsingStorageService.storeMarkdownFile(layoutParsingRequest.documentMarkdownFileStorageId()
+                                                                  .get(), new MarkdownMapper().toMarkdownContent(documentWithVisualization.document()));
+        }
+        layoutParsingStorageService.storeSimplifiedText(layoutParsingRequest, simplifiedSectionTextService.toSimplifiedText(documentWithVisualization.document()));
         layoutParsingStorageService.storeViewerDocument(layoutParsingRequest, viewerDocumentFile);
 
-        if (layoutParsingRequest.layoutParsingType().equals(LayoutParsingType.TAAS)) {
+        if (layoutParsingRequest.researchDocumentStorageId() != null) {
             log.info("Building research document data for {}", layoutParsingRequest.identifier());
-            var researchDocumentData = TaasDocumentDataMapper.fromDocument(documentGraph);
+            var researchDocumentData = TaasDocumentDataMapper.fromDocument(documentWithVisualization.document());
             layoutParsingStorageService.storeResearchDocumentData(layoutParsingRequest, researchDocumentData);
         }
 
         if (!viewerDocumentFile.equals(originFile)) {
-            viewerDocumentFile.delete();
+            assert !viewerDocumentFile.exists() || viewerDocumentFile.delete();
         }
-        originFile.delete();
+        assert !originFile.exists() || originFile.delete();
 
         return LayoutParsingFinishedEvent.builder()
                 .identifier(layoutParsingRequest.identifier())
-                .numberOfPages(documentGraph.getNumberOfPages())
+                .numberOfPages(documentWithVisualization.document().getNumberOfPages())
                 .duration(System.currentTimeMillis() - start)
                 .message(format("""
                                         Layout parsing has finished in %.02f s.
@@ -170,25 +192,26 @@ public class LayoutParsingPipeline {
                                         Viewer Doc: %s""",
                                 ((float) (System.currentTimeMillis() - start)) / 1000,
                                 layoutParsingRequest.identifier(),
-                                buildSemanticNodeCountMessage(documentGraph.getNumberOfPages(), documentGraph.buildSemanticNodeCounts()),
+                                buildSemanticNodeCountMessage(documentWithVisualization.document().getNumberOfPages(), documentWithVisualization.buildSemanticNodeCounts()),
                                 layoutParsingRequest.structureFileStorageId(),
                                 layoutParsingRequest.textBlockFileStorageId(),
                                 layoutParsingRequest.positionBlockFileStorageId(),
                                 layoutParsingRequest.pageFileStorageId(),
                                 layoutParsingRequest.simplifiedTextStorageId(),
                                 layoutParsingRequest.viewerDocumentStorageId()))
+                .layoutParserVersion(layoutParserVersion)
                 .build();
 
     }
 
 
-    private Document observeBuildDocumentGraph(ClassificationDocument classificationDocument) {
+    private DocumentWithVisualization observeBuildDocumentGraph(LayoutParsingType layoutParsingType, ClassificationDocument classificationDocument) {
 
-        AtomicReference<Document> documentReference = new AtomicReference<>();
+        AtomicReference<DocumentWithVisualization> documentReference = new AtomicReference<>();
 
         Observation.createNotStarted("LayoutParsingPipeline", observationRegistry)
                 .contextualName("build-document-graph")
-                .observe(() -> documentReference.set(DocumentGraphFactory.buildDocumentGraph(classificationDocument)));
+                .observe(() -> documentReference.set(DocumentGraphFactory.buildDocumentGraph(layoutParsingType, classificationDocument)));
 
         return documentReference.get();
     }
@@ -196,15 +219,15 @@ public class LayoutParsingPipeline {
 
     private String buildSemanticNodeCountMessage(int numberOfPages, Map<NodeType, Long> semanticNodeCounts) {
 
-        return String.format("%d pages with %d sections, %d headlines, %d paragraphs, %d tables with %d cells, %d headers, and %d footers parsed",
-                             numberOfPages,
-                             semanticNodeCounts.get(NodeType.SECTION) == null ? 0 : semanticNodeCounts.get(NodeType.SECTION),
-                             semanticNodeCounts.get(NodeType.HEADLINE) == null ? 0 : semanticNodeCounts.get(NodeType.HEADLINE),
-                             semanticNodeCounts.get(NodeType.PARAGRAPH) == null ? 0 : semanticNodeCounts.get(NodeType.PARAGRAPH),
-                             semanticNodeCounts.get(NodeType.TABLE) == null ? 0 : semanticNodeCounts.get(NodeType.TABLE),
-                             semanticNodeCounts.get(NodeType.TABLE_CELL) == null ? 0 : semanticNodeCounts.get(NodeType.TABLE_CELL),
-                             semanticNodeCounts.get(NodeType.HEADER) == null ? 0 : semanticNodeCounts.get(NodeType.HEADER),
-                             semanticNodeCounts.get(NodeType.FOOTER) == null ? 0 : semanticNodeCounts.get(NodeType.FOOTER));
+        return format("%d pages with %d sections, %d headlines, %d paragraphs, %d tables with %d cells, %d headers, and %d footers parsed",
+                      numberOfPages,
+                      semanticNodeCounts.get(NodeType.SECTION) == null ? 0 : semanticNodeCounts.get(NodeType.SECTION),
+                      semanticNodeCounts.get(NodeType.HEADLINE) == null ? 0 : semanticNodeCounts.get(NodeType.HEADLINE),
+                      semanticNodeCounts.get(NodeType.PARAGRAPH) == null ? 0 : semanticNodeCounts.get(NodeType.PARAGRAPH),
+                      semanticNodeCounts.get(NodeType.TABLE) == null ? 0 : semanticNodeCounts.get(NodeType.TABLE),
+                      semanticNodeCounts.get(NodeType.TABLE_CELL) == null ? 0 : semanticNodeCounts.get(NodeType.TABLE_CELL),
+                      semanticNodeCounts.get(NodeType.HEADER) == null ? 0 : semanticNodeCounts.get(NodeType.HEADER),
+                      semanticNodeCounts.get(NodeType.FOOTER) == null ? 0 : semanticNodeCounts.get(NodeType.FOOTER));
     }
 
 
@@ -215,16 +238,25 @@ public class LayoutParsingPipeline {
                                               ImageServiceResponse imageServiceResponse,
                                               TableServiceResponse tableServiceResponse,
                                               VisualLayoutParsingResponse visualLayoutParsingResponse,
-                                              String identifier) {
+                                              Map<String, String> identifier) {
 
         PDDocument originDocument = openDocument(originFile);
         addNumberOfPagesToTrace(originDocument.getNumberOfPages(), Files.size(originFile.toPath()));
+
         Map<Integer, List<TableCells>> pdfTableCells = cvTableParsingAdapter.buildCvParsedTablesPerPage(tableServiceResponse);
         Map<Integer, List<ClassifiedImage>> pdfImages = imageServiceResponseAdapter.buildClassifiedImagesPerPage(imageServiceResponse);
         Map<Integer, List<ClassifiedImage>> signatures = visualLayoutParsingAdapter.buildExtractedSignaturesPerPage(visualLayoutParsingResponse);
+
         ClassificationDocument classificationDocument = new ClassificationDocument();
+
+        if (settings.isDebug() || identifier.containsKey("debug")) {
+            classificationDocument.getLayoutDebugLayer().setActive(true);
+        }
+
         List<ClassificationPage> classificationPages = new ArrayList<>();
 
+        classificationDocument.setOutlineObjectTree(outlineExtractorService.getOutlineObjectTree(originDocument));
+
         long pageCount = originDocument.getNumberOfPages();
 
         for (int pageNumber = 1; pageNumber <= pageCount; pageNumber++) {
@@ -247,49 +279,73 @@ public class LayoutParsingPipeline {
             stripper.setStartPage(pageNumber);
             stripper.setEndPage(pageNumber);
             stripper.setPdpage(pdPage);
-            if (layoutParsingType.equals(LayoutParsingType.DOCUMINE)) {
-                stripper.setSortByPosition(true);
-            }
             stripper.getText(originDocument);
+            List<Word> words = stripper.getWords();
+
+//            rotateDirAdjExactly(words, pdPage); // works really well for many highly rotated documents (e.g. VV-331340.pdf), but it decreases the headline performance by 1.3%, so I am leaving it out for now
+
+            if (layoutParsingType.equals(LayoutParsingType.DOCUMINE_OLD)) {
+                var lines = TextPositionOperations.groupByLine(new HashSet<>(words));
+                classificationDocument.getLayoutDebugLayer().addLineVisualizationsFromNestedTextPosition(lines, pageNumber);
+                words = TextPositionOperations.sortWords(lines);
+            }
+            classificationDocument.getLayoutDebugLayer().addTextVisualizations(words, pageNumber);
 
             PDRectangle pdr = pdPage.getMediaBox();
 
-            int rotation = pdPage.getRotation();
-            boolean isLandscape = pdr.getWidth() > pdr.getHeight() && (rotation == 0 || rotation == 180) || pdr.getHeight() > pdr.getWidth() && (rotation == 90 || rotation == 270);
+            List<Ruling> rulings = stripper.getRulings();
+            classificationDocument.getLayoutDebugLayer().addRulingVisualization(rulings, pageNumber);
+            CleanRulings cleanRulings = rulingCleaningService.deduplicateAndStraightenRulings(pdfTableCells.get(pageNumber), rulings);
 
-            PDRectangle cropbox = pdPage.getCropBox();
-            CleanRulings cleanRulings = rulingCleaningService.getCleanRulings(pdfTableCells.get(pageNumber), stripper.getRulings());
+            PageInformation pageInformation = PageInformation.fromPDPage(pageNumber, pdPage);
+            List<Cell> emptyTableCells = TableExtractionService.findCells(cleanRulings.getHorizontals(), cleanRulings.getVerticals(), pageInformation);
+            classificationDocument.getLayoutDebugLayer().addCellVisualizations(emptyTableCells, pageNumber);
+
+            TextRulingsClassifier.classifyUnderlinedAndStrikethroughText(words, cleanRulings);
+
+            List<Box> graphics = graphicExtractorService.extractPathElementGraphics(originDocument, pdPage, pageNumber, cleanRulings, stripper.getWords(), false);
+
+            pdfImages.computeIfAbsent(pageNumber, x -> new ArrayList<>())
+                    .addAll(graphics.stream()
+                                    .map(g -> new ClassifiedImage(new Rectangle2D.Double(g.x1, g.y1, g.width(), g.height()),
+                                                                  ImageType.GRAPHIC,
+                                                                  false,
+                                                                  stripper.getPageNumber(),
+                                                                  ""))
+                                    .toList());
 
             ClassificationPage classificationPage = switch (layoutParsingType) {
-                case REDACT_MANAGER -> redactManagerBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
-                case TAAS -> taasBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
-                case DOCUMINE -> docuMineBlockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
+                case REDACT_MANAGER_OLD -> redactManagerBlockificationService.blockify(stripper.getWords(), cleanRulings, classificationDocument.getLayoutDebugLayer());
+                case DOCUMINE_OLD -> docuMineBlockificationService.blockify(words, cleanRulings);
+                case DOCUMINE, REDACT_MANAGER, REDACT_MANAGER_PARAGRAPH_DEBUG, REDACT_MANAGER_WITHOUT_DUPLICATE_PARAGRAPH ->
+                        docstrumBlockificationService.blockify(words, cleanRulings, true, classificationDocument.getLayoutDebugLayer(), layoutParsingType);
+                case CLARIFYND, CLARIFYND_PARAGRAPH_DEBUG ->
+                        docstrumBlockificationService.blockify(words, cleanRulings, false, classificationDocument.getLayoutDebugLayer(), layoutParsingType);
             };
-            classificationPage.setCleanRulings(cleanRulings);
-            classificationPage.setRotation(rotation);
-            classificationPage.setLandscape(isLandscape);
-            classificationPage.setPageNumber(pageNumber);
-            classificationPage.setPageWidth(cropbox.getWidth());
-            classificationPage.setPageHeight(cropbox.getHeight());
 
+            updateClassificationPage(pdPage, pdr, classificationPage, cleanRulings, pageNumber, pageInformation);
+
+            blockificationPostprocessingService.findHeadlinesFromOutline(classificationDocument, pageNumber, classificationPage, pageInformation);
+
+            classificationDocument.getLayoutDebugLayer().addMarkedContentVisualizations(stripper.getMarkedContents(), pageNumber);
             // MarkedContent needs to be converted at this point, otherwise it leads to GC Problems in Pdfbox.
             classificationPage.setMarkedContentBboxPerType(convertMarkedContents(stripper.getMarkedContents()));
 
             // If images is ocr needs to be calculated before textBlocks are moved into tables, otherwise findOcr algorithm needs to be adopted.
-            if (pdfImages != null && pdfImages.containsKey(pageNumber)) {
+            if (pdfImages.containsKey(pageNumber)) {
                 classificationPage.setImages(pdfImages.get(pageNumber));
                 imageServiceResponseAdapter.findOcr(classificationPage);
             }
 
             if (signatures.containsKey(pageNumber)) {
-                if (classificationPage.getImages() == null || classificationPage.getImages().size() == 0) {
+                if (classificationPage.getImages() == null || classificationPage.getImages().isEmpty()) {
                     classificationPage.setImages(signatures.get(pageNumber));
                 } else {
                     classificationPage.getImages().addAll(signatures.get(pageNumber));
                 }
             }
 
-            tableExtractionService.extractTables(cleanRulings, classificationPage);
+            tableExtractionService.extractTables(emptyTableCells, classificationPage);
 
             buildPageStatistics(classificationPage);
             increaseDocumentStatistics(classificationPage, classificationDocument);
@@ -299,22 +355,67 @@ public class LayoutParsingPipeline {
 
         originDocument.close();
 
-        log.info("Calculating BodyTextFrame for {}", identifier);
-        bodyTextFrameService.setBodyTextFrames(classificationDocument, layoutParsingType);
-        log.info("Classify TextBlocks for {}", identifier);
-        switch (layoutParsingType) {
-            case TAAS -> taasClassificationService.classifyDocument(classificationDocument);
-            case DOCUMINE -> docuMineClassificationService.classifyDocument(classificationDocument);
-            case REDACT_MANAGER -> redactManagerClassificationService.classifyDocument(classificationDocument);
-        }
+        classificationService.classify(classificationDocument, layoutParsingType, identifier);
+
+        SectionTree sectionTree = sectionTreeBuilderService.createSectionTree(classificationDocument);
+        classificationDocument.setSectionTree(sectionTree);
 
         log.info("Building Sections for {}", identifier);
-        sectionsBuilderService.buildSections(classificationDocument);
-        sectionsBuilderService.addImagesToSections(classificationDocument);
+
+        switch (layoutParsingType) {
+            case CLARIFYND_PARAGRAPH_DEBUG, REDACT_MANAGER_PARAGRAPH_DEBUG -> sectionsBuilderService.buildParagraphDebugSections(classificationDocument);
+            default -> sectionTreeEnhancementService.assignSectionBlocksAndImages(classificationDocument);
+        }
+
         return classificationDocument;
     }
 
 
+    private static void updateClassificationPage(PDPage pdPage,
+                                                 PDRectangle pdr,
+                                                 ClassificationPage classificationPage,
+                                                 CleanRulings cleanRulings,
+                                                 int pageNumber,
+                                                 PageInformation pageInformation) {
+
+        int rotation = pdPage.getRotation();
+        boolean isLandscape = pdr.getWidth() > pdr.getHeight() && (rotation == 0 || rotation == 180) || pdr.getHeight() > pdr.getWidth() && (rotation == 90 || rotation == 270);
+        classificationPage.setCleanRulings(cleanRulings);
+        classificationPage.setRotation(rotation);
+        classificationPage.setLandscape(isLandscape);
+        classificationPage.setPageNumber(pageNumber);
+        classificationPage.setPageWidth((float) pageInformation.width());
+        classificationPage.setPageHeight((float) pageInformation.height());
+    }
+
+
+    private static void rotateDirAdjExactly(List<Word> words, PDPage pdPage) {
+
+        for (TextDirection dir : TextDirection.values()) {
+            double averageRotation = words.stream()
+                    .map(Word::getCharacters)
+                    .flatMap(Collection::stream)
+                    .map(Character::getTextPosition)
+                    .filter(pos -> pos.getDir().equals(dir))
+                    .mapToDouble(RedTextPosition::getExactDir).average()
+                    .orElse(0);
+
+            if (averageRotation == 0) {
+                continue;
+            }
+
+            AffineTransform rotateInstance = AffineTransform.getRotateInstance(averageRotation, pdPage.getMediaBox().getWidth() / 2, pdPage.getMediaBox().getHeight() / 2);
+
+            for (Word word : words) {
+                if (!dir.equals(word.getDir())) {
+                    continue;
+                }
+                word.transform(rotateInstance);
+            }
+        }
+    }
+
+
     private void addNumberOfPagesToTrace(int numberOfPages, long size) {
 
         if (observationRegistry.getCurrentObservation() != null) {
@@ -356,10 +457,10 @@ public class LayoutParsingPipeline {
         // Collect all statistics for the classificationPage, except from blocks inside tables, as tables will always be added to BodyTextFrame.
         for (AbstractPageBlock textBlock : classificationPage.getTextBlocks()) {
             if (textBlock instanceof TextPageBlock) {
-                if (((TextPageBlock) textBlock).getSequences() == null) {
+                if (((TextPageBlock) textBlock).getWords() == null) {
                     continue;
                 }
-                for (TextPositionSequence word : ((TextPageBlock) textBlock).getSequences()) {
+                for (Word word : ((TextPageBlock) textBlock).getWords()) {
                     classificationPage.getTextHeightCounter().add(word.getTextHeight());
                     classificationPage.getFontCounter().add(word.getFont());
                     classificationPage.getFontSizeCounter().add(word.getFontSize());

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/LayoutParsingServiceProcessorConfiguration.java

@@ -5,7 +5,7 @@ import org.springframework.context.annotation.Bean;
 import org.springframework.context.annotation.ComponentScan;
 import org.springframework.context.annotation.Configuration;
 
-import com.knecon.fforesight.service.viewerdoc.service.ViewerDocumentService;
+import com.knecon.fforesight.service.viewerdoc.service.PDFTronViewerDocumentService;
 
 import io.micrometer.observation.ObservationRegistry;
 
@@ -13,12 +13,11 @@ import io.micrometer.observation.ObservationRegistry;
 @ComponentScan
 public class LayoutParsingServiceProcessorConfiguration {
 
-
     @Bean
     @Autowired
-    public ViewerDocumentService viewerDocumentService(ObservationRegistry registry) {
+    public PDFTronViewerDocumentService viewerDocumentService(ObservationRegistry registry) {
 
-        return new ViewerDocumentService(registry);
+        return new PDFTronViewerDocumentService(registry);
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/LayoutParsingStorageService.java

@@ -1,26 +1,31 @@
 package com.knecon.fforesight.service.layoutparser.processor;
 
+import java.io.ByteArrayInputStream;
 import java.io.File;
 import java.io.FileInputStream;
 import java.io.IOException;
 import java.io.InputStream;
+import java.nio.charset.StandardCharsets;
 import java.nio.file.Files;
 import java.nio.file.Path;
 import java.nio.file.Paths;
 import java.nio.file.StandardOpenOption;
 import java.util.Optional;
+import java.util.concurrent.CompletableFuture;
 
+import org.springframework.core.task.TaskExecutor;
 import org.springframework.stereotype.Service;
 
 import com.fasterxml.jackson.databind.ObjectMapper;
+import com.iqser.red.service.redaction.v1.server.data.DocumentData;
 import com.iqser.red.storage.commons.service.StorageService;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentData;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.SimplifiedText;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.taas.ResearchDocumentData;
 import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsingRequest;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.image.ImageServiceResponse;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
 import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
+import com.knecon.fforesight.service.viewerdoc.service.ViewerDocVersioningUtility;
 import com.knecon.fforesight.tenantcommons.TenantContext;
 
 import io.micrometer.observation.annotation.Observed;
@@ -36,6 +41,9 @@ public class LayoutParsingStorageService {
     private final StorageService storageService;
     private final ObjectMapper objectMapper;
 
+    private final TaskExecutor taskExecutor;
+
+
     @Observed(name = "LayoutParsingStorageService", contextualName = "get-origin-file")
     public File getOriginFile(String storageId) throws IOException {
 
@@ -53,11 +61,18 @@ public class LayoutParsingStorageService {
         }
         File tempFile = createTempFile("viewerDocument", ".pdf");
         storageService.downloadTo(TenantContext.getTenantId(), storageId, tempFile);
+
+        if (!ViewerDocVersioningUtility.isCurrentVersion(tempFile)) {
+            assert tempFile.delete();
+            return Optional.empty();
+        }
+
         return Optional.of(tempFile);
     }
 
 
-    public ImageServiceResponse getImagesFile(String storageId) throws IOException {
+    @SneakyThrows
+    public ImageServiceResponse getImagesFile(String storageId) {
 
         try (InputStream inputStream = getObject(storageId)) {
 
@@ -68,7 +83,8 @@ public class LayoutParsingStorageService {
     }
 
 
-    public TableServiceResponse getTablesFile(String storageId) throws IOException {
+    @SneakyThrows
+    public TableServiceResponse getTablesFile(String storageId) {
 
         try (var tableClassificationStream = getObject(storageId)) {
 
@@ -78,22 +94,45 @@ public class LayoutParsingStorageService {
         }
     }
 
-    public VisualLayoutParsingResponse getVisualLayoutParsingFile(String storageId) throws IOException {
+
+@SneakyThrows
+    public VisualLayoutParsingResponse getVisualLayoutParsingFile(String storageId) {
 
         try (InputStream inputStream = getObject(storageId)) {
-            VisualLayoutParsingResponse visualLayoutParsingResponse = objectMapper.readValue(inputStream, VisualLayoutParsingResponse.class);
-            return visualLayoutParsingResponse;
+            return objectMapper.readValue(inputStream, VisualLayoutParsingResponse.class);
         }
     }
 
 
+    @SneakyThrows
     @Observed(name = "LayoutParsingStorageService", contextualName = "store-document-data")
     public void storeDocumentData(LayoutParsingRequest layoutParsingRequest, DocumentData documentData) {
 
-        storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.structureFileStorageId(), documentData.getDocumentStructure());
-        storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.textBlockFileStorageId(), documentData.getDocumentTextData());
-        storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.positionBlockFileStorageId(), documentData.getDocumentPositions());
-        storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.pageFileStorageId(), documentData.getDocumentPages());
+        Runnable storeDocumentStructureRunnable = () -> storageService.storeProtoObject(TenantContext.getTenantId(),
+                                                                                       layoutParsingRequest.structureFileStorageId(),
+                                                                                       documentData.getDocumentStructure());
+
+        CompletableFuture<Void> storeDocumentStructureFuture = CompletableFuture.runAsync(storeDocumentStructureRunnable, taskExecutor);
+
+        Runnable storeDocumentTextDataRunnable = () -> storageService.storeProtoObject(TenantContext.getTenantId(),
+                                                                                      layoutParsingRequest.textBlockFileStorageId(),
+                                                                                      documentData.getDocumentTextData());
+
+        CompletableFuture<Void> storeDocumentTextDataFuture = CompletableFuture.runAsync(storeDocumentTextDataRunnable, taskExecutor);
+
+        Runnable storeDocumentPositionsRunnable = () -> storageService.storeProtoObject(TenantContext.getTenantId(),
+                                                                                       layoutParsingRequest.positionBlockFileStorageId(),
+                                                                                       documentData.getDocumentPositionData());
+
+        CompletableFuture<Void> storeDocumentPositionsFuture = CompletableFuture.runAsync(storeDocumentPositionsRunnable, taskExecutor);
+
+        Runnable storeDocumentPagesRunnable = () -> storageService.storeProtoObject(TenantContext.getTenantId(),
+                                                                                   layoutParsingRequest.pageFileStorageId(),
+                                                                                   documentData.getDocumentPages());
+
+        CompletableFuture<Void> storeDocumentPagesFuture = CompletableFuture.runAsync(storeDocumentPagesRunnable, taskExecutor);
+
+        CompletableFuture.allOf(storeDocumentStructureFuture, storeDocumentTextDataFuture, storeDocumentPositionsFuture, storeDocumentPagesFuture).join();
     }
 
 
@@ -154,4 +193,16 @@ public class LayoutParsingStorageService {
         }
     }
 
+
+    @SneakyThrows
+    @Observed(name = "LayoutParsingStorageService", contextualName = "store-markdown-file")
+    public void storeMarkdownFile(String markdownFileStorageId, String markdownContent) {
+
+        try (InputStream inputStream = new ByteArrayInputStream(markdownContent.getBytes(StandardCharsets.UTF_8))) {
+
+            storageService.storeObject(TenantContext.getTenantId(), markdownFileStorageId, inputStream);
+        }
+
+    }
+
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/DocstrumSegmentationService.java

@@ -0,0 +1,98 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum;
+
+import java.util.ArrayList;
+import java.util.EnumMap;
+import java.util.List;
+import java.util.stream.Collectors;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Character;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Line;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Zone;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.service.LineBuilderService;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.service.NearestNeighbourService;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.service.ReadingOrderService;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.service.SpacingService;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.service.ZoneBuilderService;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextDirection;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.Word;
+import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutDebugLayer;
+
+import lombok.RequiredArgsConstructor;
+
+@Service
+@RequiredArgsConstructor
+public class DocstrumSegmentationService {
+
+    public static final double SAME_DIRECTION_THRESHOLD = 0.9;
+    private final NearestNeighbourService nearestNeighbourService;
+    private final SpacingService spacingService;
+    private final LineBuilderService lineBuilderService;
+    private final ZoneBuilderService zoneBuilderService;
+    private final ReadingOrderService readingOrderService;
+
+
+    public List<Zone> segmentPage(List<Word> textPositions, boolean xyOrder, CleanRulings usedRulings, LayoutDebugLayer visualizations) {
+
+        EnumMap<TextDirection, Integer> directionCounts = new EnumMap<>(TextDirection.class);
+
+        List<Zone> newZones = computeZones(textPositions, usedRulings, visualizations, TextDirection.ZERO);
+        directionCounts.put(TextDirection.ZERO, newZones.size());
+        List<Zone> zones = new ArrayList<>(newZones);
+
+        newZones = computeZones(textPositions, usedRulings, visualizations, TextDirection.QUARTER_CIRCLE);
+        directionCounts.put(TextDirection.QUARTER_CIRCLE, newZones.size());
+        zones.addAll(newZones);
+
+        newZones = computeZones(textPositions, usedRulings, visualizations, TextDirection.HALF_CIRCLE);
+        directionCounts.put(TextDirection.HALF_CIRCLE, newZones.size());
+        zones.addAll(newZones);
+
+        newZones = computeZones(textPositions, usedRulings, visualizations, TextDirection.THREE_QUARTER_CIRCLE);
+        directionCounts.put(TextDirection.THREE_QUARTER_CIRCLE, newZones.size());
+        zones.addAll(newZones);
+
+        return readingOrderService.resolve(zones, xyOrder, mostSameDirection(directionCounts));
+    }
+
+
+    private boolean mostSameDirection(EnumMap<TextDirection, Integer> directionCounts) {
+
+        int total = directionCounts.values()
+                .stream()
+                .mapToInt(i -> i).sum();
+
+        if ((double) directionCounts.get(TextDirection.ZERO) / total > SAME_DIRECTION_THRESHOLD) {
+            return true;
+        } else if ((double) directionCounts.get(TextDirection.QUARTER_CIRCLE) / total > SAME_DIRECTION_THRESHOLD) {
+            return true;
+        } else if ((double) directionCounts.get(TextDirection.HALF_CIRCLE) / total > SAME_DIRECTION_THRESHOLD) {
+            return true;
+        } else if ((double) directionCounts.get(TextDirection.THREE_QUARTER_CIRCLE) / total > SAME_DIRECTION_THRESHOLD) {
+            return true;
+        }
+        return false;
+    }
+
+
+    private List<Zone> computeZones(List<Word> textPositions, CleanRulings rulings, LayoutDebugLayer visualizations, TextDirection direction) {
+
+        List<Character> characters = textPositions.stream()
+                .filter(t -> t.getDir() == direction)
+                .map(Word::getCharacters)
+                .flatMap(List::stream)
+                .toList();
+
+        nearestNeighbourService.findNearestNeighbors(characters);
+
+        double characterSpacing = spacingService.computeCharacterSpacing(characters);
+        double lineSpacing = Math.min(spacingService.computeLineSpacing(characters), 20);
+
+        List<Line> lines = lineBuilderService.buildLines(characters, characterSpacing, lineSpacing, rulings);
+        return zoneBuilderService.buildZones(lines, characterSpacing, lineSpacing, rulings);
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/AngleFilter.java

@@ -0,0 +1,31 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+public class AngleFilter {
+
+    protected double lowerAngle;
+    protected double upperAngle;
+
+
+    public AngleFilter(double lowerAngle, double upperAngle) {
+
+        this.lowerAngle = lowerAngle < -Math.PI / 2 ? lowerAngle + Math.PI : lowerAngle;
+        this.upperAngle = upperAngle >= Math.PI / 2 ? upperAngle - Math.PI : upperAngle;
+    }
+
+
+    public boolean matches(Neighbor neighbor) {
+
+        return matches(neighbor.getAngle());
+    }
+
+
+    public boolean matches(double angle) {
+
+        if (lowerAngle <= upperAngle) {
+            return lowerAngle <= angle && angle < upperAngle;
+        } else {
+            return lowerAngle <= angle || angle < upperAngle;
+        }
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/BoundingBox.java

@@ -0,0 +1,279 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+import java.awt.geom.Rectangle2D;
+import java.util.Comparator;
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
+
+import lombok.Data;
+import lombok.NoArgsConstructor;
+import lombok.experimental.SuperBuilder;
+
+@Data
+@SuperBuilder
+@NoArgsConstructor
+public abstract class BoundingBox {
+
+    // Java coordinate system: (0, 0) is always upper left, x is increasing left to right and y is increasing from top to bottom.
+    // should be used when determining reading order or other tasks which require coordinates in a harmonized system.
+    protected Rectangle2D bBox; // I would not trust this coordinate when comparing rulings and text, due to the text positions being slightly off.
+
+    // PDF coordinate system: depends on page rotation, (0, 0) is lower left corner, x is increasing left to right and y from bottom to top.
+    // This rotates completely in 90 degree steps with page rotation.
+    // Needs to be used when writing to a PDF.
+    // Also, these are definitely correct and should be used whenever possible.
+    protected Rectangle2D bBoxPdf;
+
+    protected static final float VERTICAL_COMPARISON_THRESHOLD = 0.4f;
+
+
+    public double getX() {
+
+        return bBox.getX();
+    }
+
+
+    public double getY() {
+
+        return bBox.getY();
+    }
+
+
+    public double getMinX() {
+
+        return bBox.getMinX();
+    }
+
+
+    public double getMinY() {
+
+        return bBox.getMinY();
+    }
+
+
+    public double getPdfMinX() {
+
+        return bBoxPdf.getMinX();
+    }
+
+
+    public double getPdfMaxX() {
+
+        return bBoxPdf.getMaxX();
+    }
+
+
+    public double getPdfMinY() {
+
+        return bBoxPdf.getMinY();
+    }
+
+
+    public double getPdfMaxY() {
+
+        return bBoxPdf.getMaxY();
+    }
+
+
+    public double getWidth() {
+
+        return bBox.getWidth();
+    }
+
+
+    public double getHeight() {
+
+        return bBox.getHeight();
+    }
+
+
+    public double getMaxX() {
+
+        return bBox.getMaxX();
+    }
+
+
+    public double getMaxY() {
+
+        return bBox.getMaxY();
+    }
+
+
+    public double getArea() {
+
+        return (bBox.getHeight() * bBox.getWidth());
+    }
+
+
+    public boolean contains(BoundingBox contained) {
+
+        return contains(contained, 0);
+    }
+
+
+    public boolean contains(BoundingBox contained, double tolerance) {
+
+        return getPdfMinX() <= contained.getPdfMinX() + tolerance
+               && getPdfMinY() <= contained.getPdfMinY() + tolerance
+               && getPdfMaxX() >= contained.getPdfMaxX() - tolerance
+               && getPdfMaxY() >= contained.getPdfMaxY() - tolerance;
+    }
+
+
+    public boolean intersects(BoundingBox other) {
+
+        return this.intersectsX(other) && this.intersectsY(other);
+    }
+
+
+    public boolean intersects(BoundingBox other, float yThreshold, float xThreshold) {
+
+        return this.intersectsX(other, xThreshold) && this.intersectsY(other, yThreshold);
+    }
+
+
+    public boolean intersectsX(BoundingBox other, float threshold) {
+
+        return this.getX() - threshold <= other.getMaxX() && this.getMaxX() + threshold >= other.getX();
+    }
+
+
+    public boolean intersectsPdf(BoundingBox other) {
+
+        return this.intersectsXPdf(other) && this.intersectsYPdf(other);
+    }
+
+
+    public boolean intersectsPdf(BoundingBox other, float yThreshold, float xThreshold) {
+
+        return this.intersectsXPdf(other, xThreshold) && this.intersectsYPdf(other, yThreshold);
+    }
+
+
+    public boolean intersectsYPdf(BoundingBox other) {
+
+        return this.getPdfMinY() <= other.getPdfMaxY() && this.getPdfMaxY() >= other.getPdfMinY();
+    }
+
+
+    public boolean intersectsY(BoundingBox other) {
+
+        return this.getY() <= other.getMaxY() && this.getMaxY() >= other.getY();
+    }
+
+
+    public boolean intersectsY(BoundingBox other, float threshold) {
+
+        return this.getY() - threshold <= other.getMaxY() && this.getMaxY() + threshold >= other.getY();
+    }
+
+
+    public boolean intersectsYPdf(BoundingBox other, float threshold) {
+
+        return this.getPdfMinY() - threshold <= other.getPdfMaxY() && this.getPdfMaxY() + threshold >= other.getPdfMinY();
+    }
+
+
+    public boolean intersectsXPdf(BoundingBox other) {
+
+        return this.getPdfMinX() <= other.getPdfMaxX() && this.getPdfMaxX() >= other.getPdfMinX();
+    }
+
+
+    public boolean intersectsX(BoundingBox other) {
+
+        return this.getX() <= other.getMaxX() && this.getMaxX() >= other.getMinX();
+    }
+
+
+    public boolean intersectsXPdf(BoundingBox other, float threshold) {
+
+        return this.getPdfMinX() - threshold <= other.getPdfMaxX() && this.getMaxX() + threshold >= other.getPdfMinX();
+    }
+
+
+    public void setToBBoxOfComponents(List<? extends BoundingBox> components) {
+
+        this.bBox = components.stream()
+                .map(BoundingBox::getBBox)
+                .collect(RectangleTransformations.collectBBox());
+        this.bBoxPdf = components.stream()
+                .map(BoundingBox::getBBoxPdf)
+                .collect(RectangleTransformations.collectBBox());
+    }
+
+
+    public double verticalOverlap(BoundingBox other) {
+
+        return Math.max(0, Math.min(this.getPdfMaxY(), other.getPdfMaxY()) - Math.max(this.getPdfMinY(), other.getPdfMinY()));
+    }
+
+
+    public static final Comparator<BoundingBox> ILL_DEFINED_ORDER = (o1, o2) -> {
+
+        if (o1.equals(o2)) {
+            return 0;
+        }
+        if (o1.verticalOverlap(o2) > VERTICAL_COMPARISON_THRESHOLD * ((o1.getHeight() + o2.getHeight()) / 2)) {
+            return Double.compare(o1.getPdfMinX(), o2.getPdfMinX());
+        } else {
+            return Double.compare(o1.getPdfMaxY(), o2.getPdfMaxY());
+        }
+    };
+
+
+    public double horizontalDistance(BoundingBox other) {
+
+        double rect1Right = getMaxX();
+        double rect1Left = getMinX();
+        double rect2Right = other.getMaxX();
+        double rect2Left = other.getMinX();
+
+        if (rect1Left > rect2Right || rect2Left > rect1Right) {
+            return Math.max(rect2Left - rect1Right, rect1Left - rect2Right);
+        } else {
+            return 0;
+        }
+    }
+
+
+    public double verticalDistance(BoundingBox other) {
+
+        double rect1Top = getMaxY();
+        double rect1Bottom = getMinY();
+        double rect2Top = other.getMaxY();
+        double rect2Bottom = other.getMinY();
+
+        if (rect1Bottom > rect2Top || rect2Bottom > rect1Top) {
+            return Math.max(rect2Bottom - rect1Top, rect1Bottom - rect2Top);
+        } else {
+            return 0;
+        }
+    }
+
+
+    public boolean rightOf(BoundingBox other) {
+
+        return this.intersectsY(other) && other.getMaxX() <= this.getMinX();
+    }
+
+
+    public boolean leftOf(BoundingBox other) {
+
+        return this.intersectsY(other) && other.getMinX() >= this.getMaxX();
+    }
+
+
+    public boolean isAbove(BoundingBox other) {
+
+        return this.intersectsX(other) && other.getMinY() >= this.getMaxY();
+    }
+
+
+    public boolean isBelow(BoundingBox other) {
+
+        return this.intersectsX(other) && this.getMinY() >= other.getMaxY();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/Character.java

@@ -0,0 +1,86 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.utils.FastAtan2;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition;
+
+import lombok.Data;
+import lombok.EqualsAndHashCode;
+import lombok.Setter;
+
+@Data
+@EqualsAndHashCode(onlyExplicitlyIncluded = true)
+public class Character {
+
+    @EqualsAndHashCode.Include
+    private final double x;
+    @EqualsAndHashCode.Include
+    private final double y;
+    private final RedTextPosition textPosition;
+
+    @Setter
+    private List<Neighbor> neighbors = new ArrayList<>();
+
+
+    public Character(RedTextPosition chunk) {
+
+        this.x = chunk.getBBoxDirAdj().getCenterX();
+        this.y = chunk.getBBoxDirAdj().getCenterY();
+        this.textPosition = chunk;
+    }
+
+
+    public double getHeight() {
+
+        return textPosition.getHeightDirAdj();
+    }
+
+
+    public double distance(Character character) {
+
+        double dx = getX() - character.getX();
+        double dy = getY() - character.getY();
+        return Math.sqrt(dx * dx + dy * dy);
+    }
+
+
+    public double horizontalDistance(Character character) {
+
+        return Math.abs(getX() - character.getX());
+    }
+
+
+    public double verticalDistance(Character character) {
+
+        return Math.abs(getY() - character.getY());
+    }
+
+
+    public double overlappingDistance(Character other) {
+
+        double[] xs = new double[4];
+        double s = Math.sin(-0);
+        double c = Math.cos(-0);
+        xs[0] = c * x - s * y;
+        xs[1] = c * (x + textPosition.getWidthDirAdj()) - s * (y + textPosition.getHeightDirAdj());
+        xs[2] = c * other.x - s * other.y;
+        xs[3] = c * (other.x + other.textPosition.getWidthDirAdj()) - s * (other.y + other.textPosition.getHeightDirAdj());
+        boolean overlapping = xs[1] >= xs[2] && xs[3] >= xs[0];
+        Arrays.sort(xs);
+        return Math.abs(xs[2] - xs[1]) * (overlapping ? 1 : -1);
+    }
+
+
+    public double angle(Character character) {
+
+        if (getX() > character.getX()) {
+            return FastAtan2.fastAtan2(getY() - character.getY(), getX() - character.getX());
+        } else {
+            return FastAtan2.fastAtan2(character.getY() - getY(), character.getX() - getX());
+        }
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/ColumnDetector.java

@@ -0,0 +1,324 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.Comparator;
+import java.util.HashSet;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.Set;
+
+
+/*
+WIP, mostly working, needs to be tested a bit more
+ */
+public class ColumnDetector {
+
+    public static final double MAX_VALUE_THRESHOLD = 0.5;
+    final static int bins_num = 512;
+    final static int globalStartIdx = 0; // ignore outer parts completely, we don't expect columns there
+    final static int globalEndIdx = bins_num; // i chose 7, since thirds seems a likely split for columns, therefore divided by 6 would eliminate those.
+    public static final double DERIVATIVE_ZERO_THRESHOLD = 1e-10;
+    public static final double MINIMUM_THRESHOLD_FOR_COLUMNS = 0.05;
+    public static final double NEAR_GLOBAL_THRESHOLD = 0.5;
+    double minY;
+    double maxY;
+    double midY;
+    double[] histogram;
+    double min;
+    double max;
+    double resolution;
+    double sum;
+    int N;
+
+
+    public ColumnDetector(double min, double max, double minY, double maxY) {
+
+        this.min = min;
+        this.max = max;
+        this.minY = minY;
+        this.maxY = maxY;
+        this.midY = maxY - minY;
+        this.resolution = (max - min) / bins_num;
+        this.histogram = new double[bins_num];
+    }
+
+
+    public void add(BoundingBox zone) {
+
+        N++;
+        double weight = computeWeight(zone);
+        int start = (int) ((zone.getMinX() - min) / resolution);
+        int end = (int) ((zone.getMaxX() - min) / resolution);
+        for (int i = start; i < end; i++) {
+            histogram[i] += weight;
+            sum += histogram[i];
+        }
+    }
+
+
+    private double computeWeight(BoundingBox zone) {
+
+        double areaWeight = zone.getBBox().getHeight();
+
+        double relativeDistance = relativeDistanceToMiddle(zone.getBBox().getCenterY());
+
+        double distanceWeight;
+        if (relativeDistance < 0.6) {
+            distanceWeight = 1;
+        } else if (relativeDistance < 0.8) {
+            distanceWeight = 0.8;
+        } else {
+            distanceWeight = 0.1;
+        }
+
+        return areaWeight * distanceWeight;
+    }
+
+
+    private double relativeDistanceToMiddle(double y) {
+
+        double range = (maxY - minY) / 2;
+        double mid = minY + range;
+
+        return Math.abs(y - mid) / range;
+    }
+
+
+    public double[] computeDerivative() {
+
+        int length = histogram.length;
+        double[] derivative = new double[length];
+
+        for (int i = 0; i < length; i++) {
+            if (i == 0) {
+                derivative[i] = (histogram[i + 1] - histogram[i]) / resolution;
+            } else if (i == length - 1) {
+                derivative[i] = (histogram[i] - histogram[i - 1]) / resolution;
+            } else {
+                derivative[i] = (histogram[i + 1] - histogram[i - 1]) / (2 * resolution);
+            }
+        }
+
+        return derivative;
+    }
+
+
+    public double calcMean(double[] arr, int start, int end) {
+
+        if (start == end) {
+            return 0;
+        }
+        double sum = 0;
+        for (int i = start; i < end; i++) {
+            sum += arr[i];
+        }
+        return sum / (end - start);
+    }
+
+
+    /*
+    Find columns, by finding all local maxima/minima of the derivative. Filtering them for the ones with the biggest values.
+    For each found minima, we will step to the right until we hit a 0 in the derivative, this indicates a minimum in the main histogram. If this minimum is below a threshold, it is deemed a column divider.
+    Same goes for maxima, but stepping to the left now, since minima in the function will always be to the left of a maximum in its derivative.
+     */
+    public List<Double> determineColumnsWithDerivative(double[] derivative) {
+
+        assert derivative.length == histogram.length;
+
+        Set<Integer> columnIndices = new HashSet<>();
+        double mean = calcMean(histogram, 0, histogram.length);
+        double maxDvValue = calcMax(derivative);
+        double minDvValue = calcMin(derivative);
+
+        if (maxDvValue - minDvValue < mean * MAX_VALUE_THRESHOLD) {
+            Collections.emptyList();
+        }
+
+        Extrema derivativeExtrema = calculateNearGlobalExtrema(derivative, maxDvValue, minDvValue);
+
+        List<Integer> columnsRightOfMinima = findZerosToTheRightOfMinima(derivative, derivativeExtrema.minima(), mean);
+        columnIndices.addAll(columnsRightOfMinima);
+
+        List<Integer> columnsLeftOfMaxima = findZerosToTheLeftOfMaxima(derivative, derivativeExtrema.maxima(), mean);
+        columnIndices.addAll(columnsLeftOfMaxima);
+
+        return columnIndices.stream()
+                .sorted(Comparator.naturalOrder())
+                .map(this::calculateXCoordinateFromIdx)
+                .toList();
+    }
+
+
+    private List<Integer> findZerosToTheLeftOfMaxima(double[] derivative, List<Integer> derivativeMaxima, double mean) {
+
+        List<Integer> columnsLeftOfMaxima = new ArrayList<>();
+
+        for (int i = 0; i < derivativeMaxima.size(); i++) {
+            List<Integer> consecutiveZeroes = new LinkedList<>();
+            boolean maximumFound = false;
+            int maximaIdx = derivativeMaxima.get(i) - 1; // the highest derivative will always be at least one step away from the lowest value.
+            int endIdx = (int) Math.max(globalStartIdx,
+                                        Math.min(maximaIdx - 1,
+                                                 maximaIdx - 0.1 * bins_num)); // search through 10% of array to the right, but at least one step and at most to the left edge;
+
+            for (int j = maximaIdx; j >= endIdx; j--) {
+                if (derivative[j] < DERIVATIVE_ZERO_THRESHOLD) {
+                    maximumFound = true;
+                    consecutiveZeroes.add(j);
+                } else if (maximumFound) {
+                    break;
+                }
+            }
+            if (maximumFound) {
+                int midIdx = consecutiveZeroes.size() / 2;
+                int middleMinimumIdx = consecutiveZeroes.get(midIdx);
+                if (histogram[middleMinimumIdx] < mean * MINIMUM_THRESHOLD_FOR_COLUMNS) {
+                    columnsLeftOfMaxima.add(middleMinimumIdx);
+                }
+            }
+        }
+        return columnsLeftOfMaxima;
+    }
+
+
+    private List<Integer> findZerosToTheRightOfMinima(double[] derivative, List<Integer> derivativeMinima, double mean) {
+
+        List<Integer> columnIndixes = new LinkedList<>();
+        for (int i = 0; i < derivativeMinima.size(); i++) {
+            List<Integer> consecutiveZeroes = new LinkedList<>();
+            boolean minimumFound = false;
+            int minimaIdx = derivativeMinima.get(i) + 1; // the highest derivative will always be at least one step earlier than the lowest value.
+            int endIdx = (int) Math.min(globalEndIdx,
+                                        Math.max(minimaIdx + 1,
+                                                 minimaIdx + 0.1 * bins_num)); // search through 10% of array to the right, but at least one step and at most to the right edge;
+
+            for (int j = minimaIdx; j < endIdx; j++) {
+                if (derivative[j] < DERIVATIVE_ZERO_THRESHOLD) {
+                    minimumFound = true;
+                    consecutiveZeroes.add(j);
+                } else if (minimumFound) {
+                    break;
+                }
+            }
+            if (minimumFound) {
+                int midIdx = consecutiveZeroes.size() / 2;
+                int middleMinimumIdx = consecutiveZeroes.get(midIdx);
+                if (histogram[middleMinimumIdx] < mean * MINIMUM_THRESHOLD_FOR_COLUMNS) {
+                    columnIndixes.add(middleMinimumIdx);
+                }
+            }
+        }
+        return columnIndixes;
+    }
+
+
+    private double calcMax(double[] array) {
+
+        double max = Double.NEGATIVE_INFINITY;
+        for (int i = 0; i < array.length; i++) {
+            if (array[i] > max) {
+                max = array[i];
+            }
+        }
+        return max;
+    }
+
+
+    private double calcMin(double[] array) {
+
+        double min = Double.POSITIVE_INFINITY;
+        for (int i = 0; i < array.length; i++) {
+            if (array[i] < min) {
+                min = array[i];
+            }
+        }
+        return min;
+    }
+
+
+    private Extrema calculateNearGlobalExtrema(double[] derivative, double maxDvValue, double minDvValue) {
+
+        List<Integer> nearGlobalDvMaximaIdx = new LinkedList<>();
+        List<Integer> nearGlobalDvMinimaIdx = new LinkedList<>();
+        for (int i = globalStartIdx; i < globalEndIdx; i++) {
+            if (derivative[i] <= minDvValue * NEAR_GLOBAL_THRESHOLD) {
+                nearGlobalDvMinimaIdx.add(i);
+            }
+            if (derivative[i] >= maxDvValue * NEAR_GLOBAL_THRESHOLD) {
+                nearGlobalDvMaximaIdx.add(i);
+            }
+        }
+
+        nearGlobalDvMinimaIdx = removeConsecutive(nearGlobalDvMinimaIdx);
+        nearGlobalDvMaximaIdx = removeConsecutive(nearGlobalDvMaximaIdx);
+
+        return new Extrema(nearGlobalDvMaximaIdx, nearGlobalDvMinimaIdx);
+    }
+
+
+    private record Extrema(List<Integer> maxima, List<Integer> minima) {
+
+    }
+
+
+    private Double calculateXCoordinateFromIdx(int globalMinIdx) {
+
+        return min + ((globalMinIdx + 1) * resolution);
+    }
+
+
+    public static List<Integer> removeConsecutive(List<Integer> numbers) {
+
+        List<Integer> result = new ArrayList<>();
+        if (numbers == null || numbers.isEmpty()) {
+            return result;
+        }
+
+        result.add(numbers.get(0)); // Add the first number
+
+        for (int i = 1; i < numbers.size(); i++) {
+            if (numbers.get(i) != numbers.get(i - 1) + 1) {
+                result.add(numbers.get(i)); // Add non-consecutive numbers
+            }
+        }
+
+        return result;
+    }
+
+
+    public void kernelSmooth(double[] kernel) {
+
+        double[] newFrequencies = new double[histogram.length];
+        int shift = (kernel.length - 1) / 2;
+        for (int i = 0; i < kernel.length; i++) {
+            int jStart = Math.max(0, i - shift);
+            int jEnd = Math.min(histogram.length, histogram.length + i - shift);
+            for (int j = jStart; j < jEnd; j++) {
+                newFrequencies[j - i + shift] += kernel[i] * histogram[j];
+            }
+        }
+        histogram = newFrequencies;
+    }
+
+
+    public double[] createGaussianKernel(int length, double stdDeviation) {
+
+        int r = length / 2;
+
+        int size = 2 * r + 1;
+        double[] kernel = new double[size];
+        double sum = 0;
+        double b = 2 * (stdDeviation) * (stdDeviation);
+        double a = 1 / Math.sqrt(Math.PI * b);
+        for (int i = 0; i < size; i++) {
+            kernel[i] = a * Math.exp(-(i - r) * (i - r) / b);
+            sum += kernel[i];
+        }
+        for (int i = 0; i < size; i++) {
+            kernel[i] /= sum;
+        }
+        return kernel;
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/Histogram.java

@@ -0,0 +1,90 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+public class Histogram {
+
+    private static final double EPSILON = 1.0e-6;
+    private final double min;
+    private final double resolution;
+    private double[] frequencies;
+
+
+    public Histogram(double minValue, double maxValue, double resolution) {
+
+        this.min = minValue - EPSILON;
+        double delta = maxValue - minValue + 2 * EPSILON;
+        int size = Math.max(1, (int) Math.round((maxValue - minValue) / resolution));
+        this.resolution = delta / size;
+        this.frequencies = new double[size];
+    }
+
+
+    public void kernelSmooth(double[] kernel) {
+
+        double[] newFrequencies = new double[frequencies.length];
+        int shift = (kernel.length - 1) / 2;
+        for (int i = 0; i < kernel.length; i++) {
+            int jStart = Math.max(0, i - shift);
+            int jEnd = Math.min(frequencies.length, frequencies.length + i - shift);
+            for (int j = jStart; j < jEnd; j++) {
+                newFrequencies[j - i + shift] += kernel[i] * frequencies[j];
+            }
+        }
+        frequencies = newFrequencies;
+    }
+
+
+    public double[] createGaussianKernel(double length, double stdDeviation) {
+
+        int r = (int) Math.round(length / resolution) / 2;
+
+        int size = 2 * r + 1;
+        double[] kernel = new double[size];
+        double sum = 0;
+        double b = 2 * (stdDeviation / resolution) * (stdDeviation / resolution);
+        double a = 1 / Math.sqrt(Math.PI * b);
+        for (int i = 0; i < size; i++) {
+            kernel[i] = a * Math.exp(-(i - r) * (i - r) / b);
+            sum += kernel[i];
+        }
+        for (int i = 0; i < size; i++) {
+            kernel[i] /= sum;
+        }
+        return kernel;
+    }
+
+
+    public void gaussianSmooth(double windowLength, double stdDeviation) {
+
+        kernelSmooth(createGaussianKernel(windowLength, stdDeviation));
+    }
+
+
+    public void add(double value) {
+
+        frequencies[(int) ((value - min) / resolution)] += 1.0;
+    }
+
+
+    public int getSize() {
+
+        return frequencies.length;
+    }
+
+
+    public double getPeakValue() {
+
+        int peakIndex = 0;
+        for (int i = 1; i < frequencies.length; i++) {
+            if (frequencies[i] > frequencies[peakIndex]) {
+                peakIndex = i;
+            }
+        }
+        int peakEndIndex = peakIndex + 1;
+        final double EPS = 0.0001;
+        while (peakEndIndex < frequencies.length && Math.abs(frequencies[peakEndIndex] - frequencies[peakIndex]) < EPS) {
+            peakEndIndex++;
+        }
+        return ((double) peakIndex + peakEndIndex) / 2 * resolution + min;
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/Line.java

@@ -0,0 +1,194 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+import static com.knecon.fforesight.service.layoutparser.processor.model.text.Word.BOLD;
+import static com.knecon.fforesight.service.layoutparser.processor.model.text.Word.BOLD_ITALIC;
+import static com.knecon.fforesight.service.layoutparser.processor.model.text.Word.ITALIC;
+import static com.knecon.fforesight.service.layoutparser.processor.model.text.Word.STANDARD;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.EnumMap;
+import java.util.List;
+import java.util.Map;
+import java.util.concurrent.atomic.AtomicInteger;
+
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.utils.FastAtan2;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.Word;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.FontStyle;
+
+import lombok.Data;
+import lombok.EqualsAndHashCode;
+
+@Data
+@EqualsAndHashCode(onlyExplicitlyIncluded = true, callSuper = false)
+public class Line extends TextBoundingBox {
+
+    private static final double WORD_DISTANCE_MULTIPLIER = 0.17;
+
+    @EqualsAndHashCode.Include
+    private final double x0;
+    @EqualsAndHashCode.Include
+    private final double y0;
+
+    @EqualsAndHashCode.Include
+    private final double x1;
+    @EqualsAndHashCode.Include
+    private final double y1;
+
+    private FontStyle fontStyle;
+
+    private final List<Word> words;
+
+
+    public Line(List<Character> characters, double wordSpacing) {
+
+        if (characters.size() >= 2) {
+            // linear regression
+            double sx = 0.0;
+            double sxx = 0.0;
+            double sxy = 0.0;
+            double sy = 0.0;
+            for (Character character : characters) {
+                sx += character.getX();
+                sxx += character.getX() * character.getX();
+                sxy += character.getX() * character.getY();
+                sy += character.getY();
+            }
+            double b = (characters.size() * sxy - sx * sy) / (characters.size() * sxx - sx * sx);
+            double a = (sy - b * sx) / characters.size();
+
+            this.x0 = characters.get(0).getX();
+            this.y0 = a + b * this.x0;
+            this.x1 = characters.get(characters.size() - 1).getX();
+            this.y1 = a + b * this.x1;
+        } else {
+            Character character = characters.get(0);
+            double dx = character.getTextPosition().getWidthDirAdj() / 3;
+            double dy = dx * Math.tan(0);
+            this.x0 = character.getX() - dx;
+            this.x1 = character.getX() + dx;
+            this.y0 = character.getY() - dy;
+            this.y1 = character.getY() + dy;
+        }
+        this.words = new ArrayList<>();
+        computeWords(characters, wordSpacing * WORD_DISTANCE_MULTIPLIER);
+        buildBBox();
+        computeFontStyle();
+    }
+
+
+    public Line(List<Word> words) {
+
+        this.words = words;
+        buildBBox();
+        x0 = getMinX();
+        y0 = getMinY();
+        x1 = getMaxX();
+        y1 = getMaxY();
+        computeFontStyle();
+    }
+
+
+    private void computeFontStyle() {
+
+        EnumMap<FontStyle, AtomicInteger> fontStyleCounter = new EnumMap<>(FontStyle.class);
+        for (FontStyle fontStyle : FontStyle.values()) {
+            fontStyleCounter.put(fontStyle, new AtomicInteger(0));
+        }
+        for (Word word : words) {
+            switch (word.getFontStyle()) {
+                case STANDARD -> fontStyleCounter.get(FontStyle.REGULAR).getAndIncrement();
+                case BOLD -> fontStyleCounter.get(FontStyle.BOLD).getAndIncrement();
+                case ITALIC -> fontStyleCounter.get(FontStyle.ITALIC).getAndIncrement();
+                case BOLD_ITALIC -> fontStyleCounter.get(FontStyle.BOLD_ITALIC).getAndIncrement();
+            }
+        }
+        fontStyle = fontStyleCounter.entrySet()
+                .stream()
+                .max(Comparator.comparing(entry -> entry.getValue().get()))
+                .map(Map.Entry::getKey).orElse(FontStyle.REGULAR);
+    }
+
+
+    public double getAngle() {
+
+        return FastAtan2.fastAtan2(y1 - y0, x1 - x0);
+    }
+
+
+    public double getLength() {
+
+        return Math.sqrt((x0 - x1) * (x0 - x1) + (y0 - y1) * (y0 - y1));
+    }
+
+
+    public double angularDifference(Line j) {
+
+        double diff = Math.abs(getAngle() - j.getAngle());
+        if (diff <= Math.PI / 2) {
+            return diff;
+        } else {
+            return Math.PI - diff;
+        }
+    }
+
+
+    public double horizontalDistance(Line other) {
+
+        double[] xs = new double[4];
+        xs[0] = x0;
+        xs[1] = x1;
+        xs[2] = other.x0;
+        xs[3] = other.x1;
+        boolean overlapping = xs[1] >= xs[2] && xs[3] >= xs[0];
+        Arrays.sort(xs);
+        return Math.abs(xs[2] - xs[1]) * (overlapping ? 1 : -1);
+    }
+
+
+    public double verticalDistance(Line other) {
+
+        double ym = (y0 + y1) / 2;
+        double yn = (other.y0 + other.y1) / 2;
+        return Math.abs(ym - yn);
+    }
+
+
+    private void computeWords(List<Character> characters, double wordSpacing) {
+
+        // Imo, the width of space should be scaled with the font size, but it only depends on the median distance between horizontal neighbours.
+        // If there are large differences in fontsize on a page, this might lead to missing spaces for the smaller fonts and too many for larger fonts.
+        // I've just now changed the scaling factor. If you come across this comment with missing whitespaces again, try scaling the fontsize instead of simply changing the factor again.
+        Word word = new Word();
+        Character previous = null;
+        for (Character current : characters) {
+            if (previous != null) {
+                double dist = current.getTextPosition().getXDirAdj() - previous.getTextPosition().getXDirAdj() - previous.getTextPosition().getWidthDirAdj();
+                if (dist > wordSpacing) {
+                    words.add(word);
+                    word = new Word();
+                }
+            }
+            word.add(current);
+            previous = current;
+        }
+        words.add(word);
+    }
+
+
+    private void buildBBox() {
+
+        this.setToBBoxOfComponents(words);
+    }
+
+
+    public String toString() {
+
+        StringBuilder sb = new StringBuilder();
+        words.forEach(word -> sb.append(word.toString()).append(" "));
+        return sb.toString().trim();
+    }
+
+}
+

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/Neighbor.java

@@ -0,0 +1,43 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+import lombok.Getter;
+
+public class Neighbor {
+
+    @Getter
+    private final double distance;
+    private Double angle;
+    private final Character originCharacter;
+    @Getter
+    private final Character character;
+
+
+    public Neighbor(Character neighbor, Character origin) {
+
+        this.distance = neighbor.distance(origin);
+        this.character = neighbor;
+        this.originCharacter = origin;
+    }
+
+
+    public double getHorizontalDistance() {
+
+        return character.horizontalDistance(originCharacter);
+    }
+
+
+    public double getVerticalDistance() {
+
+        return character.verticalDistance(originCharacter);
+    }
+
+
+    public double getAngle() {
+
+        if (angle != null) {
+            return angle;
+        }
+        return this.character.angle(this.originCharacter);
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/TextBoundingBox.java

@@ -0,0 +1,180 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+import java.awt.geom.Rectangle2D;
+import java.util.List;
+import java.util.Set;
+import java.util.stream.Collectors;
+
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextDirection;
+import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
+
+import lombok.EqualsAndHashCode;
+import lombok.Getter;
+import lombok.NoArgsConstructor;
+import lombok.Setter;
+import lombok.experimental.SuperBuilder;
+
+@Getter
+@Setter
+@SuperBuilder
+@NoArgsConstructor
+@EqualsAndHashCode(callSuper = false)
+public abstract class TextBoundingBox extends BoundingBox {
+
+    protected Rectangle2D bBoxDirAdj;
+
+    protected TextDirection dir;
+
+
+    @Override
+    public void setToBBoxOfComponents(List<? extends BoundingBox> components) {
+
+        super.setToBBoxOfComponents(components);
+        this.bBoxDirAdj = components.stream()
+                .filter(c -> c instanceof TextBoundingBox)
+                .map(c -> (TextBoundingBox) c)
+                .map(TextBoundingBox::getBBoxDirAdj)
+                .collect(RectangleTransformations.collectBBox());
+
+        Set<TextDirection> textDirections = components.stream()
+                .filter(c -> c instanceof TextBoundingBox)
+                .map(c -> (TextBoundingBox) c)
+                .map(TextBoundingBox::getDir)
+                .collect(Collectors.toSet());
+
+        if (textDirections.isEmpty()) {
+            dir = TextDirection.ZERO;
+        } else if (textDirections.size() > 1) {
+            throw new IllegalArgumentException("More than one text direction found");
+        } else {
+            dir = textDirections.iterator().next();
+        }
+    }
+
+
+    public double getXDirAdj() {
+
+        return this.bBoxDirAdj.getX();
+    }
+
+
+    public double getYDirAdj() {
+
+        return this.bBoxDirAdj.getY();
+    }
+
+
+    public double getWidthDirAdj() {
+
+        return this.bBoxDirAdj.getWidth();
+    }
+
+
+    public double getHeightDirAdj() {
+
+        return this.bBoxDirAdj.getHeight();
+    }
+
+
+    public double getMaxXDirAdj() {
+
+        return this.bBoxDirAdj.getMaxX();
+    }
+
+
+    public double getMaxYDirAdj() {
+
+        return this.bBoxDirAdj.getMaxY();
+    }
+
+
+    public double getCenterYDirAdj() {
+
+        return this.bBoxDirAdj.getCenterY();
+    }
+
+
+    public double getCenterXDirAdj() {
+
+        return this.bBoxDirAdj.getCenterX();
+    }
+
+
+    public double horizontalDistanceDirAdj(TextBoundingBox other) {
+
+        double rect1Right = getMaxXDirAdj();
+        double rect1Left = getXDirAdj();
+        double rect2Right = other.getMaxXDirAdj();
+        double rect2Left = other.getXDirAdj();
+
+        if (rect1Left > rect2Right || rect2Left > rect1Right) {
+            return Math.max(rect2Left - rect1Right, rect1Left - rect2Right);
+        } else {
+            return 0;
+        }
+    }
+
+
+    public double verticalDistanceDirAdj(TextBoundingBox other) {
+
+        double rect1Top = getMaxYDirAdj();
+        double rect1Bottom = getYDirAdj();
+        double rect2Top = other.getMaxYDirAdj();
+        double rect2Bottom = other.getYDirAdj();
+
+        if (rect1Bottom > rect2Top || rect2Bottom > rect1Top) {
+            return Math.max(rect2Bottom - rect1Top, rect1Bottom - rect2Top);
+        } else {
+            return 0;
+        }
+    }
+
+
+    public boolean intersectsDirAdj(TextBoundingBox other) {
+
+        return this.intersectsXDirAdj(other) && this.intersectsYDirAdj(other);
+    }
+
+
+    public boolean intersectsDirAdj(TextBoundingBox other, float yThreshold, float xThreshold) {
+
+        return this.intersectsXDirAdj(other, xThreshold) && this.intersectsYDirAdj(other, yThreshold);
+    }
+
+
+    public boolean intersectsXDirAdj(TextBoundingBox other, float threshold) {
+
+        return this.getXDirAdj() - threshold <= other.getMaxXDirAdj() && this.getMaxXDirAdj() + threshold >= other.getXDirAdj();
+    }
+
+
+    public boolean intersectsXDirAdj(TextBoundingBox other) {
+
+        return this.getXDirAdj() <= other.getMaxXDirAdj() && this.getMaxXDirAdj() >= other.getXDirAdj();
+    }
+
+
+    public boolean intersectsYDirAdj(TextBoundingBox other) {
+
+        return this.getYDirAdj() <= other.getMaxYDirAdj() && this.getMaxYDirAdj() >= other.getYDirAdj();
+    }
+
+
+    public boolean intersectsYDirAdj(TextBoundingBox other, float threshold) {
+
+        return this.getYDirAdj() - threshold <= other.getMaxYDirAdj() && this.getMaxYDirAdj() + threshold >= other.getYDirAdj();
+    }
+
+
+    public boolean isAboveDirAdj(TextBoundingBox other) {
+
+        return other.isBelow(this);
+    }
+
+
+    public boolean isBelowDirAdj(TextBoundingBox other) {
+
+        return this.intersectsXDirAdj(other) && this.getYDirAdj() >= other.getMaxYDirAdj();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/UnionFind.java

@@ -0,0 +1,37 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+import java.util.Collection;
+import java.util.LinkedHashMap;
+import java.util.LinkedHashSet;
+import java.util.Map;
+import java.util.Set;
+
+public class UnionFind<T> extends org.jgrapht.alg.util.UnionFind<T> {
+
+    public UnionFind(Set<T> elements) {
+
+        super(elements);
+    }
+
+
+    public Collection<Set<T>> getGroups() {
+
+        Map<T, Set<T>> setRep = new LinkedHashMap<>();
+        for (T t : getParentMap().keySet()) {
+            T representative = find(t);
+            if (!setRep.containsKey(representative)) {
+                setRep.put(representative, new LinkedHashSet<>());
+            }
+            setRep.get(representative).add(t);
+        }
+
+        return setRep.values();
+    }
+
+
+    public Collection<T> getElements() {
+
+        return getParentMap().keySet();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/model/Zone.java

@@ -0,0 +1,33 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.model;
+
+import java.util.Comparator;
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
+
+import lombok.Data;
+import lombok.EqualsAndHashCode;
+
+@Data
+@EqualsAndHashCode(callSuper = false)
+public class Zone extends TextBoundingBox {
+
+    private List<Line> lines;
+
+
+    @SuppressWarnings("PMD.ConstructorCallsOverridableMethod")
+    public Zone(List<Line> lines) {
+
+        this.lines = lines;
+        setToBBoxOfComponents(lines);
+    }
+
+
+    public String toString() {
+
+        StringBuilder sb = new StringBuilder();
+        lines.forEach(line -> sb.append(line.toString()).append("\n"));
+        return sb.toString().trim();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/service/LineBuilderService.java

@@ -0,0 +1,58 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.service;
+
+import java.util.Comparator;
+import java.util.HashSet;
+import java.util.List;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.AngleFilter;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Character;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Line;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.UnionFind;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
+
+@Service
+public class LineBuilderService {
+
+    private static final double CHARACTER_SPACING_DISTANCE_MULTIPLIER = 3.5;
+    private static final double LINE_SPACING_THRESHOLD_MULTIPLIER = 0.67;
+    private static final double ANGLE_TOLERANCE = Math.toRadians(5);
+
+
+    public List<Line> buildLines(List<Character> characters, double characterSpacing, double lineSpacing, CleanRulings rulings) {
+
+        double maxHorizontalDistance = characterSpacing * CHARACTER_SPACING_DISTANCE_MULTIPLIER;
+        double maxVerticalDistance = lineSpacing * LINE_SPACING_THRESHOLD_MULTIPLIER;
+
+        UnionFind<Character> unionFind = new UnionFind<>(new HashSet<>(characters));
+
+        AngleFilter angleFilter = new AngleFilter(-ANGLE_TOLERANCE, ANGLE_TOLERANCE);
+
+        characters.forEach(character -> {
+            character.getNeighbors()
+                    .forEach(neighbor -> {
+                        double normalizedHorizontalDistance = neighbor.getHorizontalDistance() / maxHorizontalDistance;
+                        double normalizedVerticalDistance = neighbor.getVerticalDistance() / maxVerticalDistance;
+
+                        if (character.getTextPosition().getDir() != neighbor.getCharacter().getTextPosition().getDir() //
+                            || !angleFilter.matches(neighbor) //
+                            || Math.pow(normalizedHorizontalDistance, 2) + Math.pow(normalizedVerticalDistance, 2) > 1 //
+                            || rulings.lineBetween(character.getTextPosition(), neighbor.getCharacter().getTextPosition())) {
+                            return;
+                        }
+
+                        unionFind.union(character, neighbor.getCharacter());
+                    });
+        });
+
+        return unionFind.getGroups()
+                .stream()
+                .map(lineCharacters -> lineCharacters.stream()
+                        .sorted(Comparator.comparingDouble(Character::getX))
+                        .toList())
+                .map(lineCharacters -> new Line(lineCharacters, characterSpacing))
+                .toList();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/service/NearestNeighbourService.java

@@ -0,0 +1,36 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.service;
+
+import java.util.List;
+
+import org.springframework.stereotype.Service;
+import org.tinspin.index.Index;
+import org.tinspin.index.kdtree.KDTree;
+
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Character;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Neighbor;
+
+@Service
+public class NearestNeighbourService {
+
+    private static final int NUMBER_OF_NEIGHBOURS = 8;
+
+
+    public void findNearestNeighbors(List<Character> characters) {
+        KDTree<Character> kdTree = KDTree.create(2);
+        characters.forEach(c -> kdTree.insert(new double[]{c.getX(), c.getY()}, c));
+
+        for(Character c : characters) {
+            Index.PointIteratorKnn<Character> iterator = kdTree.queryKnn(new double[]{c.getX(), c.getY()}, NUMBER_OF_NEIGHBOURS + 1);
+            // skip the first as this is identity
+            if(iterator.hasNext()) {
+                iterator.next();
+            }
+            while(iterator.hasNext()) {
+                c.getNeighbors().add(new Neighbor(iterator.next().value(), c));
+            }
+        }
+
+    }
+
+}
+

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/service/ReadingOrderService.java

@@ -0,0 +1,192 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.service;
+
+import java.awt.geom.Rectangle2D;
+import java.util.ArrayList;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.List;
+import java.util.ListIterator;
+import java.util.Map;
+import java.util.stream.Collectors;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.TextBoundingBox;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Zone;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.utils.DoubleUtils;
+
+@Service
+public class ReadingOrderService {
+
+    private static final double THRESHOLD = 5;
+    public static final double MULTI_COLUMN_DETECTION_THRESHOLD = 1.5;
+
+    private static final Comparator<TextBoundingBox> COMPARATOR = //
+            Comparator.comparing(TextBoundingBox::getY, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
+                    .thenComparing(TextBoundingBox::getX, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD));
+
+    private static final Comparator<TextBoundingBox> COMPARATOR_DIR_ADJ = //
+            Comparator.comparing(TextBoundingBox::getYDirAdj, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD))
+                    .thenComparing(TextBoundingBox::getXDirAdj, (o1, o2) -> DoubleUtils.compareDouble(o1, o2, THRESHOLD));
+
+
+    public List<Zone> resolve(List<Zone> zones, boolean xyReadingOrder, boolean useDirAdjCoords) {
+
+        if (zones.isEmpty() || zones.size() == 1) {
+            return zones;
+        }
+
+        if (xyReadingOrder) {
+            return resolveSingleColumnReadingOrder(zones, useDirAdjCoords);
+        }
+
+        Map<Long, Integer> histogram = new HashMap<>();
+        for (Zone zone : zones) {
+            Rectangle2D bbox = useDirAdjCoords ? zone.getBBoxDirAdj() : zone.getBBox();
+            long minY = Math.round(bbox.getMinY());
+            long maxY = Math.round(bbox.getMaxY());
+            for (long i = minY; i <= maxY; i++) {
+                histogram.put(i, histogram.getOrDefault(i, 0) + 1);
+            }
+        }
+
+        if (histogram.values()
+                    .stream()
+                    .mapToInt(Integer::intValue).average()
+                    .orElse(1) < MULTI_COLUMN_DETECTION_THRESHOLD) {
+            return resolveSingleColumnReadingOrder(zones, useDirAdjCoords);
+        } else {
+
+            return resolveMultiColumnReadingOder(zones, useDirAdjCoords);
+        }
+
+    }
+
+
+    private static List<Zone> resolveSingleColumnReadingOrder(List<Zone> zones, boolean useDirAdjCoords) {
+
+        if (useDirAdjCoords) {
+            return zones.stream()
+                    .collect(Collectors.groupingBy(TextBoundingBox::getDir)).values()
+                    .stream()
+                    .flatMap(words -> words.stream()
+                            .sorted(COMPARATOR_DIR_ADJ))
+                    .toList();
+        }
+
+        zones.sort(COMPARATOR);
+        return zones;
+    }
+
+
+    private List<Zone> resolveMultiColumnReadingOder(List<Zone> zones, boolean useDirAdjCoords) {
+
+        // Simple reading order resolver for multi column page layout as described here : https://pub.towardsai.net/advanced-rag-02-unveiling-pdf-parsing-b84ae866344e
+        // TODO implement a more fancy reading order resolver see https://github.com/BobLd/DocumentLayoutAnalysis/blob/master/README.md#reading-order
+
+        double minX = Double.POSITIVE_INFINITY;
+        double maxX = Double.NEGATIVE_INFINITY;
+
+        for (Zone zone : zones) {
+            Rectangle2D bbox = useDirAdjCoords ? zone.getBBoxDirAdj() : zone.getBBox();
+            if (bbox.getX() < minX) {
+                minX = zone.getXDirAdj();
+            }
+            if (bbox.getMaxX() > maxX) {
+                maxX = zone.getMaxXDirAdj();
+            }
+        }
+
+        double midLineXCoordinate = (minX + maxX) / 2;
+
+        List<Zone> leftOf = new ArrayList<>();
+        List<Zone> rightOf = new ArrayList<>();
+        List<Zone> middle = new ArrayList<>();
+
+        for (Zone zone : zones) {
+            Rectangle2D bbox = useDirAdjCoords ? zone.getBBoxDirAdj() : zone.getBBox();
+            if (bbox.getX() < midLineXCoordinate && bbox.getX() + bbox.getWidth() < midLineXCoordinate) {
+                leftOf.add(zone);
+            } else if (bbox.getX() > midLineXCoordinate && bbox.getX() + bbox.getWidth() > midLineXCoordinate) {
+                rightOf.add(zone);
+            } else {
+                middle.add(zone);
+            }
+        }
+
+        if (useDirAdjCoords) {
+            leftOf.sort(COMPARATOR_DIR_ADJ);
+            rightOf.sort(COMPARATOR_DIR_ADJ);
+            middle.sort(COMPARATOR_DIR_ADJ);
+        } else {
+            leftOf.sort(COMPARATOR);
+            rightOf.sort(COMPARATOR);
+            middle.sort(COMPARATOR);
+        }
+/*
+        List<Zone> leftNotIntersecting = new ArrayList<>();
+        for (Zone leftZone : leftOf) {
+            boolean intersects = false;
+            for (Zone rightZone : rightOf) {
+                if (leftZone.intersectsY(rightZone)) {
+                    intersects = true;
+                    break;
+                }
+                // early stopping
+                if (rightZone.getBBox().getMinY() > leftZone.getBBox().getMaxY()) {
+                    break;
+                }
+            }
+            if (!intersects) {
+                leftNotIntersecting.add(leftZone);
+            }
+        }
+
+        List<Zone> rightNotIntersecting = new ArrayList<>();
+        for (Zone rightZone : rightOf) {
+            boolean intersects = false;
+            for (Zone leftZone : leftOf) {
+                if (rightZone.intersectsY(leftZone)) {
+                    intersects = true;
+                    break;
+                }
+                // early stopping
+                if (leftZone.getBBox().getMinY() > rightZone.getBBox().getMaxY()) {
+                    break;
+                }
+            }
+            if (!intersects) {
+                rightNotIntersecting.add(rightZone);
+            }
+        }
+
+        leftOf.removeAll(leftNotIntersecting);
+        rightOf.removeAll(rightNotIntersecting);
+
+        middle.addAll(leftNotIntersecting);
+        middle.addAll(rightNotIntersecting);
+*/
+        List<Zone> sortedZones = new ArrayList<>();
+        sortedZones.addAll(leftOf);
+        sortedZones.addAll(rightOf);
+
+        ListIterator<Zone> itty = middle.listIterator();
+
+        while (itty.hasNext()) {
+            Zone current = itty.next();
+            Rectangle2D bbox = useDirAdjCoords ? current.getBBoxDirAdj() : current.getBBox();
+            for (int i = 0; i < sortedZones.size(); i++) {
+                if (bbox.getY() < sortedZones.get(i).getY()) {
+                    sortedZones.add(i, current);
+                    itty.remove();
+                    break;
+                }
+            }
+        }
+
+        sortedZones.addAll(middle);
+
+        return sortedZones;
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/service/SpacingService.java

@@ -0,0 +1,56 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.service;
+
+import java.util.List;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.AngleFilter;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Character;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Histogram;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Neighbor;
+
+@Service
+public class SpacingService {
+
+    private static final double SPACING_HISTOGRAM_RESOLUTION = 0.5;
+    private static final double SPACING_HISTOGRAM_SMOOTHING_LENGTH = 2.5;
+    private static final double SPACING_HIST_SMOOTHING_STANDARD_DEVIATION = 0.5;
+    private static final double ANGLE_TOLERANCE = Math.PI / 6;
+
+
+    public double computeCharacterSpacing(List<Character> characters) {
+
+        return computeSpacing(characters, 0);
+    }
+
+
+    public double computeLineSpacing(List<Character> characters) {
+
+        return computeSpacing(characters, Math.PI / 2);
+    }
+
+
+    private double computeSpacing(List<Character> characters, double angle) {
+
+        double maxDistance = Double.NEGATIVE_INFINITY;
+
+        for (Character character : characters) {
+            for (Neighbor neighbor : character.getNeighbors()) {
+                maxDistance = Math.max(maxDistance, neighbor.getDistance());
+            }
+        }
+        Histogram histogram = new Histogram(0, maxDistance, SPACING_HISTOGRAM_RESOLUTION);
+        AngleFilter angleFilter = new AngleFilter(angle - ANGLE_TOLERANCE, angle + ANGLE_TOLERANCE);
+        for (Character character : characters) {
+            for (Neighbor neighbor : character.getNeighbors()) {
+                if (angleFilter.matches(neighbor)) {
+                    histogram.add(neighbor.getDistance());
+                }
+            }
+        }
+
+        histogram.gaussianSmooth(SPACING_HISTOGRAM_SMOOTHING_LENGTH, SPACING_HIST_SMOOTHING_STANDARD_DEVIATION);
+        return histogram.getPeakValue();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/service/ZoneBuilderService.java

@@ -0,0 +1,126 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.service;
+
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+import java.util.stream.Collectors;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Line;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.UnionFind;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.Zone;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.Word;
+import com.knecon.fforesight.service.layoutparser.processor.utils.TextPositionOperations;
+
+@Service
+public class ZoneBuilderService {
+
+    private static final double MIN_HORIZONTAL_DISTANCE_MULTIPLIER = -0.5;
+    private static final double MAX_VERTICAL_DISTANCE_MULTIPLIER = 1.2;
+
+    private static final double MIN_HORIZONTAL_MERGE_DISTANCE_MULTIPLIER = -7;
+
+    private static final double MAX_VERTICAL_MERGE_DISTANCE_MULTIPLIER = 0.5;
+
+    private static final double MIN_LINE_SIZE_SCALE = 0.9;
+
+    private static final double MAX_LINE_SIZE_SCALE = 2.5;
+
+    private static final double ANGLE_TOLERANCE = Math.toRadians(5);
+
+    private static final double MAX_VERTICAL_MERGE_DISTANCE = 0.5;
+
+
+    public List<Zone> buildZones(List<Line> lines, double characterSpacing, double lineSpacing, CleanRulings rulings) {
+
+        double minHorizontalDistance = characterSpacing * MIN_HORIZONTAL_DISTANCE_MULTIPLIER;
+        double maxVerticalDistance = lineSpacing * MAX_VERTICAL_DISTANCE_MULTIPLIER;
+        double minHorizontalMergeDistance = lineSpacing * MIN_HORIZONTAL_MERGE_DISTANCE_MULTIPLIER;
+        double maxVerticalMergeDistance = lineSpacing * MAX_VERTICAL_MERGE_DISTANCE_MULTIPLIER;
+
+        UnionFind<Line> unionFind = new UnionFind<>(new HashSet<>(lines));
+
+        double meanHeight = calculateMeanHeight(lines);
+
+        lines.forEach(outerLine -> {
+            lines.forEach(innerLine -> {
+
+                if (innerLine == outerLine //
+                    || unionFind.inSameSet(outerLine, innerLine)//
+                    || outerLine.angularDifference(innerLine) > ANGLE_TOLERANCE) {
+                    return;
+                }
+
+//                if (!innerLine.getFontStyle().equals(outerLine.getFontStyle()) //
+//                    && !outerLine.intersectsY(innerLine, -2f)) {
+//                    return;
+//                }
+
+                double horizontalScale = Math.min(outerLine.getHeightDirAdj(), innerLine.getHeightDirAdj()) / meanHeight;
+                horizontalScale = Math.max(MIN_LINE_SIZE_SCALE, Math.min(horizontalScale, MAX_LINE_SIZE_SCALE));
+                double verticalScale = horizontalScale;
+
+//                if (innerLine.toString().endsWith(":")
+//                    || outerLine.toString().endsWith(":")
+//                    || numericalIdentifierPattern.matcher(innerLine.toString()).matches()
+//                    || numericalIdentifierPattern.matcher(outerLine.toString()).matches()) {
+//
+//                    horizontalScale *= 5;
+//                    verticalScale /= 10;
+//                }
+
+                double horizontalDistance = outerLine.horizontalDistance(innerLine) / horizontalScale;
+                double verticalDistance = outerLine.verticalDistance(innerLine) / verticalScale;
+
+                if ((!(minHorizontalDistance <= horizontalDistance) || !(verticalDistance <= maxVerticalDistance)) //
+                    && (!(minHorizontalMergeDistance <= horizontalDistance) || !(verticalDistance <= maxVerticalMergeDistance))) {
+                    return;
+                }
+
+                if (rulings.lineBetween(outerLine, innerLine)) {
+                    return;
+                }
+
+                unionFind.union(outerLine, innerLine);
+
+            });
+        });
+
+        return unionFind.getGroups()
+                .stream()
+                .map(group -> mergeLinesInZone(new ArrayList<>(group), characterSpacing, lineSpacing))
+                .toList();
+    }
+
+
+    private double calculateMeanHeight(List<Line> lines) {
+
+        double meanHeight = 0.0;
+        double weights = 0.0;
+        for (Line line : lines) {
+            double weight = line.getLength();
+            meanHeight += line.getHeightDirAdj() * weight;
+            weights += weight;
+        }
+        meanHeight /= weights;
+        return meanHeight;
+    }
+
+
+    private Zone mergeLinesInZone(List<Line> lines, double characterSpacing, double lineSpacing) {
+
+        Set<Word> words = lines.stream()
+                .map(Line::getWords)
+                .flatMap(Collection::stream)
+                .collect(Collectors.toSet());
+        Collection<Set<Word>> groupedLines = TextPositionOperations.groupByLine(words);
+
+        List<Line> sortedLines = TextPositionOperations.sortLines(groupedLines);
+        return new Zone(sortedLines);
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/utils/DoubleUtils.java

@@ -0,0 +1,18 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.utils;
+
+public class DoubleUtils {
+
+    public static int compareDouble(double d1, double d2, double precision) {
+
+        if (Double.isNaN(d1) || Double.isNaN(d2)) {
+            return Double.compare(d1, d2);
+        }
+
+        if (Math.abs(d1 - d2) < precision) {
+            return 0;
+        }
+
+        return Double.compare(d1, d2);
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/docstrum/utils/FastAtan2.java

@@ -0,0 +1,76 @@
+package com.knecon.fforesight.service.layoutparser.processor.docstrum.utils;
+
+public class FastAtan2 {
+
+    static final private int Size_Ac = 1000;
+    static final private int Size_Ar = Size_Ac + 1;
+    static final private double Pi = (float) Math.PI;
+    static final private double Pi_H = Pi / 2;
+
+    static final private double[] Atan2 = new double[Size_Ar];
+    static final private double[] Atan2_PM = new double[Size_Ar];
+    static final private double[] Atan2_MP = new double[Size_Ar];
+    static final private double[] Atan2_MM = new double[Size_Ar];
+
+    static final private double[] Atan2_R = new double[Size_Ar];
+    static final private double[] Atan2_RPM = new double[Size_Ar];
+    static final private double[] Atan2_RMP = new double[Size_Ar];
+    static final private double[] Atan2_RMM = new double[Size_Ar];
+
+    static {
+        for (int i = 0; i <= Size_Ac; i++) {
+            double d = (double) i / Size_Ac;
+            double x = 1;
+            double y = x * d;
+            double v = Math.atan2(y, x);
+            Atan2[i] = v;
+            Atan2_PM[i] = Pi - v;
+            Atan2_MP[i] = -v;
+            Atan2_MM[i] = -Pi + v;
+
+            Atan2_R[i] = Pi_H - v;
+            Atan2_RPM[i] = Pi_H + v;
+            Atan2_RMP[i] = -Pi_H + v;
+            Atan2_RMM[i] = -Pi_H - v;
+
+        }
+    }
+
+    @SuppressWarnings("ParameterAssignment")
+    static public double fastAtan2(double y, double x) {
+
+        if (y < 0) {
+            if (x < 0) {
+                //(y < x) because == (-y > -x)
+                if (y < x) {
+                    return Atan2_RMM[(int) (x / y * Size_Ac)];
+                } else {
+                    return Atan2_MM[(int) (y / x * Size_Ac)];
+                }
+            } else {
+                y = -y;
+                if (y > x) {
+                    return Atan2_RMP[(int) (x / y * Size_Ac)];
+                } else {
+                    return Atan2_MP[(int) (y / x * Size_Ac)];
+                }
+            }
+        } else {
+            if (x < 0) {
+                x = -x;
+                if (y > x) {
+                    return Atan2_RPM[(int) (x / y * Size_Ac)];
+                } else {
+                    return Atan2_PM[(int) (y / x * Size_Ac)];
+                }
+            } else {
+                if (y > x) {
+                    return Atan2_R[(int) (x / y * Size_Ac)];
+                } else {
+                    return Atan2[(int) (y / x * Size_Ac)];
+                }
+            }
+        }
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/AbstractPageBlock.java

@@ -1,7 +1,11 @@
 package com.knecon.fforesight.service.layoutparser.processor.model;
 
+import java.util.HashSet;
+import java.util.Set;
+
 import com.fasterxml.jackson.annotation.JsonIgnore;
-import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
+import com.iqser.red.service.redaction.v1.server.model.document.nodes.LayoutEngine;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.BoundingBox;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
 
 import lombok.AllArgsConstructor;
@@ -13,18 +17,13 @@ import lombok.NoArgsConstructor;
 @AllArgsConstructor
 @NoArgsConstructor
 @EqualsAndHashCode(callSuper = true)
-public abstract class AbstractPageBlock extends Rectangle {
+public abstract class AbstractPageBlock extends BoundingBox {
 
-    @JsonIgnore
-    protected float minX;
-    @JsonIgnore
-    protected float maxX;
-    @JsonIgnore
-    protected float minY;
-    @JsonIgnore
-    protected float maxY;
     @JsonIgnore
     protected PageBlockType classification;
+
+    Set<LayoutEngine> engines = new HashSet<>();
+
     @JsonIgnore
     protected int page;
 
@@ -41,63 +40,6 @@ public abstract class AbstractPageBlock extends Rectangle {
     }
 
 
-    public boolean containsBlock(TextPageBlock other) {
-
-        return this.minX <= other.getPdfMinX() && this.maxX >= other.getPdfMaxX() && this.minY >= other.getPdfMinY() && this.maxY <= other.getPdfMaxY();
-    }
-
-
-    public boolean contains(AbstractPageBlock other) {
-
-        return this.minX <= other.minX && this.maxX >= other.maxX && this.minY >= other.minY && this.maxY <= other.maxY;
-    }
-
-
-    public boolean contains(Rectangle other) {
-
-        return page == other.getPage() && this.minX <= other.getTopLeft().getX() && this.maxX >= other.getTopLeft().getX() + other.getWidth() && this.minY <= other.getTopLeft()
-                .getY() && this.maxY >= other.getTopLeft().getY() + other.getHeight();
-    }
-
-
-    @JsonIgnore
-    public float getHeight() {
-
-        return maxY - minY;
-    }
-
-
-    @JsonIgnore
-    public float getWidth() {
-
-        return maxX - minX;
-    }
-
-
-    public boolean intersectsY(AbstractPageBlock apb) {
-
-        return this.minY <= apb.getMaxY() && this.maxY >= apb.getMinY();
-    }
-
-
-    public boolean almostIntersects(AbstractPageBlock apb, float yThreshold, float xThreshold) {
-
-        return this.almostIntersectsX(apb, xThreshold) && this.almostIntersectsY(apb, yThreshold);
-    }
-
-
-    private boolean almostIntersectsY(AbstractPageBlock apb, float threshold) {
-
-        return this.minY - threshold <= apb.getMaxY() && this.maxY + threshold >= apb.getMinY();
-    }
-
-
-    private boolean almostIntersectsX(AbstractPageBlock apb, float threshold) {
-
-        return this.minX - threshold <= apb.getMaxX() && this.maxX + threshold >= apb.getMinX();
-    }
-
-
     public abstract boolean isEmpty();
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/ClassificationDocument.java

@@ -3,8 +3,11 @@ package com.knecon.fforesight.service.layoutparser.processor.model;
 import java.util.ArrayList;
 import java.util.List;
 
+import com.knecon.fforesight.service.layoutparser.processor.model.outline.OutlineObjectTree;
+import com.knecon.fforesight.service.layoutparser.processor.model.outline.SectionTree;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.StringFrequencyCounter;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.UnclassifiedText;
+import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutDebugLayer;
 
 import lombok.Data;
 import lombok.NoArgsConstructor;
@@ -22,8 +25,12 @@ public class ClassificationDocument {
     private FloatFrequencyCounter fontSizeCounter = new FloatFrequencyCounter();
     private StringFrequencyCounter fontCounter = new StringFrequencyCounter();
     private StringFrequencyCounter fontStyleCounter = new StringFrequencyCounter();
+    private LayoutDebugLayer layoutDebugLayer = new LayoutDebugLayer();
     private boolean headlines;
 
     private long rulesVersion;
 
+    private OutlineObjectTree outlineObjectTree;
+    private SectionTree sectionTree;
+
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/ClassificationPage.java

@@ -8,21 +8,26 @@ import java.util.Map;
 
 import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
 import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.model.outline.OutlineObject;
 import com.knecon.fforesight.service.layoutparser.processor.model.table.CleanRulings;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.StringFrequencyCounter;
 
 import lombok.Data;
 import lombok.NonNull;
 import lombok.RequiredArgsConstructor;
-import org.apache.pdfbox.pdmodel.documentinterchange.markedcontent.PDMarkedContent;
 
 @Data
 @RequiredArgsConstructor
+
 public class ClassificationPage {
 
     @NonNull
     private List<AbstractPageBlock> textBlocks;
 
+    private List<OutlineObject> outlineObjects = new ArrayList<>();
+
+    private List<AbstractPageBlock> headlines = new ArrayList<>();
+
     private List<ClassifiedImage> images = new ArrayList<>();
 
     private Rectangle bodyTextFrame;
@@ -40,7 +45,7 @@ public class ClassificationPage {
     private float pageWidth;
     private float pageHeight;
 
-    CleanRulings cleanRulings;
+    private CleanRulings cleanRulings;
 
     private Map<String, List<Rectangle2D>> markedContentBboxPerType = new HashMap<>();
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/ClassificationSection.java

@@ -12,6 +12,7 @@ import lombok.NoArgsConstructor;
 
 @Data
 @NoArgsConstructor
+@Deprecated
 public class ClassificationSection {
 
     private List<AbstractPageBlock> pageBlocks = new ArrayList<>();

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/DocumentWithVisualization.java

@@ -0,0 +1,19 @@
+package com.knecon.fforesight.service.layoutparser.processor.model;
+
+import java.util.Map;
+import java.util.stream.Collectors;
+
+import com.iqser.red.service.redaction.v1.server.model.document.nodes.Document;
+import com.iqser.red.service.redaction.v1.server.model.document.nodes.NodeType;
+import com.iqser.red.service.redaction.v1.server.model.document.nodes.SemanticNode;
+import com.knecon.fforesight.service.layoutparser.processor.visualization.LayoutDebugLayer;
+
+public record DocumentWithVisualization(Document document, LayoutDebugLayer layoutDebugLayer) {
+
+    public Map<NodeType, Long> buildSemanticNodeCounts() {
+
+        return document.streamAllSubNodes()
+                .collect(Collectors.groupingBy(SemanticNode::getType, Collectors.counting()));
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/FloatFrequencyCounter.java

@@ -1,6 +1,5 @@
 package com.knecon.fforesight.service.layoutparser.processor.model;
 
-import java.util.ArrayList;
 import java.util.Collections;
 import java.util.HashMap;
 import java.util.List;
@@ -12,10 +11,14 @@ import lombok.Getter;
 @Getter
 public class FloatFrequencyCounter {
 
-    Map<Float, Integer> countPerValue = new HashMap<>();
+    Map<Double, Integer> countPerValue = new HashMap<>();
+    boolean changed;
+    Double mostPopularCache;
 
 
-    public void add(float value) {
+    public void add(double value) {
+
+        changed = true;
 
         if (!countPerValue.containsKey(value)) {
             countPerValue.put(value, 1);
@@ -25,9 +28,11 @@ public class FloatFrequencyCounter {
     }
 
 
-    public void addAll(Map<Float, Integer> otherCounter) {
+    public void addAll(Map<Double, Integer> otherCounter) {
 
-        for (Map.Entry<Float, Integer> entry : otherCounter.entrySet()) {
+        changed = true;
+
+        for (Map.Entry<Double, Integer> entry : otherCounter.entrySet()) {
             if (countPerValue.containsKey(entry.getKey())) {
                 countPerValue.put(entry.getKey(), countPerValue.get(entry.getKey()) + entry.getValue());
             } else {
@@ -37,36 +42,36 @@ public class FloatFrequencyCounter {
     }
 
 
-    public Float getMostPopular() {
+    public Double getMostPopular() {
 
-        Map.Entry<Float, Integer> mostPopular = null;
-        for (Map.Entry<Float, Integer> entry : countPerValue.entrySet()) {
-            if (mostPopular == null || entry.getValue() >= mostPopular.getValue()) {
-                mostPopular = entry;
+        if (changed || mostPopularCache == null) {
+            Map.Entry<Double, Integer> mostPopular = null;
+            for (Map.Entry<Double, Integer> entry : countPerValue.entrySet()) {
+                if (mostPopular == null || entry.getValue() >= mostPopular.getValue()) {
+                    mostPopular = entry;
+                }
             }
+            mostPopularCache = mostPopular != null ? mostPopular.getKey() : 0;
+            changed = false;
         }
-        return mostPopular != null ? mostPopular.getKey() : null;
+
+        return mostPopularCache;
     }
 
 
-    public List<Float> getHighterThanMostPopular() {
+    public List<Double> getValuesInReverseOrder() {
 
-        Float mostPopular = getMostPopular();
-        List<Float> higher = new ArrayList<>();
-        for (Float value : countPerValue.keySet()) {
-            if (value > mostPopular) {
-                higher.add(value);
-            }
-        }
-
-        return higher.stream().sorted(Collections.reverseOrder()).collect(Collectors.toList());
+        return countPerValue.keySet()
+                .stream()
+                .sorted(Collections.reverseOrder())
+                .collect(Collectors.toList());
     }
 
 
-    public Float getHighest() {
+    public Double getHighest() {
 
-        Float highest = null;
-        for (Float value : countPerValue.keySet()) {
+        Double highest = null;
+        for (Double value : countPerValue.keySet()) {
             if (highest == null || value > highest) {
                 highest = value;
             }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/LineInformation.java

@@ -3,7 +3,7 @@ package com.knecon.fforesight.service.layoutparser.processor.model;
 import java.awt.geom.Rectangle2D;
 import java.util.List;
 
-import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.Word;
 
 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
@@ -16,8 +16,8 @@ import lombok.experimental.FieldDefaults;
 public class LineInformation {
 
     List<Rectangle2D> lineBBox;
-    List<List<TextPositionSequence>> sequencesByLines;
+    List<List<Word>> sequencesByLines;
     List<List<Rectangle2D>> bBoxWithGapsByLines;
-    List<List<List<TextPositionSequence>>> sequencesWithGapsByLines;
+    List<List<List<Word>>> sequencesWithGapsByLines;
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/PageBlockType.java

@@ -9,12 +9,14 @@ public enum PageBlockType {
     H6,
     HEADER,
     FOOTER,
-    TITLE,
     PARAGRAPH,
     PARAGRAPH_BOLD,
     PARAGRAPH_ITALIC,
     PARAGRAPH_UNKNOWN,
     OTHER,
+    TABLE_OF_CONTENTS_HEADLINE,
+    TABLE_OF_CONTENTS_ITEM,
+    LIST_ITEM,
     TABLE;
 
 
@@ -31,8 +33,21 @@ public enum PageBlockType {
     }
 
 
+    public static int getHeadlineNumber(PageBlockType pageBlockType) {
+
+        return switch (pageBlockType) {
+            case H1, TABLE_OF_CONTENTS_HEADLINE -> 1;
+            case H2 -> 2;
+            case H3 -> 3;
+            case H4 -> 4;
+            case H5 -> 5;
+            default -> 6;
+        };
+    }
+
+
     public boolean isHeadline() {
 
-        return this.equals(H1) || this.equals(H2) || this.equals(H3) || this.equals(H4) || this.equals(H5) || this.equals(H6);
+        return this.equals(H1) || this.equals(H2) || this.equals(H3) || this.equals(H4) || this.equals(H5) || this.equals(H6) || this.equals(TABLE_OF_CONTENTS_HEADLINE);
     }
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/PageContents.java

@@ -4,7 +4,7 @@ import java.awt.geom.Rectangle2D;
 import java.util.List;
 
 import com.knecon.fforesight.service.layoutparser.processor.model.table.Ruling;
-import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.Word;
 
 import lombok.AllArgsConstructor;
 import lombok.Builder;
@@ -15,7 +15,7 @@ import lombok.Getter;
 @AllArgsConstructor
 public class PageContents {
 
-    List<TextPositionSequence> sortedTextPositionSequences;
+    List<Word> sortedWords;
     Rectangle2D cropBox;
     Rectangle2D mediaBox;
     List<Ruling> rulings;

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/SectionIdentifier.java

@@ -3,26 +3,32 @@ package com.knecon.fforesight.service.layoutparser.processor.model;
 import java.util.Collections;
 import java.util.LinkedList;
 import java.util.List;
+import java.util.Locale;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
 import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
+import lombok.Getter;
 import lombok.experimental.FieldDefaults;
 
 @AllArgsConstructor
 @FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
 public class SectionIdentifier {
 
-    static Pattern numericalIdentifierPattern = Pattern.compile("^[\\s]?(\\d+)[\\s.,;]?(\\d+)?[\\s.,;]?(\\d+)?[\\s.,;]?(\\d+)?");
+    public static Pattern numericalIdentifierPattern = Pattern.compile("^[\\s]?(\\d{1,2})(?:[\\s.,;](\\d{1,2}))?(?:[\\s.,;](\\d{1,2}))?(?:[\\s.,;](\\d{1,2}))?");
+    public static Pattern alphanumericIdentifierPattern = Pattern.compile("^[\\s]?[A-Za-z][\\s.,;]?(\\d{1,2})[\\s.,;]?(\\d{1,2})?[\\s.,;]?(\\d{1,2})?[\\s.,;]?(\\d{1,2})?[\\s.,;]?");
 
-    private enum Format {
+    public enum Format {
         EMPTY,
         NUMERICAL,
+        ALPHANUMERIC,
         DOCUMENT
     }
 
+    @Getter
     Format format;
+    @Getter
     String identifierString;
     List<Integer> identifiers;
     boolean asChild;
@@ -38,6 +44,10 @@ public class SectionIdentifier {
         if (numericalIdentifierMatcher.find()) {
             return buildNumericalSectionIdentifier(headline, numericalIdentifierMatcher);
         }
+        Matcher alphanumericIdentifierMatcher = alphanumericIdentifierPattern.matcher(headline);
+        if (alphanumericIdentifierMatcher.find()) {
+            return buildAlphanumericSectionIdentifier(headline, alphanumericIdentifierMatcher);
+        }
         // more formats here
         return SectionIdentifier.empty();
     }
@@ -72,7 +82,36 @@ public class SectionIdentifier {
             }
             identifiers.add(Integer.parseInt(numericalIdentifier.trim()));
         }
-        return new SectionIdentifier(Format.NUMERICAL, identifierString, identifiers.stream().toList(), false);
+        return new SectionIdentifier(Format.NUMERICAL,
+                                     identifierString,
+                                     identifiers.stream()
+                                             .toList(),
+                                     false);
+    }
+
+
+    private static SectionIdentifier buildAlphanumericSectionIdentifier(String headline, Matcher alphanumericIdentifierMatcher) {
+
+        String identifierString = headline.substring(alphanumericIdentifierMatcher.start(), alphanumericIdentifierMatcher.end());
+
+        String alphanumericIdentifier = alphanumericIdentifierMatcher.group(0).substring(0, 1).toUpperCase(Locale.ENGLISH);
+        int mappedCharacterValue = alphanumericIdentifier.charAt(0) - 'A' + 1;
+        List<Integer> identifiers = new LinkedList<>();
+        identifiers.add(mappedCharacterValue);
+
+        for (int i = 1; i <= 3; i++) {
+            String numericalIdentifier = alphanumericIdentifierMatcher.group(i);
+            if (numericalIdentifier == null || numericalIdentifier.equals("0") || numericalIdentifier.isEmpty() || numericalIdentifier.isBlank()) {
+                break;
+            }
+            identifiers.add(Integer.parseInt(numericalIdentifier.trim()));
+        }
+
+        return new SectionIdentifier(Format.ALPHANUMERIC,
+                                     identifierString,
+                                     identifiers.stream()
+                                             .toList(),
+                                     false);
     }
 
 
@@ -120,4 +159,22 @@ public class SectionIdentifier {
         return identifierString;
     }
 
+
+    public boolean isEmpty() {
+
+        return this.format.equals(Format.EMPTY);
+    }
+
+
+    public int level() {
+
+        return identifiers.size();
+    }
+
+
+    protected List<Integer> getIdentifiers() {
+
+        return identifiers;
+    }
+
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/Boundary.java

@@ -1,145 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph;
-
-import static java.lang.String.format;
-
-import java.util.Collection;
-import java.util.LinkedList;
-import java.util.List;
-import java.util.stream.IntStream;
-
-import lombok.EqualsAndHashCode;
-import lombok.Setter;
-
-@Setter
-@EqualsAndHashCode
-@SuppressWarnings("PMD.AvoidFieldNameMatchingMethodName")
-public class Boundary implements Comparable<Boundary> {
-
-    private int start;
-    private int end;
-
-
-    public Boundary(int start, int end) {
-
-        if (start > end) {
-            throw new IllegalArgumentException(format("start: %d > end: %d", start, end));
-        }
-        this.start = start;
-        this.end = end;
-    }
-
-
-    public int length() {
-
-        return end - start;
-    }
-
-
-    public int start() {
-
-        return start;
-    }
-
-
-    public int end() {
-
-        return end;
-    }
-
-
-    public boolean contains(Boundary boundary) {
-
-        return start <= boundary.start() && boundary.end() <= end;
-    }
-
-
-    public boolean containedBy(Boundary boundary) {
-
-        return boundary.contains(this);
-    }
-
-
-    public boolean contains(int start, int end) {
-
-        if (start > end) {
-            throw new IllegalArgumentException(format("start: %d > end: %d", start, end));
-        }
-        return this.start <= start && end <= this.end;
-    }
-
-
-    public boolean containedBy(int start, int end) {
-
-        if (start > end) {
-            throw new IllegalArgumentException(format("start: %d > end: %d", start, end));
-        }
-        return start <= this.start && this.end <= end;
-    }
-
-
-    public boolean contains(int index) {
-
-        return start <= index && index < end;
-    }
-
-
-    public boolean intersects(Boundary boundary) {
-
-        return boundary.start() < this.end && this.start < boundary.end();
-    }
-
-
-    public List<Boundary> split(List<Integer> splitIndices) {
-
-        if (splitIndices.stream().anyMatch(idx -> !this.contains(idx))) {
-            throw new IndexOutOfBoundsException(format("%s splitting indices are out of range for %s", splitIndices.stream().filter(idx -> !this.contains(idx)).toList(), this));
-        }
-        List<Boundary> splitBoundaries = new LinkedList<>();
-        int previousIndex = start;
-        for (int splitIndex : splitIndices) {
-
-            // skip split if it would produce a boundary of length 0
-            if (splitIndex == previousIndex) {
-                continue;
-            }
-            splitBoundaries.add(new Boundary(previousIndex, splitIndex));
-            previousIndex = splitIndex;
-        }
-        splitBoundaries.add(new Boundary(previousIndex, end));
-        return splitBoundaries;
-    }
-
-    public IntStream intStream() {
-
-        return IntStream.range(start, end);
-    }
-
-    public static Boundary merge(Collection<Boundary> boundaries) {
-
-        int minStart = boundaries.stream().mapToInt(Boundary::start).min().orElseThrow(IllegalArgumentException::new);
-        int maxEnd = boundaries.stream().mapToInt(Boundary::end).max().orElseThrow(IllegalArgumentException::new);
-        return new Boundary(minStart, maxEnd);
-    }
-
-
-    @Override
-    public String toString() {
-
-        return format("Boundary [%d|%d)", start, end);
-    }
-
-
-    @Override
-    public int compareTo(Boundary boundary) {
-
-        if (end < boundary.end() && start < boundary.start()) {
-            return -1;
-        }
-        if (start > boundary.start() && end > boundary.end()) {
-            return 1;
-        }
-
-        return 0;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/DocumentTree.java

@@ -1,217 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph;
-
-import static java.lang.String.format;
-
-import java.util.Collections;
-import java.util.LinkedList;
-import java.util.List;
-import java.util.stream.Stream;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.GenericSemanticNode;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Table;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.TableCell;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.Getter;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@EqualsAndHashCode
-public class DocumentTree {
-
-    private final Entry root;
-
-
-    public DocumentTree(Document document) {
-
-        root = Entry.builder().treeId(Collections.emptyList()).children(new LinkedList<>()).node(document).build();
-    }
-
-
-    public TextBlock buildTextBlock() {
-
-        return allEntriesInOrder().map(Entry::getNode).filter(SemanticNode::isLeaf).map(SemanticNode::getLeafTextBlock).collect(new TextBlockCollector());
-    }
-
-
-    public List<Integer> createNewMainEntryAndReturnId(GenericSemanticNode node) {
-
-        return createNewChildEntryAndReturnIdImpl(Collections.emptyList(), node);
-    }
-
-
-    public List<Integer> createNewChildEntryAndReturnId(GenericSemanticNode parentNode, GenericSemanticNode node) {
-
-        return createNewChildEntryAndReturnIdImpl(parentNode.getTreeId(), node);
-    }
-
-
-    public List<Integer> createNewChildEntryAndReturnId(GenericSemanticNode parentNode, Table node) {
-
-        return createNewChildEntryAndReturnIdImpl(parentNode.getTreeId(), node);
-    }
-
-
-    public List<Integer> createNewTableChildEntryAndReturnId(Table parentTable, TableCell tableCell) {
-
-        return createNewChildEntryAndReturnIdImpl(parentTable.getTreeId(), tableCell);
-    }
-
-
-    @SuppressWarnings("PMD.UnusedPrivateMethod") // PMD actually flags this wrong
-    private List<Integer> createNewChildEntryAndReturnIdImpl(List<Integer> parentId, SemanticNode node) {
-
-        if (!entryExists(parentId)) {
-            throw new IllegalArgumentException(format("parentId %s does not exist!", parentId));
-        }
-
-        Entry parent = getEntryById(parentId);
-        List<Integer> newId = new LinkedList<>(parentId);
-        newId.add(parent.children.size());
-        parent.children.add(Entry.builder().treeId(newId).node(node).build());
-
-        return newId;
-    }
-
-
-    private boolean entryExists(List<Integer> treeId) {
-
-        if (treeId.isEmpty()) {
-            return root != null;
-        }
-        Entry entry = root.children.get(treeId.get(0));
-        for (int id : treeId.subList(1, treeId.size())) {
-            if (id >= entry.children.size() || 0 > id) {
-                return false;
-            }
-            entry = entry.children.get(id);
-        }
-        return true;
-    }
-
-
-    public Entry getParentEntryById(List<Integer> treeId) {
-
-        return getEntryById(getParentId(treeId));
-    }
-
-
-    public boolean hasParentById(List<Integer> treeId) {
-
-        return !treeId.isEmpty();
-    }
-
-
-    public Stream<SemanticNode> childNodes(List<Integer> treeId) {
-
-        return getEntryById(treeId).children.stream().map(Entry::getNode);
-    }
-
-
-    public Stream<SemanticNode> childNodesOfType(List<Integer> treeId, NodeType nodeType) {
-
-        return getEntryById(treeId).children.stream().filter(entry -> entry.node.getType().equals(nodeType)).map(Entry::getNode);
-    }
-
-
-    private static List<Integer> getParentId(List<Integer> treeId) {
-
-        if (treeId.isEmpty()) {
-            throw new UnsupportedOperationException("Root has no parent!");
-        }
-        if (treeId.size() < 2) {
-            return Collections.emptyList();
-        }
-        return treeId.subList(0, treeId.size() - 1);
-    }
-
-
-    public Entry getEntryById(List<Integer> treeId) {
-
-        if (treeId.isEmpty()) {
-            return root;
-        }
-        Entry entry = root.children.get(treeId.get(0));
-        for (int id : treeId.subList(1, treeId.size())) {
-            entry = entry.children.get(id);
-        }
-        return entry;
-    }
-
-
-    public Stream<Entry> mainEntries() {
-
-        return root.children.stream();
-    }
-
-
-    public Stream<Entry> allEntriesInOrder() {
-
-        return Stream.of(root).flatMap(DocumentTree::flatten);
-    }
-
-
-    public Stream<Entry> allSubEntriesInOrder(List<Integer> parentId) {
-
-        return getEntryById(parentId).children.stream().flatMap(DocumentTree::flatten);
-    }
-
-
-    @Override
-    public String toString() {
-
-        return String.join("\n", allEntriesInOrder().map(Entry::toString).toList());
-    }
-
-
-    private static Stream<Entry> flatten(Entry entry) {
-
-        return Stream.concat(Stream.of(entry), entry.children.stream().flatMap(DocumentTree::flatten));
-    }
-
-
-    public SemanticNode getHighestParentById(List<Integer> treeId) {
-
-        if (treeId.isEmpty()) {
-            return root.node;
-        }
-        return root.children.get(treeId.get(0)).node;
-    }
-
-
-    @Builder
-    @Getter
-    @AllArgsConstructor
-    @FieldDefaults(level = AccessLevel.PRIVATE, makeFinal = true)
-    public static class Entry {
-
-        List<Integer> treeId;
-        SemanticNode node;
-        @Builder.Default
-        List<Entry> children = new LinkedList<>();
-
-
-        @Override
-        public String toString() {
-
-            return node.toString();
-        }
-
-
-        public NodeType getType() {
-
-            return node.getType();
-        }
-
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/entity/EntityType.java

@@ -1,8 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.entity;
-
-public enum EntityType {
-    ENTITY,
-    RECOMMENDATION,
-    FALSE_POSITIVE,
-    FALSE_RECOMMENDATION
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/entity/RedactionEntity.java

@@ -1,228 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.entity;
-
-import java.awt.geom.Rectangle2D;
-import java.util.Collection;
-import java.util.Comparator;
-import java.util.Deque;
-import java.util.HashSet;
-import java.util.LinkedList;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-
-import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Engine;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.Boundary;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
-import com.knecon.fforesight.service.layoutparser.processor.utils.IdBuilder;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-@EqualsAndHashCode(onlyExplicitlyIncluded = true)
-public class RedactionEntity {
-
-    // initial values
-    @EqualsAndHashCode.Include
-    final Boundary boundary;
-    @EqualsAndHashCode.Include
-    final String type;
-    @EqualsAndHashCode.Include
-    final EntityType entityType;
-
-    // empty defaults
-    boolean redaction;
-    boolean removed;
-    boolean ignored;
-    boolean resized;
-    boolean skipRemoveEntitiesContainedInLarger;
-    boolean dictionaryEntry;
-    boolean dossierDictionaryEntry;
-    Set<Engine> engines;
-    Set<RedactionEntity> references;
-    @Builder.Default
-    Deque<Integer> matchedRules = new LinkedList<>();
-    String redactionReason;
-    String legalBasis;
-
-    // inferred on graph insertion
-    @EqualsAndHashCode.Include
-    String value;
-    String textBefore;
-    String textAfter;
-    @Builder.Default
-    Set<Page> pages = new HashSet<>();
-    List<RedactionPosition> redactionPositionsPerPage;
-    @Builder.Default
-    List<SemanticNode> intersectingNodes = new LinkedList<>();
-    SemanticNode deepestFullyContainingNode;
-
-
-    public static RedactionEntity initialEntityNode(Boundary boundary, String type, EntityType entityType) {
-
-        return RedactionEntity.builder().type(type).entityType(entityType).boundary(boundary).engines(new HashSet<>()).references(new HashSet<>()).build();
-    }
-
-
-    public boolean occursInNodeOfType(Class<? extends SemanticNode> clazz) {
-
-        return intersectingNodes.stream().anyMatch(clazz::isInstance);
-    }
-
-
-    public boolean occursInNode(SemanticNode semanticNode) {
-
-        return intersectingNodes.stream().anyMatch(node -> node.equals(semanticNode));
-    }
-
-
-    public boolean isType(String type) {
-
-        return this.type.equals(type);
-    }
-
-
-    public boolean isAnyType(List<String> types) {
-
-        return types.contains(type);
-    }
-
-
-    public void addIntersectingNode(SemanticNode containingNode) {
-
-        intersectingNodes.add(containingNode);
-    }
-
-
-    public void removeFromGraph() {
-
-        intersectingNodes.forEach(node -> node.getEntities().remove(this));
-        pages.forEach(page -> page.getEntities().remove(this));
-        intersectingNodes = new LinkedList<>();
-        deepestFullyContainingNode = null;
-        pages = new HashSet<>();
-        removed = true;
-        ignored = true;
-    }
-
-
-    public void addMatchedRule(int ruleNumber) {
-
-        matchedRules.add(ruleNumber);
-    }
-
-
-    public int getMatchedRule() {
-
-        if (matchedRules.isEmpty()) {
-            return 0;
-        }
-        return matchedRules.getLast();
-    }
-
-
-    public List<RedactionPosition> getRedactionPositionsPerPage() {
-
-        if (redactionPositionsPerPage == null || redactionPositionsPerPage.isEmpty()) {
-            Map<Page, List<Rectangle2D>> rectanglesPerLinePerPage = deepestFullyContainingNode.getTextBlock().getPositionsPerPage(boundary);
-
-            Page firstPage = rectanglesPerLinePerPage.keySet()
-                    .stream()
-                    .min(Comparator.comparingInt(Page::getNumber))
-                    .orElseThrow(() -> new RuntimeException("No Positions found on any page!"));
-            String id = IdBuilder.buildId(pages, rectanglesPerLinePerPage.values().stream().flatMap(Collection::stream).toList());
-            redactionPositionsPerPage = rectanglesPerLinePerPage.entrySet().stream().map(entry -> buildRedactionPosition(firstPage, id, entry)).toList();
-        }
-        return redactionPositionsPerPage;
-    }
-
-
-    private static RedactionPosition buildRedactionPosition(Page firstPage, String id, Map.Entry<Page, List<Rectangle2D>> entry) {
-
-        if (entry.getKey().equals(firstPage)) {
-            return new RedactionPosition(id, entry.getKey(), entry.getValue());
-        } else {
-            return new RedactionPosition(id + "-" + entry.getKey().getNumber(), entry.getKey(), entry.getValue());
-        }
-    }
-
-
-    public boolean containedBy(RedactionEntity redactionEntity) {
-
-        return this.boundary.containedBy(redactionEntity.getBoundary());
-    }
-
-
-    public boolean contains(RedactionEntity redactionEntity) {
-
-        return this.boundary.contains(redactionEntity.getBoundary());
-    }
-
-
-    public boolean intersects(RedactionEntity redactionEntity) {
-
-        return this.boundary.intersects(redactionEntity.getBoundary());
-    }
-
-
-    public void addEngine(Engine engine) {
-
-        engines.add(engine);
-    }
-
-
-    public void addEngines(Set<Engine> engines) {
-
-        this.engines.addAll(engines);
-    }
-
-
-    public void addReference(RedactionEntity reference) {
-
-        references.add(reference);
-    }
-
-
-    public void addReferences(List<RedactionEntity> references) {
-
-        this.references.addAll(references);
-    }
-
-
-    public boolean matchesAnnotationId(String manualRedactionId) {
-
-        return getRedactionPositionsPerPage().stream().anyMatch(entityPosition -> entityPosition.getId().equals(manualRedactionId));
-    }
-
-
-    @Override
-    public String toString() {
-
-        StringBuilder sb = new StringBuilder();
-        sb.append("Entity[\"");
-        sb.append(value);
-        sb.append("\", ");
-        sb.append(boundary);
-        sb.append(", pages[");
-        pages.forEach(page -> {
-            sb.append(page.getNumber());
-            sb.append(", ");
-        });
-        sb.delete(sb.length() - 2, sb.length());
-        sb.append("], type = \"");
-        sb.append(type);
-        sb.append("\", EntityType.");
-        sb.append(entityType);
-        sb.append("]");
-        return sb.toString();
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/entity/RedactionPosition.java

@@ -1,24 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.entity;
-
-import java.awt.geom.Rectangle2D;
-import java.util.List;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Data;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class RedactionPosition {
-
-    final String id;
-    Page page;
-    // Each entry in this list corresponds to an entry in the redaction log, this means:
-    // An entity might be represented by multiple redaction log entries
-    List<Rectangle2D> rectanglePerLine;
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Document.java

@@ -1,134 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.awt.geom.Rectangle2D;
-import java.util.Collections;
-import java.util.HashMap;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-import java.util.stream.Collectors;
-import java.util.stream.Stream;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@NoArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class Document implements GenericSemanticNode {
-
-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-
-    Set<Page> pages;
-    DocumentTree documentTree;
-    Integer numberOfPages;
-    TextBlock textBlock;
-    @Builder.Default
-    Set<RedactionEntity> entities = new HashSet<>();
-
-
-    @Override
-    public NodeType getType() {
-
-        return NodeType.DOCUMENT;
-    }
-
-
-    public TextBlock getTextBlock() {
-
-        if (textBlock == null) {
-            textBlock = streamTerminalTextBlocksInOrder().collect(new TextBlockCollector());
-        }
-        return textBlock;
-    }
-
-
-    public List<Section> getMainSections() {
-
-        return streamChildrenOfType(NodeType.SECTION).map(node -> (Section) node)
-                .collect(Collectors.toList());
-    }
-
-
-    public Stream<TextBlock> streamTerminalTextBlocksInOrder() {
-
-        return streamAllNodes().filter(SemanticNode::isLeaf)
-                .map(SemanticNode::getLeafTextBlock);
-    }
-
-
-    @Override
-    public List<Integer> getTreeId() {
-
-        return Collections.emptyList();
-    }
-
-
-    @Override
-    public void setTreeId(List<Integer> tocId) {
-
-        throw new UnsupportedOperationException("Document is always the root of the TablePageBlock of Contents");
-    }
-
-
-    @Override
-    public Headline getHeadline() {
-
-        return streamAllSubNodesOfType(NodeType.HEADLINE).map(node -> (Headline) node)
-                .findFirst()
-                .orElse(Headline.builder().build());
-    }
-
-
-    private Stream<SemanticNode> streamAllNodes() {
-
-        return documentTree.allEntriesInOrder()
-                .map(DocumentTree.Entry::getNode);
-    }
-
-
-    public Stream<Image> streamAllImages() {
-
-        return streamAllSubNodesOfType(NodeType.IMAGE).map(node -> (Image) node);
-    }
-
-
-    public Map<NodeType, Long> buildSemanticNodeCounts() {
-
-        return streamAllSubNodes().collect(Collectors.groupingBy(SemanticNode::getType, Collectors.counting()));
-    }
-
-
-    @Override
-    public String toString() {
-
-        return NodeType.DOCUMENT + ": " + this.getTextBlock().buildSummary();
-    }
-
-
-    @Override
-    public Map<Page, Rectangle2D> getBBox() {
-
-        Map<Page, Rectangle2D> bBox = new HashMap<>();
-        for (Page page : pages) {
-            bBox.put(page, new Rectangle2D.Double(0, 0, page.getWidth(), page.getHeight()));
-        }
-        return bBox;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Footer.java

@@ -1,84 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.awt.geom.Rectangle2D;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.NoArgsConstructor;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@NoArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class Footer implements GenericSemanticNode {
-
-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-
-    List<Integer> treeId;
-    TextBlock leafTextBlock;
-
-    @EqualsAndHashCode.Exclude
-    DocumentTree documentTree;
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<RedactionEntity> entities = new HashSet<>();
-
-    @EqualsAndHashCode.Exclude
-    Map<Page, Rectangle2D> bBoxCache;
-
-
-    @Override
-    public NodeType getType() {
-
-        return NodeType.FOOTER;
-    }
-
-
-    @Override
-    public boolean isLeaf() {
-
-        return true;
-    }
-
-
-    @Override
-    public TextBlock getTextBlock() {
-
-        return leafTextBlock;
-    }
-
-
-    @Override
-    public String toString() {
-
-        return treeId + ": " + NodeType.FOOTER + ": " + leafTextBlock.buildSummary();
-    }
-
-
-    @Override
-    public Map<Page, Rectangle2D> getBBox() {
-
-        if (bBoxCache == null) {
-            bBoxCache = GenericSemanticNode.super.getBBox();
-        }
-        return bBoxCache;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/GenericSemanticNode.java

@@ -1,5 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-public interface GenericSemanticNode extends SemanticNode {
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Header.java

@@ -1,83 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.awt.geom.Rectangle2D;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.NoArgsConstructor;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@NoArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class Header implements GenericSemanticNode {
-
-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-    List<Integer> treeId;
-    TextBlock leafTextBlock;
-
-    @EqualsAndHashCode.Exclude
-    DocumentTree documentTree;
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<RedactionEntity> entities = new HashSet<>();
-
-    @EqualsAndHashCode.Exclude
-    Map<Page, Rectangle2D> bBoxCache;
-
-
-    @Override
-    public boolean isLeaf() {
-
-        return true;
-    }
-
-
-    @Override
-    public NodeType getType() {
-
-        return NodeType.HEADER;
-    }
-
-
-    @Override
-    public TextBlock getTextBlock() {
-
-        return leafTextBlock;
-    }
-
-
-    @Override
-    public String toString() {
-
-        return treeId + ": " + NodeType.HEADER + ": " + leafTextBlock.buildSummary();
-    }
-
-
-    @Override
-    public Map<Page, Rectangle2D> getBBox() {
-
-        if (bBoxCache == null) {
-            bBoxCache = GenericSemanticNode.super.getBBox();
-        }
-        return bBoxCache;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Headline.java

@@ -1,90 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.awt.geom.Rectangle2D;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.NoArgsConstructor;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@NoArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class Headline implements GenericSemanticNode {
-
-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-    List<Integer> treeId;
-    TextBlock leafTextBlock;
-
-    @EqualsAndHashCode.Exclude
-    DocumentTree documentTree;
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<RedactionEntity> entities = new HashSet<>();
-
-    @EqualsAndHashCode.Exclude
-    Map<Page, Rectangle2D> bBoxCache;
-
-
-    @Override
-    public NodeType getType() {
-
-        return NodeType.HEADLINE;
-    }
-
-
-    @Override
-    public boolean isLeaf() {
-
-        return true;
-    }
-
-
-    @Override
-    public TextBlock getTextBlock() {
-
-        return leafTextBlock;
-    }
-
-
-    @Override
-    public String toString() {
-
-        return treeId + ": " + NodeType.HEADLINE + ": " + leafTextBlock.buildSummary();
-    }
-
-
-    @Override
-    public Headline getHeadline() {
-
-        return this;
-    }
-
-
-    @Override
-    public Map<Page, Rectangle2D> getBBox() {
-
-        if (bBoxCache == null) {
-            bBoxCache = GenericSemanticNode.super.getBBox();
-        }
-        return bBoxCache;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Image.java

@@ -1,101 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.awt.geom.Rectangle2D;
-import java.util.Collections;
-import java.util.HashMap;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.NoArgsConstructor;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@NoArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class Image implements GenericSemanticNode {
-
-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-
-    List<Integer> treeId;
-    String id;
-
-    ImageType imageType;
-    boolean transparent;
-    Rectangle2D position;
-
-    boolean redaction;
-    boolean ignored;
-    @Builder.Default
-    String redactionReason = "";
-    @Builder.Default
-    String legalBasis = "";
-    @Builder.Default
-    int matchedRule = -1;
-
-    @EqualsAndHashCode.Exclude
-    Page page;
-
-    @EqualsAndHashCode.Exclude
-    DocumentTree documentTree;
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<RedactionEntity> entities = new HashSet<>();
-
-
-    @Override
-    public NodeType getType() {
-
-        return NodeType.IMAGE;
-    }
-
-
-    @Override
-    public TextBlock getTextBlock() {
-
-        return streamAllSubNodes().filter(SemanticNode::isLeaf)
-                .map(SemanticNode::getLeafTextBlock)
-                .collect(new TextBlockCollector());
-    }
-
-
-    @Override
-    public Set<Page> getPages() {
-
-        return Collections.singleton(page);
-    }
-
-
-    @Override
-    public String toString() {
-
-        return treeId + ": " + NodeType.IMAGE + ": " + imageType.toString() + " " + position;
-    }
-
-
-    @Override
-    public Map<Page, Rectangle2D> getBBox() {
-
-        Map<Page, Rectangle2D> bBoxPerPage = new HashMap<>();
-        bBoxPerPage.put(page, position);
-        return bBoxPerPage;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/ImageType.java

@@ -1,25 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.util.Locale;
-
-public enum ImageType {
-    LOGO,
-    FORMULA,
-    SIGNATURE,
-
-    SIGNATURE_VISUAL,
-    OTHER,
-    OCR;
-
-
-    public static ImageType fromString(String imageType) {
-
-        return switch (imageType.toLowerCase(Locale.ROOT)) {
-            case "logo" -> ImageType.LOGO;
-            case "formula" -> ImageType.FORMULA;
-            case "signature" -> ImageType.SIGNATURE;
-            case "ocr" -> ImageType.OCR;
-            default -> ImageType.OTHER;
-        };
-    }
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Page.java

@@ -1,87 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.util.HashSet;
-import java.util.LinkedList;
-import java.util.List;
-import java.util.Set;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.EqualsAndHashCode;
-import lombok.Getter;
-import lombok.Setter;
-import lombok.experimental.FieldDefaults;
-
-@Getter
-@Setter
-@Builder
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class Page {
-
-    Integer number;
-    Integer height;
-    Integer width;
-    Integer rotation;
-
-    @EqualsAndHashCode.Exclude
-    List<SemanticNode> mainBody;
-    @EqualsAndHashCode.Exclude
-    Header header;
-    @EqualsAndHashCode.Exclude
-    Footer footer;
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<RedactionEntity> entities = new HashSet<>();
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<Image> images = new HashSet<>();
-
-
-    public static Page fromClassificationPage(ClassificationPage classificationPage) {
-
-        return Page.builder()
-                .height((int) classificationPage.getPageHeight())
-                .width((int) classificationPage.getPageWidth())
-                .number(classificationPage.getPageNumber())
-                .rotation(classificationPage.getRotation())
-                .mainBody(new LinkedList<>())
-                .build();
-    }
-
-
-    public TextBlock getMainBodyTextBlock() {
-
-        return mainBody.stream().filter(SemanticNode::isLeaf).map(SemanticNode::getLeafTextBlock).collect(new TextBlockCollector());
-    }
-
-
-    @Override
-    public String toString() {
-
-        return String.valueOf(number);
-    }
-
-
-    @Override
-    public int hashCode() {
-
-        return number;
-    }
-
-
-    @Override
-    public boolean equals(Object o) {
-
-        return o instanceof Page && o.hashCode() == this.hashCode();
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Paragraph.java

@@ -1,82 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.awt.geom.Rectangle2D;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class Paragraph implements GenericSemanticNode {
-
-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-
-    List<Integer> treeId;
-    TextBlock leafTextBlock;
-
-    @EqualsAndHashCode.Exclude
-    DocumentTree documentTree;
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<RedactionEntity> entities = new HashSet<>();
-
-    @EqualsAndHashCode.Exclude
-    Map<Page, Rectangle2D> bBoxCache;
-
-
-    @Override
-    public NodeType getType() {
-
-        return NodeType.PARAGRAPH;
-    }
-
-
-    @Override
-    public boolean isLeaf() {
-
-        return true;
-    }
-
-
-    @Override
-    public TextBlock getTextBlock() {
-
-        return leafTextBlock;
-    }
-
-
-    @Override
-    public String toString() {
-
-        return treeId + ": " + NodeType.PARAGRAPH + ": " + leafTextBlock.buildSummary();
-    }
-
-
-    @Override
-    public Map<Page, Rectangle2D> getBBox() {
-
-        if (bBoxCache == null) {
-            bBoxCache = GenericSemanticNode.super.getBBox();
-        }
-        return bBoxCache;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Section.java

@@ -1,98 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.awt.geom.Rectangle2D;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.experimental.FieldDefaults;
-import lombok.extern.slf4j.Slf4j;
-
-@Slf4j
-@Data
-@Builder
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class Section implements GenericSemanticNode {
-
-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-    List<Integer> treeId;
-
-    TextBlock textBlock;
-    @EqualsAndHashCode.Exclude
-    DocumentTree documentTree;
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<RedactionEntity> entities = new HashSet<>();
-
-    @EqualsAndHashCode.Exclude
-    Map<Page, Rectangle2D> bBoxCache;
-
-
-    @Override
-    public NodeType getType() {
-
-        return NodeType.SECTION;
-    }
-
-
-    public boolean hasTables() {
-
-        return streamAllSubNodesOfType(NodeType.TABLE).findAny()
-                .isPresent();
-    }
-
-
-    @Override
-    public TextBlock getTextBlock() {
-
-        if (textBlock == null) {
-            textBlock = streamAllSubNodes().filter(SemanticNode::isLeaf)
-                    .map(SemanticNode::getLeafTextBlock)
-                    .collect(new TextBlockCollector());
-        }
-        return textBlock;
-    }
-
-
-    @Override
-    public String toString() {
-
-        return treeId.toString() + ": " + NodeType.SECTION + ": " + this.getTextBlock().buildSummary();
-    }
-
-
-    public Headline getHeadline() {
-
-        return streamChildrenOfType(NodeType.HEADLINE)//
-                .map(node -> (Headline) node)//
-                .findFirst()//
-                .orElseGet(() -> getParent().getHeadline());
-    }
-
-
-    @Override
-    public Map<Page, Rectangle2D> getBBox() {
-
-        if (bBoxCache == null) {
-            bBoxCache = GenericSemanticNode.super.getBBox();
-        }
-        return bBoxCache;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/SemanticNode.java

@@ -1,477 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import static java.lang.String.format;
-
-import java.awt.geom.Rectangle2D;
-import java.util.Comparator;
-import java.util.HashMap;
-import java.util.List;
-import java.util.Locale;
-import java.util.Map;
-import java.util.Set;
-import java.util.stream.Collectors;
-import java.util.stream.Stream;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.Boundary;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.EntityType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.AtomicTextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
-
-public interface SemanticNode {
-
-    /**
-     * Returns the type of this node, such as Section, Paragraph, etc.
-     *
-     * @return NodeType of this node
-     */
-    NodeType getType();
-
-
-    /**
-     * Searches all Nodes located underneath this Node in the DocumentTree and concatenates their AtomicTextBlocks into a single TextBlock.
-     * So, for a Section all TextBlocks of Subsections, Paragraphs, and Tables are concatenated into a single TextBlock
-     * If the Node is a Leaf, the LeafTextBlock will be returned instead.
-     *
-     * @return TextBlock containing all AtomicTextBlocks that are located under this Node.
-     */
-    TextBlock getTextBlock();
-
-
-    /**
-     * Any Node maintains its own Set of Entities.
-     * This Set contains all Entities whose boundary intersects the boundary of this node.
-     *
-     * @return Set of all Entities associated with this Node
-     */
-    Set<RedactionEntity> getEntities();
-
-
-    /**
-     * Each AtomicTextBlock is assigned a page, so to get the pages this node appears on, it collects the PageNodes from each AtomicTextBlock belonging to this node's TextBlock.
-     *
-     * @return Set of PageNodes this node appears on.
-     */
-    default Set<Page> getPages() {
-
-        return getTextBlock().getPages();
-    }
-
-
-    default Page getFirstPage() {
-
-        return getTextBlock().getPages().stream().min(Comparator.comparingInt(Page::getNumber)).orElseThrow(() -> new IllegalStateException("SemanticNode has no Page!"));
-    }
-
-
-    /**
-     * Each AtomicTextBlock is assigned a page, so to get the pages for this boundary, it collects the PageNodes from each AtomicTextBlock belonging to this node's TextBlock.
-     *
-     * @return Set of PageNodes this node appears on.
-     */
-    default Set<Page> getPages(Boundary boundary) {
-
-        if (!getBoundary().contains(boundary)) {
-            throw new IllegalArgumentException(format("%s which was used to query for pages is not contained in the %s of this node!", boundary, getBoundary()));
-        }
-        return getTextBlock().getPages(boundary);
-    }
-
-
-    default boolean isOnPage(int pageNumber) {
-
-        return getPages().stream().anyMatch(page -> page.getNumber() == pageNumber);
-    }
-
-
-    /**
-     * Returns the DocumentTree Object.
-     *
-     * @return the DocumentTree of the Document this node belongs to
-     */
-    DocumentTree getDocumentTree();
-
-
-    /**
-     * The id is a List of Integers uniquely identifying this node in the DocumentTree.
-     *
-     * @return the DocumentTree ID
-     */
-    List<Integer> getTreeId();
-
-
-    /**
-     * This should only be used during graph construction.
-     *
-     * @param tocId List of Integers
-     */
-    void setTreeId(List<Integer> tocId);
-
-
-    /**
-     * Traverses the Tree up, until it hits a Headline or hits a Section which will then return the first Headline from its children.
-     * Throws NotFoundException if no Headline is found this way
-     *
-     * @return First Headline found
-     */
-    default Headline getHeadline() {
-
-        return getParent().getHeadline();
-    }
-
-
-    /**
-     * Checks if its TocId has a length greater than zero.
-     *
-     * @return boolean indicating whether this Node has a Parent in the DocumentTree
-     */
-    default boolean hasParent() {
-
-        return getDocumentTree().hasParentById(getTreeId());
-    }
-
-
-    /**
-     * @return The SemanticNode representing the Parent in the DocumentTree
-     * throws NotFoundException, when no parent is present
-     */
-    default SemanticNode getParent() {
-
-        return getDocumentTree().getParentEntryById(getTreeId()).getNode();
-    }
-
-
-    /**
-     * @return The SemanticNode which is directly underneath the document and also under which this node is.
-     * if this is the highest child node or the document itself, it returns itself.
-     */
-    default SemanticNode getHighestParent() {
-
-        return getDocumentTree().getHighestParentById(getTreeId());
-    }
-
-
-    /**
-     * Leaf means a SemanticNode has direct access to a TextBlock, by default this is false and must be overridden.
-     * Currently only Sections, Images, and Tables are not leaves.
-     * A TableCell might be a leaf depending on its area compared to the page.
-     *
-     * @return boolean, indicating if a Node has direct access to a TextBlock
-     */
-    default boolean isLeaf() {
-
-        return false;
-    }
-
-
-    /**
-     * Leaf means a SemanticNode has direct access to a TextBlock, by default this is false and must be overridden.
-     * Currently only Sections and Tables are no leaves.
-     *
-     * @return AtomicTextBlock
-     */
-    default TextBlock getLeafTextBlock() {
-
-        throw new UnsupportedOperationException("Only leaf Nodes have access to LeafTextBlocks!");
-    }
-
-
-    /**
-     * Should only be used during construction of the Graph. Sets the LeafTextBlock of this SemanticNode.
-     *
-     * @param textBlock the TextBlock to set as the LeafTextBlock of this SemanticNode
-     */
-    default void setLeafTextBlock(TextBlock textBlock) {
-
-        throw new UnsupportedOperationException();
-    }
-
-
-    /**
-     * Checks whether this SemanticNode has any Entity with EntityType.ENTITY of the provided type.
-     *
-     * @param type string representing the type of entity to check for
-     * @return true, if this SemanticNode has at least one Entity of the provided type
-     */
-    default boolean hasEntitiesOfType(String type) {
-
-        return getEntities().stream().filter(entity -> entity.getEntityType().equals(EntityType.ENTITY)).anyMatch(redactionEntity -> redactionEntity.getType().equals(type));
-    }
-
-
-    /**
-     * Returns a List of Entities in this SemanticNode which are of the provided type such as "CBI_author".
-     *
-     * @param type string representing the type of entities to return
-     * @return List of RedactionEntities of any the type
-     */
-    default List<RedactionEntity> getEntitiesOfType(String type) {
-
-        return getEntities().stream().filter(redactionEntity -> redactionEntity.getType().equals(type)).toList();
-    }
-
-
-    /**
-     * Returns a List of Entities in this SemanticNode which have any of the provided types such as "CBI_author".
-     *
-     * @param types A list of strings representing the types of entities to return
-     * @return List of RedactionEntities of any provided type
-     */
-    default List<RedactionEntity> getEntitiesOfType(List<String> types) {
-
-        return getEntities().stream().filter(redactionEntity -> redactionEntity.isAnyType(types)).toList();
-    }
-
-
-    /**
-     * Each AtomicTextBlock has an index on its page, this returns the number of the first AtomicTextBlock underneath this node.
-     * If this node does not have any AtomicTexBlocks underneath it, e.g. an empty TableCell. It returns -1.
-     *
-     * @return Integer representing the number on the page
-     */
-    default Integer getNumberOnPage() {
-
-        TextBlock textBlock = getTextBlock();
-        if (!textBlock.getAtomicTextBlocks().isEmpty()) {
-            return getTextBlock().getAtomicTextBlocks().get(0).getNumberOnPage();
-        } else {
-            return -1;
-        }
-    }
-
-
-    /**
-     * Checks if the SemanticNode contains any text.
-     *
-     * @return true, if this node's TextBlock is not empty
-     */
-    default boolean hasText() {
-
-        return !getTextBlock().isEmpty();
-    }
-
-
-    /**
-     * Checks whether this SemanticNode contains the provided String.
-     *
-     * @param string A String which the TextBlock might contain
-     * @return true, if this node's TextBlock contains the string
-     */
-    default boolean containsString(String string) {
-
-        return getTextBlock().getSearchText().contains(string);
-    }
-
-
-    /**
-     * Checks whether this SemanticNode contains all the provided Strings.
-     *
-     * @param strings A List of Strings which the TextBlock might contain
-     * @return true, if this node's TextBlock contains all strings
-     */
-    default boolean containsStrings(List<String> strings) {
-
-        return strings.stream().allMatch(this::containsString);
-    }
-
-
-    /**
-     * Checks whether this SemanticNode contains all the provided Strings ignoring case.
-     *
-     * @param string A String which the TextBlock might contain
-     * @return true, if this node's TextBlock contains the string ignoring case
-     */
-    default boolean containsStringIgnoreCase(String string) {
-
-        return getTextBlock().getSearchText().toLowerCase(Locale.ROOT).contains(string.toLowerCase(Locale.ROOT));
-    }
-
-
-    /**
-     * Checks whether this SemanticNode contains any of the provided Strings.
-     *
-     * @param strings A List of Strings which the TextBlock might contain
-     * @return true, if this node's TextBlock contains any of the strings
-     */
-    default boolean containsAnyString(List<String> strings) {
-
-        return strings.stream().anyMatch(this::containsString);
-    }
-
-
-    /**
-     * Checks whether this SemanticNode contains any of the provided Strings ignoring case.
-     *
-     * @param strings A List of Strings which the TextBlock might contain
-     * @return true, if this node's TextBlock contains any of the strings
-     */
-    default boolean containsAnyStringIgnoreCase(List<String> strings) {
-
-        return strings.stream().anyMatch(this::containsStringIgnoreCase);
-    }
-
-
-    /**
-     * This function is used during insertion of EntityNodes into the graph, it checks if the boundary of the RedactionEntity intersects or even contains the RedactionEntity.
-     * It sets the fields accordingly and recursively calls this function on all its children.
-     *
-     * @param redactionEntity RedactionEntity, which is being inserted into the graph
-     */
-    default void addThisToEntityIfIntersects(RedactionEntity redactionEntity) {
-
-        TextBlock textBlock = getTextBlock();
-        if (textBlock.getBoundary().intersects(redactionEntity.getBoundary())) {
-            if (textBlock.containsBoundary(redactionEntity.getBoundary())) {
-                redactionEntity.setDeepestFullyContainingNode(this);
-            }
-
-            redactionEntity.addIntersectingNode(this);
-            streamChildren().filter(semanticNode -> semanticNode.getBoundary().intersects(redactionEntity.getBoundary()))
-                    .forEach(node -> node.addThisToEntityIfIntersects(redactionEntity));
-        }
-    }
-
-
-    /**
-     * returns the set of layoutengines.
-     *
-     * @return set of layoutengines.
-     */
-    Set<LayoutEngine> getEngines();
-
-
-    /**
-     * adds a layoutengine to the set.
-     */
-    default void addEngine(LayoutEngine engine) {
-
-        getEngines().add(engine);
-    }
-
-
-    /**
-     * Streams all children located directly underneath this node in the DocumentTree.
-     *
-     * @return Stream of all children
-     */
-    default Stream<SemanticNode> streamChildren() {
-
-        return getDocumentTree().childNodes(getTreeId());
-    }
-
-
-    /**
-     * Streams all children located directly underneath this node in the DocumentTree of the provided type.
-     *
-     * @return Stream of all children
-     */
-    default Stream<SemanticNode> streamChildrenOfType(NodeType nodeType) {
-
-        return getDocumentTree().childNodesOfType(getTreeId(), nodeType);
-    }
-
-
-    /**
-     * Recursively streams all SemanticNodes located underneath this node in the DocumentTree in order.
-     *
-     * @return Stream of all SubNodes
-     */
-    default Stream<SemanticNode> streamAllSubNodes() {
-
-        return getDocumentTree().allSubEntriesInOrder(getTreeId()).map(DocumentTree.Entry::getNode);
-    }
-
-
-    /**
-     * Recursively streams all SemanticNodes of the provided type located underneath this node in the DocumentTree in order.
-     *
-     * @return Stream of all SubNodes
-     */
-    default Stream<SemanticNode> streamAllSubNodesOfType(NodeType nodeType) {
-
-        return getDocumentTree().allSubEntriesInOrder(getTreeId()).filter(entry -> entry.getType().equals(nodeType)).map(DocumentTree.Entry::getNode);
-    }
-
-
-    /**
-     * The Boundary is the start and end string offsets in the reading order of the document.
-     *
-     * @return Boundary of this Node's TextBlock
-     */
-    default Boundary getBoundary() {
-
-        return getTextBlock().getBoundary();
-    }
-
-
-    /**
-     * If this Node is a Leaf it will calculate the boundingBox of its LeafTextBlock, otherwise it will calculate the Union of the BoundingBoxes of all its Children.
-     * If called on the Document, it will return the cropbox of each page
-     *
-     * @return Rectangle2D fully encapsulating this Node for each page.
-     */
-    default Map<Page, Rectangle2D> getBBox() {
-
-        if (isLeaf()) {
-            return getBBoxFromLeafTextBlock();
-        }
-
-        return getBBoxFromChildren();
-    }
-
-
-    /**
-     * Checks whether the Bounding Box of this SemanticNode contains the provided rectangle on the provided page.
-     *
-     * @param rectangle2D The rectangle to check if it is contained
-     * @param pageNumber  The Page number on which the rectangle should be checked
-     * @return boolean
-     */
-    default boolean containsRectangle(Rectangle2D rectangle2D, Integer pageNumber) {
-
-        Page helperPage = Page.builder().number(pageNumber).build();
-        if (!getPages().contains(helperPage)) {
-            return false;
-        }
-        return getBBox().get(helperPage).contains(rectangle2D);
-    }
-
-
-    /**
-     * TODO: this produces unwanted results for sections spanning multiple columns.
-     * Computes the Union of the bounding boxes of all children recursively.
-     *
-     * @return The union of the BoundingBoxes of all children
-     */
-    private Map<Page, Rectangle2D> getBBoxFromChildren() {
-
-        Map<Page, Rectangle2D> bBoxPerPage = new HashMap<>();
-        List<Map<Page, Rectangle2D>> childrenBBoxes = streamChildren().map(SemanticNode::getBBox).toList();
-        Set<Page> pages = childrenBBoxes.stream().flatMap(map -> map.keySet().stream()).collect(Collectors.toSet());
-        for (Page page : pages) {
-            Rectangle2D bBoxOnPage = childrenBBoxes.stream()
-                    .filter(childBboxPerPage -> childBboxPerPage.containsKey(page))
-                    .map(childBboxPerPage -> childBboxPerPage.get(page))
-                    .collect(RectangleTransformations.collectBBox());
-            bBoxPerPage.put(page, bBoxOnPage);
-        }
-        return bBoxPerPage;
-    }
-
-
-    /**
-     * @return The union of all BoundingBoxes of the TextBlock of this node
-     */
-    private Map<Page, Rectangle2D> getBBoxFromLeafTextBlock() {
-
-        Map<Page, Rectangle2D> bBoxPerPage = new HashMap<>();
-        Map<Page, List<AtomicTextBlock>> atomicTextBlockPerPage = getTextBlock().getAtomicTextBlocks().stream().collect(Collectors.groupingBy(AtomicTextBlock::getPage));
-        atomicTextBlockPerPage.forEach((page, atbs) -> bBoxPerPage.put(page, RectangleTransformations.bBoxUnionAtomicTextBlock(atbs)));
-        return bBoxPerPage;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Table.java

@@ -1,359 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import static java.lang.String.format;
-
-import java.awt.geom.Rectangle2D;
-import java.util.Collection;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Locale;
-import java.util.Map;
-import java.util.Set;
-import java.util.stream.IntStream;
-import java.util.stream.Stream;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class Table implements SemanticNode {
-
-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-    List<Integer> treeId;
-    DocumentTree documentTree;
-
-    int numberOfRows;
-    int numberOfCols;
-    TextBlock textBlock;
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<RedactionEntity> entities = new HashSet<>();
-
-    @EqualsAndHashCode.Exclude
-    Map<Page, Rectangle2D> bBoxCache;
-
-
-    /**
-     * Streams all entities in this table, that appear in a row, which contains any of the provided strings.
-     *
-     * @param strings Strings to check whether a row contains them
-     * @return Stream of all entities in this table, that appear in a row, which contains any of the provided strings
-     */
-    public Stream<RedactionEntity> streamEntitiesWhereRowContainsStringsIgnoreCase(List<String> strings) {
-
-        return IntStream.range(0, numberOfRows).boxed()
-                .filter(row -> rowContainsStringsIgnoreCase(row, strings))
-                .flatMap(this::streamRow)
-                .map(TableCell::getEntities)
-                .flatMap(Collection::stream);
-    }
-
-
-    /**
-     * Checks whether the specified row contains all the provided strings.
-     *
-     * @param row     the row to check as an Integer, must be smaller than numberOfRows
-     * @param strings a list of strings to check for
-     * @return true, if all strings appear in the provided row
-     */
-    public boolean rowContainsStringsIgnoreCase(Integer row, List<String> strings) {
-
-        String rowText = streamRow(row).map(TableCell::getTextBlock)
-                .collect(new TextBlockCollector()).getSearchText().toLowerCase(Locale.ROOT);
-        return strings.stream()
-                .map(String::toLowerCase)
-                .allMatch(rowText::contains);
-    }
-
-
-    /**
-     * Streams all entities which appear in a row where at least one cell has the provided header and the provided value.
-     *
-     * @param header the header value to search for
-     * @param value  the string which the table cell should contain
-     * @return a stream of all entities, which appear in a row where at least one cell has the provided header and the provided value.
-     */
-    public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndValue(String header, String value) {
-
-        List<Integer> vertebrateStudyCols = streamHeaders().filter(headerNode -> headerNode.containsString(header))
-                .map(TableCell::getCol)
-                .toList();
-        return streamTableCells().filter(tableCellNode -> vertebrateStudyCols.stream()
-                        .anyMatch(vertebrateStudyCol -> getCell(tableCellNode.getRow(), vertebrateStudyCol).containsString(value)))
-                .map(TableCell::getEntities)
-                .flatMap(Collection::stream);
-    }
-
-
-    /**
-     * Streams all entities which appear in a row where at least one cell has the provided header and any provided value.
-     *
-     * @param header the header value to search for
-     * @param values the strings which the table cell should contain
-     * @return a stream of all entities, which appear in a row where at least one cell has the provided header and any provided value.
-     */
-    public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndAnyValue(String header, List<String> values) {
-
-        List<Integer> colsWithHeader = streamHeaders().filter(headerNode -> headerNode.containsString(header))
-                .map(TableCell::getCol)
-                .toList();
-        return streamTableCells().filter(tableCellNode -> colsWithHeader.stream()
-                        .anyMatch(colWithHeader -> getCell(tableCellNode.getRow(), colWithHeader).containsAnyString(values)))
-                .map(TableCell::getEntities)
-                .flatMap(Collection::stream);
-    }
-
-
-    /**
-     * Streams all entities in this table, that appear in a row, which contains at least one entity with any of the provided types.
-     *
-     * @param types type strings to check whether a row contains an entity like them
-     * @return Stream of all entities in this table, that appear in a row, which contains at least one entity with any of the provided types.
-     */
-    public Stream<RedactionEntity> streamEntitiesWhereRowContainsEntitiesOfType(List<String> types) {
-
-        List<Integer> rowsWithEntityOfType = IntStream.range(0, numberOfRows).boxed()
-                .filter(rowNumber -> streamEntityTypesInRow(rowNumber).anyMatch(existingType -> types.stream()
-                        .anyMatch(typeToCheck -> typeToCheck.equals(existingType))))
-                .toList();
-
-        return rowsWithEntityOfType.stream()
-                .flatMap(this::streamRow)
-                .map(TableCell::getEntities)
-                .flatMap(Collection::stream);
-    }
-
-
-    /**
-     * Streams all entities in this table, that appear in a row, which does not contain any entity with any of the provided types.
-     *
-     * @param types type strings to check whether a row doesn't contain an entity like it
-     * @return Stream of all entities in this table, that appear in a row, which does not contain any entity with any of the provided types.
-     */
-    public Stream<RedactionEntity> streamEntitiesWhereRowContainsNoEntitiesOfType(List<String> types) {
-
-        List<Integer> rowsWithNoEntityOfType = IntStream.range(0, numberOfRows).boxed()
-                .filter(rowNumber -> streamEntityTypesInRow(rowNumber).noneMatch(existingType -> types.stream()
-                        .anyMatch(typeToCheck -> typeToCheck.equals(existingType))))
-                .toList();
-
-        return rowsWithNoEntityOfType.stream()
-                .flatMap(this::streamRow)
-                .map(TableCell::getEntities)
-                .flatMap(Collection::stream);
-    }
-
-
-    private Stream<String> streamEntityTypesInRow(Integer rowNumber) {
-
-        return streamRow(rowNumber).map(TableCell::getEntities)
-                .flatMap(Collection::stream)
-                .map(RedactionEntity::getType)
-                .distinct();
-    }
-
-
-    /**
-     * Returns a TableCell at the provided row and column location.
-     *
-     * @param row int representing the row, must be smaller than numberOfRows
-     * @param col int representing the col, must be smaller than numberOfCols
-     * @return TableCell at the provided location in the table
-     */
-    public TableCell getCell(int row, int col) {
-
-        if (numberOfRows - row < 0 || numberOfCols - col < 0) {
-            throw new IllegalArgumentException(format("row %d, col %d is out of bounds for number of rows of %d and number of cols %d", row, col, numberOfRows, numberOfCols));
-        }
-        int idx = row * numberOfCols + col;
-        return (TableCell) documentTree.getEntryById(treeId).getChildren()
-                .get(idx).getNode();
-    }
-
-
-    /**
-     * Streams all TableCells in this Table row-wise.
-     *
-     * @return Stream of all TableCells
-     */
-    public Stream<TableCell> streamTableCells() {
-
-        return streamChildrenOfType(NodeType.TABLE_CELL).map(node -> (TableCell) node);
-    }
-
-
-    /**
-     * Streams all TableCells in this Table which have the provided header row-wise.
-     *
-     * @return Stream of all TableCells which have the provided header
-     */
-    public Stream<TableCell> streamTableCellsWithHeader(String header) {
-
-        return streamHeaders().filter(tableCellNode -> tableCellNode.getTextBlock().getSearchText().contains(header))
-                .map(TableCell::getCol)
-                .flatMap(this::streamCol)
-                .filter(tableCellNode -> !tableCellNode.isHeader());
-    }
-
-
-    /**
-     * Streams all TableCells belonging to the provided column from top down.
-     *
-     * @param col int representing the column
-     * @return Stream of all TableCell in the provided column
-     */
-    public Stream<TableCell> streamCol(int col) {
-
-        return IntStream.range(0, numberOfRows).boxed()
-                .map(row -> getCell(row, col));
-    }
-
-
-    /**
-     * Streams all TableCells belonging to the provided row from left to right.
-     *
-     * @param row int representing the row
-     * @return Stream of all TableCell in the provided row
-     */
-    public Stream<TableCell> streamRow(int row) {
-
-        return IntStream.range(0, numberOfCols).boxed()
-                .map(col -> getCell(row, col));
-    }
-
-
-    /**
-     * Streams all TableCells row-wise and filters them with header == true.
-     *
-     * @return Stream of all TableCells with header == true
-     */
-    public Stream<TableCell> streamHeaders() {
-
-        return streamTableCells().filter(TableCell::isHeader);
-    }
-
-
-    /**
-     * Streams all TableCells of the provided row and column and filters them with header == true.
-     *
-     * @param row int representing the row
-     * @param col int representing the column
-     * @return Stream of all TableCells with header == true in the provided row or col
-     */
-    public Stream<TableCell> streamHeadersForCell(int row, int col) {
-
-        return Stream.concat(streamRow(row), streamCol(col))
-                .filter(TableCell::isHeader);
-    }
-
-
-    /**
-     * Streams all Headers and checks if any equal the provided string.
-     *
-     * @param header string to check the headers for
-     * @return true, if at least one header equals the provided string
-     */
-    public boolean hasHeader(String header) {
-
-        return streamHeaders().anyMatch(tableCellNode -> tableCellNode.getTextBlock().getSearchText().strip().equals(header));
-    }
-
-
-    /**
-     * Checks if this table has a column with the provided header and any of the table cells in that column contain the provided value.
-     *
-     * @param header string to find header cells
-     * @param value  string to check cells with provided header
-     * @return true, if this table has a column with the provided header and any of the table cells in that column contain the provided value
-     */
-    public boolean hasRowWithHeaderAndValue(String header, String value) {
-
-        return streamTableCellsWithHeader(header).anyMatch(tableCellNode -> tableCellNode.containsString(value));
-    }
-
-
-    /**
-     * Checks if this table has a column with the provided header and any of the table cells in that column contains any of the provided values.
-     *
-     * @param header string to find header cells
-     * @param values List of strings to check cells with provided header
-     * @return true, if this table has a column with the provided header and any of the table cells in that column contains any of the provided values.
-     */
-    public boolean hasRowWithHeaderAndAnyValue(String header, List<String> values) {
-
-        return streamTableCellsWithHeader(header).anyMatch(tableCellNode -> tableCellNode.containsAnyString(values));
-    }
-
-
-    /**
-     * Finds all entities of the provided type, which appear in the same row that the provided entity appears in.
-     *
-     * @param type            the type of entities to search for
-     * @param redactionEntity the entity, which appears in the row to search
-     * @return List of all entities of the provided type, which appear in the same row that the provided entity appears in.
-     */
-    public List<RedactionEntity> getEntitiesOfTypeInSameRow(String type, RedactionEntity redactionEntity) {
-
-        return redactionEntity.getIntersectingNodes()
-                .stream()
-                .filter(node -> node instanceof TableCell)
-                .map(node -> (TableCell) node)
-                .flatMap(tableCellNode -> streamRow(tableCellNode.getRow()))
-                .map(cell -> cell.getEntitiesOfType(type))
-                .flatMap(Collection::stream)
-                .toList();
-    }
-
-
-    @Override
-    public NodeType getType() {
-
-        return NodeType.TABLE;
-    }
-
-
-    @Override
-    public TextBlock getTextBlock() {
-
-        if (textBlock == null) {
-            textBlock = streamAllSubNodes().filter(SemanticNode::isLeaf)
-                    .map(SemanticNode::getLeafTextBlock)
-                    .collect(new TextBlockCollector());
-        }
-        return textBlock;
-    }
-
-
-    @Override
-    public String toString() {
-
-        return treeId.toString() + ": " + NodeType.TABLE + ": #cols: " + numberOfCols + ", #rows: " + numberOfRows + ", " + this.getTextBlock().buildSummary();
-    }
-
-
-    @Override
-    public Map<Page, Rectangle2D> getBBox() {
-
-        if (bBoxCache == null) {
-            bBoxCache = SemanticNode.super.getBBox();
-        }
-        return bBoxCache;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/TableCell.java

@@ -1,102 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
-
-import java.awt.geom.Rectangle2D;
-import java.util.HashMap;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlockCollector;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class TableCell implements GenericSemanticNode {
-
-    @Builder.Default
-    Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
-    List<Integer> treeId;
-    int row;
-    int col;
-    boolean header;
-
-    Rectangle2D bBox;
-
-    TextBlock leafTextBlock;
-
-    TextBlock textBlock;
-
-    @EqualsAndHashCode.Exclude
-    DocumentTree documentTree;
-
-    @Builder.Default
-    @EqualsAndHashCode.Exclude
-    Set<RedactionEntity> entities = new HashSet<>();
-
-
-    @Override
-    public Map<Page, Rectangle2D> getBBox() {
-
-        Map<Page, Rectangle2D> bBoxPerPage = new HashMap<>();
-        getPages().forEach(page -> bBoxPerPage.put(page, bBox));
-        return bBoxPerPage;
-    }
-
-
-    @Override
-    public NodeType getType() {
-
-        return NodeType.TABLE_CELL;
-    }
-
-
-    @Override
-    public boolean isLeaf() {
-
-        return getDocumentTree().getEntryById(getTreeId()).getChildren().isEmpty();
-    }
-
-
-    @Override
-    public TextBlock getTextBlock() {
-
-        if (isLeaf()) {
-            return leafTextBlock;
-        }
-
-        if (textBlock == null) {
-            textBlock = buildTextBlock();
-        }
-        return textBlock;
-    }
-
-
-    private TextBlock buildTextBlock() {
-
-        return streamAllSubNodes().filter(SemanticNode::isLeaf)
-                .map(SemanticNode::getLeafTextBlock)
-                .collect(new TextBlockCollector());
-    }
-
-
-    @Override
-    public String toString() {
-
-        return treeId + ": " + NodeType.TABLE_CELL + ": " + this.getTextBlock().buildSummary();
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/textblock/AtomicTextBlock.java

@@ -1,232 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock;
-
-import static java.lang.String.format;
-
-import java.awt.geom.Rectangle2D;
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.Collection;
-import java.util.Collections;
-import java.util.HashMap;
-import java.util.List;
-import java.util.Map;
-
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentPositionData;
-import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.DocumentTextData;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.Boundary;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
-import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
-
-import lombok.AccessLevel;
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.EqualsAndHashCode;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@Builder
-@AllArgsConstructor
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class AtomicTextBlock implements TextBlock {
-
-    Long id;
-    Integer numberOnPage;
-    Page page;
-
-    //string coordinates
-    Boundary boundary;
-    String searchText;
-    @Builder.Default
-    List<Integer> lineBreaks = new ArrayList<>();
-    @Builder.Default
-    List<Boundary> boldTextBoundaries = new ArrayList<>();
-    @Builder.Default
-    List<Boundary> italicTextBoundaries = new ArrayList<>();
-    String orientation;
-    int textDirection;
-
-    //position coordinates
-    @Builder.Default
-    List<Integer> stringIdxToPositionIdx = new ArrayList<>();
-    @Builder.Default
-    List<Rectangle2D> positions = new ArrayList<>();
-
-    @EqualsAndHashCode.Exclude
-    SemanticNode parent;
-
-
-    @Override
-    public int numberOfLines() {
-
-        return lineBreaks.size() + 1;
-    }
-
-
-    public static AtomicTextBlock fromSearchTextWithTextPosition(String searchText,
-                                                                 List<Integer> lineBreaks,
-                                                                 List<Boundary> boldTextBoundaries,
-                                                                 List<Boundary> italicTextBoundaries,
-                                                                 List<Rectangle2D> positions,
-                                                                 List<Integer> stringIdxToPositionIdx,
-                                                                 long idx,
-                                                                 SemanticNode parent,
-                                                                 int numberOnPage,
-                                                                 Page page,
-                                                                 int offset,
-                                                                 String orientation,
-                                                                 int textDirection) {
-
-        return AtomicTextBlock.builder()
-                .id(idx)
-                .parent(parent)
-                .searchText(searchText)
-                .numberOnPage(numberOnPage)
-                .page(page)
-                .lineBreaks(lineBreaks)
-                .boldTextBoundaries(boldTextBoundaries)
-                .italicTextBoundaries(italicTextBoundaries)
-                .positions(positions)
-                .stringIdxToPositionIdx(stringIdxToPositionIdx)
-                .boundary(new Boundary(offset, offset + searchText.length()))
-                .textDirection(textDirection)
-                .orientation(orientation)
-                .build();
-    }
-
-
-    public static AtomicTextBlock empty(Long textBlockIdx, int stringOffset, Page page, int numberOnPage, SemanticNode parent) {
-
-        return AtomicTextBlock.builder()
-                .id(textBlockIdx)
-                .boundary(new Boundary(stringOffset, stringOffset))
-                .searchText("")
-                .page(page)
-                .numberOnPage(numberOnPage)
-                .parent(parent)
-                .build();
-    }
-
-
-    public static AtomicTextBlock fromAtomicTextBlockData(DocumentTextData documentTextData,
-                                                          DocumentPositionData documentPositionData,
-                                                          SemanticNode parent,
-                                                          Page page) {
-
-        return AtomicTextBlock.builder()
-                .id(documentTextData.getId())
-                .numberOnPage(documentTextData.getNumberOnPage())
-                .page(page)
-                .boundary(new Boundary(documentTextData.getStart(), documentTextData.getEnd()))
-                .searchText(documentTextData.getSearchText())
-                .lineBreaks(Arrays.stream(documentTextData.getLineBreaks()).boxed().toList())
-                .stringIdxToPositionIdx(Arrays.stream(documentPositionData.getStringIdxToPositionIdx()).boxed().toList())
-                .positions(toRectangle2DList(documentPositionData.getPositions()))
-                .parent(parent)
-                .build();
-    }
-
-
-    private static List<Rectangle2D> toRectangle2DList(float[][] positions) {
-
-        return Arrays.stream(positions).map(floatArr -> (Rectangle2D) new Rectangle2D.Float(floatArr[0], floatArr[1], floatArr[2], floatArr[3])).toList();
-    }
-
-
-    public CharSequence getLine(int lineNumber) {
-
-        if (lineNumber >= numberOfLines() || lineNumber < 0) {
-            throw new IndexOutOfBoundsException(format("line %d out of range for AtomicTextBlock with %d lines", lineNumber, numberOfLines()));
-        }
-        if (lineNumber == 0) {
-            return subSequence(boundary.start(), lineBreaks.get(0) + boundary.start());
-        } else if (lineNumber == numberOfLines() - 1) {
-            return subSequence(lineBreaks.get(lineBreaks.size() - 1) + boundary.start(), boundary.end());
-        }
-        return subSequence(lineBreaks.get(lineNumber - 1) + boundary.start(), lineBreaks.get(lineNumber) + boundary.start());
-    }
-
-
-    @Override
-    public List<AtomicTextBlock> getAtomicTextBlocks() {
-
-        return List.of(this);
-    }
-
-
-    @Override
-    public int getNextLinebreak(int fromIndex) {
-
-        return lineBreaks.stream()//
-                .filter(linebreak -> linebreak > fromIndex - boundary.start()) //
-                .findFirst() //
-                .orElse(searchText.length()) + boundary.start();
-    }
-
-
-    @Override
-    public int getPreviousLinebreak(int fromIndex) {
-
-        return lineBreaks.stream()//
-                .filter(linebreak -> linebreak <= fromIndex - boundary.start())//
-                .reduce((a, b) -> b)//
-                .orElse(0) + boundary.start();
-    }
-
-
-    @Override
-    public Rectangle2D getPosition(int stringIdx) {
-
-        return positions.get(stringIdxToPositionIdx.get(stringIdx - boundary.start()));
-    }
-
-
-    @Override
-    public List<Rectangle2D> getPositions(Boundary stringBoundary) {
-
-        if (!containsBoundary(stringBoundary)) {
-            throw new IndexOutOfBoundsException(format("%s is out of bounds for %s", stringBoundary, this.boundary));
-        }
-        if (stringBoundary.length() == 0) {
-            return Collections.emptyList();
-        }
-
-        int startPositionIdx = stringIdxToPositionIdx.get(stringBoundary.start() - this.boundary.start());
-
-        if (stringBoundary.end() == this.boundary.end()) {
-            return positions.subList(startPositionIdx, positions.size());
-        }
-
-        return positions.subList(startPositionIdx, stringIdxToPositionIdx.get(stringBoundary.end() - this.boundary.start()));
-
-    }
-
-
-    public Map<Page, List<Rectangle2D>> getPositionsPerPage(Boundary stringBoundary) {
-
-        List<Rectangle2D> rectanglesPerLine = stringBoundary.split(getAllLineBreaksInBoundary(stringBoundary))
-                .stream()
-                .map(this::getPositions)
-                .map(RectangleTransformations::rectangleBBoxWithGaps)
-                .flatMap(Collection::stream)
-                .toList();
-        Map<Page, List<Rectangle2D>> rectanglePerLinePerPage = new HashMap<>();
-        rectanglePerLinePerPage.put(page, rectanglesPerLine);
-        return rectanglePerLinePerPage;
-    }
-
-
-    private List<Integer> getAllLineBreaksInBoundary(Boundary boundary) {
-
-        return getLineBreaks().stream().map(linebreak -> linebreak + this.boundary.start()).filter(boundary::contains).toList();
-    }
-
-
-    @Override
-    public String toString() {
-
-        return searchText;
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/textblock/ConcatenatedTextBlock.java

@@ -1,220 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock;
-
-import static java.lang.String.format;
-
-import java.awt.geom.Rectangle2D;
-import java.util.Collection;
-import java.util.Collections;
-import java.util.HashMap;
-import java.util.LinkedList;
-import java.util.List;
-import java.util.Map;
-import java.util.stream.Stream;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.Boundary;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
-
-import lombok.AccessLevel;
-import lombok.Data;
-import lombok.experimental.FieldDefaults;
-
-@Data
-@FieldDefaults(level = AccessLevel.PRIVATE)
-public class ConcatenatedTextBlock implements TextBlock {
-
-    List<AtomicTextBlock> atomicTextBlocks;
-    String searchText;
-    Boundary boundary;
-
-
-    public static ConcatenatedTextBlock empty() {
-
-        return new ConcatenatedTextBlock(Collections.emptyList());
-    }
-
-
-    public ConcatenatedTextBlock(List<AtomicTextBlock> atomicTextBlocks) {
-
-        this.atomicTextBlocks = new LinkedList<>();
-        if (atomicTextBlocks.isEmpty()) {
-            boundary = new Boundary(-1, -1);
-            return;
-        }
-        var firstTextBlock = atomicTextBlocks.get(0);
-        this.atomicTextBlocks.add(firstTextBlock);
-        boundary = new Boundary(firstTextBlock.getBoundary().start(), firstTextBlock.getBoundary().end());
-
-        atomicTextBlocks.subList(1, atomicTextBlocks.size()).forEach(this::concat);
-    }
-
-
-    public ConcatenatedTextBlock concat(TextBlock textBlock) {
-
-        if (this.atomicTextBlocks.isEmpty()) {
-            boundary.setStart(textBlock.getBoundary().start());
-            boundary.setEnd(textBlock.getBoundary().end());
-        } else if (boundary.end() != textBlock.getBoundary().start()) {
-            throw new UnsupportedOperationException(format("Can only concat consecutive TextBlocks, trying to concat %s and %s", boundary, textBlock.getBoundary()));
-        }
-        this.atomicTextBlocks.addAll(textBlock.getAtomicTextBlocks());
-        boundary.setEnd(textBlock.getBoundary().end());
-        this.searchText = null;
-        return this;
-    }
-
-
-    private AtomicTextBlock getAtomicTextBlockByStringIndex(int stringIdx) {
-
-        return atomicTextBlocks.stream().filter(textBlock -> textBlock.getBoundary().contains(stringIdx)).findAny().orElseThrow(IndexOutOfBoundsException::new);
-    }
-
-
-    private List<AtomicTextBlock> getAllAtomicTextBlocksPartiallyInStringBoundary(Boundary boundary) {
-
-        return atomicTextBlocks.stream().filter(tb -> tb.getBoundary().intersects(boundary)).toList();
-    }
-
-
-    @Override
-    public String getSearchText() {
-
-        if (searchText == null) {
-            StringBuilder sb = new StringBuilder();
-            getAtomicTextBlocks().forEach(atb -> sb.append(atb.getSearchText()));
-            searchText = sb.toString();
-        }
-        return searchText;
-    }
-
-
-    @Override
-    public int numberOfLines() {
-
-        return atomicTextBlocks.stream().map(AtomicTextBlock::getLineBreaks).mapToInt(List::size).sum();
-    }
-
-
-    @Override
-    public int getNextLinebreak(int fromIndex) {
-
-        return getAtomicTextBlockByStringIndex(fromIndex).getNextLinebreak(fromIndex);
-    }
-
-
-    @Override
-    public int getPreviousLinebreak(int fromIndex) {
-
-        return getAtomicTextBlockByStringIndex(fromIndex).getPreviousLinebreak(fromIndex);
-    }
-
-
-    @Override
-    public List<Integer> getLineBreaks() {
-
-        return getAtomicTextBlocks().stream().flatMap(atomicTextBlock -> atomicTextBlock.getLineBreaks().stream()).toList();
-    }
-
-
-    @Override
-    public Rectangle2D getPosition(int stringIdx) {
-
-        return getAtomicTextBlockByStringIndex(stringIdx).getPosition(stringIdx);
-    }
-
-
-    @Override
-    public List<Rectangle2D> getPositions(Boundary stringBoundary) {
-
-        List<AtomicTextBlock> textBlocks = getAllAtomicTextBlocksPartiallyInStringBoundary(stringBoundary);
-
-        if (textBlocks.size() == 1) {
-            return textBlocks.get(0).getPositions(stringBoundary);
-        }
-
-        AtomicTextBlock firstTextBlock = textBlocks.get(0);
-        List<Rectangle2D> positions = new LinkedList<>(firstTextBlock.getPositions(new Boundary(stringBoundary.start(), firstTextBlock.getBoundary().end())));
-
-        for (AtomicTextBlock textBlock : textBlocks.subList(1, textBlocks.size() - 1)) {
-            positions.addAll(textBlock.getPositions());
-        }
-
-        var lastTextBlock = textBlocks.get(textBlocks.size() - 1);
-        positions.addAll(lastTextBlock.getPositions(new Boundary(lastTextBlock.getBoundary().start(), stringBoundary.end())));
-
-        return positions;
-    }
-
-
-    @Override
-    public Map<Page, List<Rectangle2D>> getPositionsPerPage(Boundary stringBoundary) {
-
-        List<AtomicTextBlock> textBlocks = getAllAtomicTextBlocksPartiallyInStringBoundary(stringBoundary);
-
-        if (textBlocks.size() == 1) {
-            return textBlocks.get(0).getPositionsPerPage(stringBoundary);
-        }
-
-        AtomicTextBlock firstTextBlock = textBlocks.get(0);
-        Map<Page, List<Rectangle2D>> rectanglesPerLinePerPage = firstTextBlock.getPositionsPerPage(new Boundary(stringBoundary.start(), firstTextBlock.getBoundary().end()));
-
-        for (AtomicTextBlock textBlock : textBlocks.subList(1, textBlocks.size() - 1)) {
-            rectanglesPerLinePerPage = mergeEntityPositionsWithSamePageNode(rectanglesPerLinePerPage, textBlock.getPositionsPerPage(textBlock.getBoundary()));
-        }
-
-        AtomicTextBlock lastTextBlock = textBlocks.get(textBlocks.size() - 1);
-        rectanglesPerLinePerPage = mergeEntityPositionsWithSamePageNode(rectanglesPerLinePerPage,
-                lastTextBlock.getPositionsPerPage(new Boundary(lastTextBlock.getBoundary().start(), stringBoundary.end())));
-
-        return rectanglesPerLinePerPage;
-    }
-
-
-    private Map<Page, List<Rectangle2D>> mergeEntityPositionsWithSamePageNode(Map<Page, List<Rectangle2D>> map1, Map<Page, List<Rectangle2D>> map2) {
-
-        Map<Page, List<Rectangle2D>> mergedMap = new HashMap<>(map1);
-        map2.forEach((pageNode, rectangles) -> mergedMap.merge(pageNode, rectangles, (l1, l2) -> Stream.concat(l1.stream(), l2.stream()).toList()));
-        return mergedMap;
-    }
-
-
-    @Override
-    public String toString() {
-
-        return getSearchText();
-    }
-
-
-    @Override
-    public List<Boundary> getBoldTextBoundaries() {
-
-        return getAtomicTextBlocks().stream().map(AtomicTextBlock::getBoldTextBoundaries).flatMap(Collection::stream).toList();
-    }
-
-
-    @Override
-    public List<Boundary> getItalicTextBoundaries() {
-
-        return getAtomicTextBlocks().stream().map(AtomicTextBlock::getItalicTextBoundaries).flatMap(Collection::stream).toList();
-    }
-
-
-    @Override
-    public String getOrientation() {
-
-        if (atomicTextBlocks.isEmpty()) {
-            return "";
-        }
-        return atomicTextBlocks.get(0).getOrientation();
-    }
-
-
-    @Override
-    public int getTextDirection() {
-
-        if (atomicTextBlocks.isEmpty()) {
-            return 0;
-        }
-        return atomicTextBlocks.get(0).getTextDirection();
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/textblock/TextBlock.java

@@ -1,148 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock;
-
-import static java.lang.String.format;
-
-import java.awt.geom.Rectangle2D;
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-import java.util.stream.Collectors;
-
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.Boundary;
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
-
-public interface TextBlock extends CharSequence {
-
-    String getSearchText();
-
-
-    List<AtomicTextBlock> getAtomicTextBlocks();
-
-
-    List<Boundary> getBoldTextBoundaries();
-
-
-    List<Boundary> getItalicTextBoundaries();
-
-
-    String getOrientation();
-
-
-    int getTextDirection();
-
-
-    Boundary getBoundary();
-
-
-    int getNextLinebreak(int fromIndex);
-
-
-    int getPreviousLinebreak(int fromIndex);
-
-
-    List<Integer> getLineBreaks();
-
-
-    Rectangle2D getPosition(int stringIdx);
-
-
-    List<Rectangle2D> getPositions(Boundary stringBoundary);
-
-
-    Map<Page, List<Rectangle2D>> getPositionsPerPage(Boundary stringBoundary);
-
-
-    int numberOfLines();
-
-
-    default int indexOf(String searchTerm) {
-
-        return indexOf(searchTerm, getBoundary().start());
-    }
-
-
-    default Set<Page> getPages() {
-
-        return getAtomicTextBlocks().stream().map(AtomicTextBlock::getPage).collect(Collectors.toUnmodifiableSet());
-    }
-
-
-    default Set<Page> getPages(Boundary boundary) {
-
-        return getAtomicTextBlocks().stream()
-                .filter(atomicTextBlock -> atomicTextBlock.getBoundary().intersects(boundary))
-                .map(AtomicTextBlock::getPage)
-                .collect(Collectors.toUnmodifiableSet());
-    }
-
-
-    default int indexOf(String searchTerm, int startOffset) {
-
-        int start = getSearchText().indexOf(searchTerm, startOffset - getBoundary().start());
-        if (start == -1) {
-            return -1;
-        }
-        return start + getBoundary().start();
-    }
-
-
-    default CharSequence getFirstLine() {
-
-        return subSequence(getBoundary().start(), getNextLinebreak(getBoundary().start()));
-    }
-
-
-    default boolean containsBoundary(Boundary boundary) {
-
-        if (boundary.end() < boundary.start()) {
-            throw new IllegalArgumentException(format("Invalid %s, StartIndex must be smaller than EndIndex", boundary));
-        }
-        return getBoundary().contains(boundary);
-    }
-
-
-    default boolean containsIndex(int stringIndex) {
-
-        return getBoundary().contains(stringIndex);
-    }
-
-
-    default CharSequence subSequence(Boundary boundary) {
-
-        return subSequence(boundary.start(), boundary.end());
-    }
-
-
-    default String buildSummary() {
-
-        String[] words = getSearchText().split(" ");
-        int bound = Math.min(words.length, 4);
-        List<String> list = new ArrayList<>(Arrays.asList(words).subList(0, bound));
-
-        return String.join(" ", list);
-    }
-
-
-    @Override
-    default CharSequence subSequence(int start, int end) {
-
-        return getSearchText().substring(start - getBoundary().start(), end - getBoundary().start());
-    }
-
-
-    @Override
-    default int length() {
-
-        return getBoundary().length();
-    }
-
-
-    @Override
-    default char charAt(int index) {
-
-        return getSearchText().charAt(index - getBoundary().start());
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/textblock/TextBlockCollector.java

@@ -1,49 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock;
-
-import java.util.Set;
-import java.util.function.BiConsumer;
-import java.util.function.BinaryOperator;
-import java.util.function.Function;
-import java.util.function.Supplier;
-import java.util.stream.Collector;
-
-import lombok.NoArgsConstructor;
-
-@NoArgsConstructor
-public class TextBlockCollector implements Collector<TextBlock, ConcatenatedTextBlock, TextBlock> {
-
-    @Override
-    public Supplier<ConcatenatedTextBlock> supplier() {
-
-        return ConcatenatedTextBlock::empty;
-    }
-
-
-    @Override
-    public BiConsumer<ConcatenatedTextBlock, TextBlock> accumulator() {
-
-        return ConcatenatedTextBlock::concat;
-    }
-
-
-    @Override
-    public BinaryOperator<ConcatenatedTextBlock> combiner() {
-
-        return ConcatenatedTextBlock::concat;
-    }
-
-
-    @Override
-    public Function<ConcatenatedTextBlock, TextBlock> finisher() {
-
-        return a -> a;
-    }
-
-
-    @Override
-    public Set<Characteristics> characteristics() {
-
-        return Set.of(Characteristics.IDENTITY_FINISH, Characteristics.CONCURRENT);
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/image/ClassifiedImage.java

@@ -2,7 +2,7 @@ package com.knecon.fforesight.service.layoutparser.processor.model.image;
 
 import java.awt.geom.Rectangle2D;
 
-import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.ImageType;
+import com.iqser.red.service.redaction.v1.server.model.document.nodes.ImageType;
 
 import lombok.AllArgsConstructor;
 import lombok.Data;
@@ -10,8 +10,8 @@ import lombok.NonNull;
 import lombok.RequiredArgsConstructor;
 
 @Data
-@RequiredArgsConstructor
 @AllArgsConstructor
+@RequiredArgsConstructor
 public class ClassifiedImage {
 
     @NonNull
@@ -20,9 +20,18 @@ public class ClassifiedImage {
     private ImageType imageType;
     private boolean sourceByAi;
     private boolean isAppendedToSection;
-    @NonNull
     private boolean hasTransparency;
-    @NonNull
     private int page;
+    private String representation;
+
+
+    public ClassifiedImage(@NonNull Rectangle2D position, @NonNull ImageType imageType, boolean hasTransparency, int page, String representation) {
+
+        this.position = position;
+        this.imageType = imageType;
+        this.hasTransparency = hasTransparency;
+        this.page = page;
+        this.representation = representation;
+    }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/OutlineExtractorService.java

@@ -0,0 +1,229 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.outline;
+
+import java.awt.geom.AffineTransform;
+import java.awt.geom.Point2D;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Locale;
+import java.util.Optional;
+
+import org.apache.pdfbox.cos.COSArray;
+import org.apache.pdfbox.cos.COSBase;
+import org.apache.pdfbox.cos.COSDictionary;
+import org.apache.pdfbox.cos.COSName;
+import org.apache.pdfbox.cos.COSString;
+import org.apache.pdfbox.pdmodel.PDDestinationNameTreeNode;
+import org.apache.pdfbox.pdmodel.PDDocument;
+import org.apache.pdfbox.pdmodel.PDDocumentNameDictionary;
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.pdmodel.interactive.action.PDAction;
+import org.apache.pdfbox.pdmodel.interactive.documentnavigation.destination.PDDestination;
+import org.apache.pdfbox.pdmodel.interactive.documentnavigation.destination.PDPageFitHeightDestination;
+import org.apache.pdfbox.pdmodel.interactive.documentnavigation.destination.PDPageFitRectangleDestination;
+import org.apache.pdfbox.pdmodel.interactive.documentnavigation.destination.PDPageFitWidthDestination;
+import org.apache.pdfbox.pdmodel.interactive.documentnavigation.destination.PDPageXYZDestination;
+import org.apache.pdfbox.pdmodel.interactive.documentnavigation.outline.PDDocumentOutline;
+import org.apache.pdfbox.pdmodel.interactive.documentnavigation.outline.PDOutlineItem;
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.utils.CoordinateTransforms;
+import com.knecon.fforesight.service.layoutparser.processor.utils.PageInformation;
+
+import lombok.SneakyThrows;
+import lombok.extern.slf4j.Slf4j;
+
+@Service
+@Slf4j
+public class OutlineExtractorService {
+
+    private static final String PDDESTINATION_TYPE_FIT = "Fit";
+    private static final String PDDESTINATION_TYPE_FIT_B = "FitB";
+    private static final String PDDESTINATION_TYPE_FIT_H = "FitH";
+    private static final String PDDESTINATION_TYPE_FIT_V = "FitV";
+    private static final String PDDESTINATION_TYPE_FIT_R = "FitR";
+    private static final String PDDESTINATION_TYPE_FIT_BH = "FitBH";
+    private static final String PDDESTINATION_TYPE_FIT_BV = "FitBV";
+    private static final String PDDESTINATION_TYPE_XYZ = "XYZ";
+
+
+    @SneakyThrows
+    public OutlineObjectTree getOutlineObjectTree(PDDocument document) {
+
+        PDDocumentOutline documentOutline = document.getDocumentCatalog().getDocumentOutline();
+
+        List<OutlineObjectTreeNode> rootNodes = new ArrayList<>();
+        if (documentOutline != null) {
+            for (PDOutlineItem child : documentOutline.children()) {
+                Optional<OutlineObjectTreeNode> outlineObjectWithChildren = createOutlineObjectWithChildren(child, document, 1);
+                outlineObjectWithChildren.ifPresent(rootNodes::add);
+            }
+        }
+
+        return new OutlineObjectTree(rootNodes);
+    }
+
+
+    @SneakyThrows
+    private Optional<OutlineObjectTreeNode> createOutlineObjectWithChildren(PDOutlineItem item, PDDocument document, int depth) {
+
+        Optional<OutlineObjectTreeNode> outlineObject = createOutlineObject(item, document, depth);
+        if (outlineObject.isPresent()) {
+            for (var child : item.children()) {
+                Optional<OutlineObjectTreeNode> outlineObjectWithChildren = createOutlineObjectWithChildren(child, document, depth + 1);
+                outlineObjectWithChildren.ifPresent(outlineObjectTreeNode -> outlineObject.get().addChild(outlineObjectTreeNode));
+            }
+        }
+
+        return outlineObject;
+    }
+
+
+    // if the structure elements are processed beforehand, another case can be handled here as well:
+    // outline objects can reference structure elements (see pdf documentation)
+    @SneakyThrows
+    private Optional<OutlineObjectTreeNode> createOutlineObject(PDOutlineItem item, PDDocument document, int depth) {
+
+        String title = item.getTitle();
+
+        PDPage page;
+        try {
+            // Can throw: "Error: can't convert to Destination COSArray" for some OCR'd PDFs
+            page = item.findDestinationPage(document);
+            if (page == null) {
+                return Optional.empty();
+            }
+        } catch (IOException e) {
+            log.info(String.format("Error occurred during position resolution for outline item with title %s: " + e, title));
+            return Optional.empty();
+        }
+
+        int pageNumber = document.getPages().indexOf(page) + 1;
+        AffineTransform userSpaceToPageCoords = CoordinateTransforms.calculateInitialUserSpaceCoordsToPageCoords(PageInformation.fromPDPage(pageNumber, page));
+
+        Optional<Point2D> outlinePosition = Optional.empty();
+
+        try {
+            PDDocumentNameDictionary names = document.getDocumentCatalog().getNames();
+            PDDestinationNameTreeNode destinations = null;
+            if (names != null) {
+                destinations = names.getDests();
+            }
+
+            PDDestination destination = item.getDestination();
+            if (destination != null) {
+                outlinePosition = getLocationFromCOSBase(destinations, destination.getCOSObject());
+            }
+
+            if (outlinePosition.isEmpty()) {
+
+                PDAction action = item.getAction();
+                if (action != null) {
+                    outlinePosition = extractOutlineLocationGoTo(destinations, action.getCOSObject());
+                }
+
+            }
+
+        } catch (Exception e) {
+            log.info(String.format("Error occurred during position resolution for outline item on page %s with title %s: " + e, pageNumber, title));
+        }
+
+        return Optional.of(new OutlineObjectTreeNode(new OutlineObject(title,
+                                                                       pageNumber,
+                                                                       transformPointToPageCoords(outlinePosition, userSpaceToPageCoords), depth)));
+    }
+
+
+    private static Point2D transformPointToPageCoords(Optional<Point2D> outlinePosition, AffineTransform userSpaceToPageCoords) {
+
+        return outlinePosition.map(point -> userSpaceToPageCoords.transform(point, null)).orElse(null);
+    }
+
+
+    @SneakyThrows
+    private static Optional<Point2D> extractOutlineLocationGoTo(PDDestinationNameTreeNode destinations, COSDictionary cosDictionary) {
+
+        if (isGoToAction(cosDictionary)) {
+            COSBase cosBase = cosDictionary.getItem(COSName.D);
+            return getLocationFromCOSBase(destinations, cosBase);
+        }
+
+        return Optional.empty();
+    }
+
+
+    private static Optional<Point2D> getLocationFromCOSBase(PDDestinationNameTreeNode destinations, COSBase cosBase) throws IOException {
+
+        if (cosBase != null) {
+            if (cosBase instanceof COSArray cosArray) {
+                return getLocationFromCosArray(cosArray);
+            }
+
+            if (cosBase instanceof COSString cosString) {
+                String destinationName = cosString.getString();
+                COSArray cosArray = destinations.getValue(destinationName).getCOSObject();
+                return getLocationFromCosArray(cosArray);
+            }
+
+        }
+        return Optional.empty();
+    }
+
+
+    private static Optional<Point2D> getLocationFromCosArray(COSArray cosArray) {
+
+        boolean located = false;
+        float x = 0;
+        float y = 0;
+
+        try {
+
+            PDDestination destination = PDDestination.create(cosArray);
+            COSName type = (COSName) cosArray.getObject(1);
+            String typeString = type.getName();
+
+            switch (typeString) {
+                case PDDESTINATION_TYPE_FIT_V:
+                case PDDESTINATION_TYPE_FIT_BV:
+                    PDPageFitHeightDestination fitHeightDestination = (PDPageFitHeightDestination) destination;
+                    x = fitHeightDestination.getLeft();
+                    located = true;
+                    break;
+                case PDDESTINATION_TYPE_FIT_R:
+                    PDPageFitRectangleDestination fitRectangleDestination = (PDPageFitRectangleDestination) destination;
+                    x = fitRectangleDestination.getLeft();
+                    y = fitRectangleDestination.getTop();
+                    located = true;
+                    break;
+                case PDDESTINATION_TYPE_FIT_H:
+                case PDDESTINATION_TYPE_FIT_BH:
+                    PDPageFitWidthDestination fitWidthDestination = (PDPageFitWidthDestination) destination;
+                    y = fitWidthDestination.getTop();
+                    located = true;
+                    break;
+                case PDDESTINATION_TYPE_XYZ:
+                    PDPageXYZDestination xyzDestination = (PDPageXYZDestination) destination;
+                    x = xyzDestination.getLeft();
+                    y = xyzDestination.getTop();
+                    located = true;
+                    break;
+                case PDDESTINATION_TYPE_FIT:
+                case PDDESTINATION_TYPE_FIT_B:
+                default:
+            }
+
+        } catch (IOException e) {
+            throw new RuntimeException(e);
+        }
+
+        return located ? Optional.of(new Point2D.Float(x, y)) : Optional.empty();
+
+    }
+
+
+    private static boolean isGoToAction(COSDictionary cosDictionary) {
+
+        return cosDictionary.getNameAsString("S").toLowerCase(Locale.ROOT).equals("goto");
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/OutlineObject.java

@@ -0,0 +1,77 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.outline;
+
+import java.awt.geom.Point2D;
+import java.util.Optional;
+
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.BoundingBox;
+
+import lombok.Getter;
+import lombok.Setter;
+
+public class OutlineObject {
+
+    @Getter
+    private final String title;
+    @Getter
+    private final int pageNumber;
+    @Getter
+    private final int treeDepth;
+
+    private Point2D point; // java coordinates, (0, 0) is always top left
+
+    @Getter
+    @Setter
+    private boolean found;
+
+
+    public OutlineObject(String title, int pageNumber, Point2D point2D, int depth) {
+
+        this.title = title;
+        this.pageNumber = pageNumber;
+        this.treeDepth = depth;
+        this.point = point2D;
+    }
+
+
+    @Override
+    public String toString() {
+
+        return "OutlineObject{" + "title='" + title + '\'' + '}';
+    }
+
+
+    public Optional<Point2D> getPoint() {
+
+        return Optional.ofNullable(point);
+    }
+
+
+    public boolean isAbove(BoundingBox boundingBox) {
+
+        if (point == null) {
+            return true;
+        }
+        return point.getY() <= boundingBox.getMaxY();
+    }
+
+
+    public double distance(BoundingBox boundingBox) {
+
+        if (point == null) {
+            return 0;
+        }
+        if (boundingBox.getBBox().contains(point)) {
+            return 0;
+        }
+        double deltaX = Math.min(Math.abs(boundingBox.getMinX() - point.getX()), Math.abs(boundingBox.getMaxX() - point.getX()));
+        double deltaY = Math.min(Math.abs(boundingBox.getMinY() - point.getY()), Math.abs(boundingBox.getMaxY() - point.getY()));
+        return Math.sqrt(deltaX * deltaX + deltaY * deltaY);
+    }
+
+
+    public void resetPoint() {
+
+        this.point = null;
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/OutlineObjectTree.java

@@ -0,0 +1,66 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.outline;
+
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import lombok.Data;
+import lombok.RequiredArgsConstructor;
+
+@Data
+@RequiredArgsConstructor
+public class OutlineObjectTree {
+
+    private List<OutlineObjectTreeNode> rootNodes = new ArrayList<>();
+
+    private Map<Integer, List<OutlineObject>> outlineObjectsPerPage = new HashMap<>();
+
+
+    public OutlineObjectTree(List<OutlineObjectTreeNode> rootNodes) {
+
+        this.rootNodes = rootNodes;
+        flattenNodesAndGroupByPage(rootNodes);
+    }
+
+
+    private void flattenNodesAndGroupByPage(List<OutlineObjectTreeNode> outlineObjectTreeNodes) {
+
+        for (OutlineObjectTreeNode node : outlineObjectTreeNodes) {
+            int pageNumber = node.getOutlineObject().getPageNumber();
+            if (!this.outlineObjectsPerPage.containsKey(pageNumber)) {
+                outlineObjectsPerPage.put(pageNumber, new ArrayList<>());
+            }
+            outlineObjectsPerPage.get(pageNumber).add(node.getOutlineObject());
+
+            if (!node.getChildren().isEmpty()) {
+                flattenNodesAndGroupByPage(node.getChildren());
+            }
+        }
+    }
+
+    @Override
+    public String toString() {
+
+        StringBuilder sb = new StringBuilder();
+        sb.append("OutlineObjectTree(\n");
+        for (OutlineObjectTreeNode node : rootNodes) {
+            buildString(node, sb, 1);
+        }
+        sb.append(")");
+        return sb.toString();
+    }
+
+    private void buildString(OutlineObjectTreeNode node, StringBuilder sb, int depth) {
+
+        for (int i = 0; i < depth; i++) {
+            sb.append("  ");
+        }
+        sb.append(node.getOutlineObject().getTitle()).append("\n");
+
+        for (OutlineObjectTreeNode child : node.getChildren()) {
+            buildString(child, sb, depth + 1);
+        }
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/OutlineObjectTreeNode.java

@@ -0,0 +1,34 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.outline;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import lombok.Data;
+
+@Data
+public class OutlineObjectTreeNode {
+
+    private OutlineObject outlineObject;
+
+    private List<OutlineObjectTreeNode> children = new ArrayList<>();
+
+
+    public OutlineObjectTreeNode(OutlineObject outlineObject) {
+
+        this.outlineObject = outlineObject;
+    }
+
+
+    public void addChild(OutlineObjectTreeNode outlineObject) {
+
+        children.add(outlineObject);
+    }
+
+
+    @Override
+    public String toString() {
+
+        return "OutlineObjectTreeNode{" + "outlineObject=" + outlineObject + '}';
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/SectionTree.java

@@ -0,0 +1,136 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.outline;
+
+import java.util.ArrayList;
+import java.util.Iterator;
+import java.util.List;
+import java.util.Stack;
+
+import org.springframework.lang.NonNull;
+
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
+
+import lombok.Data;
+import lombok.RequiredArgsConstructor;
+
+@Data
+@RequiredArgsConstructor
+public class SectionTree implements Iterable<SectionTreeEntry> {
+
+    private List<SectionTreeEntry> mainSections = new ArrayList<>();
+
+
+    public SectionTree(List<SectionTreeEntry> mainSections) {
+
+        this.mainSections = mainSections;
+    }
+
+
+    public List<TextPageBlock> getAllTextPageBlocks() {
+
+        List<TextPageBlock> allTextPageBlocks = new ArrayList<>();
+        for (SectionTreeEntry item : mainSections) {
+            collectTextPageBlocks(item, allTextPageBlocks);
+        }
+        return allTextPageBlocks;
+    }
+
+
+    private void collectTextPageBlocks(SectionTreeEntry item, List<TextPageBlock> textPageBlocks) {
+
+        textPageBlocks.add(item.getHeadline());
+        for (SectionTreeEntry child : item.getChildren()) {
+            collectTextPageBlocks(child, textPageBlocks);
+        }
+    }
+
+
+    public List<SectionTreeEntry> getAllTableOfContentItems() {
+
+        List<SectionTreeEntry> allItems = new ArrayList<>();
+        for (SectionTreeEntry item : mainSections) {
+            collectTableOfContentItems(item, allItems);
+        }
+        return allItems;
+    }
+
+
+    private void collectTableOfContentItems(SectionTreeEntry item, List<SectionTreeEntry> allItems) {
+
+        allItems.add(item);
+        for (SectionTreeEntry child : item.getChildren()) {
+            collectTableOfContentItems(child, allItems);
+        }
+    }
+
+
+    private boolean containsBlock(TextPageBlock block) {
+
+        for (SectionTreeEntry existingItem : this.getMainSections()) {
+            if (existingItem.getHeadline().equals(block) || existingItem.contains(block)) {
+                return true;
+            }
+        }
+        return false;
+    }
+
+
+    private boolean containsItem(SectionTreeEntry tocItem) {
+
+        for (SectionTreeEntry existingItem : this.getMainSections()) {
+            if (existingItem.equals(tocItem) || existingItem.contains(tocItem)) {
+                return true;
+            }
+        }
+        return false;
+    }
+
+
+    @Override
+    public @NonNull Iterator<SectionTreeEntry> iterator() {
+
+        return new SectionTreeEntryIterator(mainSections);
+    }
+
+
+    private static class SectionTreeEntryIterator implements Iterator<SectionTreeEntry> {
+
+        private final Stack<Iterator<SectionTreeEntry>> stack = new Stack<>();
+
+
+        SectionTreeEntryIterator(List<SectionTreeEntry> mainSections) {
+
+            stack.push(mainSections.iterator());
+        }
+
+
+        @Override
+        public boolean hasNext() {
+
+            ensureStackTopIsCurrent();
+            return !stack.isEmpty() && stack.peek().hasNext();
+        }
+
+
+        @Override
+        public SectionTreeEntry next() {
+
+            ensureStackTopIsCurrent();
+            SectionTreeEntry currentItem = stack.peek().next();
+            if (currentItem.getChildren() != null && !currentItem.getChildren().isEmpty()) {
+                stack.push(currentItem.getChildren()
+                                   .iterator());
+            }
+            return currentItem;
+        }
+
+
+        private void ensureStackTopIsCurrent() {
+
+            while (!stack.isEmpty() && !stack.peek().hasNext()) {
+                stack.pop();
+            }
+        }
+
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/SectionTreeBuilderService.java

@@ -0,0 +1,82 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.outline;
+
+import static com.knecon.fforesight.service.layoutparser.processor.model.PageBlockType.TABLE_OF_CONTENTS_HEADLINE;
+import static com.knecon.fforesight.service.layoutparser.processor.model.PageBlockType.getHeadlineNumber;
+
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.TreeSet;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationDocument;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
+
+import io.micrometer.observation.annotation.Observed;
+import lombok.extern.slf4j.Slf4j;
+
+@Service
+@Slf4j
+public class SectionTreeBuilderService {
+
+    @Observed(name = "OutlineValidationService", contextualName = "create-toc")
+    public SectionTree createSectionTree(ClassificationDocument classificationDocument) {
+
+        List<TextPageBlock> headlines = extractHeadlines(classificationDocument);
+
+        List<SectionTreeEntry> mainSections = new ArrayList<>();
+        Map<Integer, SectionTreeEntry> lastItemsPerDepth = new HashMap<>();
+        SectionTreeEntry last = null;
+        TreeSet<Integer> depths = new TreeSet<>();
+
+        for (TextPageBlock current : headlines) {
+            int currentDepth = getHeadlineNumber(current.getClassification());
+            Integer parentDepth = depths.floor(currentDepth - 1);
+
+            var tocItem = new SectionTreeEntry(current);
+
+            if (parentDepth == null) {
+                mainSections.add(tocItem);
+                lastItemsPerDepth = new HashMap<>();
+                depths = new TreeSet<>();
+
+            } else {
+                assert last != null;
+                int lastDepth = getHeadlineNumber(last.getHeadline().getClassification());
+                if (last.getHeadline().getClassification().equals(TABLE_OF_CONTENTS_HEADLINE) && !current.getClassification().equals(TABLE_OF_CONTENTS_HEADLINE)) {
+                    // headline after toc should always start a main section
+                    parentDepth = 1;
+                } else if (lastDepth < parentDepth) {
+                    parentDepth = lastDepth;
+                } else if (lastDepth == currentDepth && last.getParent() != null) {
+                    parentDepth = getHeadlineNumber(last.getParent().getHeadline().getClassification());
+                }
+
+                SectionTreeEntry parent = lastItemsPerDepth.get(parentDepth);
+                parent.addChild(tocItem);
+            }
+
+            last = tocItem;
+            lastItemsPerDepth.put(currentDepth, tocItem);
+            depths.add(currentDepth);
+        }
+
+        return new SectionTree(mainSections);
+
+    }
+
+
+    private static List<TextPageBlock> extractHeadlines(ClassificationDocument classificationDocument) {
+
+        return classificationDocument.getPages()
+                .stream()
+                .flatMap(classificationPage -> classificationPage.getTextBlocks()
+                        .stream()
+                        .filter(tb -> tb instanceof TextPageBlock && tb.getClassification() != null && tb.getClassification().isHeadline())
+                        .map(tb -> (TextPageBlock) tb))
+                .toList();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/SectionTreeEnhancementService.java

@@ -0,0 +1,252 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.outline;
+
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.Collections;
+import java.util.Iterator;
+import java.util.List;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationDocument;
+import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationFooter;
+import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationHeader;
+import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.model.PageBlockType;
+import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.Cell;
+import com.knecon.fforesight.service.layoutparser.processor.model.table.TablePageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
+
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+public class SectionTreeEnhancementService {
+
+    public void assignSectionBlocksAndImages(ClassificationDocument document) {
+
+        SectionTree toc = document.getSectionTree();
+        Iterator<SectionTreeEntry> iterator = toc.iterator();
+        SectionTreeEntry currentTOCItem = null;
+        if (iterator.hasNext()) {
+            currentTOCItem = iterator.next();
+        }
+        List<AbstractPageBlock> startBlocks = new ArrayList<>();
+        List<ClassifiedImage> startImages = new ArrayList<>();
+        SectionTreeEntry currentSection = null;
+        boolean foundFirstHeadline = false;
+
+        List<ClassificationHeader> headers = new ArrayList<>();
+        List<ClassificationFooter> footers = new ArrayList<>();
+        TablePageBlock previousTable = null;
+        List<SectionTreeEntry> lastFoundTOCItems = new ArrayList<>();
+
+        for (ClassificationPage page : document.getPages()) {
+            List<SectionTreeEntry> currentPageTOCItems = new ArrayList<>();
+            List<TextPageBlock> header = new ArrayList<>();
+            List<TextPageBlock> footer = new ArrayList<>();
+            for (AbstractPageBlock current : page.getTextBlocks()) {
+
+                if (current.getClassification() == null) {
+                    continue;
+                }
+
+                current.setPage(page.getPageNumber());
+
+                if (current.getClassification().equals(PageBlockType.HEADER)) {
+                    header.add((TextPageBlock) current);
+                    continue;
+                }
+
+                if (current.getClassification().equals(PageBlockType.FOOTER)) {
+                    footer.add((TextPageBlock) current);
+                    continue;
+                }
+
+                if (current instanceof TablePageBlock table) {
+                    if (previousTable != null) {
+                        mergeTableMetadata(table, previousTable);
+                    }
+                    previousTable = table;
+                }
+
+                if (current instanceof TextPageBlock && currentTOCItem != null && currentTOCItem.getHeadline().getText().equals(current.getText())) {
+                    if (!foundFirstHeadline) {
+                        foundFirstHeadline = true;
+                    }
+                    currentSection = currentTOCItem;
+                    currentTOCItem.getSectionBlocks().add(current);
+                    currentPageTOCItems.add(currentTOCItem);
+
+                    if (iterator.hasNext()) {
+                        currentTOCItem = iterator.next();
+                    }
+                } else if (!foundFirstHeadline) {
+                    startBlocks.add(current);
+                } else {
+                    currentSection.getSectionBlocks().add(current);
+                }
+            }
+
+            if (!currentPageTOCItems.isEmpty()) {
+                lastFoundTOCItems = currentPageTOCItems;
+            }
+
+            for (ClassifiedImage image : page.getImages()) {
+
+                Double xMin = null;
+                Double yMin = null;
+                Double xMax = null;
+                Double yMax = null;
+
+                for (SectionTreeEntry tocItem : lastFoundTOCItems) {
+                    var headline = tocItem.getHeadline();
+
+                    if (headline.getPage() != page.getPageNumber()) {
+                        continue;
+                    }
+
+                    if (headline.getMinX() < headline.getMaxX()) {
+                        if (xMin == null || headline.getMinX() < xMin) {
+                            xMin = headline.getMinX();
+                        }
+                        if (xMax == null || headline.getMaxX() > xMax) {
+                            xMax = headline.getMaxX();
+                        }
+                    } else {
+                        if (xMin == null || headline.getMaxX() < xMin) {
+                            xMin = headline.getMaxX();
+                        }
+                        if (xMax == null || headline.getMinX() > xMax) {
+                            xMax = headline.getMinX();
+                        }
+                    }
+
+                    if (headline.getMinY() < headline.getMaxY()) {
+                        if (yMin == null || headline.getMinY() < yMin) {
+                            yMin = headline.getMinY();
+                        }
+                        if (yMax == null || headline.getMaxY() > yMax) {
+                            yMax = headline.getMaxY();
+                        }
+                    } else {
+                        if (yMin == null || headline.getMaxY() < yMin) {
+                            yMin = headline.getMaxY();
+                        }
+                        if (yMax == null || headline.getMinY() > yMax) {
+                            yMax = headline.getMinY();
+                        }
+                    }
+
+                    log.debug("Image position x: {}, y: {}", image.getPosition().getX(), image.getPosition().getY());
+                    log.debug("Headline position xMin: {}, xMax: {}, yMin: {}, yMax: {}", xMin, xMax, yMin, yMax);
+
+                    if (image.getPosition().getX() >= xMin && image.getPosition().getX() <= xMax && image.getPosition().getY() >= yMin && image.getPosition().getY() <= yMax) {
+                        tocItem.getImages().add(image);
+                        image.setAppendedToSection(true);
+                        break;
+                    }
+                }
+                if (!image.isAppendedToSection()) {
+                    log.debug("Image uses last found section");
+                    if (!lastFoundTOCItems.isEmpty()) {
+                        lastFoundTOCItems.get(lastFoundTOCItems.size() - 1).getImages().add(image);
+                    } else {
+                        startImages.add(image);
+                    }
+                    image.setAppendedToSection(true);
+                }
+            }
+
+            if (!header.isEmpty()) {
+                headers.add(new ClassificationHeader(header));
+            }
+            if (!footer.isEmpty()) {
+                footers.add(new ClassificationFooter(footer));
+            }
+        }
+
+        if (!startBlocks.isEmpty() || !startImages.isEmpty()) {
+            SectionTreeEntry unassigned = new SectionTreeEntry(null);
+            unassigned.setSectionBlocks(startBlocks);
+            unassigned.setImages(startImages);
+            document.getSectionTree().getMainSections().add(0, unassigned);
+        }
+        document.setHeaders(headers);
+        document.setFooters(footers);
+    }
+
+
+    private void mergeTableMetadata(TablePageBlock currentTable, TablePageBlock previousTable) {
+
+        // Distribute header information for subsequent tables
+        if (previousTable != null && hasInvalidHeaderInformation(currentTable) && hasValidHeaderInformation(previousTable)) {
+            List<Cell> previousTableNonHeaderRow = getRowWithNonHeaderCells(previousTable);
+            List<Cell> tableNonHeaderRow = getRowWithNonHeaderCells(currentTable);
+            // Allow merging of tables if header row is separated from first logical non-header row
+            if (previousTableNonHeaderRow.isEmpty() && previousTable.getRowCount() == 1 && previousTable.getRows().get(0).size() == tableNonHeaderRow.size()) {
+                previousTableNonHeaderRow = previousTable.getRows().get(0)
+                        .stream()
+                        .map(cell -> {
+                            Cell fakeCell = Cell.copy(cell);
+                            fakeCell.setHeaderCells(Collections.singletonList(cell));
+                            return fakeCell;
+                        })
+                        .toList();
+            }
+            if (previousTableNonHeaderRow.size() == tableNonHeaderRow.size()) {
+                for (int i = currentTable.getRowCount() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of table
+                    List<Cell> row = currentTable.getRows().get(i);
+                    if (row.size() == tableNonHeaderRow.size() && row.stream()
+                            .allMatch(cell -> cell.getHeaderCells().isEmpty())) {
+                        for (int j = 0; j < row.size(); j++) {
+                            row.get(j).setHeaderCells(previousTableNonHeaderRow.get(j).getHeaderCells());
+                        }
+                    }
+                }
+            }
+        }
+    }
+
+
+    private boolean hasValidHeaderInformation(TablePageBlock table) {
+
+        return !hasInvalidHeaderInformation(table);
+    }
+
+
+    private boolean hasInvalidHeaderInformation(TablePageBlock table) {
+
+        return table.getRows()
+                .stream()
+                .flatMap(Collection::stream)
+                .allMatch(cell -> cell.getHeaderCells().isEmpty());
+    }
+
+
+    private List<Cell> getRowWithNonHeaderCells(TablePageBlock table) {
+
+        for (int i = table.getRowCount() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of table
+            List<Cell> row = table.getRows().get(i);
+            if (row.size() == 1) {
+                continue;
+            }
+            boolean allNonHeader = true;
+            for (Cell cell : row) {
+                if (cell.isHeaderCell()) {
+                    allNonHeader = false;
+                    break;
+                }
+            }
+            if (allNonHeader) {
+                return row;
+            }
+        }
+
+        return Collections.emptyList();
+
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/SectionTreeEntry.java

@@ -0,0 +1,129 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.outline;
+
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Objects;
+import java.util.stream.Collectors;
+
+import com.iqser.red.service.redaction.v1.server.model.document.nodes.GenericSemanticNode;
+import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.PageBlockType;
+import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
+
+import lombok.Data;
+import lombok.EqualsAndHashCode;
+
+@Data
+@EqualsAndHashCode(onlyExplicitlyIncluded = true)
+public class SectionTreeEntry {
+
+    public enum Type {
+        SECTION,
+        SUPER_SECTION,
+        TOC_SECTION
+    }
+
+    @EqualsAndHashCode.Include
+    private TextPageBlock headline;
+    private List<SectionTreeEntry> children = new ArrayList<>();
+    private SectionTreeEntry parent;
+
+    private List<AbstractPageBlock> sectionBlocks = new ArrayList<>();
+    private List<ClassifiedImage> images = new ArrayList<>();
+
+    private GenericSemanticNode section;
+
+
+    public SectionTreeEntry(TextPageBlock headline) {
+
+        this.headline = headline;
+    }
+
+
+    public Type getType() {
+
+        if (!Objects.isNull(headline) && headline.getClassification().equals(PageBlockType.TABLE_OF_CONTENTS_HEADLINE)) {
+            return Type.TOC_SECTION;
+        }
+        if (children.isEmpty()) {
+            return Type.SECTION;
+        }
+        return Type.SUPER_SECTION;
+    }
+
+
+    public void addChild(SectionTreeEntry sectionTreeEntry) {
+
+        children.add(sectionTreeEntry);
+        sectionTreeEntry.setParent(this);
+    }
+
+
+    public SectionTreeEntry getSiblingBefore() {
+
+        if (parent != null) {
+            int index = parent.getChildren().indexOf(this);
+            if (index > 0) {
+                return parent.getChildren().get(index - 1);
+            }
+        }
+        return null;
+    }
+
+
+    public SectionTreeEntry getSiblingAfter() {
+
+        if (parent != null) {
+            int index = parent.getChildren().indexOf(this);
+            if (index >= 0 && index < parent.getChildren().size() - 1) {
+                return parent.getChildren().get(index + 1);
+            }
+        }
+        return null;
+    }
+
+
+    public boolean contains(TextPageBlock block) {
+
+        if (headline.equals(block)) {
+            return true;
+        }
+        for (SectionTreeEntry child : children) {
+            if (child.contains(block)) {
+                return true;
+            }
+        }
+        return false;
+    }
+
+
+    public boolean contains(SectionTreeEntry tocItem) {
+
+        if (this.equals(tocItem)) {
+            return true;
+        }
+        for (SectionTreeEntry child : children) {
+            if (child.contains(tocItem)) {
+                return true;
+            }
+        }
+        return false;
+    }
+
+
+    public List<AbstractPageBlock> getNonEmptySectionBlocks() {
+
+        return sectionBlocks.stream()
+                .filter(pageBlock -> !pageBlock.isEmpty())
+                .collect(Collectors.toList());
+    }
+
+
+    @Override
+    public String toString() {
+
+        return "OutlineObjectTreeNode{" + "textPageBlock=" + headline + '}';
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/table/Cell.java

@@ -1,13 +1,15 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.table;
 
+import java.awt.geom.AffineTransform;
 import java.awt.geom.Point2D;
 import java.awt.geom.Rectangle2D;
 import java.util.ArrayList;
 import java.util.Iterator;
 import java.util.List;
 
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.BoundingBox;
 import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.model.text.Word;
 import com.knecon.fforesight.service.layoutparser.processor.utils.TextNormalizationUtilities;
 
 import lombok.Data;
@@ -18,7 +20,7 @@ import lombok.NoArgsConstructor;
 @Data
 @EqualsAndHashCode(callSuper = true)
 @NoArgsConstructor
-public class Cell extends Rectangle {
+public class Cell extends BoundingBox {
 
     private List<TextPageBlock> textBlocks = new ArrayList<>();
 
@@ -33,13 +35,24 @@ public class Cell extends Rectangle {
 
     public Cell(Point2D topLeft, Point2D bottomRight) {
 
-        super((float) topLeft.getY(), (float) topLeft.getX(), (float) (bottomRight.getX() - topLeft.getX()), (float) (bottomRight.getY() - topLeft.getY()));
+        this.bBoxPdf = new Rectangle2D.Double(topLeft.getX(), topLeft.getY(), (bottomRight.getX() - topLeft.getX()), (bottomRight.getY() - topLeft.getY()));
+        this.bBox = bBoxPdf;
     }
 
 
-    public Cell(Rectangle2D r) {
+    public Cell(Rectangle2D bBoxInitialUserSpace, AffineTransform initialUserSpaceToJava) {
 
-        super((float) r.getY(), (float) r.getX(), (float) r.getWidth(), (float) r.getHeight());
+        this.bBoxPdf = bBoxInitialUserSpace;
+        this.bBox = initialUserSpaceToJava.createTransformedShape(bBoxInitialUserSpace).getBounds2D();
+    }
+
+
+    public static Cell copy(Cell cell) {
+
+        Cell copy = new Cell();
+        copy.bBoxPdf = cell.bBoxPdf;
+        copy.bBox = cell.bBox;
+        return copy;
     }
 
 
@@ -55,12 +68,12 @@ public class Cell extends Rectangle {
         StringBuilder sb = new StringBuilder();
 
         Iterator<TextPageBlock> itty = textBlocks.iterator();
-        TextPositionSequence previous = null;
+        Word previous = null;
         while (itty.hasNext()) {
 
             TextPageBlock textBlock = itty.next();
 
-            for (TextPositionSequence word : textBlock.getSequences()) {
+            for (Word word : textBlock.getWords()) {
                 if (previous != null) {
                     if (Math.abs(previous.getMaxYDirAdj() - word.getMaxYDirAdj()) > word.getTextHeight()) {
                         sb.append('\n');
@@ -74,7 +87,7 @@ public class Cell extends Rectangle {
 
         }
 
-        return TextNormalizationUtilities.removeHyphenLineBreaks(sb.toString()).replaceAll("\n", " ").replaceAll(" {2}", " ");
+        return TextNormalizationUtilities.cleanString(sb.toString());
     }
 
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/table/CleanRulings.java

@@ -1,15 +1,206 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.table;
 
+import java.awt.geom.Line2D;
+import java.awt.geom.Point2D;
+import java.awt.geom.Rectangle2D;
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.Collections;
+import java.util.Comparator;
+import java.util.LinkedList;
 import java.util.List;
+import java.util.stream.Stream;
 
-import lombok.Builder;
-import lombok.Data;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.BoundingBox;
 
-@Data
-@Builder
+import lombok.Getter;
+
+@Getter
 public class CleanRulings {
 
-    List<Ruling> horizontal;
-    List<Ruling> vertical;
+    List<Ruling> horizontals; // unmodifiable sorted by Y list
+    List<Ruling> verticals; // unmodifiable sorted by X list
+
+
+    public CleanRulings(List<Ruling> horizontals, List<Ruling> verticals) {
+
+        this.horizontals = horizontals.stream()
+                .peek(Ruling::assertHorizontal)
+                .sorted(Comparator.comparing(Line2D.Float::getY1))
+                .toList();
+        this.verticals = verticals.stream()
+                .peek(Ruling::assertVertical)
+                .sorted(Comparator.comparing(Line2D.Float::getX1))
+                .toList();
+    }
+
+
+    public CleanRulings getTableLines() {
+
+        return new CleanRulings(horizontals.stream()
+                                        .filter(ruling -> ruling.getClassification().equals(Ruling.Classification.TABLE_LINE))
+                                        .toList(),
+                                verticals.stream()
+                                        .filter(ruling -> ruling.getClassification().equals(Ruling.Classification.TABLE_LINE))
+                                        .toList());
+    }
+
+
+    public CleanRulings withoutTextRulings() {
+
+        return new CleanRulings(horizontals.stream()
+                                        .filter(ruling -> !(ruling.getClassification().equals(Ruling.Classification.UNDERLINE) || ruling.getClassification()
+                                                .equals(Ruling.Classification.STRIKETROUGH)))
+                                        .toList(),
+                                verticals.stream()
+                                        .filter(ruling -> !(ruling.getClassification().equals(Ruling.Classification.UNDERLINE) || ruling.getClassification()
+                                                .equals(Ruling.Classification.STRIKETROUGH)))
+                                        .toList());
+    }
+
+
+    public List<Ruling> buildAll() {
+
+        ArrayList<Ruling> rulings = new ArrayList<>(horizontals.size() + verticals.size());
+        rulings.addAll(horizontals);
+        rulings.addAll(verticals);
+        return rulings;
+    }
+
+
+    public boolean lineBetween(BoundingBox a, BoundingBox b) {
+
+        return lineBetween(a.getBBoxPdf(), b.getBBoxPdf());
+    }
+
+
+    public boolean lineBetween(Rectangle2D a, Rectangle2D b) {
+
+        return lineBetween(new Point2D.Double(a.getCenterX(), a.getCenterY()), new Point2D.Double(b.getCenterX(), b.getCenterY()));
+    }
+
+
+    public boolean lineBetween(Point2D p1, Point2D p2) {
+
+        Ruling ruling = new Ruling(p1, p2);
+
+        if (ruling.isHorizontal()) {
+            return getVerticalsInXInterval(ruling.x1, ruling.x2).stream()
+                    .anyMatch(vertical -> vertical.intersectsLine(ruling));
+
+        }
+
+        if (ruling.isVertical()) {
+            return getHorizontalsInYInterval(ruling.y1, ruling.y2).stream()
+                    .anyMatch(horizontal -> horizontal.intersectsLine(ruling));
+
+        }
+
+        return Stream.of(getVerticalsInXInterval(ruling.x1, ruling.x2), getHorizontalsInYInterval(ruling.y1, ruling.y2))
+                .flatMap(Collection::stream)
+                .anyMatch(other -> other.intersectsLine(ruling));
+    }
+
+
+    public List<Ruling> getHorizontalsInYInterval(float y1, float y2) {
+
+        float startY = Math.min(y1, y2);
+        float endY = Math.max(y1, y2);
+
+        if (horizontals.isEmpty() || Float.isNaN(startY) || Float.isNaN(endY)) {
+            return Collections.emptyList();
+        }
+
+        int firstGreaterThanIdx = findFirstHorizontalRulingIdxAbove(startY);
+
+        if (firstGreaterThanIdx == -1) {
+            return Collections.emptyList();
+        }
+
+        List<Ruling> result = new LinkedList<>();
+        for (int i = firstGreaterThanIdx; i < horizontals.size(); i++) {
+            Ruling horizontal = horizontals.get(i);
+            if (horizontal.y1 > endY) {
+                break;
+            }
+            result.add(horizontal);
+        }
+        return result;
+    }
+
+
+    private int findFirstHorizontalRulingIdxAbove(float y) {
+
+        int low = 0;
+        int high = horizontals.size() - 1;
+
+        while (low <= high) {
+            int mid = low + (high - low) / 2;
+            Line2D.Float midLine = horizontals.get(mid);
+            float midY = midLine.y1;
+
+            if (midY == y) {
+                return mid;
+            } else if (midY > y) {
+                high = mid - 1;
+            } else {
+                low = mid + 1;
+            }
+        }
+
+        // Return the index of the first element greater than y or -1 if not found
+        return horizontals.size() > low && horizontals.get(low).y1 > y ? low : -1;
+    }
+
+
+    public List<Ruling> getVerticalsInXInterval(float x1, float x2) {
+
+        float startX = Math.min(x1, x2);
+        float endX = Math.max(x1, x2);
+
+        if (verticals.isEmpty() || Float.isNaN(startX) || Float.isNaN(endX)) {
+            return Collections.emptyList();
+        }
+
+        int firstGreaterThanIdx = findFirstVerticalRulingIdxRightOf(startX);
+
+        if (firstGreaterThanIdx == -1) {
+            return Collections.emptyList();
+        }
+
+        List<Ruling> result = new LinkedList<>();
+        for (int i = firstGreaterThanIdx; i < verticals.size(); i++) {
+            Ruling horizontal = verticals.get(i);
+            if (horizontal.x1 > endX) {
+                break;
+            }
+            result.add(horizontal);
+        }
+        return result;
+    }
+
+
+    private int findFirstVerticalRulingIdxRightOf(float x) {
+
+        int low = 0;
+        int high = verticals.size() - 1;
+
+        while (low <= high) {
+            int mid = low + (high - low) / 2;
+            Line2D.Float midLine = verticals.get(mid);
+            float midX = midLine.x1;
+
+            if (midX == x) {
+                return mid;
+            } else if (midX > x) {
+                high = mid - 1;
+            } else {
+                low = mid + 1;
+            }
+        }
+
+        // Return the index of the first element greater than y or -1 if not found
+        return verticals.size() > low && verticals.get(low).x1 > x ? low : -1;
+    }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/table/Rectangle.java

@@ -1,218 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.table;
-
-import java.awt.geom.Point2D;
-import java.awt.geom.Rectangle2D;
-import java.util.Comparator;
-import java.util.List;
-
-@SuppressWarnings("all")
-public class Rectangle extends Rectangle2D.Float {
-
-    protected static final float VERTICAL_COMPARISON_THRESHOLD = 0.4f;
-    /**
-     * Ill-defined comparator, from when Rectangle was Comparable.
-     * <p>
-     * see https://github.com/tabulapdf/tabula-java/issues/116
-     *
-     * @deprecated with no replacement
-     */
-    @Deprecated
-    public static final Comparator<Rectangle> ILL_DEFINED_ORDER = new Comparator<Rectangle>() {
-        @Override
-        public int compare(Rectangle o1, Rectangle o2) {
-
-            if (o1.equals(o2)) {
-                return 0;
-            }
-            if (o1.verticalOverlap(o2) > VERTICAL_COMPARISON_THRESHOLD) {
-                return o1.isLtrDominant() == -1 && o2.isLtrDominant() == -1 ? -java.lang.Double.compare(o1.getX(), o2.getX()) : java.lang.Double.compare(o1.getX(), o2.getX());
-            } else {
-                return java.lang.Float.compare(o1.getBottom(), o2.getBottom());
-            }
-        }
-    };
-
-
-    public Rectangle() {
-
-        super();
-    }
-
-
-    public Rectangle(float top, float left, float width, float height) {
-
-        super();
-        this.setRect(left, top, width, height);
-    }
-
-
-    /**
-     * @param rectangles
-     * @return minimum bounding box that contains all the rectangles
-     */
-    public static Rectangle boundingBoxOf(List<? extends Rectangle> rectangles) {
-
-        float minx = java.lang.Float.MAX_VALUE;
-        float miny = java.lang.Float.MAX_VALUE;
-        float maxx = java.lang.Float.MIN_VALUE;
-        float maxy = java.lang.Float.MIN_VALUE;
-
-        for (Rectangle r : rectangles) {
-            minx = (float) Math.min(r.getMinX(), minx);
-            miny = (float) Math.min(r.getMinY(), miny);
-            maxx = (float) Math.max(r.getMaxX(), maxx);
-            maxy = (float) Math.max(r.getMaxY(), maxy);
-        }
-        return new Rectangle(miny, minx, maxx - minx, maxy - miny);
-    }
-
-
-    public int compareTo(Rectangle other) {
-
-        return ILL_DEFINED_ORDER.compare(this, other);
-    }
-
-
-    // I'm bad at Java and need this for fancy sorting in
-    // technology.tabula.TextChunk.
-    public int isLtrDominant() {
-
-        return 0;
-    }
-
-
-    public float getArea() {
-
-        return this.width * this.height;
-    }
-
-
-    public float verticalOverlap(Rectangle other) {
-
-        return Math.max(0, Math.min(this.getBottom(), other.getBottom()) - Math.max(this.getTop(), other.getTop()));
-    }
-
-
-    public boolean verticallyOverlaps(Rectangle other) {
-
-        return verticalOverlap(other) > 0;
-    }
-
-
-    public float horizontalOverlap(Rectangle other) {
-
-        return Math.max(0, Math.min(this.getRight(), other.getRight()) - Math.max(this.getLeft(), other.getLeft()));
-    }
-
-
-    public boolean horizontallyOverlaps(Rectangle other) {
-
-        return horizontalOverlap(other) > 0;
-    }
-
-
-    public float verticalOverlapRatio(Rectangle other) {
-
-        float rv = 0, delta = Math.min(this.getBottom() - this.getTop(), other.getBottom() - other.getTop());
-
-        if (other.getTop() <= this.getTop() && this.getTop() <= other.getBottom() && other.getBottom() <= this.getBottom()) {
-            rv = (other.getBottom() - this.getTop()) / delta;
-        } else if (this.getTop() <= other.getTop() && other.getTop() <= this.getBottom() && this.getBottom() <= other.getBottom()) {
-            rv = (this.getBottom() - other.getTop()) / delta;
-        } else if (this.getTop() <= other.getTop() && other.getTop() <= other.getBottom() && other.getBottom() <= this.getBottom()) {
-            rv = (other.getBottom() - other.getTop()) / delta;
-        } else if (other.getTop() <= this.getTop() && this.getTop() <= this.getBottom() && this.getBottom() <= other.getBottom()) {
-            rv = (this.getBottom() - this.getTop()) / delta;
-        }
-
-        return rv;
-
-    }
-
-
-    public float overlapRatio(Rectangle other) {
-
-        double intersectionWidth = Math.max(0, Math.min(this.getRight(), other.getRight()) - Math.max(this.getLeft(), other.getLeft()));
-        double intersectionHeight = Math.max(0, Math.min(this.getBottom(), other.getBottom()) - Math.max(this.getTop(), other.getTop()));
-        double intersectionArea = Math.max(0, intersectionWidth * intersectionHeight);
-        double unionArea = this.getArea() + other.getArea() - intersectionArea;
-
-        return (float) (intersectionArea / unionArea);
-    }
-
-
-    public Rectangle merge(Rectangle other) {
-
-        this.setRect(this.createUnion(other));
-        return this;
-    }
-
-
-    public float getTop() {
-
-        return (float) this.getMinY();
-    }
-
-
-    public void setTop(float top) {
-
-        float deltaHeight = top - this.y;
-        this.setRect(this.x, top, this.width, this.height - deltaHeight);
-    }
-
-
-    public float getRight() {
-
-        return (float) this.getMaxX();
-    }
-
-
-    public void setRight(float right) {
-
-        this.setRect(this.x, this.y, right - this.x, this.height);
-    }
-
-
-    public float getLeft() {
-
-        return (float) this.getMinX();
-    }
-
-
-    public void setLeft(float left) {
-
-        float deltaWidth = left - this.x;
-        this.setRect(left, this.y, this.width - deltaWidth, this.height);
-    }
-
-
-    public float getBottom() {
-
-        return (float) this.getMaxY();
-    }
-
-
-    public void setBottom(float bottom) {
-
-        this.setRect(this.x, this.y, this.width, bottom - this.y);
-    }
-
-
-    public Point2D[] getPoints() {
-
-        return new Point2D[]{new Point2D.Float(this.getLeft(), this.getTop()), new Point2D.Float(this.getRight(), this.getTop()), new Point2D.Float(this.getRight(),
-                this.getBottom()), new Point2D.Float(this.getLeft(), this.getBottom())};
-    }
-
-
-    @Override
-    public String toString() {
-
-        StringBuilder sb = new StringBuilder();
-        String s = super.toString();
-        sb.append(s.substring(0, s.length() - 1));
-        sb.append(String.format(",bottom=%f,right=%f]", this.getBottom(), this.getRight()));
-        return sb.toString();
-    }
-
-}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/table/Ruling.java

@@ -4,16 +4,14 @@ import java.awt.geom.Line2D;
 import java.awt.geom.Point2D;
 import java.awt.geom.Rectangle2D;
 import java.util.ArrayList;
-import java.util.Collections;
-import java.util.Comparator;
 import java.util.Formatter;
 import java.util.List;
-import java.util.Map;
-import java.util.TreeMap;
 
 import com.knecon.fforesight.service.layoutparser.processor.utils.CohenSutherlandClipping;
 import com.knecon.fforesight.service.layoutparser.processor.utils.DoubleComparisons;
 
+import lombok.Getter;
+import lombok.Setter;
 import lombok.extern.slf4j.Slf4j;
 
 @Slf4j
@@ -23,10 +21,24 @@ public class Ruling extends Line2D.Float {
     public static final int PERPENDICULAR_UNIT_EXPAND_AMOUNT = 2;
     public static final int COLINEAR_OR_PARALLEL_UNIT_EXPAND_AMOUNT = 2;
 
+    public enum Classification {
+        TABLE_LINE,
+        UNDERLINE,
+        STRIKETROUGH,
+        HEADER_SEPARATOR,
+        FOOTER_SEPARATOR,
+        OTHER
+    }
+
+    @Getter
+    @Setter
+    private Classification classification;
+
 
     public Ruling(Point2D p1, Point2D p2) {
 
         super(p1, p2);
+        this.classification = Classification.OTHER;
     }
 
 
@@ -60,126 +72,32 @@ public class Ruling extends Line2D.Float {
     }
 
 
-    // log(n) implementation of find_intersections
-    // based on http://people.csail.mit.edu/indyk/6.838-old/handouts/lec2.pdf
-    public static Map<Point2D, Ruling[]> findIntersections(List<Ruling> horizontals, List<Ruling> verticals) {
-
-        class SortObject {
-
-            protected SOType type;
-            protected float position;
-            protected Ruling ruling;
-
-
-            public SortObject(SOType type, float position, Ruling ruling) {
-
-                this.type = type;
-                this.position = position;
-                this.ruling = ruling;
-            }
+    public void assertHorizontal() {
 
+        if (isHorizontal()) {
+            return;
         }
-
-        List<SortObject> sos = new ArrayList<>();
-
-        TreeMap<Ruling, Boolean> tree = new TreeMap<>(new Comparator<Ruling>() {
-            @Override
-            public int compare(Ruling o1, Ruling o2) {
-
-                return java.lang.Double.compare(o1.getTop(), o2.getTop());
-            }
-        });
-
-        TreeMap<Point2D, Ruling[]> rv = new TreeMap<>(new Comparator<Point2D>() {
-            @Override
-            public int compare(Point2D o1, Point2D o2) {
-
-                if (o1.getY() > o2.getY()) {
-                    return 1;
-                }
-                if (o1.getY() < o2.getY()) {
-                    return -1;
-                }
-                if (o1.getX() > o2.getX()) {
-                    return 1;
-                }
-                if (o1.getX() < o2.getX()) {
-                    return -1;
-                }
-                return 0;
-            }
-        });
-
-        for (Ruling h : horizontals) {
-            sos.add(new SortObject(SOType.HLEFT, h.getLeft() - PERPENDICULAR_UNIT_EXPAND_AMOUNT, h));
-            sos.add(new SortObject(SOType.HRIGHT, h.getRight() + PERPENDICULAR_UNIT_EXPAND_AMOUNT, h));
-        }
-
-        for (Ruling v : verticals) {
-            sos.add(new SortObject(SOType.VERTICAL, v.getLeft(), v));
-        }
-
-        Collections.sort(sos, new Comparator<SortObject>() {
-            @Override
-            public int compare(SortObject a, SortObject b) {
-
-                int rv;
-                if (DoubleComparisons.feq(a.position, b.position)) {
-                    if (a.type == SOType.VERTICAL && b.type == SOType.HLEFT) {
-                        rv = 1;
-                    } else if (a.type == SOType.VERTICAL && b.type == SOType.HRIGHT) {
-                        rv = -1;
-                    } else if (a.type == SOType.HLEFT && b.type == SOType.VERTICAL) {
-                        rv = -1;
-                    } else if (a.type == SOType.HRIGHT && b.type == SOType.VERTICAL) {
-                        rv = 1;
-                    } else {
-                        rv = java.lang.Double.compare(a.position, b.position);
-                    }
-                } else {
-                    return java.lang.Double.compare(a.position, b.position);
-                }
-                return rv;
-            }
-        });
-
-        for (SortObject so : sos) {
-            switch (so.type) {
-                case VERTICAL:
-                    for (Map.Entry<Ruling, Boolean> h : tree.entrySet()) {
-                        try {
-                            Point2D i = h.getKey().intersectionPoint(so.ruling);
-                            if (i == null) {
-                                continue;
-                            }
-                            rv.put(i, new Ruling[]{h.getKey().expand(PERPENDICULAR_UNIT_EXPAND_AMOUNT), so.ruling.expand(PERPENDICULAR_UNIT_EXPAND_AMOUNT)});
-                        } catch (UnsupportedOperationException e) {
-                            log.info("Some line are oblique, ignoring...");
-                            continue;
-                        }
-                    }
-                    break;
-                case HRIGHT:
-                    tree.remove(so.ruling);
-                    break;
-                case HLEFT:
-                    tree.put(so.ruling, true);
-                    break;
-            }
-        }
-
-        return rv;
+        throw new IllegalArgumentException("Ruling " + this + " is not horizontal");
 
     }
 
 
-    public boolean vertical() {
+    public void assertVertical() {
+
+        if (isVertical()) {
+            return;
+        }
+        throw new IllegalArgumentException("Ruling " + this + " is not vertical");
+    }
+
+
+    public boolean isVertical() {
 
         return this.length() > 0 && DoubleComparisons.feq(this.x1, this.x2); //diff < ORIENTATION_CHECK_THRESHOLD;
     }
 
 
-    public boolean horizontal() {
+    public boolean isHorizontal() {
 
         return this.length() > 0 && DoubleComparisons.feq(this.y1, this.y2); //diff < ORIENTATION_CHECK_THRESHOLD;
     }
@@ -188,36 +106,36 @@ public class Ruling extends Line2D.Float {
     // these are used to have a single collapse method (in page, currently)
 
 
-    public boolean oblique() {
+    public boolean isOblique() {
 
-        return !(this.vertical() || this.horizontal());
+        return !(this.isVertical() || this.isHorizontal());
     }
 
 
     public float getPosition() {
 
-        if (this.oblique()) {
+        if (this.isOblique()) {
             throw new UnsupportedOperationException();
         }
-        return this.vertical() ? this.getLeft() : this.getTop();
+        return this.isVertical() ? this.getLeft() : this.getTop();
     }
 
 
     public float getStart() {
 
-        if (this.oblique()) {
+        if (this.isOblique()) {
             throw new UnsupportedOperationException();
         }
-        return this.vertical() ? this.getTop() : this.getLeft();
+        return this.isVertical() ? this.getTop() : this.getLeft();
     }
 
 
     public void setStart(float v) {
 
-        if (this.oblique()) {
+        if (this.isOblique()) {
             throw new UnsupportedOperationException();
         }
-        if (this.vertical()) {
+        if (this.isVertical()) {
             this.setTop(v);
         } else {
             this.setLeft(v);
@@ -227,19 +145,19 @@ public class Ruling extends Line2D.Float {
 
     public float getEnd() {
 
-        if (this.oblique()) {
+        if (this.isOblique()) {
             throw new UnsupportedOperationException();
         }
-        return this.vertical() ? this.getBottom() : this.getRight();
+        return this.isVertical() ? this.getBottom() : this.getRight();
     }
 
 
     public void setEnd(float v) {
 
-        if (this.oblique()) {
+        if (this.isOblique()) {
             throw new UnsupportedOperationException();
         }
-        if (this.vertical()) {
+        if (this.isVertical()) {
             this.setBottom(v);
         } else {
             this.setRight(v);
@@ -249,10 +167,10 @@ public class Ruling extends Line2D.Float {
 
     public void setStartEnd(float start, float end) {
 
-        if (this.oblique()) {
+        if (this.isOblique()) {
             throw new UnsupportedOperationException();
         }
-        if (this.vertical()) {
+        if (this.isVertical()) {
             this.setTop(start);
             this.setBottom(end);
         } else {
@@ -264,7 +182,7 @@ public class Ruling extends Line2D.Float {
 
     public boolean perpendicularTo(Ruling other) {
 
-        return this.vertical() == other.horizontal();
+        return this.isVertical() == other.isHorizontal();
     }
 
 
@@ -318,30 +236,6 @@ public class Ruling extends Line2D.Float {
     }
 
 
-    public Point2D intersectionPoint(Ruling other) {
-
-        Ruling this_l = this.expand(PERPENDICULAR_UNIT_EXPAND_AMOUNT);
-        Ruling other_l = other.expand(PERPENDICULAR_UNIT_EXPAND_AMOUNT);
-        Ruling horizontal, vertical;
-
-        if (!this_l.intersectsLine(other_l)) {
-            return null;
-        }
-
-        if (this_l.horizontal() && other_l.vertical()) {
-            horizontal = this_l;
-            vertical = other_l;
-        } else if (this_l.vertical() && other_l.horizontal()) {
-            vertical = this_l;
-            horizontal = other_l;
-        } else {
-            log.warn("lines must be orthogonal, vertical and horizontal");
-            return null;
-        }
-        return new Point2D.Float(vertical.getLeft(), horizontal.getTop());
-    }
-
-
     @Override
     public boolean equals(Object other) {
 
@@ -451,16 +345,9 @@ public class Ruling extends Line2D.Float {
 
         final float TOLERANCE = 1;
         return Math.abs(ruling.getX1() - x1) < TOLERANCE &&//
-                Math.abs(ruling.getY1() - y1) < TOLERANCE &&//
-                Math.abs(ruling.getX2() - x2) < TOLERANCE &&//
-                Math.abs(ruling.getY2() - y2) < TOLERANCE;
-    }
-
-
-    private enum SOType {
-        VERTICAL,
-        HRIGHT,
-        HLEFT
+               Math.abs(ruling.getY1() - y1) < TOLERANCE &&//
+               Math.abs(ruling.getX2() - x2) < TOLERANCE &&//
+               Math.abs(ruling.getY2() - y2) < TOLERANCE;
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/table/TablePageBlock.java

@@ -36,14 +36,11 @@ public class TablePageBlock extends AbstractPageBlock {
     private List<Cell> cells;
 
 
-    public TablePageBlock(List<Cell> cells, Rectangle area, int rotation) {
+    public TablePageBlock(List<Cell> cells, int rotation) {
 
+        setToBBoxOfComponents(cells);
         this.cells = cells;
         addCells(cells);
-        minX = area.getLeft();
-        minY = area.getBottom();
-        maxX = area.getRight();
-        maxY = area.getTop();
         classification = PageBlockType.TABLE;
         this.rotation = rotation;
     }
@@ -230,15 +227,15 @@ public class TablePageBlock extends AbstractPageBlock {
             return new ArrayList<>();
         }
 
-        Set<Float> uniqueX = new HashSet<>();
-        Set<Float> uniqueY = new HashSet<>();
+        Set<Double> uniqueX = new HashSet<>();
+        Set<Double> uniqueY = new HashSet<>();
         cells.stream()
                 .filter(c -> !c.getTextBlocks().isEmpty() || c.getHeight() > 3 && c.getWidth() > 3)
                 .forEach(c -> {
-                    uniqueX.add(c.getLeft());
-                    uniqueX.add(c.getRight());
-                    uniqueY.add(c.getBottom());
-                    uniqueY.add(c.getTop());
+                    uniqueX.add(c.getPdfMinX());
+                    uniqueX.add(c.getPdfMaxX());
+                    uniqueY.add(c.getPdfMinY());
+                    uniqueY.add(c.getPdfMaxY());
                 });
 
         var sortedUniqueX = uniqueX.stream()
@@ -250,22 +247,24 @@ public class TablePageBlock extends AbstractPageBlock {
 
         List<List<Cell>> rowsOfCells = new ArrayList<>();
 
-        Float prevY = null;
+        Double prevY = null;
 
-        for (Float y : sortedUniqueY) {
+        for (Double y : sortedUniqueY) {
 
             List<Cell> row = new ArrayList<>();
 
-            Float prevX = null;
-            for (Float x : sortedUniqueX) {
+            Double prevX = null;
+            for (Double x : sortedUniqueX) {
 
                 if (prevY != null && prevX != null) {
-                    var cellFromGridStructure = new Cell(new Point2D.Float(prevX, prevY), new Point2D.Float(x, y));
+                    var cellFromGridStructure = new Cell(new Point2D.Double(prevX, prevY), new Point2D.Double(x, y));
 
                     if (cellFromGridStructure.hasMinimumSize()) {
 
                         cells.stream()
-                                .map(originalCell -> new CellWithIntersection(originalCell, RectangleTransformations.calculateIntersectedArea(cellFromGridStructure, originalCell)))
+                                .map(originalCell -> new CellWithIntersection(originalCell,
+                                                                              RectangleTransformations.calculateIntersectedArea(cellFromGridStructure.getBBoxPdf(),
+                                                                                                                                originalCell.getBBoxPdf())))
                                 .filter(cellWithIntersection -> cellWithIntersection.intersectedArea > 0)
                                 .filter(cellWithIntersection -> cellWithIntersection.originalCell.getArea() > cellWithIntersection.intersectedArea * CELL_AREA_CONTAINED_THRESHOLD)
                                 .max(Comparator.comparing(CellWithIntersection::intersectedArea))

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/AbstractBlockOnPage.java

@@ -0,0 +1,8 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.text;
+
+import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
+import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
+
+public record AbstractBlockOnPage(AbstractPageBlock block, ClassificationPage page) {
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/FontStyle.java

@@ -0,0 +1,9 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.text;
+
+
+public enum FontStyle {
+    REGULAR,
+    BOLD,
+    ITALIC,
+    BOLD_ITALIC;
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/FrequencyCounters.java

@@ -0,0 +1,21 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.text;
+
+import com.knecon.fforesight.service.layoutparser.processor.model.FloatFrequencyCounter;
+
+import lombok.AccessLevel;
+import lombok.Getter;
+import lombok.NoArgsConstructor;
+import lombok.experimental.FieldDefaults;
+
+@Getter
+@NoArgsConstructor
+@FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
+public class FrequencyCounters {
+
+    FloatFrequencyCounter lineHeightFrequencyCounter = new FloatFrequencyCounter();
+    FloatFrequencyCounter fontSizeFrequencyCounter = new FloatFrequencyCounter();
+    FloatFrequencyCounter spaceFrequencyCounter = new FloatFrequencyCounter();
+    StringFrequencyCounter fontFrequencyCounter = new StringFrequencyCounter();
+    StringFrequencyCounter styleFrequencyCounter = new StringFrequencyCounter();
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/ListIdentifier.java

@@ -0,0 +1,107 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.text;
+
+import java.util.List;
+import java.util.Optional;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+import lombok.AccessLevel;
+import lombok.AllArgsConstructor;
+import lombok.Getter;
+import lombok.experimental.FieldDefaults;
+
+@AllArgsConstructor
+@FieldDefaults(makeFinal = true, level = AccessLevel.PRIVATE)
+public class ListIdentifier {
+
+    public static final Pattern NUMBER_WITH_DOT = Pattern.compile("^\\s*([1-9]{1,4})\\.\\s+");
+    public static final Pattern NUMBER_IN_PARENTHESES = Pattern.compile("^\\s*\\(([1-9]{1,4})\\)\\s+");
+
+    enum Format {
+        NUMBER_WITH_DOT,
+        NUMBER_IN_PARENTHESES
+    }
+
+    Format format;
+    @Getter
+    Word word;
+    @Getter
+    int page;
+    int representation;
+
+
+    public static Optional<ListIdentifier> parse(TextPageBlock textPageBlock, int page) {
+
+        return parse(textPageBlock.getWords().subList(0, Math.min(5, textPageBlock.getWords().size())), page);
+    }
+
+
+    public static Optional<ListIdentifier> parse(List<Word> sequences, int page) {
+
+        StringBuilder sb = new StringBuilder();
+        for (Word sequence : sequences) {
+            sb.append(sequence.toString());
+            sb.append(" ");
+        }
+        sb.replace(sb.length() - 1, sb.length(), "");
+        String text = sb.toString();
+
+        Matcher numberMatcher = NUMBER_WITH_DOT.matcher(text);
+
+        if (numberMatcher.find()) {
+            Optional<Integer> representation = parseInteger(numberMatcher.group(1));
+            if (representation.isPresent()) {
+                return Optional.of(new ListIdentifier(Format.NUMBER_WITH_DOT, sequences.get(0), page, representation.get()));
+            }
+        }
+
+        Matcher parenthesisMatcher = NUMBER_IN_PARENTHESES.matcher(text);
+        if (parenthesisMatcher.find()) {
+            Optional<Integer> representation = parseInteger(parenthesisMatcher.group(1));
+            if (representation.isPresent()) {
+                return Optional.of(new ListIdentifier(Format.NUMBER_IN_PARENTHESES, sequences.get(0), page, representation.get()));
+            }
+        }
+        return Optional.empty();
+    }
+
+
+    private static Optional<Integer> parseInteger(String text) {
+
+        try {
+            return Optional.of(Integer.parseInt(text));
+        } catch (NumberFormatException e) {
+            return Optional.empty();
+        }
+    }
+
+
+    public static boolean isInOrder(List<ListIdentifier> listIdentifiers) {
+
+        if (listIdentifiers.size() <= 1) {
+            return true;
+        }
+
+        for (int i = 1; i < listIdentifiers.size(); i++) {
+            ListIdentifier current = listIdentifiers.get(i);
+            ListIdentifier previous = listIdentifiers.get(i - 1);
+            if (current.format != previous.format) {
+                return false;
+            }
+            if (current.representation <= previous.representation) {
+                return false;
+            }
+            if (!current.word.intersectsXDirAdj(previous.word, 2)) {
+                return false;
+            }
+            if (current.page == previous.page && !current.word.isBelowDirAdj(previous.word)) {
+                return false;
+            }
+            if (current.page < previous.page) {
+                return false;
+            }
+        }
+        return true;
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/RedTextPosition.java

@@ -1,97 +1,94 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.text;
 
+import java.awt.geom.AffineTransform;
+import java.awt.geom.Rectangle2D;
+
 import org.apache.pdfbox.text.TextPosition;
-import org.springframework.beans.BeanUtils;
 
-import com.fasterxml.jackson.annotation.JsonIgnore;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.model.TextBoundingBox;
+import com.knecon.fforesight.service.layoutparser.processor.docstrum.utils.FastAtan2;
+import com.knecon.fforesight.service.layoutparser.processor.utils.CoordinateTransforms;
+import com.knecon.fforesight.service.layoutparser.processor.utils.PageInformation;
 
+import lombok.AccessLevel;
 import lombok.AllArgsConstructor;
-import lombok.Builder;
 import lombok.Data;
+import lombok.EqualsAndHashCode;
 import lombok.NoArgsConstructor;
 import lombok.SneakyThrows;
+import lombok.experimental.FieldDefaults;
+import lombok.experimental.SuperBuilder;
 
 @Data
-@Builder
+@SuperBuilder
 @NoArgsConstructor
 @AllArgsConstructor
-public class RedTextPosition {
+@EqualsAndHashCode(callSuper = true)
+@FieldDefaults(level = AccessLevel.PRIVATE)
+public class RedTextPosition extends TextBoundingBox {
 
-    private float[] position;
+    public final static int HEIGHT_PADDING = 2;
 
-    @JsonIgnore
-    private int rotation;
+    String unicode;
 
-    @JsonIgnore
-    private float pageHeight;
+    // estimated using the TextMatrix in radians
+    float exactDir;
 
-    @JsonIgnore
-    private float pageWidth;
+    float widthOfSpace;
 
-    private String unicode;
+    float fontSizeInPt;
 
-    @JsonIgnore
-    private float dir;
-
-    // not used in reanalysis
-    @JsonIgnore
-    private float widthOfSpace;
-
-    // not used in reanalysis
-    @JsonIgnore
-    private float fontSizeInPt;
-
-    // not used in reanalysis
-    @JsonIgnore
-    private String fontName;
+    String fontName;
 
 
     @SneakyThrows
     public static RedTextPosition fromTextPosition(TextPosition textPosition) {
 
         var pos = new RedTextPosition();
-        BeanUtils.copyProperties(textPosition, pos);
-        pos.setFontName(textPosition.getFont().getName());
-
+        pos.setUnicode(textPosition.getUnicode());
+        pos.setWidthOfSpace(textPosition.getWidthOfSpace());
         pos.setFontSizeInPt(textPosition.getFontSizeInPt());
+        pos.setFontName(textPosition.getFont().getName());
+        pos.setExactDir((float) FastAtan2.fastAtan2(textPosition.getTextMatrix().getShearY(), textPosition.getTextMatrix().getScaleX()));
+        pos.setDir(TextDirection.fromDegrees(textPosition.getDir()));
 
-        var position = new float[4];
+        //TODO: There is a mismatch in the java coords of the text and the rulings,
+        // I guess if we start with the initial user space positions and transform them the same way we do the rulings it would work.
+        pos.setBBox(new Rectangle2D.Float(textPosition.getX(), textPosition.getY(), textPosition.getWidthDirAdj(), textPosition.getHeight()));
 
-        position[0] = textPosition.getXDirAdj();
-        position[1] = textPosition.getYDirAdj();
-        position[2] = textPosition.getWidthDirAdj();
-        position[3] = textPosition.getHeightDir();
+        float textHeight = textPosition.getHeight() + 2 * HEIGHT_PADDING;
+        Rectangle2D.Float dirAdjPosition = new Rectangle2D.Float(textPosition.getXDirAdj(),
+                                                                 textPosition.getYDirAdj() - textHeight,
+                                                                 textPosition.getWidthDirAdj(),
+                                                                 textHeight + HEIGHT_PADDING);
+
+        pos.setBBoxDirAdj(dirAdjPosition);
+
+        AffineTransform affineTransform = getRotationMatrix(TextDirection.fromDegrees(textPosition.getDir()), textPosition.getPageWidth(), textPosition.getPageHeight());
+        Rectangle2D bBoxInitialUserSpace = affineTransform.createTransformedShape(dirAdjPosition).getBounds2D();
+
+        pos.setBBoxPdf(bBoxInitialUserSpace); // These are definitely correct
 
-        pos.setPosition(position);
         return pos;
     }
 
 
-    @JsonIgnore
-    public float getXDirAdj() {
+    private static AffineTransform getRotationMatrix(TextDirection textDirection, float pageWidth, float pageHeight) {
 
-        return position[0];
-    }
+        AffineTransform transform = new AffineTransform();
 
-
-    @JsonIgnore
-    public float getYDirAdj() {
-
-        return position[1];
-    }
-
-
-    @JsonIgnore
-    public float getWidthDirAdj() {
-
-        return position[2];
-    }
-
-
-    @JsonIgnore
-    public float getHeightDir() {
-
-        return position[3];
+        if (textDirection == TextDirection.ZERO || textDirection == TextDirection.HALF_CIRCLE) {
+            transform.rotate(textDirection.getRadians(), pageWidth / 2f, pageHeight / 2f);
+            transform.translate(0f, pageHeight);
+        } else if (textDirection == TextDirection.QUARTER_CIRCLE) {
+            transform.rotate(textDirection.getRadians(), pageWidth / 2f, pageWidth / 2f);
+            transform.translate(0f, pageWidth);
+        } else {
+            transform.rotate(textDirection.getRadians(), pageHeight / 2f, pageHeight / 2f);
+            transform.translate(0f, pageWidth);
+        }
+        transform.scale(1., -1.);
+        return transform;
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/SearchableText.java

@@ -2,6 +2,7 @@ package com.knecon.fforesight.service.layoutparser.processor.model.text;
 
 import java.util.ArrayList;
 import java.util.List;
+
 import com.knecon.fforesight.service.layoutparser.processor.utils.TextNormalizationUtilities;
 
 import lombok.Getter;
@@ -9,18 +10,18 @@ import lombok.Getter;
 @Getter
 public class SearchableText {
 
-    private final List<TextPositionSequence> sequences = new ArrayList<>();
+    private final List<Word> sequences = new ArrayList<>();
 
 
-    public void add(TextPositionSequence textPositionSequence) {
+    public void add(Word word) {
 
-        sequences.add(textPositionSequence);
+        sequences.add(word);
     }
 
 
-    public void addAll(List<TextPositionSequence> textPositionSequences) {
+    public void addAll(List<Word> words) {
 
-        sequences.addAll(textPositionSequences);
+        sequences.addAll(words);
     }
 
 
@@ -31,18 +32,14 @@ public class SearchableText {
     }
 
 
-    public static String buildString(List<TextPositionSequence> sequences) {
+    public static String buildString(List<Word> sequences) {
 
         StringBuilder sb = new StringBuilder();
-        for (TextPositionSequence word : sequences) {
+        for (Word word : sequences) {
             sb.append(word);
             sb.append(' ');
         }
-        String text = sb.toString();
-        text = TextNormalizationUtilities.removeHyphenLineBreaks(text);
-        text = TextNormalizationUtilities.removeLineBreaks(text);
-        text = TextNormalizationUtilities.removeRepeatingWhitespaces(text);
-        return text;
+        return TextNormalizationUtilities.cleanString(sb.toString());
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/StringFrequencyCounter.java

@@ -9,10 +9,14 @@ public class StringFrequencyCounter {
 
     @Getter
     private final Map<String, Integer> countPerValue = new HashMap<>();
+    boolean changed;
+    String mostPopularCache;
 
 
     public void add(String value) {
 
+        changed = true;
+
         if (!countPerValue.containsKey(value)) {
             countPerValue.put(value, 1);
         } else {
@@ -23,6 +27,8 @@ public class StringFrequencyCounter {
 
     public void addAll(Map<String, Integer> otherCounter) {
 
+        changed = true;
+
         for (Map.Entry<String, Integer> entry : otherCounter.entrySet()) {
             if (countPerValue.containsKey(entry.getKey())) {
                 countPerValue.put(entry.getKey(), countPerValue.get(entry.getKey()) + entry.getValue());
@@ -35,13 +41,18 @@ public class StringFrequencyCounter {
 
     public String getMostPopular() {
 
-        Map.Entry<String, Integer> mostPopular = null;
-        for (Map.Entry<String, Integer> entry : countPerValue.entrySet()) {
-            if (mostPopular == null || entry.getValue() > mostPopular.getValue()) {
-                mostPopular = entry;
+        if (changed || mostPopularCache == null) {
+            Map.Entry<String, Integer> mostPopular = null;
+            for (Map.Entry<String, Integer> entry : countPerValue.entrySet()) {
+                if (mostPopular == null || entry.getValue() >= mostPopular.getValue()) {
+                    mostPopular = entry;
+                }
             }
+            mostPopularCache = mostPopular != null ? mostPopular.getKey() : null;
+            changed = false;
         }
-        return mostPopular != null ? mostPopular.getKey() : null;
+
+        return mostPopularCache;
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/TextBlockOnPage.java

@@ -0,0 +1,7 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.text;
+
+import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationPage;
+
+public record TextBlockOnPage(TextPageBlock textBlock, ClassificationPage page) {
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/TextDirection.java

@@ -44,4 +44,15 @@ public enum TextDirection {
 
         throw new IllegalArgumentException(String.format("A value of %f is not supported by TextDirection", degrees));
     }
+
+
+    public int getRotation() {
+
+        return switch (this) {
+            case ZERO -> 0;
+            case QUARTER_CIRCLE -> 1;
+            case HALF_CIRCLE -> 2;
+            case THREE_QUARTER_CIRCLE -> 3;
+        };
+    }
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/TextPageBlock.java

@@ -1,16 +1,14 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.text;
 
-import static java.util.stream.Collectors.toSet;
-
+import java.awt.geom.Rectangle2D;
 import java.util.ArrayList;
-import java.util.Comparator;
+import java.util.Collections;
 import java.util.List;
 
 import com.fasterxml.jackson.annotation.JsonIgnore;
 import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
-import com.knecon.fforesight.service.layoutparser.processor.model.FloatFrequencyCounter;
 import com.knecon.fforesight.service.layoutparser.processor.model.PageBlockType;
-import com.knecon.fforesight.service.layoutparser.processor.utils.DoubleComparisons;
+import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
 import com.knecon.fforesight.service.layoutparser.processor.utils.TextNormalizationUtilities;
 
 import lombok.AllArgsConstructor;
@@ -27,234 +25,112 @@ import lombok.NoArgsConstructor;
 public class TextPageBlock extends AbstractPageBlock {
 
     @Builder.Default
-    private List<TextPositionSequence> sequences = new ArrayList<>();
+    private List<Word> words = new ArrayList<>();
+    @Builder.Default
+    private FrequencyCounters frequencyCounters = new FrequencyCounters();
 
-    @JsonIgnore
-    private int rotation;
+    private Rectangle2D bBoxDirAdj;
 
-    @JsonIgnore
-    private String mostPopularWordFont;
+    private boolean underlined;
 
-    @JsonIgnore
-    private String mostPopularWordStyle;
-
-    @JsonIgnore
-    private float mostPopularWordFontSize;
-
-    @JsonIgnore
-    private float mostPopularWordHeight;
-
-    @JsonIgnore
-    private float mostPopularWordSpaceWidth;
-
-    @JsonIgnore
-    private float highestFontSize;
-
-    @JsonIgnore
     private PageBlockType classification;
 
+    private boolean toDuplicate;
+
+    private String text;
+    private boolean changed;
+
+
+    public TextPageBlock(List<Word> words) {
+
+        this.words = new ArrayList<>(words);
+        this.frequencyCounters = new FrequencyCounters();
+
+        if (!words.isEmpty()) {
+            addToFrequencyCounters(words);
+        }
+        calculateBBox();
+    }
+
+
+    public List<Word> getWords() {
+
+        return Collections.unmodifiableList(words);
+    }
+
 
-    @JsonIgnore
     public TextDirection getDir() {
 
-        return sequences.get(0).getDir();
+        return words.get(0).getDir();
     }
 
 
-    @JsonIgnore
-    private float getPageHeight() {
+    private void calculateBBox() {
 
-        return sequences.get(0).getPageHeight();
+        if (words == null) {
+            this.bBox = new Rectangle2D.Double();
+            this.bBoxPdf = new Rectangle2D.Double();
+            this.bBoxDirAdj = new Rectangle2D.Double();
+            return;
+        }
+        this.bBoxDirAdj = words.stream()
+                .map(Word::getBBoxDirAdj)
+                .collect(RectangleTransformations.collectBBox());
+
+        setToBBoxOfComponents(words);
     }
 
 
-    @JsonIgnore
-    private float getPageWidth() {
+    public void recalculateBBox() {
 
-        return sequences.get(0).getPageWidth();
+        calculateBBox();
     }
-    
+
 
     public static TextPageBlock merge(List<TextPageBlock> textBlocksToMerge) {
 
-        List<TextPositionSequence> sequences = textBlocksToMerge.stream().map(TextPageBlock::getSequences).flatMap(java.util.Collection::stream).toList();
+        if (textBlocksToMerge.isEmpty()) {
+            throw new IllegalArgumentException("Need to provide at least one TextPageBlock.");
+        }
+        if (textBlocksToMerge.stream()
+                    .map(AbstractPageBlock::getPage)
+                    .distinct()
+                    .count() != 1) {
+            throw new IllegalArgumentException("Cannot merge textBlocks on different pages.");
+        }
+
+        List<Word> sequences = textBlocksToMerge.stream()
+                .map(TextPageBlock::getWords)
+                .flatMap(java.util.Collection::stream)
+                .toList();
         sequences = new ArrayList<>(sequences);
-        return fromTextPositionSequences(sequences);
-    }
 
-    public static TextPageBlock fromTextPositionSequences(List<TextPositionSequence> wordBlockList) {
-
-        TextPageBlock textBlock = null;
-
-        FloatFrequencyCounter lineHeightFrequencyCounter = new FloatFrequencyCounter();
-        FloatFrequencyCounter fontSizeFrequencyCounter = new FloatFrequencyCounter();
-        FloatFrequencyCounter spaceFrequencyCounter = new FloatFrequencyCounter();
-        StringFrequencyCounter fontFrequencyCounter = new StringFrequencyCounter();
-        StringFrequencyCounter styleFrequencyCounter = new StringFrequencyCounter();
-
-        for (TextPositionSequence wordBlock : wordBlockList) {
-
-            lineHeightFrequencyCounter.add(wordBlock.getTextHeight());
-            fontSizeFrequencyCounter.add(wordBlock.getFontSize());
-            spaceFrequencyCounter.add(wordBlock.getSpaceWidth());
-            fontFrequencyCounter.add(wordBlock.getFont());
-            styleFrequencyCounter.add(wordBlock.getFontStyle());
-
-            if (textBlock == null) {
-                textBlock = new TextPageBlock(wordBlock.getMinXDirAdj(),
-                        wordBlock.getMaxXDirAdj(),
-                        wordBlock.getMinYDirAdj(),
-                        wordBlock.getMaxYDirAdj(),
-                        wordBlockList,
-                        wordBlock.getRotation());
-            } else {
-                TextPageBlock spatialEntity = textBlock.union(wordBlock);
-                textBlock.resize(spatialEntity.getMinX(), spatialEntity.getMinY(), spatialEntity.getWidth(), spatialEntity.getHeight());
-            }
-        }
-
-        if (textBlock != null) {
-            textBlock.setMostPopularWordFont(fontFrequencyCounter.getMostPopular());
-            textBlock.setMostPopularWordStyle(styleFrequencyCounter.getMostPopular());
-            textBlock.setMostPopularWordFontSize(fontSizeFrequencyCounter.getMostPopular());
-            textBlock.setMostPopularWordHeight(lineHeightFrequencyCounter.getMostPopular());
-            textBlock.setMostPopularWordSpaceWidth(spaceFrequencyCounter.getMostPopular());
-            textBlock.setHighestFontSize(fontSizeFrequencyCounter.getHighest());
-        }
-
-        if (textBlock != null && textBlock.getSequences() != null && textBlock.getSequences()
-                .stream()
-                .map(t -> DoubleComparisons.round(t.getMinYDirAdj(), 3))
-                .collect(toSet())
-                .size() == 1) {
-            textBlock.getSequences().sort(Comparator.comparing(TextPositionSequence::getMinXDirAdj));
-        }
-        return textBlock;
+        return new TextPageBlock(sequences);
     }
 
 
+    private void addToFrequencyCounters(List<Word> sequences) {
 
-    /**
-     * Returns the minX value in pdf coordinate system.
-     * Note: This needs to use Pdf Coordinate System where {0,0} rotated with the page rotation.
-     * 0 -> LowerLeft
-     * 90 -> UpperLeft
-     * 180 -> UpperRight
-     * 270 -> LowerRight
-     *
-     * @return the minX value in pdf coordinate system
-     */
-    @JsonIgnore
-    public float getPdfMinX() {
+        for (Word wordBlock : sequences) {
 
-        if (getDir().getDegrees() == 90) {
-            return minY;
-        } else if (getDir().getDegrees() == 180) {
-            return getPageWidth() - maxX;
-
-        } else if (getDir().getDegrees() == 270) {
-
-            return getPageWidth() - maxY;
-        } else {
-            return minX;
+            frequencyCounters.getLineHeightFrequencyCounter().add(wordBlock.getTextHeight());
+            frequencyCounters.getFontSizeFrequencyCounter().add(wordBlock.getFontSize());
+            frequencyCounters.getSpaceFrequencyCounter().add(wordBlock.getSpaceWidth());
+            frequencyCounters.getFontFrequencyCounter().add(wordBlock.getFont());
+            frequencyCounters.getStyleFrequencyCounter().add(wordBlock.getFontStyle());
         }
+
+        setUnderlined(this.words.stream()
+                              .allMatch(Word::isUnderline));
     }
 
 
-    /**
-     * Returns the maxX value in pdf coordinate system.
-     * Note: This needs to use Pdf Coordinate System where {0,0} rotated with the page rotation.
-     * 0 -> LowerLeft
-     * 90 -> UpperLeft
-     * 180 -> UpperRight
-     * 270 -> LowerRight
-     *
-     * @return the maxX value in pdf coordinate system
-     */
-    @JsonIgnore
-    public float getPdfMaxX() {
-
-        if (getDir().getDegrees() == 90) {
-            return maxY;
-        } else if (getDir().getDegrees() == 180) {
-            return getPageWidth() - minX;
-        } else if (getDir().getDegrees() == 270) {
-            return getPageWidth() - minY;
-
-        } else {
-            return maxX;
-        }
-    }
-
-
-    /**
-     * Returns the minY value in pdf coordinate system.
-     * Note: This needs to use Pdf Coordinate System where {0,0} rotated with the page rotation.
-     * 0 -> LowerLeft
-     * 90 -> UpperLeft
-     * 180 -> UpperRight
-     * 270 -> LowerRight
-     *
-     * @return the minY value in pdf coordinate system
-     */
-    @JsonIgnore
-    public float getPdfMinY() {
-
-        if (getDir().getDegrees() == 90) {
-            return minX;
-        } else if (getDir().getDegrees() == 180) {
-            return maxY;
-
-        } else if (getDir().getDegrees() == 270) {
-            return getPageHeight() - maxX;
-
-        } else {
-            return getPageHeight() - maxY;
-        }
-    }
-
-
-    /**
-     * Returns the maxY value in pdf coordinate system.
-     * Note: This needs to use Pdf Coordinate System where {0,0} rotated with the page rotation.
-     * 0 -> LowerLeft
-     * 90 -> UpperLeft
-     * 180 -> UpperRight
-     * 270 -> LowerRight
-     *
-     * @return the maxY value in pdf coordinate system
-     */
-    @JsonIgnore
-    public float getPdfMaxY() {
-
-        if (getDir().getDegrees() == 90) {
-            return maxX;
-        } else if (getDir().getDegrees() == 180) {
-
-            return minY;
-        } else if (getDir().getDegrees() == 270) {
-            return getPageHeight() - minX;
-        } else {
-            return getPageHeight() - minY;
-        }
-    }
-
-
-    public TextPageBlock(float minX, float maxX, float minY, float maxY, List<TextPositionSequence> sequences, int rotation) {
-
-        this.minX = minX;
-        this.maxX = maxX;
-        this.minY = minY;
-        this.maxY = maxY;
-        this.sequences = sequences;
-        this.rotation = rotation;
-    }
-
-
-    public TextPageBlock union(TextPositionSequence r) {
+    public TextPageBlock union(Word r) {
 
         TextPageBlock union = this.copy();
         union.add(r);
+        addToFrequencyCounters(List.of(r));
+        calculateBBox();
         return union;
     }
 
@@ -262,83 +138,50 @@ public class TextPageBlock extends AbstractPageBlock {
     public TextPageBlock union(TextPageBlock r) {
 
         TextPageBlock union = this.copy();
-        union.add(r);
+        union.addAll(r.getWords());
+        addToFrequencyCounters(r.getWords());
+        calculateBBox();
         return union;
     }
 
 
-    public void add(TextPageBlock r) {
+    public void add(TextPageBlock textPageBlock) {
 
-        if (r.getMinX() < minX) {
-            minX = r.getMinX();
-        }
-        if (r.getMaxX() > maxX) {
-            maxX = r.getMaxX();
-        }
-        if (r.getMinY() < minY) {
-            minY = r.getMinY();
-        }
-        if (r.getMaxY() > maxY) {
-            maxY = r.getMaxY();
-        }
-        sequences.addAll(r.getSequences());
+        changed = true;
+        words.addAll(textPageBlock.getWords());
+        addToFrequencyCounters(textPageBlock.getWords());
+        calculateBBox();
     }
 
 
-    public void add(TextPositionSequence r) {
+    public void add(Word word) {
 
-        if (r.getMinXDirAdj() < minX) {
-            minX = r.getMinXDirAdj();
-        }
-        if (r.getMaxXDirAdj() > maxX) {
-            maxX = r.getMaxXDirAdj();
-        }
-        if (r.getMinYDirAdj() < minY) {
-            minY = r.getMinYDirAdj();
-        }
-        if (r.getMaxYDirAdj() > maxY) {
-            maxY = r.getMaxYDirAdj();
-        }
+        changed = true;
+        words.add(word);
+        addToFrequencyCounters(List.of(word));
+        calculateBBox();
+    }
+
+
+    public void addAll(List<Word> words) {
+
+        changed = true;
+        this.words.addAll(words);
+        addToFrequencyCounters(words);
+        calculateBBox();
     }
 
 
     public TextPageBlock copy() {
 
-        return new TextPageBlock(minX, maxX, minY, maxY, sequences, rotation);
-    }
-
-
-    public void resize(float x1, float y1, float width, float height) {
-
-        set(x1, y1, x1 + width, y1 + height);
-    }
-
-
-    public void set(float x1, float y1, float x2, float y2) {
-
-        this.minX = Math.min(x1, x2);
-        this.maxX = Math.max(x1, x2);
-        this.minY = Math.min(y1, y2);
-        this.maxY = Math.max(y1, y2);
+        return new TextPageBlock(new ArrayList<>(words));
     }
 
 
     @Override
     public String toString() {
 
-        StringBuilder builder = new StringBuilder();
-
-        for (int i = 0; i < sequences.size(); i++) {
-            String sequenceAsString = sequences.get(i).toString();
-            // Fix for missing Whitespace. This is recognized in getSequences method. See PDFTextStripper Line 1730.
-            if (i != 0 && sequences.get(i - 1).charAt(sequences.get(i - 1).length() - 1) != ' ' && sequenceAsString.charAt(0) != ' ') {
-                builder.append(' ');
-            }
-            builder.append(sequenceAsString);
-        }
-
-        return builder.toString();
-
+        return getText();
     }
 
 
@@ -346,30 +189,88 @@ public class TextPageBlock extends AbstractPageBlock {
     @JsonIgnore
     public String getText() {
 
-        StringBuilder sb = new StringBuilder();
+        if (text == null || changed) {
 
-        TextPositionSequence previous = null;
-        for (TextPositionSequence word : sequences) {
-            if (previous != null) {
-                if (Math.abs(previous.getMaxYDirAdj() - word.getMaxYDirAdj()) > word.getTextHeight()) {
-                    sb.append('\n');
-                } else {
-                    sb.append(' ');
+            StringBuilder sb = new StringBuilder();
+
+            Word previous = null;
+            for (Word word : words) {
+                if (previous != null) {
+                    if (Math.abs(previous.getMaxYDirAdj() - word.getMaxYDirAdj()) > word.getTextHeight()) {
+                        sb.append('\n');
+                    } else {
+                        sb.append(' ');
+                    }
                 }
+                sb.append(word.toString());
+                previous = word;
             }
-            sb.append(word.toString());
-            previous = word;
+
+            text = TextNormalizationUtilities.removeHyphenLinebreaks(sb.toString());
+            changed = false;
         }
 
-        return TextNormalizationUtilities.removeHyphenLineBreaks(sb.toString());
+        return text;
+    }
 
+
+    public int getNumberOfLines() {
+
+        int numberOfLines = 1;
+        Word previous = null;
+        for (Word word : words) {
+            if (previous != null) {
+                if (word.getMaxYDirAdj() - previous.getMaxYDirAdj() > word.getTextHeight()) {
+                    numberOfLines++;
+                }
+            }
+            previous = word;
+        }
+        return numberOfLines;
+    }
+
+
+    public String getMostPopularWordFont() {
+
+        return frequencyCounters.getFontFrequencyCounter().getMostPopular();
+    }
+
+
+    public String getMostPopularWordStyle() {
+
+        return frequencyCounters.getStyleFrequencyCounter().getMostPopular();
+    }
+
+
+    public double getMostPopularWordFontSize() {
+
+        return frequencyCounters.getFontSizeFrequencyCounter().getMostPopular();
+    }
+
+
+    public double getMostPopularWordHeight() {
+
+        return frequencyCounters.getLineHeightFrequencyCounter().getMostPopular();
+    }
+
+
+    public double getMostPopularWordSpaceWidth() {
+
+        return frequencyCounters.getSpaceFrequencyCounter().getMostPopular();
+    }
+
+
+    public double getHighestFontSize() {
+
+        Double highest = frequencyCounters.getFontSizeFrequencyCounter().getHighest();
+        return highest == null ? 0 : highest;
     }
 
 
     @Override
     public boolean isEmpty() {
 
-        return sequences.isEmpty();
+        return words.isEmpty();
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/TextPositionSequence.java

@@ -1,300 +0,0 @@
-package com.knecon.fforesight.service.layoutparser.processor.model.text;
-
-import java.awt.geom.AffineTransform;
-import java.awt.geom.Point2D;
-import java.util.ArrayList;
-import java.util.List;
-import java.util.Locale;
-import java.util.stream.Collectors;
-
-import org.apache.pdfbox.text.TextPosition;
-
-import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Point;
-import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
-
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.Data;
-import lombok.NoArgsConstructor;
-import lombok.SneakyThrows;
-import lombok.extern.slf4j.Slf4j;
-
-@Slf4j
-@Data
-@Builder
-@NoArgsConstructor
-@AllArgsConstructor
-public class TextPositionSequence implements CharSequence {
-
-    public static final int HEIGHT_PADDING = 2;
-    private int page;
-    private List<RedTextPosition> textPositions = new ArrayList<>();
-
-    private TextDirection dir;
-    private int rotation;
-    private float pageHeight;
-    private float pageWidth;
-    private boolean isParagraphStart;
-
-
-    public TextPositionSequence(int page) {
-
-        this.page = page;
-    }
-
-
-    public TextPositionSequence(List<TextPosition> textPositions, int page, boolean isParagraphStart) {
-
-        this.textPositions = textPositions.stream().map(RedTextPosition::fromTextPosition).collect(Collectors.toList());
-        this.page = page;
-        this.dir = TextDirection.fromDegrees(textPositions.get(0).getDir());
-        this.rotation = textPositions.get(0).getRotation();
-        this.pageHeight = textPositions.get(0).getPageHeight();
-        this.pageWidth = textPositions.get(0).getPageWidth();
-        this.isParagraphStart = isParagraphStart;
-    }
-
-
-    @Override
-    public int length() {
-
-        return textPositions.size();
-    }
-
-
-    @Override
-    public char charAt(int index) {
-
-        RedTextPosition textPosition = textPositionAt(index);
-        String text = textPosition.getUnicode();
-        return text.charAt(0);
-    }
-
-
-    public char charAt(int index, boolean caseInSensitive) {
-
-        RedTextPosition textPosition = textPositionAt(index);
-        String text = textPosition.getUnicode();
-        return caseInSensitive ? text.toLowerCase(Locale.ROOT).charAt(0) : text.charAt(0);
-    }
-
-
-    @Override
-    public TextPositionSequence subSequence(int start, int end) {
-
-        var textPositionSequence = new TextPositionSequence();
-        textPositionSequence.textPositions = textPositions.subList(start, end);
-        textPositionSequence.page = page;
-        textPositionSequence.dir = dir;
-        textPositionSequence.rotation = rotation;
-        textPositionSequence.pageHeight = pageHeight;
-        textPositionSequence.pageWidth = pageWidth;
-
-        return textPositionSequence;
-    }
-
-
-    @Override
-    public String toString() {
-
-        StringBuilder builder = new StringBuilder(length());
-        for (int i = 0; i < length(); i++) {
-            builder.append(charAt(i));
-        }
-        return builder.toString();
-    }
-
-
-    public RedTextPosition textPositionAt(int index) {
-
-        return textPositions.get(index);
-    }
-
-
-    public void add(TextPositionSequence textPositionSequence, RedTextPosition textPosition) {
-
-        this.textPositions.add(textPosition);
-        this.page = textPositionSequence.getPage();
-        this.dir = textPositionSequence.getDir();
-        this.rotation = textPositionSequence.getRotation();
-        this.pageHeight = textPositionSequence.getPageHeight();
-        this.pageWidth = textPositionSequence.getPageWidth();
-    }
-
-
-    public void add(TextPosition textPosition) {
-
-        this.textPositions.add(RedTextPosition.fromTextPosition(textPosition));
-
-        this.dir = TextDirection.fromDegrees(textPositions.get(0).getDir());
-        this.rotation = textPositions.get(0).getRotation();
-        this.pageHeight = textPositions.get(0).getPageHeight();
-        this.pageWidth = textPositions.get(0).getPageWidth();
-
-    }
-
-
-    /**
-     * This value is adjusted so that 0,0 is upper left and it is adjusted based on the text direction.
-     * This method ignores the page rotation but takes the text rotation and adjusts the coordinates to awt.
-     *
-     * @return the text direction adjusted minX value
-     */
-
-    public float getMinXDirAdj() {
-
-        return textPositions.get(0).getXDirAdj();
-
-    }
-
-
-    /**
-     * This value is adjusted so that 0,0 is upper left and it is adjusted based on the text direction.
-     * This method ignores the page rotation but takes the text rotation and adjusts the coordinates to awt.
-     *
-     * @return the text direction adjusted maxX value
-     */
-
-    public float getMaxXDirAdj() {
-
-        return textPositions.get(textPositions.size() - 1).getXDirAdj() + textPositions.get(textPositions.size() - 1).getWidthDirAdj() + HEIGHT_PADDING;
-
-    }
-
-
-    /**
-     * This value is adjusted so that 0,0 is upper left and it is adjusted based on the text direction.
-     * This method ignores the page rotation but takes the text rotation and adjusts the coordinates to awt.
-     *
-     * @return the text direction adjusted minY value. The upper border of the bounding box of the word.
-     */
-
-    public float getMinYDirAdj() {
-
-        return textPositions.get(0).getYDirAdj() - getTextHeight();
-
-    }
-
-
-    /**
-     * This value is adjusted so that 0,0 is upper left and it is adjusted based on the text direction.
-     * This method ignores the page rotation but takes the text rotation and adjusts the coordinates to awt.
-     *
-     * @return the text direction adjusted maxY value. The lower border of the bounding box of the word.
-     */
-
-    public float getMaxYDirAdj() {
-
-        return textPositions.get(0).getYDirAdj();
-
-    }
-
-
-    public float getTextHeight() {
-
-        return textPositions.get(0).getHeightDir() + HEIGHT_PADDING;
-    }
-
-
-    public float getHeight() {
-
-        return getMaxYDirAdj() - getMinYDirAdj();
-    }
-
-
-    public float getWidth() {
-
-        return getMaxXDirAdj() - getMinXDirAdj();
-    }
-
-
-    public String getFont() {
-
-        if (textPositions.get(0).getFontName() == null) {
-            return "none";
-        }
-        return textPositions.get(0).getFontName().toLowerCase(Locale.ROOT).replaceAll(",bold", "").replaceAll(",italic", "");
-    }
-
-
-    public String getFontStyle() {
-
-        if (textPositions.get(0).getFontName() == null) {
-            return "standard";
-        }
-        String lowercaseFontName = textPositions.get(0).getFontName().toLowerCase(Locale.ROOT);
-
-        if (lowercaseFontName.contains("bold") && lowercaseFontName.contains("italic")) {
-            return "bold, italic";
-        } else if (lowercaseFontName.contains("bold")) {
-            return "bold";
-        } else if (lowercaseFontName.contains("italic")) {
-            return "italic";
-        } else {
-            return "standard";
-        }
-    }
-
-
-    public float getFontSize() {
-
-        return textPositions.get(0).getFontSizeInPt();
-    }
-
-
-    public float getSpaceWidth() {
-
-        return textPositions.get(0).getWidthOfSpace();
-    }
-
-
-    /**
-     * This returns the bounding box of the word in Pdf Coordinate System where {0,0} rotated with the page rotation.
-     * 0 -> LowerLeft
-     * 90 -> UpperLeft
-     * 180 -> UpperRight
-     * 270 -> LowerRight
-     *
-     * @return bounding box of the word in Pdf Coordinate System
-     */
-
-    @SneakyThrows
-    public Rectangle getRectangle() {
-
-        log.debug("Page: '{}', Word: '{}', Rotation: '{}', textRotation {}", page, this, rotation, dir);
-
-        float textHeight = getTextHeight();
-
-        RedTextPosition firstTextPos = textPositions.get(0);
-        RedTextPosition lastTextPos = textPositions.get(textPositions.size() - 1);
-
-        Point2D bottomLeft = new Point2D.Double(firstTextPos.getXDirAdj(), firstTextPos.getYDirAdj() - HEIGHT_PADDING);
-        Point2D topRight = new Point2D.Double(lastTextPos.getXDirAdj() + lastTextPos.getWidthDirAdj(), lastTextPos.getYDirAdj() + textHeight + HEIGHT_PADDING);
-
-        AffineTransform transform = new AffineTransform();
-        if (dir == TextDirection.ZERO || dir == TextDirection.HALF_CIRCLE) {
-            transform.rotate(dir.getRadians(), pageWidth / 2f, pageHeight / 2f);
-            transform.translate(0f, pageHeight + textHeight);
-            transform.scale(1., -1.);
-        } else if (dir == TextDirection.QUARTER_CIRCLE) {
-            transform.rotate(dir.getRadians(), pageWidth / 2f, pageWidth / 2f);
-            transform.translate(0f, pageWidth + textHeight);
-            transform.scale(1., -1.);
-        } else {
-            transform.rotate(dir.getRadians(), pageHeight / 2f, pageHeight / 2f);
-            transform.translate(0f, pageWidth + textHeight);
-            transform.scale(1., -1.);
-        }
-
-        bottomLeft = transform.transform(bottomLeft, null);
-        topRight = transform.transform(topRight, null);
-
-        return new Rectangle( //
-                new Point((float) bottomLeft.getX(), (float) bottomLeft.getY()),
-                (float) (topRight.getX() - bottomLeft.getX()),
-                (float) (topRight.getY() - bottomLeft.getY()),
-                page);
-    }
-
-}
-

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/text/TocNumberComparator.java

@@ -0,0 +1,36 @@
+package com.knecon.fforesight.service.layoutparser.processor.model.text;
+
+import java.util.Comparator;
+import java.util.HashMap;
+
+import com.knecon.fforesight.service.layoutparser.processor.services.classification.NumberWord;
+
+public class TocNumberComparator implements Comparator<NumberWord> {
+
+    private HashMap<NumberWord, TextBlockOnPage> lookup;
+
+
+    public TocNumberComparator(HashMap<NumberWord, TextBlockOnPage> lookup) {
+
+        this.lookup = lookup;
+    }
+
+
+    @Override
+    public int compare(NumberWord number1, NumberWord number2) {
+
+        int page1 = lookup.get(number1).page().getPageNumber();
+        int page2 = lookup.get(number2).page().getPageNumber();
+
+        if (page1 != page2) {
+            return Integer.compare(page1, page2);
+        }
+
+        if (number1.word().getY() != number2.word().getY()) {
+            return Double.compare(number1.word().getY(), number2.word().getY());
+        }
+
+        return Integer.compare(number1.number(), number2.number());
+    }
+
+}