RED-6009: Document Tree Structure

*moved all layoutparsing code to separate project
*wip (some dependency issues)

.gitignore

@@ -0,0 +1,33 @@
+HELP.md
+target/
+!.mvn/wrapper/maven-wrapper.jar
+!**/src/main/**/target/
+!**/src/test/**/target/
+
+### STS ###
+.apt_generated
+.classpath
+.factorypath
+.project
+.settings
+.springBeans
+.sts4-cache
+
+### IntelliJ IDEA ###
+.idea
+*.iws
+*.iml
+*.ipr
+
+### NetBeans ###
+/nbproject/private/
+/nbbuild/
+/dist/
+/nbdist/
+/.nb-gradle/
+build/
+!**/src/main/**/build/
+!**/src/test/**/build/
+
+### VS Code ###
+.vscode/

.mvn/wrapper/maven-wrapper.properties

@@ -0,0 +1,18 @@
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements.  See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership.  The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License.  You may obtain a copy of the License at
+#
+#   https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied.  See the License for the
+# specific language governing permissions and limitations
+# under the License.
+distributionUrl=https://repo.maven.apache.org/maven2/org/apache/maven/apache-maven/3.8.7/apache-maven-3.8.7-bin.zip
+wrapperUrl=https://repo.maven.apache.org/maven2/org/apache/maven/wrapper/maven-wrapper/3.1.1/maven-wrapper-3.1.1.jar

layoutparser-service-image/pom.xml

@@ -0,0 +1,17 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+         xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 https://maven.apache.org/xsd/maven-4.0.0.xsd">
+    <modelVersion>4.0.0</modelVersion>
+
+
+
+    <parent>
+        <groupId>com.knecon.fforesight</groupId>
+        <artifactId>layoutparser</artifactId>
+        <version>1.0.0</version>
+    </parent>
+
+    <artifactId>layoutparser-service-image</artifactId>
+    <version>1.0.0</version>
+
+</project>

layoutparser-service/layoutparser-service-internal-api/pom.xml

@@ -0,0 +1,49 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+         xmlns="http://maven.apache.org/POM/4.0.0"
+         xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
+    <modelVersion>4.0.0</modelVersion>
+
+
+    <parent>
+        <groupId>com.knecon.fforesight</groupId>
+        <artifactId>layoutparser-service</artifactId>
+        <version>1.0.0</version>
+    </parent>
+
+    <artifactId>layoutparser-service-internal-api</artifactId>
+    <version>1.0.0</version>
+
+    <packaging>pom</packaging>
+
+    <dependencies>
+        <dependency>
+            <groupId>org.projectlombok</groupId>
+            <artifactId>lombok</artifactId>
+            <version>1.18.26</version>
+        </dependency>
+        <dependency>
+            <groupId>com.google.guava</groupId>
+            <artifactId>guava</artifactId>
+            <version>31.1-jre</version>
+        </dependency>
+    </dependencies>
+
+    <build>
+        <plugins>
+            <plugin>
+                <groupId>org.springframework.boot</groupId>
+                <artifactId>spring-boot-maven-plugin</artifactId>
+                <configuration>
+                    <excludes>
+                        <exclude>
+                            <groupId>org.projectlombok</groupId>
+                            <artifactId>lombok</artifactId>
+                        </exclude>
+                    </excludes>
+                </configuration>
+            </plugin>
+        </plugins>
+    </build>
+
+</project>

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

@@ -0,0 +1,19 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.data;
+
+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)
+public class AtomicPositionBlockData {
+
+    Long id;
+    int[] stringIdxToPositionIdx;
+    float[][] positions;
+
+}

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

@@ -0,0 +1,23 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.data;
+
+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)
+public class AtomicTextBlockData {
+
+    Long id;
+    Long page;
+    String searchText;
+    int numberOnPage;
+    int start;
+    int end;
+    int[] lineBreaks;
+
+}

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

@@ -0,0 +1,20 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.data;
+
+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)
+public class DocumentData {
+
+    PageData[] pages;
+    AtomicTextBlockData[] atomicTextBlocks;
+    AtomicPositionBlockData[] atomicPositionBlocks;
+    TableOfContentsData tableOfContents;
+
+}

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

@@ -0,0 +1,20 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.data;
+
+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)
+public class PageData {
+
+    int number;
+    int height;
+    int width;
+    int rotation;
+
+}

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

@@ -0,0 +1,90 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.data;
+
+import java.util.Arrays;
+import java.util.List;
+import java.util.Map;
+import java.util.stream.Stream;
+
+import javax.management.openmbean.InvalidKeyException;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.NodeType;
+
+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 TableOfContentsData {
+
+    List<EntryData> entries;
+
+
+    public EntryData get(List<Integer> tocId) {
+
+        if (tocId.size() < 1) {
+            throw new InvalidKeyException(String.format("ClassificationSection Identifier: \"%s\" is not valid.", tocId));
+        }
+        EntryData entry = entries.get(tocId.get(0));
+        for (int id : tocId.subList(1, tocId.size())) {
+            entry = entry.subEntries().get(id);
+        }
+        return entry;
+    }
+
+
+    public Stream<EntryData> streamAllEntries() {
+
+        return entries.stream().flatMap(TableOfContentsData::flatten);
+    }
+
+
+    private static List<Integer> getIds(String idsAsString) {
+
+        return Arrays.stream(idsAsString.split("\\.")).map(Integer::valueOf).toList();
+    }
+
+
+    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.subEntries().stream().flatMap(TableOfContentsData::flatten));
+    }
+
+
+    @Builder
+    public record EntryData(NodeType type, int[] tocId, Long[] atomicBlocks, Long[] pages, Map<String, String> properties, List<EntryData> subEntries) {
+
+        @Override
+        public String toString() {
+
+            StringBuilder sb = new StringBuilder();
+            sb.append("[");
+            for (int i : tocId) {
+                sb.append(i);
+                sb.append(",");
+            }
+            sb.delete(sb.length() - 1, sb.length());
+            sb.append("]: ");
+
+            sb.append(type);
+            sb.append(" atbs = ");
+            sb.append(atomicBlocks.length);
+
+            return sb.toString();
+        }
+
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/Boundary.java

@@ -0,0 +1,148 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph;
+
+import java.util.LinkedList;
+import java.util.List;
+
+import lombok.Setter;
+
+@Setter
+public class Boundary implements Comparable<Boundary> {
+
+    private int start;
+    private int end;
+
+
+    public Boundary(int start, int end) {
+
+        if (start > end) {
+            throw new IllegalArgumentException(String.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(String.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(String.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 contains(boundary.start()) || contains(boundary.end() - 1);
+    }
+
+
+    public List<Boundary> split(List<Integer> splitIndices) {
+
+        if (splitIndices.stream().anyMatch(idx -> !this.contains(idx))) {
+            throw new IndexOutOfBoundsException(String.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 static Boundary merge(List<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 String.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;
+    }
+
+
+    @Override
+    public int hashCode() {
+
+        return toString().hashCode();
+    }
+
+
+    @Override
+    public boolean equals(Object object) {
+
+        return hashCode() == object.hashCode();
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/DocumentGraph.java

@@ -0,0 +1,98 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph;
+
+import java.awt.geom.Rectangle2D;
+import java.util.Collections;
+import java.util.HashMap;
+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.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SectionNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SemanticNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlockCollector;
+
+import lombok.AccessLevel;
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.Data;
+import lombok.experimental.FieldDefaults;
+
+@Data
+@Builder
+@AllArgsConstructor
+@FieldDefaults(level = AccessLevel.PRIVATE)
+public class DocumentGraph implements SemanticNode {
+
+    Set<PageNode> pages;
+    TableOfContents tableOfContents;
+    Integer numberOfPages;
+    TextBlock textBlock;
+
+
+    public TextBlock buildTextBlock() {
+
+        return streamTerminalTextBlocksInOrder().collect(new TextBlockCollector());
+    }
+
+
+    public List<SectionNode> getMainSections() {
+
+        return tableOfContents.entries.stream().filter(entry -> entry.node() instanceof SectionNode).map(entry -> (SectionNode) entry.node()).collect(Collectors.toList());
+    }
+
+
+    public Stream<TextBlock> streamTerminalTextBlocksInOrder() {
+
+        return streamAllNodes().filter(SemanticNode::isTerminal).map(SemanticNode::getTerminalTextBlock);
+    }
+
+
+    public Set<EntityNode> getEntities() {
+
+        return streamAllNodes().map(SemanticNode::getEntities).flatMap(Set::stream).collect(Collectors.toUnmodifiableSet());
+    }
+
+
+    @Override
+    public List<Integer> getTocId() {
+
+        return Collections.emptyList();
+    }
+
+
+    @Override
+    public void setTocId(List<Integer> tocId) {
+
+        throw new UnsupportedOperationException("DocumentGraph is always the root of the Table of Contents");
+    }
+
+
+    private Stream<SemanticNode> streamAllNodes() {
+
+        return tableOfContents.streamEntriesInOrder().map(TableOfContents.Entry::node);
+    }
+
+
+    @Override
+    public String toString() {
+
+        return tableOfContents.toString();
+    }
+
+
+    @Override
+    public Map<PageNode, Rectangle2D> getBBox() {
+
+        Map<PageNode, Rectangle2D> bBox = new HashMap<>();
+        for (PageNode page : pages) {
+            bBox.put(page, new Rectangle2D.Double(0, 0, page.getWidth(), page.getHeight()));
+        }
+        return bBox;
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/TableOfContents.java

@@ -0,0 +1,164 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph;
+
+import java.nio.charset.StandardCharsets;
+import java.util.Collections;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.NoSuchElementException;
+import java.util.stream.Stream;
+
+import com.google.common.hash.Hashing;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.NodeType;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SemanticNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlockCollector;
+
+import lombok.Builder;
+import lombok.Data;
+
+@Data
+public class TableOfContents {
+
+    List<Entry> entries;
+
+
+    public TableOfContents() {
+
+        entries = new LinkedList<>();
+    }
+
+
+    public TextBlock buildTextBlock() {
+
+        return streamEntriesInOrder().map(Entry::node).filter(SemanticNode::isTerminal).map(SemanticNode::getTerminalTextBlock).collect(new TextBlockCollector());
+    }
+
+
+    public List<Integer> createNewEntryAndReturnId(NodeType nodeType, SemanticNode node) {
+
+        return createNewChildEntryAndReturnId(Collections.emptyList(), nodeType, node);
+    }
+
+
+    public List<Integer> createNewChildEntryAndReturnId(List<Integer> parentId, NodeType nodeType, SemanticNode node) {
+
+        List<Integer> newId;
+        if (entryExists(parentId)) {
+            Entry parent = getEntryById(parentId);
+            newId = new LinkedList<>(parentId);
+            newId.add(parent.children().size());
+            parent.children().add(Entry.builder().tocId(newId).node(node).type(nodeType).children(new LinkedList<>()).build());
+        } else {
+            newId = List.of(entries.size());
+            entries.add(Entry.builder().tocId(newId).node(node).type(nodeType).children(new LinkedList<>()).build());
+        }
+
+        return newId;
+    }
+
+
+    private boolean entryExists(List<Integer> tocId) {
+
+        if (tocId.size() < 1) {
+            return false;
+        }
+        Entry entry = entries.get(tocId.get(0));
+        for (int id : tocId.subList(1, tocId.size())) {
+            if (id >= entry.children.size() || 0 > id) {
+                return false;
+            }
+            entry = entry.children().get(id);
+        }
+        return true;
+    }
+
+
+    public Entry getParentEntryById(List<Integer> tocId) {
+
+        List<Integer> parentIds = getParentId(tocId);
+        if (parentIds.size() < 1) {
+            throw new NoSuchElementException(String.format("Node with tocId \"%s\" has no parent!", tocId));
+        }
+        return getEntryById(parentIds);
+    }
+
+
+    public boolean hasParentById(List<Integer> tocId) {
+
+        List<Integer> parentId = getParentId(tocId);
+        return entryExists(parentId);
+    }
+
+
+    public Stream<SemanticNode> streamChildren(List<Integer> tocId) {
+
+        return getEntryById(tocId).children().stream().map(Entry::node);
+    }
+
+
+    private static List<Integer> getParentId(List<Integer> tocId) {
+
+        return tocId.subList(0, tocId.size() - 1);
+    }
+
+
+    public Entry getEntryById(List<Integer> tocId) {
+
+        Entry entry = entries.get(tocId.get(0));
+        for (int id : tocId.subList(1, tocId.size())) {
+            entry = entry.children().get(id);
+        }
+        return entry;
+    }
+
+
+    public Stream<Entry> streamEntriesInOrder() {
+
+        return entries.stream().flatMap(TableOfContents::flatten);
+    }
+
+
+    public Stream<Entry> streamSubEntriesInOrder(List<Integer> parentId) {
+
+        return Stream.of(getEntryById(parentId)).flatMap(TableOfContents::flatten);
+    }
+
+
+    @Override
+    public String toString() {
+
+        return String.join("\n", streamEntriesInOrder().map(Entry::toString).toList());
+    }
+
+
+    public String toString(List<Integer> id) {
+
+        return String.join("\n", streamSubEntriesInOrder(id).map(Entry::toString).toList());
+    }
+
+
+    private static Stream<Entry> flatten(Entry entry) {
+
+        return Stream.concat(Stream.of(entry), entry.children().stream().flatMap(TableOfContents::flatten));
+    }
+
+
+    @Builder
+    public record Entry(List<Integer> tocId, NodeType type, SemanticNode node, List<Entry> children) {
+
+        @Override
+        public String toString() {
+
+            return node().toString();
+        }
+
+
+        @Override
+        public int hashCode() {
+
+            return Hashing.murmur3_32_fixed().hashString(toString(), StandardCharsets.UTF_8).hashCode();
+        }
+
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/entity/EntityNode.java

@@ -0,0 +1,76 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.entity;
+
+import java.util.Collections;
+import java.util.List;
+import java.util.Set;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.Boundary;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SemanticNode;
+
+public interface EntityNode {
+
+    /**
+     * This represents the text, which is contained within the boundary of the Entity.
+     *
+     * @return String
+     */
+    String getValue();
+
+
+    /**
+     * The Boundary primarily defines the Entity, all other values may be inferred from it.
+     *
+     * @return Boundary, uniquely identifying this Entity
+     */
+    Boundary getBoundary();
+
+
+    /**
+     * The deepest fully containing node represents the node which is the deepest node in the document tree structure,
+     * whose boundary also fully contains the boundary of this entity
+     *
+     * @return the deepest fully containing node
+     */
+    SemanticNode getDeepestFullyContainingNode();
+
+
+    /**
+     * The intersecting nodes represent all nodes, whose boundary intersects the boundary of this entity.
+     *
+     * @return all intersecting Nodes
+     */
+    List<SemanticNode> getIntersectingNodes();
+
+
+    void setDeepestFullyContainingNode(SemanticNode semanticNode);
+
+
+    void addIntersectingNode(SemanticNode semanticNode);
+
+
+    void setIntersectingNodes(List<SemanticNode> semanticNodes);
+
+
+    /**
+     * @return all pages this entity intersects.
+     */
+    Set<PageNode> getPages();
+
+
+    void setPages(Set<PageNode> pages);
+
+
+    /**
+     * removes all occurrences of this node in the graph and resets all graph specific fields
+     */
+    default void removeFromGraph() {
+
+        getIntersectingNodes().forEach(node -> node.getEntities().remove(this));
+        getPages().forEach(page -> page.getEntities().remove(this));
+        setPages(Collections.emptySet());
+        setDeepestFullyContainingNode(null);
+        setIntersectingNodes(Collections.emptyList());
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/entity/EntityPosition.java

@@ -0,0 +1,39 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.entity;
+
+import java.awt.geom.Rectangle2D;
+import java.nio.charset.StandardCharsets;
+import java.util.List;
+
+import com.google.common.hash.Hashing;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+
+import lombok.AccessLevel;
+import lombok.Builder;
+import lombok.Data;
+import lombok.experimental.FieldDefaults;
+
+@Data
+@Builder
+@FieldDefaults(level = AccessLevel.PRIVATE)
+public class EntityPosition {
+
+    PageNode pageNode;
+    List<Rectangle2D> rectanglePerLine;
+
+
+    public String getId() {
+
+        return String.valueOf(hashCode());
+    }
+
+
+    @Override
+    public int hashCode() {
+
+        StringBuilder sb = new StringBuilder();
+        sb.append(pageNode.getNumber());
+        rectanglePerLine.forEach(r -> sb.append(r.getX()).append(r.getY()).append(r.getWidth()).append(r.getHeight()));
+        return Hashing.murmur3_128().hashString(sb.toString(), StandardCharsets.UTF_8).hashCode();
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/FooterNode.java

@@ -0,0 +1,53 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.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 FooterNode implements SemanticNode {
+
+    List<Integer> tocId;
+    TextBlock terminalTextBlock;
+
+    @Builder.Default
+    boolean terminal = true;
+
+    @EqualsAndHashCode.Exclude
+    TableOfContents tableOfContents;
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<EntityNode> entities = new HashSet<>();
+
+
+    @Override
+    public TextBlock buildTextBlock() {
+
+        return terminalTextBlock;
+    }
+
+
+    @Override
+    public String toString() {
+
+        return tocId + ": " + NodeType.FOOTER + ": " + terminalTextBlock.buildSummary();
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/HeaderNode.java

@@ -0,0 +1,53 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.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 HeaderNode implements SemanticNode {
+
+    List<Integer> tocId;
+    TextBlock terminalTextBlock;
+
+    @Builder.Default
+    boolean terminal = true;
+
+    @EqualsAndHashCode.Exclude
+    TableOfContents tableOfContents;
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<EntityNode> entities = new HashSet<>();
+
+
+    @Override
+    public TextBlock buildTextBlock() {
+
+        return terminalTextBlock;
+    }
+
+
+    @Override
+    public String toString() {
+
+        return tocId + ": " + NodeType.HEADER + ": " + terminalTextBlock.buildSummary();
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/HeadlineNode.java

@@ -0,0 +1,60 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.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 HeadlineNode implements SemanticNode {
+
+    List<Integer> tocId;
+    TextBlock terminalTextBlock;
+
+    @Builder.Default
+    boolean terminal = true;
+
+    @EqualsAndHashCode.Exclude
+    TableOfContents tableOfContents;
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<EntityNode> entities = new HashSet<>();
+
+
+    @Override
+    public TextBlock buildTextBlock() {
+
+        return terminalTextBlock;
+    }
+
+
+    @Override
+    public String toString() {
+
+        return tocId + ": " + NodeType.HEADLINE + ": " + terminalTextBlock.buildSummary();
+    }
+
+
+    @Override
+    public SemanticNode getHeadline() {
+
+        return this;
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/ImageNode.java

@@ -0,0 +1,88 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.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.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.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 ImageNode implements SemanticNode {
+
+    List<Integer> tocId;
+
+    ImageType imageType;
+    boolean transparency;
+    Rectangle2D position;
+
+    @Builder.Default
+    boolean redaction = false;
+    @Builder.Default
+    boolean ignored = false;
+    @Builder.Default
+    String redactionReason = "";
+    @Builder.Default
+    String legalBasis = "";
+    @Builder.Default
+    int matchedRule = -1;
+
+    @EqualsAndHashCode.Exclude
+    PageNode page;
+
+    @EqualsAndHashCode.Exclude
+    TableOfContents tableOfContents;
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<EntityNode> entities = new HashSet<>();
+
+
+    @Override
+    public TextBlock buildTextBlock() {
+
+        return streamAllSubNodes().filter(SemanticNode::isTerminal).map(SemanticNode::getTerminalTextBlock).collect(new TextBlockCollector());
+    }
+
+
+    @Override
+    public Set<PageNode> getPages() {
+
+        return Collections.singleton(page);
+    }
+
+
+    @Override
+    public String toString() {
+
+        return tocId + ": " + NodeType.IMAGE + ": " + imageType.toString() + " " + position;
+    }
+
+
+    @Override
+    public Map<PageNode, Rectangle2D> getBBox() {
+
+        Map<PageNode, Rectangle2D> bBoxPerPage = new HashMap<>();
+        bBoxPerPage.put(page, position);
+        return bBoxPerPage;
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/ImageType.java

@@ -0,0 +1,9 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+public enum ImageType {
+    LOGO,
+    FORMULA,
+    SIGNATURE,
+    OTHER,
+    OCR
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/NodeType.java

@@ -0,0 +1,12 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+public enum NodeType {
+    SECTION,
+    HEADLINE,
+    PARAGRAPH,
+    TABLE,
+    TABLE_CELL,
+    IMAGE,
+    HEADER,
+    FOOTER
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/PageNode.java

@@ -0,0 +1,66 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.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 PageNode {
+
+    Integer number;
+    Integer height;
+    Integer width;
+    Integer rotation;
+
+    @EqualsAndHashCode.Exclude
+    List<SemanticNode> mainBody;
+    @EqualsAndHashCode.Exclude
+    HeaderNode header;
+    @EqualsAndHashCode.Exclude
+    FooterNode footer;
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<EntityNode> entities = new HashSet<>();
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<ImageNode> images = new HashSet<>();
+
+
+    public TextBlock getMainBodyTextBlock() {
+
+        return mainBody.stream().filter(SemanticNode::isTerminal).map(SemanticNode::getTerminalTextBlock).collect(new TextBlockCollector());
+    }
+
+
+    @Override
+    public String toString() {
+
+        return String.valueOf(number);
+    }
+
+
+    @Override
+    public int hashCode() {
+
+        return number;
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/ParagraphNode.java

@@ -0,0 +1,51 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.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 ParagraphNode implements SemanticNode {
+
+    List<Integer> tocId;
+    TextBlock terminalTextBlock;
+
+    @Builder.Default
+    boolean terminal = true;
+
+    @EqualsAndHashCode.Exclude
+    TableOfContents tableOfContents;
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<EntityNode> entities = new HashSet<>();
+
+
+    @Override
+    public TextBlock buildTextBlock() {
+
+        return terminalTextBlock;
+    }
+
+
+    @Override
+    public String toString() {
+
+        return tocId + ": " + NodeType.PARAGRAPH + ": " + terminalTextBlock.buildSummary();
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/SectionNode.java

@@ -0,0 +1,63 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+import java.util.HashSet;
+import java.util.List;
+import java.util.NoSuchElementException;
+import java.util.Set;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.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 SectionNode implements SemanticNode {
+
+    List<Integer> tocId;
+
+    TextBlock textBlock;
+    @EqualsAndHashCode.Exclude
+    TableOfContents tableOfContents;
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<EntityNode> entities = new HashSet<>();
+
+
+    @Override
+    public TextBlock buildTextBlock() {
+
+        if (textBlock == null) {
+            textBlock = streamAllSubNodes().filter(SemanticNode::isTerminal).map(SemanticNode::getTerminalTextBlock).collect(new TextBlockCollector());
+        }
+        return textBlock;
+    }
+
+
+    @Override
+    public String toString() {
+
+        return tocId.toString() + ": " + NodeType.SECTION + ": " + buildTextBlock().buildSummary();
+    }
+
+
+    public HeadlineNode getHeadline() {
+
+        return streamChildren().filter(node -> node instanceof HeadlineNode)
+                .map(node -> (HeadlineNode) node)
+                .findFirst()
+                .orElseThrow(() -> new NoSuchElementException("ClassificationSection has no Headline!"));
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/SemanticNode.java

@@ -0,0 +1,275 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+import java.awt.geom.Rectangle2D;
+import java.util.HashMap;
+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.graph.Boundary;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.AtomicTextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.services.RectangleTransformations;
+
+public interface SemanticNode {
+
+    /**
+     * Searches all Nodes located underneath this Node in the TableOfContents and concatenates their AtomicTextBlocks into a single TextBlockEntity.
+     * So, for a ClassificationSection all TextBlocks of Subsections, Paragraphs, and Tables are concatenated into a single TextBlockEntity
+     * If the Node is Terminal, the TerminalTextBlock will be returned instead.
+     *
+     * @return ClassificationTextBlock containing all AtomicTextBlocks that are located under this Node.
+     */
+    TextBlock buildTextBlock();
+
+
+    /**
+     * 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<EntityNode> 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 ClassificationTextBlock
+     *
+     * @return Set of PageNodes this node appears on.
+     */
+    default Set<PageNode> getPages() {
+
+        return buildTextBlock().getPages();
+    }
+
+
+    /**
+     * @return the TableOfContents of the ClassificationDocument this node belongs to
+     */
+    TableOfContents getTableOfContents();
+
+
+    /**
+     * The id is a List of Integers uniquely identifying this node in the TableOfContents
+     *
+     * @return the TableOfContents ID
+     */
+    List<Integer> getTocId();
+
+
+    /**
+     * This should only be used during graph construction
+     *
+     * @param tocId List of Integers
+     */
+    void setTocId(List<Integer> tocId);
+
+
+    /**
+     * Traverses the Tree up, until it hits a HeadlineNode or hits a SectionNode which will then return the first HeadlineNode from its children.
+     * Throws NotFoundException if no Headline is found this way
+     *
+     * @return First HeadlineNode found
+     */
+    default SemanticNode getHeadline() {
+
+        return getParent().getHeadline();
+    }
+
+
+    /**
+     * @return boolean indicating wether this Node has a Parent in the TableOfContents
+     */
+    default boolean hasParent() {
+
+        return getTableOfContents().hasParentById(getTocId());
+    }
+
+
+    /**
+     * @return The SemanticNode representing the Parent in the TableOfContents
+     * throws NotFoundException, when no parent is present
+     */
+    default SemanticNode getParent() {
+
+        return getTableOfContents().getParentEntryById(getTocId()).node();
+    }
+
+
+    /**
+     * Terminal means a SemanticNode has direct access to a ClassificationTextBlock, by default this is false and must be overridden.
+     * Currently only Sections, Images, and Tables are not terminal.
+     * A TableCell might be Terminal depending on its area compared to the page.
+     *
+     * @return boolean, indicating if a Node has direct access to a ClassificationTextBlock
+     */
+    default boolean isTerminal() {
+
+        return false;
+    }
+
+
+    /**
+     * Terminal means a SemanticNode has direct access to a ClassificationTextBlock, by default this is false and must be overridden.
+     * Currently only Sections and Tables are not terminal.
+     *
+     * @return AtomicTextBlock
+     */
+    default TextBlock getTerminalTextBlock() {
+
+        throw new UnsupportedOperationException("Only terminal Nodes have access to TerminalTextBlocks!");
+    }
+
+
+    default void setTerminalTextBlock(TextBlock textBlock) {
+
+        throw new UnsupportedOperationException();
+    }
+
+
+    /**
+     * 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 = buildTextBlock();
+        if (textBlock.getAtomicTextBlocks().size() > 0) {
+            return buildTextBlock().getAtomicTextBlocks().get(0).getNumberOnPage();
+        } else {
+            return -1;
+        }
+    }
+
+
+    /**
+     * @return true, if this node's ClassificationTextBlock is not empty
+     */
+    default boolean hasText() {
+
+        return buildTextBlock().length() > 0;
+    }
+
+
+    /**
+     * @param string A String which the ClassificationTextBlock might contain
+     * @return true, if this node's ClassificationTextBlock contains the string
+     */
+    default boolean containsString(String string) {
+
+        return buildTextBlock().getSearchText().contains(string);
+    }
+
+
+    /**
+     * @param strings A List of Strings which the ClassificationTextBlock might contain
+     * @return true, if this node's ClassificationTextBlock contains any of the strings
+     */
+    default boolean containsAnyString(List<String> strings) {
+
+        return strings.stream().anyMatch(this::containsString);
+    }
+
+
+    /**
+     * This function is used during insertion of EntityNodes into the graph, it checks if the boundary of the EntityNode intersects or even contains the EntityNode.
+     * It sets the fields accordingly and recursively calls this function on all its children.
+     *
+     * @param entityNode EntityNode, which is being inserted into the graph
+     */
+    default void addThisToEntityIfIntersects(EntityNode entityNode) {
+
+        TextBlock textBlock = buildTextBlock();
+        if (textBlock.getBoundary().intersects(entityNode.getBoundary())) {
+
+            if (textBlock.containsBoundary(entityNode.getBoundary())) {
+                entityNode.setDeepestFullyContainingNode(this);
+            }
+
+            entityNode.addIntersectingNode(this);
+            streamChildren().forEach(node -> node.addThisToEntityIfIntersects(entityNode));
+        }
+    }
+
+
+    /**
+     * Streams all children located directly underneath this node in the TableOfContents
+     *
+     * @return Stream of all children
+     */
+    default Stream<SemanticNode> streamChildren() {
+
+        return getTableOfContents().streamChildren(getTocId());
+    }
+
+
+    /**
+     * recursively streams all SemanticNodes located underneath this node in the TableOfContents in order.
+     *
+     * @return Stream of all SubNodes
+     */
+    default Stream<SemanticNode> streamAllSubNodes() {
+
+        return getTableOfContents().streamSubEntriesInOrder(getTocId()).map(TableOfContents.Entry::node);
+    }
+
+
+    /**
+     * @return Boundary of this Node's ClassificationTextBlock
+     */
+    default Boundary getBoundary() {
+
+        return buildTextBlock().getBoundary();
+    }
+
+
+    /**
+     * If this Node is Terminal it will calculate the boundingBox of its TerminalTextBlock, otherwise it will calculate the Union of the BoundingBoxes of all its Children.
+     * If called on the ClassificationDocument, it will return the cropbox of each page
+     *
+     * @return Rectangle2D fully encapsulating this Node for each page.
+     */
+    default Map<PageNode, Rectangle2D> getBBox() {
+
+        Map<PageNode, Rectangle2D> bBoxPerPage = new HashMap<>();
+        if (isTerminal()) {
+            return getBBoxFromTerminalTextBlock(bBoxPerPage);
+        }
+
+        return getBBoxFromChildren(bBoxPerPage);
+    }
+
+
+    /**
+     * TODO this does not yet work for sections spanning multiple columns
+     *
+     * @param bBoxPerPage initial empty BoundingBox
+     * @return The union of the BoundingBoxes of all children
+     */
+    private Map<PageNode, Rectangle2D> getBBoxFromChildren(Map<PageNode, Rectangle2D> bBoxPerPage) {
+
+        return streamChildren().map(SemanticNode::getBBox).reduce((map1, map2) -> {
+            map1.forEach((page, rectangle) -> map2.merge(page, rectangle, (rect1, rect2) -> rect1.createUnion(rect2).getBounds2D()));
+            return map2;
+        }).orElse(bBoxPerPage);
+    }
+
+
+    /**
+     * @param bBoxPerPage initial empty BoundingBox
+     * @return The union of all BoundingBoxes of the ClassificationTextBlock of this node
+     */
+    private Map<PageNode, Rectangle2D> getBBoxFromTerminalTextBlock(Map<PageNode, Rectangle2D> bBoxPerPage) {
+
+        Map<PageNode, List<AtomicTextBlock>> atomicTextBlockPerPage = buildTextBlock().getAtomicTextBlocks().stream().collect(Collectors.groupingBy(AtomicTextBlock::getPage));
+        atomicTextBlockPerPage.forEach((page, atbs) -> bBoxPerPage.put(page, RectangleTransformations.bBoxUnionAtomicTextBlock(atbs)));
+        return bBoxPerPage;
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/TableCellNode.java

@@ -0,0 +1,92 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.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 java.util.stream.Stream;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.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 TableCellNode implements SemanticNode {
+
+    List<Integer> tocId;
+    int row;
+    int col;
+    boolean header;
+
+    Rectangle2D bBox;
+
+    @Builder.Default
+    boolean terminal = true;
+    TextBlock terminalTextBlock;
+
+    TextBlock textBlock;
+
+    @EqualsAndHashCode.Exclude
+    TableOfContents tableOfContents;
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<EntityNode> entities = new HashSet<>();
+
+
+    @Override
+    public Map<PageNode, Rectangle2D> getBBox() {
+
+        Map<PageNode, Rectangle2D> bBoxPerPage = new HashMap<>();
+        getPages().forEach(page -> bBoxPerPage.put(page, bBox));
+        return bBoxPerPage;
+    }
+
+
+    @Override
+    public TextBlock buildTextBlock() {
+
+        if (terminal) {
+            return terminalTextBlock;
+        }
+
+        if (textBlock == null) {
+            textBlock = streamAllSubNodes().filter(SemanticNode::isTerminal).map(SemanticNode::getTerminalTextBlock).collect(new TextBlockCollector());
+        }
+        return textBlock;
+    }
+
+
+    @Override
+    public String toString() {
+
+        return tocId + ": " + NodeType.TABLE_CELL + ": " + buildTextBlock().buildSummary();
+    }
+
+
+    public boolean hasHeader(String headerString) {
+
+        return getHeaders().anyMatch(header -> header.buildTextBlock().getSearchText().strip().equals(headerString));
+    }
+
+
+    private Stream<TableCellNode> getHeaders() {
+
+        TableNode tableNode = (TableNode) getParent();
+        return tableNode.streamHeadersForCell(row, col);
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/nodes/TableNode.java

@@ -0,0 +1,73 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes;
+
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+import java.util.stream.Stream;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.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 TableNode implements SemanticNode {
+
+    List<Integer> tocId;
+    TableOfContents tableOfContents;
+
+    Integer numberOfRows;
+    Integer numberOfCols;
+
+    TextBlock textBlock;
+
+    @Builder.Default
+    @EqualsAndHashCode.Exclude
+    Set<EntityNode> entities = new HashSet<>();
+
+
+    public Stream<TableCellNode> streamTableCells() {
+
+        return streamChildren().map(node -> (TableCellNode) node);
+    }
+
+
+    public Stream<TableCellNode> streamHeaders() {
+
+        return streamTableCells().filter(TableCellNode::isHeader);
+    }
+
+
+    public Stream<TableCellNode> streamHeadersForCell(int row, int col) {
+
+        return streamHeaders().filter(cell -> cell.getRow() == row || cell.getCol() == col);
+    }
+
+
+    @Override
+    public TextBlock buildTextBlock() {
+
+        if (textBlock == null) {
+            textBlock = streamAllSubNodes().filter(SemanticNode::isTerminal).map(SemanticNode::getTerminalTextBlock).collect(new TextBlockCollector());
+        }
+        return textBlock;
+    }
+
+
+    @Override
+    public String toString() {
+
+        return tocId.toString() + ": " + NodeType.TABLE + ": " + buildTextBlock().buildSummary();
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/textblock/AtomicTextBlock.java

@@ -0,0 +1,131 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock;
+
+import java.awt.geom.Rectangle2D;
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.Boundary;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityPosition;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SemanticNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.services.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;
+    PageNode page;
+
+    //string coordinates
+    Boundary boundary;
+    String searchText;
+    List<Integer> lineBreaks;
+
+    //position coordinates
+    List<Integer> stringIdxToPositionIdx;
+    List<Rectangle2D> positions;
+
+    @EqualsAndHashCode.Exclude
+    SemanticNode parent;
+
+
+    @Override
+    public int numberOfLines() {
+
+        return lineBreaks.size() + 1;
+    }
+
+
+    public CharSequence getLine(int lineNumber) {
+
+        if (lineNumber >= numberOfLines() || lineNumber < 0) {
+            throw new IndexOutOfBoundsException(String.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(String.format("%s is out of bounds for %s", stringBoundary, this.boundary));
+        }
+
+        if (stringBoundary.end() == this.boundary.end()) {
+            return positions.subList(stringIdxToPositionIdx.get(stringBoundary.start() - this.boundary.start()), positions.size());
+        }
+
+        return positions.subList(stringIdxToPositionIdx.get(stringBoundary.start() - this.boundary.start()),
+                stringIdxToPositionIdx.get(stringBoundary.end() - this.boundary.start()));
+    }
+
+
+    public List<EntityPosition> getEntityPositionsPerPage(Boundary stringBoundary) {
+
+        List<Rectangle2D> positionsPerLine = stringBoundary.split(getLineBreaks().stream().map(lb -> lb + boundary.start()).filter(stringBoundary::contains).toList())
+                .stream()
+                .map(this::getPositions)
+                .map(RectangleTransformations::rectangleUnion)
+                .toList();
+
+        return List.of(EntityPosition.builder().rectanglePerLine(positionsPerLine).pageNode(page).build());
+    }
+
+
+    @Override
+    public String toString() {
+
+        return searchText;
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/textblock/ConcatenatedTextBlock.java

@@ -0,0 +1,179 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock;
+
+import java.awt.geom.Rectangle2D;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.Map;
+import java.util.stream.Collectors;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.Boundary;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityPosition;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+
+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 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(String.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());
+        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 List<EntityPosition> getEntityPositionsPerPage(Boundary stringBoundary) {
+
+        List<AtomicTextBlock> textBlocks = getAllAtomicTextBlocksPartiallyInStringBoundary(stringBoundary);
+
+        if (textBlocks.size() == 1) {
+            return textBlocks.get(0).getEntityPositionsPerPage(stringBoundary);
+        }
+
+        AtomicTextBlock firstTextBlock = textBlocks.get(0);
+        List<EntityPosition> positions = new LinkedList<>(firstTextBlock.getEntityPositionsPerPage(new Boundary(stringBoundary.start(), firstTextBlock.getBoundary().end())));
+
+        for (AtomicTextBlock textBlock : textBlocks.subList(1, textBlocks.size() - 1)) {
+            positions.addAll(textBlock.getEntityPositionsPerPage(textBlock.getBoundary()));
+        }
+
+        AtomicTextBlock lastTextBlock = textBlocks.get(textBlocks.size() - 1);
+        positions.addAll(lastTextBlock.getEntityPositionsPerPage(new Boundary(lastTextBlock.getBoundary().start(), stringBoundary.end())));
+
+        return mergeEntityPositionsWithSamePageNode(positions);
+    }
+
+
+    private List<EntityPosition> mergeEntityPositionsWithSamePageNode(List<EntityPosition> positions) {
+
+        Map<PageNode, List<Rectangle2D>> entityPositionsPerPage = positions.stream().collect(//
+                Collectors.groupingBy(EntityPosition::getPageNode, //
+                        Collectors.flatMapping(entityPosition -> entityPosition.getRectanglePerLine().stream(), Collectors.toList())));
+
+        return entityPositionsPerPage.entrySet().stream()//
+                .map(entry -> EntityPosition.builder().pageNode(entry.getKey()).rectanglePerLine(entry.getValue()).build())//
+                .toList();
+
+    }
+
+
+    @Override
+    public String toString() {
+
+        return getSearchText();
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/textblock/TextBlock.java

@@ -0,0 +1,125 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock;
+
+import java.awt.geom.Rectangle2D;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
+import java.util.Set;
+import java.util.stream.Collectors;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.Boundary;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityPosition;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+
+public interface TextBlock extends CharSequence {
+
+    String getSearchText();
+
+
+    List<AtomicTextBlock> getAtomicTextBlocks();
+
+
+    Boundary getBoundary();
+
+
+    int getNextLinebreak(int fromIndex);
+
+
+    int getPreviousLinebreak(int fromIndex);
+
+
+    List<Integer> getLineBreaks();
+
+
+    Rectangle2D getPosition(int stringIdx);
+
+
+    List<Rectangle2D> getPositions(Boundary stringBoundary);
+
+
+    List<EntityPosition> getEntityPositionsPerPage(Boundary stringBoundary);
+
+
+    int numberOfLines();
+
+
+    default int indexOf(String searchTerm) {
+
+        return indexOf(searchTerm, getBoundary().start());
+    }
+
+
+    default Set<PageNode> getPages() {
+
+        return getAtomicTextBlocks().stream().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(String.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-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/graph/textblock/TextBlockCollector.java

@@ -0,0 +1,50 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock;
+
+import java.util.Collections;
+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 () -> new ConcatenatedTextBlock(Collections.emptyList());
+    }
+
+
+    @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-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/mapper/DocumentDataMapper.java

@@ -0,0 +1,143 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.mapper;
+
+import java.awt.geom.Rectangle2D;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.data.AtomicPositionBlockData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.AtomicTextBlockData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.DocumentData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.PageData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.TableOfContentsData;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.DocumentGraph;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.ImageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.TableCellNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.TableNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.AtomicTextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+
+public class DocumentDataMapper {
+
+    public DocumentData toDocumentData(DocumentGraph documentGraph) {
+
+        List<AtomicTextBlockData> atomicTextBlockData = documentGraph.streamTerminalTextBlocksInOrder()
+                .flatMap(textBlock -> textBlock.getAtomicTextBlocks().stream())
+                .distinct()
+                .map(this::toAtomicTextBlockData)
+                .toList();
+
+        List<AtomicPositionBlockData> atomicPositionBlockData = documentGraph.streamTerminalTextBlocksInOrder()
+                .flatMap(textBlock -> textBlock.getAtomicTextBlocks().stream())
+                .distinct()
+                .map(this::toAtomicPositionBlockData)
+                .toList();
+
+        List<PageData> pageData = documentGraph.getPages().stream().map(this::toPageData).toList();
+        TableOfContentsData tableOfContentsData = toTableOfContentsData(documentGraph.getTableOfContents());
+        return DocumentData.builder()
+                .atomicTextBlocks(atomicTextBlockData.toArray(new AtomicTextBlockData[0]))
+                .atomicPositionBlocks(atomicPositionBlockData.toArray(new AtomicPositionBlockData[0]))
+                .pages(pageData.toArray(new PageData[0]))
+                .tableOfContents(tableOfContentsData)
+                .build();
+    }
+
+
+    private TableOfContentsData toTableOfContentsData(TableOfContents tableOfContents) {
+
+        return new TableOfContentsData(tableOfContents.getEntries().stream().map(this::toEntryData).toList());
+    }
+
+
+    private TableOfContentsData.EntryData toEntryData(TableOfContents.Entry entry) {
+
+        Long[] atomicTextBlocks;
+
+        if (entry.node().isTerminal()) {
+            atomicTextBlocks = toAtomicTextBlockIds(entry.node().getTerminalTextBlock());
+        } else {
+            atomicTextBlocks = new Long[]{};
+        }
+
+        Map<String, String> properties = switch (entry.type()) {
+            case TABLE -> PropertiesMapper.buildTableProperties((TableNode) entry.node());
+            case TABLE_CELL -> PropertiesMapper.buildTableCellProperties((TableCellNode) entry.node());
+            case IMAGE -> PropertiesMapper.buildImageProperties((ImageNode) entry.node());
+            default -> new HashMap<>();
+        };
+
+        return TableOfContentsData.EntryData.builder()
+                .tocId(toPrimitiveIntArray(entry.tocId()))
+                .subEntries(entry.children().stream().map(this::toEntryData).toList())
+                .type(entry.type())
+                .atomicBlocks(atomicTextBlocks)
+                .pages(entry.node().getPages().stream().map(PageNode::getNumber).map(Integer::longValue).toArray(Long[]::new))
+                .properties(properties)
+                .build();
+    }
+
+
+    private Long[] toAtomicTextBlockIds(TextBlock textBlock) {
+
+        return textBlock.getAtomicTextBlocks().stream().map(AtomicTextBlock::getId).toArray(Long[]::new);
+    }
+
+
+    private PageData toPageData(PageNode p) {
+
+        return PageData.builder().rotation(p.getRotation()).height(p.getHeight()).width(p.getWidth()).number(p.getNumber()).build();
+    }
+
+
+    private AtomicTextBlockData toAtomicTextBlockData(AtomicTextBlock atomicTextBlock) {
+
+        return AtomicTextBlockData.builder()
+                .id(atomicTextBlock.getId())
+                .page(atomicTextBlock.getPage().getNumber().longValue())
+                .searchText(atomicTextBlock.getSearchText())
+                .numberOnPage(atomicTextBlock.getNumberOnPage())
+                .start(atomicTextBlock.getBoundary().start())
+                .end(atomicTextBlock.getBoundary().end())
+                .lineBreaks(toPrimitiveIntArray(atomicTextBlock.getLineBreaks()))
+                .build();
+    }
+
+
+    private AtomicPositionBlockData toAtomicPositionBlockData(AtomicTextBlock atomicTextBlock) {
+
+        return AtomicPositionBlockData.builder()
+                .id(atomicTextBlock.getId())
+                .positions(toPrimitiveFloatMatrix(atomicTextBlock.getPositions()))
+                .stringIdxToPositionIdx(toPrimitiveIntArray(atomicTextBlock.getStringIdxToPositionIdx()))
+                .build();
+    }
+
+
+    private float[][] toPrimitiveFloatMatrix(List<Rectangle2D> positions) {
+
+        float[][] positionMatrix = new float[positions.size()][];
+        for (int i = 0; i < positions.size(); i++) {
+            float[] singlePositions = new float[4];
+            singlePositions[0] = (float) positions.get(i).getMinX();
+            singlePositions[1] = (float) positions.get(i).getMinY();
+            singlePositions[2] = (float) positions.get(i).getWidth();
+            singlePositions[3] = (float) positions.get(i).getHeight();
+            positionMatrix[i] = singlePositions;
+        }
+        return positionMatrix;
+    }
+
+
+    private int[] toPrimitiveIntArray(List<Integer> list) {
+
+        int[] array = new int[list.size()];
+        for (int i = 0; i < list.size(); i++) {
+            array[i] = list.get(i);
+        }
+        return array;
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/mapper/DocumentGraphMapper.java

@@ -0,0 +1,225 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.mapper;
+
+import static com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.NodeType.FOOTER;
+import static com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.NodeType.HEADER;
+
+import java.awt.geom.Rectangle2D;
+import java.util.Arrays;
+import java.util.HashSet;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.Map;
+import java.util.NoSuchElementException;
+
+import com.google.common.primitives.Ints;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.AtomicPositionBlockData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.AtomicTextBlockData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.DocumentData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.PageData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.TableOfContentsData;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.Boundary;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.DocumentGraph;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.FooterNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.HeaderNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.HeadlineNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.ImageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.ParagraphNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SectionNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SemanticNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.TableCellNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.TableNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.AtomicTextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlockCollector;
+
+public class DocumentGraphMapper {
+
+    public DocumentGraph toDocumentGraph(DocumentData documentData) {
+
+        Context context = new Context(documentData,
+                new TableOfContents(),
+                new LinkedList<>(),
+                new LinkedList<>(),
+                Arrays.stream(documentData.getAtomicTextBlocks()).toList(),
+                Arrays.stream(documentData.getAtomicPositionBlocks()).toList());
+
+        context.pages.addAll(Arrays.stream(documentData.getPages()).map(this::buildPage).toList());
+
+        context.tableOfContents.setEntries(buildEntries(documentData.getTableOfContents().getEntries(), context));
+
+        DocumentGraph documentGraph = DocumentGraph.builder()
+                .numberOfPages(documentData.getPages().length)
+                .pages(new HashSet<>(context.pages))
+                .tableOfContents(context.tableOfContents)
+                .build();
+        documentGraph.setTextBlock(documentGraph.buildTextBlock());
+        return documentGraph;
+    }
+
+
+    private List<TableOfContents.Entry> buildEntries(List<TableOfContentsData.EntryData> entries,
+                                                     Context context) {
+
+        List<TableOfContents.Entry> newEntries = new LinkedList<>();
+        for (TableOfContentsData.EntryData entryData : entries) {
+
+            boolean terminal = isTerminal(entryData);
+            List<PageNode> pages = Arrays.stream(entryData.pages()).map(pageNumber -> getPage(pageNumber, context)).toList();
+
+            SemanticNode node = switch (entryData.type()) {
+                case SECTION -> buildSection(context);
+                case PARAGRAPH -> buildParagraph(context, terminal);
+                case HEADLINE -> buildHeadline(context, terminal);
+                case HEADER -> buildHeader(context, terminal);
+                case FOOTER -> buildFooter(context, terminal);
+                case TABLE -> buildTable(context, entryData.properties());
+                case TABLE_CELL -> buildTableCell(context, entryData.properties(), terminal);
+                case IMAGE -> buildImage(context, entryData.properties());
+                default -> throw new UnsupportedOperationException("Not yet implemented for type " + entryData.type());
+            };
+
+            if (node.isTerminal()) {
+                TextBlock textBlock = toTextBlock(entryData.atomicBlocks(), context, node);
+                node.setTerminalTextBlock(textBlock);
+            }
+            List<Integer> tocId = Arrays.stream(entryData.tocId()).boxed().toList();
+            node.setTocId(tocId);
+
+            if (entryData.type() == HEADER) {
+                pages.forEach(page -> page.setHeader((HeaderNode) node));
+            } else if (entryData.type() == FOOTER) {
+                pages.forEach(page -> page.setFooter((FooterNode) node));
+            } else {
+                pages.forEach(page -> page.getMainBody().add(node));
+            }
+            newEntries.add(TableOfContents.Entry.builder().tocId(tocId).type(entryData.type()).children(buildEntries(entryData.subEntries(), context)).node(node).build());
+        }
+        return newEntries;
+    }
+
+
+    private HeadlineNode buildHeadline(Context context, boolean terminal) {
+
+        return HeadlineNode.builder().terminal(terminal).tableOfContents(context.tableOfContents()).build();
+    }
+
+
+    private static boolean isTerminal(TableOfContentsData.EntryData entryData) {
+
+        return entryData.atomicBlocks().length > 0;
+    }
+
+
+    private ImageNode buildImage(Context context, Map<String, String> properties) {
+
+        var builder = ImageNode.builder();
+        PropertiesMapper.parseImageProperties(properties, builder);
+        return builder.tableOfContents(context.tableOfContents()).build();
+    }
+
+
+    private TableCellNode buildTableCell(Context context, Map<String, String> properties, boolean terminal) {
+
+        TableCellNode.TableCellNodeBuilder builder = TableCellNode.builder();
+        PropertiesMapper.parseTableCellProperties(properties, builder);
+        return builder.terminal(terminal).tableOfContents(context.tableOfContents()).build();
+    }
+
+
+    private TableNode buildTable(Context context, Map<String, String> properties) {
+
+        TableNode.TableNodeBuilder builder = TableNode.builder();
+        PropertiesMapper.parseTableProperties(properties, builder);
+        return TableNode.builder().tableOfContents(context.tableOfContents()).build();
+    }
+
+
+    private FooterNode buildFooter(Context context, boolean terminal) {
+
+        return FooterNode.builder().terminal(terminal).tableOfContents(context.tableOfContents()).build();
+    }
+
+
+    private HeaderNode buildHeader(Context context, boolean terminal) {
+
+        return HeaderNode.builder().terminal(terminal).tableOfContents(context.tableOfContents()).build();
+    }
+
+
+    private SectionNode buildSection(Context context) {
+
+        return SectionNode.builder().tableOfContents(context.tableOfContents()).build();
+
+    }
+
+
+    private ParagraphNode buildParagraph(Context context, boolean terminal) {
+
+        return ParagraphNode.builder().terminal(terminal).tableOfContents(context.tableOfContents()).build();
+    }
+
+
+    private TextBlock toTextBlock(Long[] atomicTextBlockIds, Context context, SemanticNode parent) {
+
+        return Arrays.stream(atomicTextBlockIds)
+                .map(atomicTextBlockId -> toAtomicTextBlock(context.atomicTextBlockData.get(Math.toIntExact(atomicTextBlockId)),
+                        context.atomicPositionBlockData.get(Math.toIntExact(atomicTextBlockId)),
+                        parent,
+                        context))
+                .collect(new TextBlockCollector());
+    }
+
+
+    private PageNode buildPage(PageData p) {
+
+        return PageNode.builder().rotation(p.getRotation()).height(p.getHeight()).width(p.getWidth()).number(p.getNumber()).mainBody(new LinkedList<>()).build();
+    }
+
+
+    private AtomicTextBlock toAtomicTextBlock(AtomicTextBlockData atomicTextBlockData,
+                                              AtomicPositionBlockData atomicPositionBlockData,
+                                              SemanticNode parent,
+                                              Context context) {
+
+        return AtomicTextBlock.builder()
+                .id(atomicTextBlockData.getId())
+                .numberOnPage(atomicTextBlockData.getNumberOnPage())
+                .page(getPage(atomicTextBlockData.getPage(), context))
+                .boundary(new Boundary(atomicTextBlockData.getStart(), atomicTextBlockData.getEnd()))
+                .searchText(atomicTextBlockData.getSearchText())
+                .lineBreaks(Ints.asList(atomicTextBlockData.getLineBreaks()))
+                .stringIdxToPositionIdx(Ints.asList(atomicPositionBlockData.getStringIdxToPositionIdx()))
+                .positions(toRectangle2DList(atomicPositionBlockData.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();
+    }
+
+
+    private PageNode getPage(Long pageIndex, Context context) {
+
+        return context.pages.stream()
+                .filter(page -> page.getNumber() == Math.toIntExact(pageIndex))
+                .findFirst()
+                .orElseThrow(() -> new NoSuchElementException(String.format("ClassificationPage with number %d not found", pageIndex)));
+    }
+
+
+    record Context(
+            DocumentData layoutParsingModel,
+            TableOfContents tableOfContents,
+            List<PageNode> pages,
+            List<SectionNode> sections,
+            List<AtomicTextBlockData> atomicTextBlockData,
+            List<AtomicPositionBlockData> atomicPositionBlockData) {
+
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/mapper/PropertiesMapper.java

@@ -0,0 +1,101 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.mapper;
+
+import java.awt.geom.Rectangle2D;
+import java.util.Arrays;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.ImageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.ImageType;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.TableCellNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.TableNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.services.RectangleTransformations;
+
+public class PropertiesMapper {
+
+    public static Map<String, String> buildImageProperties(ImageNode image) {
+
+        Map<String, String> properties = new HashMap<>();
+        properties.put("imageType", image.getImageType().toString());
+        properties.put("transparency", String.valueOf(image.isTransparency()));
+        properties.put("position", RectangleTransformations.toString(image.getPosition()));
+        return properties;
+    }
+
+
+    public static Map<String, String> buildTableCellProperties(TableCellNode tableCell) {
+
+        Map<String, String> properties = new HashMap<>();
+        properties.put("row", String.valueOf(tableCell.getRow()));
+        properties.put("col", String.valueOf(tableCell.getCol()));
+        properties.put("header", String.valueOf(tableCell.isHeader()));
+
+        if (tableCell.getPages().size() > 1 || tableCell.getBBox().keySet().size() > 1) {
+            throw new IllegalArgumentException("TableCell can only occur on a single page!");
+        }
+        String bBoxString = RectangleTransformations.toString(tableCell.getBBox().get(tableCell.getPages().stream().findFirst().get()));
+        properties.put("bBox", bBoxString);
+
+        return properties;
+    }
+
+
+    public static Map<String, String> buildTableProperties(TableNode table) {
+
+        Map<String, String> properties = new HashMap<>();
+        properties.put("numberOfRows", String.valueOf(table.getNumberOfRows()));
+        properties.put("numberOfCols", String.valueOf(table.getNumberOfCols()));
+        return properties;
+    }
+
+
+    public static void parseImageProperties(Map<String, String> properties, ImageNode.ImageNodeBuilder builder) {
+
+        builder.imageType(parseImageType(properties.get("imageType")));
+        builder.transparency(Boolean.parseBoolean(properties.get("transparency")));
+        builder.position(parseRectangle2D(properties.get("position")));
+    }
+
+
+    public static void parseTableCellProperties(Map<String, String> properties, TableCellNode.TableCellNodeBuilder builder) {
+
+        builder.row(Integer.parseInt(properties.get("row")));
+        builder.col(Integer.parseInt(properties.get("col")));
+        builder.header(Boolean.parseBoolean(properties.get("header")));
+        builder.bBox(parseRectangle2D(properties.get("bBox")));
+    }
+
+
+    public static void parseTableProperties(Map<String, String> properties, TableNode.TableNodeBuilder builder) {
+
+        builder.numberOfRows(Integer.parseInt(properties.get("numberOfRows")));
+        builder.numberOfCols(Integer.parseInt(properties.get("numberOfCols")));
+    }
+
+
+    private static ImageType parseImageType(String imageType) {
+
+        return switch (imageType) {
+            case "LOGO" -> ImageType.LOGO;
+            case "FORMULA" -> ImageType.FORMULA;
+            case "SIGNATURE" -> ImageType.SIGNATURE;
+            case "OCR" -> ImageType.OCR;
+            default -> ImageType.OTHER;
+        };
+    }
+
+
+    public static String toString(Rectangle2D rectangle2D) {
+
+        return String.format("%f,%f,%f,%f", rectangle2D.getX(), rectangle2D.getY(), rectangle2D.getWidth(), rectangle2D.getHeight());
+    }
+
+
+    public static Rectangle2D parseRectangle2D(String bBox) {
+
+        List<Float> floats = Arrays.stream(bBox.split(",")).map(Float::parseFloat).toList();
+        return new Rectangle2D.Float(floats.get(0), floats.get(1), floats.get(2), floats.get(3));
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/services/EntityEnrichmentService.java

@@ -0,0 +1,10 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.services;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+
+public interface EntityEnrichmentService {
+
+    void enrichEntity(EntityNode entity, TextBlock textBlock);
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/services/EntityInsertionService.java

@@ -0,0 +1,65 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.services;
+
+import java.util.NoSuchElementException;
+import java.util.Set;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.Boundary;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.entity.EntityNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SemanticNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+
+import lombok.RequiredArgsConstructor;
+
+@RequiredArgsConstructor
+public class EntityInsertionService {
+
+    private final EntityEnrichmentService entityEnrichmentService;
+
+
+    public void addEntityToGraph(EntityNode entity, TableOfContents tableOfContents) {
+
+        try {
+            SemanticNode containingNode = tableOfContents.getEntries()
+                    .stream()
+                    .map(TableOfContents.Entry::node)
+                    .filter(node -> node.buildTextBlock().containsBoundary(entity.getBoundary()))
+                    .findFirst()
+                    .orElseThrow(() -> new NoSuchElementException("No containing Node found!"));
+
+            containingNode.addThisToEntityIfIntersects(entity);
+
+            TextBlock textBlock = entity.getDeepestFullyContainingNode().buildTextBlock();
+            entityEnrichmentService.enrichEntity(entity, textBlock);
+
+            addToPages(entity);
+            addToNodeEntitySets(entity);
+
+        } catch (NoSuchElementException e) {
+            entityEnrichmentService.enrichEntity(entity, tableOfContents.buildTextBlock());
+            entity.removeFromGraph();
+        }
+    }
+
+
+    private void addToPages(EntityNode entity) {
+
+        Set<PageNode> pages = entity.getDeepestFullyContainingNode().getPages();
+        entity.getPages().addAll(pages);
+        pages.forEach(page -> page.getEntities().add(entity));
+    }
+
+
+    private void addToNodeEntitySets(EntityNode entity) {
+
+        entity.getIntersectingNodes().forEach(node -> node.getEntities().add(entity));
+    }
+
+
+    private static Boundary toLineAfterBoundary(TextBlock textBlock, Boundary boundary) {
+
+        return new Boundary(boundary.end(), textBlock.getNextLinebreak(boundary.end()));
+    }
+
+}

layoutparser-service/layoutparser-service-internal-api/src/main/java/com.knecon.fforesight.service.layoutparser.internal.api/services/RectangleTransformations.java

@@ -0,0 +1,95 @@
+package com.knecon.fforesight.service.layoutparser.internal.api.services;
+
+import static java.lang.String.format;
+
+import java.awt.geom.Area;
+import java.awt.geom.Rectangle2D;
+import java.util.Arrays;
+import java.util.List;
+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 com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.AtomicTextBlock;
+
+import lombok.experimental.UtilityClass;
+
+@UtilityClass
+public class RectangleTransformations {
+
+    public static Rectangle2D pad(Rectangle2D rectangle2D, int deltaX, int deltaY) {
+
+        return new Rectangle2D.Double(rectangle2D.getMinX() - deltaX, rectangle2D.getMinY() - deltaY, rectangle2D.getWidth() + 2 * deltaX, rectangle2D.getHeight() + 2 * deltaY);
+    }
+
+
+    public static Rectangle2D bBoxUnionAtomicTextBlock(List<AtomicTextBlock> atomicTextBlocks) {
+
+        return atomicTextBlocks.stream().flatMap(atomicTextBlock -> atomicTextBlock.getPositions().stream()).collect(new Rectangle2DUnion());
+    }
+
+
+    public static Rectangle2D rectangleUnion(List<Rectangle2D> rectangle2DList) {
+
+        return rectangle2DList.stream().collect(new Rectangle2DUnion());
+    }
+
+
+    public static String toString(Rectangle2D rectangle2D) {
+
+        return format("%f,%f,%f,%f", rectangle2D.getX(), rectangle2D.getY(), rectangle2D.getWidth(), rectangle2D.getHeight());
+    }
+
+
+    public static Rectangle2D parseRectangle2D(String bBox) {
+
+        List<Float> floats = Arrays.stream(bBox.split(",")).map(Float::parseFloat).toList();
+        return new Rectangle2D.Float(floats.get(0), floats.get(1), floats.get(2), floats.get(3));
+    }
+
+
+    private static class Rectangle2DUnion implements Collector<Rectangle2D, Area, Rectangle2D> {
+
+        @Override
+        public Supplier<Area> supplier() {
+
+            return Area::new;
+        }
+
+
+        @Override
+        public BiConsumer<Area, Rectangle2D> accumulator() {
+
+            return (area, rectangle2D) -> area.add(new Area(rectangle2D));
+        }
+
+
+        @Override
+        public BinaryOperator<Area> combiner() {
+
+            return (area1, area2) -> {
+                area1.add(area2);
+                return area1;
+            };
+        }
+
+
+        @Override
+        public Function<Area, Rectangle2D> finisher() {
+
+            return Area::getBounds2D;
+        }
+
+
+        @Override
+        public Set<Characteristics> characteristics() {
+
+            return Set.of(Characteristics.CONCURRENT, Characteristics.UNORDERED);
+        }
+
+    }
+
+}

layoutparser-service/layoutparser-service-processor/pom.xml

@@ -0,0 +1,161 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+         xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 https://maven.apache.org/xsd/maven-4.0.0.xsd">
+    <modelVersion>4.0.0</modelVersion>
+
+    <parent>
+        <groupId>com.knecon.fforesight</groupId>
+        <artifactId>layoutparser-service</artifactId>
+        <version>1.0.0</version>
+    </parent>
+
+    <artifactId>layoutparser-service-processor</artifactId>
+    <version>1.0.0</version>
+
+    <dependencies>
+
+        <dependency>
+            <groupId>com.iqser.red.service</groupId>
+            <artifactId>persistence-service-internal-api-v1</artifactId>
+            <version>2.36.0</version>
+        </dependency>
+
+        <dependency>
+            <groupId>com.knecon.fforesight</groupId>
+            <artifactId>layoutparser-service-internal-api</artifactId>
+            <version>${project.version}</version>
+        </dependency>
+
+        <dependency>
+            <groupId>com.iqser.red.commons</groupId>
+            <artifactId>spring-commons</artifactId>
+            <version>6.2.0</version>
+        </dependency>
+
+        <dependency>
+            <groupId>com.iqser.red.commons</groupId>
+            <artifactId>storage-commons</artifactId>
+            <version>1.13.0</version>
+        </dependency>
+
+        <dependency>
+            <groupId>com.dslplatform</groupId>
+            <artifactId>dsl-json-java8</artifactId>
+            <version>1.10.0</version>
+        </dependency>
+
+        <dependency>
+            <groupId>org.apache.pdfbox</groupId>
+            <artifactId>pdfbox</artifactId>
+            <version>${pdfbox.version}</version>
+        </dependency>
+
+        <dependency>
+            <groupId>org.apache.pdfbox</groupId>
+            <artifactId>pdfbox-tools</artifactId>
+            <version>${pdfbox.version}</version>
+        </dependency>
+
+        <dependency>
+            <groupId>com.google.guava</groupId>
+            <artifactId>guava</artifactId>
+            <version>31.1-jre</version>
+        </dependency>
+
+        <dependency>
+            <groupId>com.fasterxml.jackson.module</groupId>
+            <artifactId>jackson-module-afterburner</artifactId>
+            <version>${jackson.version}</version>
+        </dependency>
+
+        <dependency>
+            <groupId>com.fasterxml.jackson.datatype</groupId>
+            <artifactId>jackson-datatype-jsr310</artifactId>
+            <version>${jackson.version}</version>
+        </dependency>
+
+        <dependency>
+            <groupId>org.springframework.boot</groupId>
+            <artifactId>spring-boot-starter-security</artifactId>
+            <version>${spring.version}</version>
+        </dependency>
+
+        <dependency>
+            <groupId>org.springframework.boot</groupId>
+            <artifactId>spring-boot-starter-web</artifactId>
+            <version>${spring.version}</version>
+        </dependency>
+
+        <dependency>
+            <groupId>org.projectlombok</groupId>
+            <artifactId>lombok</artifactId>
+            <optional>true</optional>
+        </dependency>
+
+        <dependency>
+            <groupId>org.springframework.cloud</groupId>
+            <artifactId>spring-cloud-starter-openfeign</artifactId>
+            <version>4.0.2</version>
+        </dependency>
+
+        <dependency>
+            <groupId>org.springframework.boot</groupId>
+            <artifactId>spring-boot-starter-amqp</artifactId>
+            <version>${spring.version}</version>
+        </dependency>
+    </dependencies>
+
+    <build>
+        <plugins>
+            <plugin>
+                <groupId>org.springframework.boot</groupId>
+                <artifactId>spring-boot-maven-plugin</artifactId>
+                <configuration>
+                    <excludes>
+                        <exclude>
+                            <groupId>org.projectlombok</groupId>
+                            <artifactId>lombok</artifactId>
+                        </exclude>
+                    </excludes>
+                </configuration>
+            </plugin>
+        </plugins>
+    </build>
+    <repositories>
+        <repository>
+            <id>spring-milestones</id>
+            <name>Spring Milestones</name>
+            <url>https://repo.spring.io/milestone</url>
+            <snapshots>
+                <enabled>false</enabled>
+            </snapshots>
+        </repository>
+        <repository>
+            <id>spring-snapshots</id>
+            <name>Spring Snapshots</name>
+            <url>https://repo.spring.io/snapshot</url>
+            <releases>
+                <enabled>false</enabled>
+            </releases>
+        </repository>
+    </repositories>
+    <pluginRepositories>
+        <pluginRepository>
+            <id>spring-milestones</id>
+            <name>Spring Milestones</name>
+            <url>https://repo.spring.io/milestone</url>
+            <snapshots>
+                <enabled>false</enabled>
+            </snapshots>
+        </pluginRepository>
+        <pluginRepository>
+            <id>spring-snapshots</id>
+            <name>Spring Snapshots</name>
+            <url>https://repo.spring.io/snapshot</url>
+            <releases>
+                <enabled>false</enabled>
+            </releases>
+        </pluginRepository>
+    </pluginRepositories>
+
+</project>

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

@@ -0,0 +1,114 @@
+package com.knecon.fforesight.service.layoutparser.processor;
+
+import static java.lang.String.format;
+
+import java.io.IOException;
+
+import org.apache.pdfbox.pdmodel.PDDocument;
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.DocumentGraph;
+import com.knecon.fforesight.service.layoutparser.internal.api.mapper.DocumentDataMapper;
+import com.knecon.fforesight.service.layoutparser.processor.adapter.CvTableParsingAdapter;
+import com.knecon.fforesight.service.layoutparser.processor.adapter.ImageServiceResponseAdapter;
+import com.knecon.fforesight.service.layoutparser.processor.adapter.model.image.ImageServiceResponse;
+import com.knecon.fforesight.service.layoutparser.processor.adapter.model.table.TableServiceResponse;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationDocument;
+import com.knecon.fforesight.service.layoutparser.processor.classification.service.ClassificationService;
+import com.knecon.fforesight.service.layoutparser.processor.classification.service.PdfParsingService;
+import com.knecon.fforesight.service.layoutparser.processor.classification.service.SectionsBuilderService;
+import com.knecon.fforesight.service.layoutparser.processor.factory.DocumentGraphFactory;
+import com.knecon.fforesight.service.layoutparser.processor.queue.LayoutParsingFinishedEvent;
+import com.knecon.fforesight.service.layoutparser.processor.queue.LayoutParsingRequest;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class LayoutParsingService {
+
+    private final ImageServiceResponseAdapter imageServiceResponseAdapter;
+    private final CvTableParsingAdapter cvTableParsingAdapter;
+    private final LayoutParsingStorageService layoutParsingStorageService;
+    private final PdfParsingService pdfParsingService;
+    private final ClassificationService classificationService;
+    private final SectionsBuilderService sectionsBuilderService;
+    private final DocumentGraphFactory documentGraphFactory;
+    private final DocumentDataMapper documentDataMapper;
+
+
+    public LayoutParsingFinishedEvent parseLayoutAndSaveFilesToStorage(LayoutParsingRequest layoutParsingRequest) {
+
+        PDDocument originDocument;
+        try {
+            originDocument = layoutParsingStorageService.getOriginFile(layoutParsingRequest.getOriginFileStorageId());
+        } catch (IOException e) {
+            log.error(e.toString());
+            return LayoutParsingFinishedEvent.builder()
+                    .status(400)
+                    .message(format("Origin PDF File with id %s could not be loaded!", layoutParsingRequest.getPageFileStorageId()))
+                    .build();
+        }
+
+        ImageServiceResponse imageServiceResponse = new ImageServiceResponse();
+        if (layoutParsingRequest.getImagesFileStorageId().isPresent()) {
+            try {
+                imageServiceResponse = layoutParsingStorageService.getImagesFile(layoutParsingRequest.getPageFileStorageId());
+            } catch (IOException e) {
+                log.error(e.toString());
+                return LayoutParsingFinishedEvent.builder()
+                        .status(400)
+                        .message(format("Image Service File with id %s could not be loaded!", layoutParsingRequest.getImagesFileStorageId()))
+                        .build();
+            }
+        }
+
+        TableServiceResponse tableServiceResponse = new TableServiceResponse();
+        if (layoutParsingRequest.getTablesFileStorageId().isPresent()) {
+            try {
+                tableServiceResponse = layoutParsingStorageService.getTablesFile(layoutParsingRequest.getPageFileStorageId());
+            } catch (IOException e) {
+                log.error(e.toString());
+                return LayoutParsingFinishedEvent.builder()
+                        .status(400)
+                        .message(format("CV Table Parsing File with id %s could not be loaded!", layoutParsingRequest.getPageFileStorageId()))
+                        .build();
+            }
+        }
+
+        DocumentGraph documentGraph = parseLayout(originDocument, imageServiceResponse, tableServiceResponse);
+
+        try {
+            layoutParsingStorageService.storeDocumentData(layoutParsingRequest, documentDataMapper.toDocumentData(documentGraph));
+        } catch (IOException e) {
+            log.error("Parsed Document files could not be saved!");
+            log.error(e.getMessage());
+            return LayoutParsingFinishedEvent.builder().status(500).message("Files could not be saved").build();
+        }
+        return LayoutParsingFinishedEvent.builder()
+                .status(200)
+                .message(format("Layout parsing is finished and files have been saved with Ids:\n Structure: %s\nText: %s\nPositions: %s\nPageData: %s",
+                        layoutParsingRequest.getStructureFileStorageId(),
+                        layoutParsingRequest.getTextBlockFileStorageId(),
+                        layoutParsingRequest.getPositionBlockFileStorageId(),
+                        layoutParsingRequest.getPageFileStorageId()))
+                .build();
+    }
+
+
+    public DocumentGraph parseLayout(PDDocument originDocument, ImageServiceResponse imageServiceResponse, TableServiceResponse tableServiceResponse) {
+
+        ClassificationDocument classificationDocument = pdfParsingService.parseDocument(originDocument,
+                cvTableParsingAdapter.buildCvParsedTablesPerPage(tableServiceResponse),
+                imageServiceResponseAdapter.buildClassifiedImagesPerPage(imageServiceResponse));
+
+        classificationService.classifyDocument(classificationDocument);
+
+        sectionsBuilderService.buildSections(classificationDocument);
+
+        return documentGraphFactory.buildDocumentGraph(classificationDocument);
+    }
+
+}

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

@@ -0,0 +1,126 @@
+package com.knecon.fforesight.service.layoutparser.processor;
+
+import java.io.File;
+import java.io.FileOutputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.nio.file.Files;
+
+import org.apache.commons.io.IOUtils;
+import org.apache.pdfbox.Loader;
+import org.apache.pdfbox.io.MemoryUsageSetting;
+import org.apache.pdfbox.pdmodel.PDDocument;
+import org.springframework.stereotype.Service;
+
+import com.fasterxml.jackson.databind.ObjectMapper;
+import com.iqser.red.storage.commons.service.StorageService;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.AtomicPositionBlockData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.AtomicTextBlockData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.DocumentData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.PageData;
+import com.knecon.fforesight.service.layoutparser.internal.api.data.TableOfContentsData;
+import com.knecon.fforesight.service.layoutparser.processor.adapter.model.image.ImageServiceResponse;
+import com.knecon.fforesight.service.layoutparser.processor.adapter.model.table.TableServiceResponse;
+import com.knecon.fforesight.service.layoutparser.processor.multitenancy.TenantContext;
+import com.knecon.fforesight.service.layoutparser.processor.queue.LayoutParsingRequest;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class LayoutParsingStorageService {
+
+    private final StorageService storageService;
+    private final ObjectMapper objectMapper;
+
+
+    public PDDocument getOriginFile(String storageId) throws IOException {
+
+        try (var originDocumentInputStream = storageService.getObject(TenantContext.getTenantId(), storageId).getInputStream()) {
+            File tempFile = createTempFile("document", ".pdf");
+            try (var tempFileOutputStream = new FileOutputStream(tempFile)) {
+                IOUtils.copy(originDocumentInputStream, tempFileOutputStream);
+            }
+            return Loader.loadPDF(tempFile, MemoryUsageSetting.setupMixed(67108864L));
+        }
+    }
+
+
+    public ImageServiceResponse getImagesFile(String storageId) throws IOException {
+
+        try (InputStream inputStream = storageService.getObject(TenantContext.getTenantId(), storageId).getInputStream()) {
+
+            return objectMapper.readValue(inputStream, ImageServiceResponse.class);
+        }
+    }
+
+
+    public TableServiceResponse getTablesFile(String storageId) throws IOException {
+
+        try (var tableClassificationStream = storageService.getObject(TenantContext.getTenantId(), storageId).getInputStream()) {
+
+            return objectMapper.readValue(tableClassificationStream, TableServiceResponse.class);
+
+        }
+    }
+
+
+    public void storeDocumentData(LayoutParsingRequest layoutParsingRequest, DocumentData documentData) throws IOException {
+
+        storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.getStructureFileStorageId(), documentData.getTableOfContents());
+        storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.getTextBlockFileStorageId(), documentData.getAtomicTextBlocks());
+        storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.getPositionBlockFileStorageId(), documentData.getAtomicPositionBlocks());
+        storageService.storeJSONObject(TenantContext.getTenantId(), layoutParsingRequest.getPageFileStorageId(), documentData.getPages());
+
+    }
+
+
+    public DocumentData readDocumentData(LayoutParsingRequest layoutParsingRequest) throws IOException {
+
+        PageData[] pageData = storageService.readJSONObject(TenantContext.getTenantId(), layoutParsingRequest.getPageFileStorageId(), PageData[].class);
+        AtomicTextBlockData[] atomicTextBlockData = storageService.readJSONObject(TenantContext.getTenantId(),
+                layoutParsingRequest.getTextBlockFileStorageId(),
+                AtomicTextBlockData[].class);
+        AtomicPositionBlockData[] atomicPositionBlockData = storageService.readJSONObject(TenantContext.getTenantId(),
+                layoutParsingRequest.getPositionBlockFileStorageId(),
+                AtomicPositionBlockData[].class);
+        TableOfContentsData tableOfContentsData = storageService.readJSONObject(TenantContext.getTenantId(),
+                layoutParsingRequest.getStructureFileStorageId(),
+                TableOfContentsData.class);
+
+        return DocumentData.builder()
+                .tableOfContents(tableOfContentsData)
+                .atomicPositionBlocks(atomicPositionBlockData)
+                .atomicTextBlocks(atomicTextBlockData)
+                .pages(pageData)
+                .build();
+    }
+
+
+    private File createTempFile(String filenamePrefix, String filenameSuffix) throws IOException {
+
+        File tempFile = Files.createTempFile(filenamePrefix, filenameSuffix).toFile();
+        setRWPermissionsOnlyForOwner(tempFile);
+
+        return tempFile;
+    }
+
+
+    // We don't need to check the results of the permission setters below,
+    // since we're manipulating a file we created ourselves.
+    @SuppressWarnings({"ResultOfMethodCallIgnored", "squid:S899"})
+    private void setRWPermissionsOnlyForOwner(File tempFile) {
+
+        try {
+            tempFile.setReadable(true, true);
+            tempFile.setWritable(true, true);
+            tempFile.setExecutable(false);
+        } catch (SecurityException ex) {
+            // This should never happen since we're creating a temp file ourselves.
+            log.warn("Caught an exception during temp file creation.  This should not happend.  Check the code.", ex);
+        }
+    }
+
+}

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

@@ -0,0 +1,10 @@
+package com.knecon.fforesight.service.layoutparser.processor;
+
+import org.springframework.context.annotation.ComponentScan;
+import org.springframework.context.annotation.Configuration;
+
+@Configuration
+@ComponentScan
+public class LayoutparserServiceProcessorConfiguration {
+
+}

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

@@ -0,0 +1,49 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter;
+
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.adapter.model.table.CvParsedTableCell;
+import com.knecon.fforesight.service.layoutparser.processor.adapter.model.table.TableServiceResponse;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class CvTableParsingAdapter {
+
+    public Map<Integer, List<com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CvParsedTableCell>> buildCvParsedTablesPerPage(TableServiceResponse tableServiceResponse) {
+
+        Map<Integer, List<com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CvParsedTableCell>> tableCells = new HashMap<>();
+        tableServiceResponse.getData()
+                .forEach(tableData -> tableCells.computeIfAbsent(tableData.getPageInfo().getNumber(), tableCell -> new ArrayList<>())
+                        .addAll(convertTableCells(tableData.getTableCells())));
+
+        return tableCells;
+    }
+
+
+    private Collection<? extends com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CvParsedTableCell> convertTableCells(List<CvParsedTableCell> tableCells) {
+
+        List<com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CvParsedTableCell> cvParsedTableCells = new ArrayList<>();
+
+        tableCells.forEach(t -> cvParsedTableCells.add(com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CvParsedTableCell.builder()
+                .y0(t.getY0())
+                .x1(t.getX1())
+                .y1(t.getY1())
+                .x0(t.getX0())
+                .width(t.getWidth())
+                .height(t.getHeight())
+                .build()));
+
+        return cvParsedTableCells;
+    }
+
+}

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

@@ -0,0 +1,67 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter;
+
+import java.awt.geom.Rectangle2D;
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Locale;
+import java.util.Map;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.ImageType;
+import com.knecon.fforesight.service.layoutparser.processor.adapter.model.image.ImageServiceResponse;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.image.ClassifiedImage;
+
+import lombok.RequiredArgsConstructor;
+
+@Service
+@RequiredArgsConstructor
+public class ImageServiceResponseAdapter {
+
+
+    public Map<Integer, List<ClassifiedImage>> buildClassifiedImagesPerPage(ImageServiceResponse imageServiceResponse ) {
+
+        Map<Integer, List<ClassifiedImage>> images = new HashMap<>();
+        imageServiceResponse.getData().forEach(imageMetadata -> {
+            var classification = imageMetadata.getFilters().isAllPassed() ? ImageType.valueOf(imageMetadata.getClassification()
+                    .getLabel()
+                    .toUpperCase(Locale.ROOT)) : ImageType.OTHER;
+            images.computeIfAbsent(imageMetadata.getPosition().getPageNumber(), x -> new ArrayList<>())
+                    .add(new ClassifiedImage(new Rectangle2D.Double(imageMetadata.getPosition().getX1(),
+                            imageMetadata.getPosition().getY1(),
+                            imageMetadata.getGeometry().getWidth(),
+                            imageMetadata.getGeometry().getHeight()), classification, imageMetadata.isAlpha(), imageMetadata.getPosition().getPageNumber()));
+        });
+
+        // Currently This is a copy but, it will be changed later because i don' t think that we should unclassified images.
+        imageServiceResponse.getDataCV().forEach(imageMetadata -> {
+            var classification = imageMetadata.getFilters().isAllPassed() ? ImageType.valueOf(imageMetadata.getClassification()
+                    .getLabel()
+                    .toUpperCase(Locale.ROOT)) : ImageType.OTHER;
+            images.computeIfAbsent(imageMetadata.getPosition().getPageNumber(), x -> new ArrayList<>())
+                    .add(new ClassifiedImage(new Rectangle2D.Double(imageMetadata.getPosition().getX1(),
+                            imageMetadata.getPosition().getY1(),
+                            imageMetadata.getGeometry().getWidth(),
+                            imageMetadata.getGeometry().getHeight()), classification, imageMetadata.isAlpha(), imageMetadata.getPosition().getPageNumber()));
+        });
+
+        return images;
+    }
+
+
+    public void findOcr(ClassificationPage classificationPage) {
+
+        classificationPage.getImages().forEach(image -> {
+            if (image.getImageType().equals(ImageType.OTHER)) {
+                classificationPage.getTextBlocks().forEach(textblock -> {
+                    if (image.getPosition().contains(textblock.getMinX(), textblock.getMinY(), textblock.getWidth(), textblock.getHeight())) {
+                        image.setImageType(ImageType.OCR);
+                    }
+                });
+            }
+        });
+    }
+
+}

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

@@ -0,0 +1,17 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import java.util.HashMap;
+import java.util.Map;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class Classification {
+
+    private Map<String, Float> probabilities = new HashMap<>();
+    private String label;
+
+}

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

@@ -0,0 +1,14 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class FilterGeometry {
+
+    private ImageSize imageSize;
+    private Format imageFormat;
+
+}

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

@@ -0,0 +1,15 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class Filters {
+
+    private FilterGeometry geometry;
+    private Probability probability;
+    private boolean allPassed;
+
+}

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

@@ -0,0 +1,15 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class Format {
+
+    private float quotient;
+    private boolean tooTall;
+    private boolean tooWide;
+
+}

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

@@ -0,0 +1,14 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class Geometry {
+
+    private float width;
+    private float height;
+
+}

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

@@ -0,0 +1,33 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import com.dslplatform.json.CompiledJson;
+import com.dslplatform.json.JsonAttribute;
+import com.fasterxml.jackson.annotation.JsonAlias;
+import com.fasterxml.jackson.annotation.JsonProperty;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class ImageServiceResponse {
+
+    private String dossierId;
+    private String fileId;
+
+    @JsonProperty(value = "imageMetadata")
+    @JsonAlias("data")
+    @JsonAttribute(alternativeNames = {"imageMetadata"})
+    private List<Metadata> data = new ArrayList<>();
+
+    private List<Metadata> dataCV = new ArrayList<>();
+
+
+    @JsonProperty(value = "imageMetadata")
+    @JsonAlias("data")
+    @JsonAttribute(alternativeNames = {"imageMetadata"})
+    public void setData(List<Metadata> data) {this.data = data;}
+
+}

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

@@ -0,0 +1,15 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class ImageSize {
+
+    private float quotient;
+    private boolean tooLarge;
+    private boolean tooSmall;
+
+}

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

@@ -0,0 +1,17 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class Metadata {
+
+    private Classification classification;
+    private Position position;
+    private Geometry geometry;
+    private Filters filters;
+    private boolean alpha;
+
+}

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

@@ -0,0 +1,17 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class Position {
+
+    private float x1;
+    private float x2;
+    private float y1;
+    private float y2;
+    private int pageNumber;
+
+}

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

@@ -0,0 +1,13 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.image;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class Probability {
+
+    private boolean unconfident;
+
+}

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

@@ -0,0 +1,16 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.table;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class CvParsedPageInfo {
+
+    private int number;
+    private int rotation;
+    private float width;
+    private float height;
+
+}

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

@@ -0,0 +1,18 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.table;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class CvParsedTableCell {
+
+    private float x0;
+    private float y0;
+    private float x1;
+    private float y1;
+    private float width;
+    private float height;
+
+}

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

@@ -0,0 +1,17 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.table;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class CvParsedTableModel {
+
+    private CvParsedPageInfo pageInfo;
+    private List<CvParsedTableCell> tableCells = new ArrayList<>();
+
+}

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

@@ -0,0 +1,22 @@
+package com.knecon.fforesight.service.layoutparser.processor.adapter.model.table;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import com.dslplatform.json.CompiledJson;
+
+import lombok.Data;
+
+@Data
+@CompiledJson
+public class TableServiceResponse {
+
+    private String dossierId;
+    private String fileId;
+    private String operation;
+    private String targetFileExtension;
+    private String responseFileExtension;
+
+    private List<CvParsedTableModel> data = new ArrayList<>();
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/AbstractTextContainer.java

@@ -0,0 +1,71 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto;
+
+import java.awt.geom.Rectangle2D;
+
+import com.dslplatform.json.JsonAttribute;
+import com.fasterxml.jackson.annotation.JsonIgnore;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.TextBlockOrientation;
+
+import lombok.AllArgsConstructor;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+@Data
+@AllArgsConstructor
+@NoArgsConstructor
+public abstract class AbstractTextContainer {
+
+    protected float minX;
+    protected float maxX;
+    protected float minY;
+    protected float maxY;
+    protected String classification;
+    protected int page;
+
+    private TextBlockOrientation orientation = TextBlockOrientation.NONE;
+
+
+    public abstract String getText();
+
+
+    public boolean containsBlock(ClassificationTextBlock other) {
+
+        return this.minX <= other.getPdfMinX() && this.maxX >= other.getPdfMaxX() && this.minY >= other.getPdfMinY() && this.maxY <= other.getPdfMaxY();
+    }
+
+
+    public boolean contains(AbstractTextContainer other) {
+
+        return this.minX <= other.minX && this.maxX >= other.maxX && this.minY >= other.minY && this.maxY <= other.maxY;
+    }
+
+
+    public boolean contains(Rectangle2D other) {
+
+        return other.contains(minX, minY, getWidth(), getHeight());
+    }
+
+
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    public float getHeight() {
+
+        return maxY - minY;
+    }
+
+
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    public float getWidth() {
+
+        return maxX - minX;
+    }
+
+
+    public boolean intersectsY(AbstractTextContainer atc) {
+
+        return this.minY <= atc.getMaxY() && this.maxY >= atc.getMinY();
+    }
+
+}

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

@@ -0,0 +1,27 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.StringFrequencyCounter;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.UnclassifiedText;
+
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+@Data
+@NoArgsConstructor
+public class ClassificationDocument {
+
+    private List<ClassificationPage> pages = new ArrayList<>();
+    private List<ClassificationSection> sections = new ArrayList<>();
+    private List<ClassificationHeader> headers = new ArrayList<>();
+    private List<ClassificationFooter> footers = new ArrayList<>();
+    private List<UnclassifiedText> unclassifiedTexts = new ArrayList<>();
+    private FloatFrequencyCounter textHeightCounter = new FloatFrequencyCounter();
+    private FloatFrequencyCounter fontSizeCounter = new FloatFrequencyCounter();
+    private StringFrequencyCounter fontCounter = new StringFrequencyCounter();
+    private StringFrequencyCounter fontStyleCounter = new StringFrequencyCounter();
+    private boolean headlines;
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/ClassificationFooter.java

@@ -0,0 +1,16 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto;
+
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+
+import lombok.AllArgsConstructor;
+import lombok.Data;
+
+@Data
+@AllArgsConstructor
+public class ClassificationFooter {
+
+    private List<ClassificationTextBlock> textBlocks;
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/ClassificationHeader.java

@@ -0,0 +1,16 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto;
+
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+
+import lombok.AllArgsConstructor;
+import lombok.Data;
+
+@Data
+@AllArgsConstructor
+public class ClassificationHeader {
+
+    private List<ClassificationTextBlock> textBlocks;
+
+}

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

@@ -0,0 +1,38 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.StringFrequencyCounter;
+
+import lombok.Data;
+import lombok.NonNull;
+import lombok.RequiredArgsConstructor;
+
+@Data
+@RequiredArgsConstructor
+public class ClassificationPage {
+
+    @NonNull
+    private List<AbstractTextContainer> textBlocks;
+
+    private List<ClassifiedImage> images = new ArrayList<>();
+
+    private Rectangle bodyTextFrame;
+
+    private boolean landscape;
+    private int rotation;
+
+    private int pageNumber;
+
+    private FloatFrequencyCounter textHeightCounter = new FloatFrequencyCounter();
+    private FloatFrequencyCounter fontSizeCounter = new FloatFrequencyCounter();
+    private StringFrequencyCounter fontCounter = new StringFrequencyCounter();
+    private StringFrequencyCounter fontStyleCounter = new StringFrequencyCounter();
+
+    private float pageWidth;
+    private float pageHeight;
+
+}

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

@@ -0,0 +1,38 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Table;
+
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+@Data
+@NoArgsConstructor
+public class ClassificationSection implements Comparable {
+
+    private List<AbstractTextContainer> pageBlocks = new ArrayList<>();
+    private List<ClassifiedImage> images = new ArrayList<>();
+    private String headline;
+
+
+    public List<Table> getTables() {
+
+        List<Table> tables = new ArrayList<>();
+        pageBlocks.forEach(block -> {
+            if (block instanceof Table) {
+                tables.add((Table) block);
+            }
+        });
+        return tables;
+    }
+
+
+    @Override
+    public int compareTo(Object o) {
+
+        return 0;
+    }
+}

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

@@ -0,0 +1,77 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto;
+
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.stream.Collectors;
+
+import lombok.Getter;
+
+public class FloatFrequencyCounter {
+
+    @Getter
+    Map<Float, Integer> countPerValue = new HashMap<>();
+
+
+    public void add(float value) {
+
+        if (!countPerValue.containsKey(value)) {
+            countPerValue.put(value, 1);
+        } else {
+            countPerValue.put(value, countPerValue.get(value) + 1);
+        }
+    }
+
+
+    public void addAll(Map<Float, Integer> otherCounter) {
+
+        for (Map.Entry<Float, Integer> entry : otherCounter.entrySet()) {
+            if (countPerValue.containsKey(entry.getKey())) {
+                countPerValue.put(entry.getKey(), countPerValue.get(entry.getKey()) + entry.getValue());
+            } else {
+                countPerValue.put(entry.getKey(), entry.getValue());
+            }
+        }
+    }
+
+
+    public Float getMostPopular() {
+
+        Map.Entry<Float, Integer> mostPopular = null;
+        for (Map.Entry<Float, Integer> entry : countPerValue.entrySet()) {
+            if (mostPopular == null || entry.getValue() >= mostPopular.getValue()) {
+                mostPopular = entry;
+            }
+        }
+        return mostPopular != null ? mostPopular.getKey() : null;
+    }
+
+
+    public List<Float> getHighterThanMostPopular() {
+
+        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());
+    }
+
+
+    public Float getHighest() {
+
+        Float highest = null;
+        for (Float value : countPerValue.keySet()) {
+            if (highest == null || value > highest) {
+                highest = value;
+            }
+        }
+        return highest;
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/Rectangle.java

@@ -0,0 +1,218 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto;
+
+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/classification/dto/image/ClassifiedImage.java

@@ -0,0 +1,25 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.image;
+
+import java.awt.geom.Rectangle2D;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.ImageType;
+
+import lombok.Data;
+import lombok.NonNull;
+import lombok.RequiredArgsConstructor;
+
+@Data
+@RequiredArgsConstructor
+public class ClassifiedImage {
+
+    @NonNull
+    private Rectangle2D position;
+    @NonNull
+    private ImageType imageType;
+    private boolean isAppendedToSection;
+    @NonNull
+    private boolean hasTransparency;
+    @NonNull
+    private int page;
+
+}

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

@@ -0,0 +1,17 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.table;
+
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Ruling;
+
+import lombok.Builder;
+import lombok.Data;
+
+@Data
+@Builder
+public class CleanRulings {
+
+    List<Ruling> horizontal;
+    List<Ruling> vertical;
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/table/CvParsedTableCell.java

@@ -0,0 +1,21 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.table;
+
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.Data;
+import lombok.RequiredArgsConstructor;
+
+@Data
+@Builder
+@AllArgsConstructor
+@RequiredArgsConstructor
+public class CvParsedTableCell {
+
+    private float x0;
+    private float y0;
+    private float x1;
+    private float y1;
+    private float width;
+    private float height;
+
+}

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

@@ -0,0 +1,437 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.table;
+
+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.classification.utils.CohenSutherlandClipping;
+import com.knecon.fforesight.service.layoutparser.processor.classification.utils.DoubleComparisons;
+
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@SuppressWarnings("all")
+public class Ruling extends Line2D.Float {
+
+    private static int PERPENDICULAR_PIXEL_EXPAND_AMOUNT = 2;
+
+
+    public Ruling(Point2D p1, Point2D p2) {
+
+        super(p1, p2);
+    }
+
+
+    public static List<Ruling> cropRulingsToArea(List<Ruling> rulings, Rectangle2D area) {
+
+        ArrayList<Ruling> rv = new ArrayList<>();
+        for (Ruling r : rulings) {
+            if (r.intersects(area)) {
+                rv.add(r.intersect(area));
+            }
+        }
+        return rv;
+    }
+
+
+    // 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;
+            }
+
+        }
+
+        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_PIXEL_EXPAND_AMOUNT, h));
+            sos.add(new SortObject(SOType.HRIGHT, h.getRight() + PERPENDICULAR_PIXEL_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_PIXEL_EXPAND_AMOUNT), so.ruling.expand(PERPENDICULAR_PIXEL_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;
+
+    }
+
+
+    public boolean vertical() {
+
+        return this.length() > 0 && DoubleComparisons.feq(this.x1, this.x2); //diff < ORIENTATION_CHECK_THRESHOLD;
+    }
+
+
+    public boolean horizontal() {
+
+        return this.length() > 0 && DoubleComparisons.feq(this.y1, this.y2); //diff < ORIENTATION_CHECK_THRESHOLD;
+    }
+
+    // attributes that make sense only for non-oblique lines
+    // these are used to have a single collapse method (in page, currently)
+
+
+    public boolean oblique() {
+
+        return !(this.vertical() || this.horizontal());
+    }
+
+
+    public float getPosition() {
+
+        if (this.oblique()) {
+            throw new UnsupportedOperationException();
+        }
+        return this.vertical() ? this.getLeft() : this.getTop();
+    }
+
+
+    public float getStart() {
+
+        if (this.oblique()) {
+            throw new UnsupportedOperationException();
+        }
+        return this.vertical() ? this.getTop() : this.getLeft();
+    }
+
+
+    public void setStart(float v) {
+
+        if (this.oblique()) {
+            throw new UnsupportedOperationException();
+        }
+        if (this.vertical()) {
+            this.setTop(v);
+        } else {
+            this.setLeft(v);
+        }
+    }
+
+
+    public float getEnd() {
+
+        if (this.oblique()) {
+            throw new UnsupportedOperationException();
+        }
+        return this.vertical() ? this.getBottom() : this.getRight();
+    }
+
+
+    public void setEnd(float v) {
+
+        if (this.oblique()) {
+            throw new UnsupportedOperationException();
+        }
+        if (this.vertical()) {
+            this.setBottom(v);
+        } else {
+            this.setRight(v);
+        }
+    }
+
+
+    public void setStartEnd(float start, float end) {
+
+        if (this.oblique()) {
+            throw new UnsupportedOperationException();
+        }
+        if (this.vertical()) {
+            this.setTop(start);
+            this.setBottom(end);
+        } else {
+            this.setLeft(start);
+            this.setRight(end);
+        }
+    }
+
+
+    public boolean perpendicularTo(Ruling other) {
+
+        return this.vertical() == other.horizontal();
+    }
+
+
+    public boolean nearlyIntersects(Ruling another, int colinearOrParallelExpandAmount) {
+
+        if (this.intersectsLine(another)) {
+            return true;
+        }
+
+        boolean rv = false;
+
+        if (this.perpendicularTo(another)) {
+            rv = this.expand(PERPENDICULAR_PIXEL_EXPAND_AMOUNT).intersectsLine(another);
+        } else {
+            rv = this.expand(colinearOrParallelExpandAmount).intersectsLine(another.expand(colinearOrParallelExpandAmount));
+        }
+
+        return rv;
+    }
+
+
+    public double length() {
+
+        return Math.sqrt(Math.pow(this.x1 - this.x2, 2) + Math.pow(this.y1 - this.y2, 2));
+    }
+
+
+    public Ruling intersect(Rectangle2D clip) {
+
+        Float clipee = (Float) this.clone();
+        boolean clipped = new CohenSutherlandClipping(clip).clip(clipee);
+
+        if (clipped) {
+            return new Ruling(clipee.getP1(), clipee.getP2());
+        } else {
+            return this;
+        }
+    }
+
+
+    public Ruling expand(float amount) {
+
+        Ruling r = (Ruling) this.clone();
+        try {
+            r.setStart(this.getStart() - amount);
+            r.setEnd(this.getEnd() + amount);
+        } catch (UnsupportedOperationException e) {
+            log.warn("Could not expand ruling!");
+        }
+        return r;
+    }
+
+
+    public Point2D intersectionPoint(Ruling other) {
+
+        Ruling this_l = this.expand(PERPENDICULAR_PIXEL_EXPAND_AMOUNT);
+        Ruling other_l = other.expand(PERPENDICULAR_PIXEL_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) {
+
+        if (this == other) {
+            return true;
+        }
+
+        if (!(other instanceof Ruling)) {
+            return false;
+        }
+
+        Ruling o = (Ruling) other;
+        return this.getP1().equals(o.getP1()) && this.getP2().equals(o.getP2());
+    }
+
+
+    @Override
+    public int hashCode() {
+
+        return super.hashCode();
+    }
+
+
+    public float getTop() {
+
+        return this.y1;
+    }
+
+
+    public void setTop(float v) {
+
+        setLine(this.getLeft(), v, this.getRight(), this.getBottom());
+    }
+
+
+    public float getLeft() {
+
+        return this.x1;
+    }
+
+
+    public void setLeft(float v) {
+
+        setLine(v, this.getTop(), this.getRight(), this.getBottom());
+    }
+
+
+    public float getBottom() {
+
+        return this.y2;
+    }
+
+
+    public void setBottom(float v) {
+
+        setLine(this.getLeft(), this.getTop(), this.getRight(), v);
+    }
+
+
+    public float getRight() {
+
+        return this.x2;
+    }
+
+
+    public void setRight(float v) {
+
+        setLine(this.getLeft(), this.getTop(), v, this.getBottom());
+    }
+
+
+    public float getWidth() {
+
+        return this.getRight() - this.getLeft();
+    }
+
+
+    public float getHeight() {
+
+        return this.getBottom() - this.getTop();
+    }
+
+
+    public double getAngle() {
+
+        double angle = Math.toDegrees(Math.atan2(this.getP2().getY() - this.getP1().getY(), this.getP2().getX() - this.getP1().getX()));
+
+        if (angle < 0) {
+            angle += 360;
+        }
+        return angle;
+    }
+
+
+    @Override
+    public String toString() {
+
+        StringBuilder sb = new StringBuilder();
+        Formatter formatter = new Formatter(sb);
+        String rv = formatter.format("%s[minX=%f minY=%f maxX=%f maxY=%f]", this.getClass().toString(), this.x1, this.y1, this.x2, this.y2).toString();
+        formatter.close();
+        return rv;
+    }
+
+
+    private enum SOType {
+        VERTICAL,
+        HRIGHT,
+        HLEFT
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/table/Table.java

@@ -0,0 +1,350 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.table;
+
+import java.awt.geom.Point2D;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+import java.util.TreeMap;
+import java.util.stream.Collectors;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.Rectangle;
+
+import lombok.Getter;
+import lombok.Setter;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+public class Table extends AbstractTextContainer {
+
+    private final TreeMap<TableCellPosition, TableCell> cells = new TreeMap<>();
+
+    private final int rotation;
+    @Getter
+    @Setter
+    private String headline;
+    private int unrotatedRowCount;
+    private int unrotatedColCount;
+    private int rowCount = -1;
+    private int colCount = -1;
+    private List<List<TableCell>> rows;
+
+
+    public Table(List<TableCell> cells, Rectangle area, int rotation) {
+
+        addCells(cells);
+        minX = area.getLeft();
+        minY = area.getBottom();
+        maxX = area.getRight();
+        maxY = area.getTop();
+        classification = "Table";
+        this.rotation = rotation;
+
+    }
+
+
+    public List<List<TableCell>> getRows() {
+
+        if (rows == null) {
+            rows = computeRows();
+
+            // Ignore rows that does not contain any cells and values.
+            List<List<TableCell>> rowsToRemove = new ArrayList<>();
+            for (List<TableCell> row : rows) {
+                if (row.size() == 1 && row.get(0).getTextBlocks().isEmpty()) {
+                    rowsToRemove.add(row);
+                }
+            }
+            rows.removeAll(rowsToRemove);
+
+            computeHeaders();
+        }
+
+        return rows;
+
+    }
+
+
+    public int getRowCount() {
+
+        if (rowCount == -1) {
+            rowCount = getRows().size();
+        }
+        return rowCount;
+    }
+
+
+    public int getColCount() {
+
+        if (colCount == -1) {
+            colCount = getRows().stream().mapToInt(List::size).max().orElse(0);
+        }
+        return colCount;
+
+    }
+
+
+    /**
+     * Detect header cells (either first row or first column):
+     * Column is marked as header if cell text is bold and row cell text is not bold.
+     * Defaults to row.
+     */
+    private void computeHeaders() {
+
+        if (rows == null) {
+            rows = computeRows();
+        }
+        // A bold cell is a header cell as long as every cell to the left/top is bold, too
+        // we move from left to right and top to bottom
+        for (int rowIndex = 0; rowIndex < rows.size(); rowIndex++) {
+            List<TableCell> rowCells = rows.get(rowIndex);
+            if (rowCells.size() == 1) {
+                continue;
+            }
+
+            for (int colIndex = 0; colIndex < rowCells.size(); colIndex++) {
+                TableCell cell = rowCells.get(colIndex);
+                List<TableCell> cellsToTheLeft = rowCells.subList(0, colIndex);
+                TableCell lastHeaderCell = null;
+                for (TableCell leftCell : cellsToTheLeft) {
+                    if (leftCell.isHeaderCell()) {
+                        lastHeaderCell = leftCell;
+                    } else {
+                        break;
+                    }
+                }
+                if (lastHeaderCell != null) {
+                    cell.getHeaderCells().add(lastHeaderCell);
+                }
+                List<TableCell> cellsToTheTop = new ArrayList<>();
+                for (int i = 0; i < rowIndex; i++) {
+                    try {
+                        cellsToTheTop.add(rows.get(i).get(colIndex));
+                    } catch (IndexOutOfBoundsException e) {
+                        log.debug("No cell {} in row {}, ignoring.", colIndex, rowIndex);
+                    }
+                }
+                for (TableCell topCell : cellsToTheTop) {
+                    if (topCell.isHeaderCell()) {
+                        lastHeaderCell = topCell;
+                    } else {
+                        break;
+                    }
+                }
+                if (lastHeaderCell != null) {
+                    cell.getHeaderCells().add(lastHeaderCell);
+                }
+                if (!cell.getTextBlocks().isEmpty() && cell.getTextBlocks().get(0).getMostPopularWordStyle().equals("bold")) {
+                    cell.setHeaderCell(true);
+                }
+            }
+        }
+
+    }
+
+
+    private List<List<TableCell>> computeRows() {
+
+        List<List<TableCell>> rows = new ArrayList<>();
+        if (rotation == 90) {
+            for (int i = 0; i < unrotatedColCount; i++) { // rows
+                List<TableCell> lastRow = new ArrayList<>();
+                for (int j = unrotatedRowCount - 1; j >= 0; j--) { // cols
+                    TableCell cell = cells.get(new TableCellPosition(j, i));
+                    if (cell != null) {
+                        lastRow.add(cell);
+                    }
+                }
+                rows.add(lastRow);
+            }
+        } else if (rotation == 270) {
+            for (int i = unrotatedColCount - 1; i >= 0; i--) { // rows
+                List<TableCell> lastRow = new ArrayList<>();
+                for (int j = 0; j < unrotatedRowCount; j++) { // cols
+                    TableCell cell = cells.get(new TableCellPosition(j, i));
+                    if (cell != null) {
+                        lastRow.add(cell);
+                    }
+                }
+                rows.add(lastRow);
+            }
+        } else {
+            for (int i = 0; i < unrotatedRowCount; i++) {
+                List<TableCell> lastRow = new ArrayList<>();
+                for (int j = 0; j < unrotatedColCount; j++) {
+                    TableCell cell = cells.get(new TableCellPosition(i, j)); // JAVA_8 use getOrDefault()
+                    if (cell != null) {
+                        lastRow.add(cell);
+                    }
+                }
+                rows.add(lastRow);
+            }
+        }
+
+        return rows;
+
+    }
+
+
+    private void add(TableCell chunk, int row, int col) {
+
+        unrotatedRowCount = Math.max(unrotatedRowCount, row + 1);
+        unrotatedColCount = Math.max(unrotatedColCount, col + 1);
+
+        TableCellPosition cp = new TableCellPosition(row, col);
+        cells.put(cp, chunk);
+
+    }
+
+
+    private void addCells(List<TableCell> cells) {
+
+        if (cells.isEmpty()) {
+            return;
+        }
+
+        cells.removeIf(cell -> cell.getWidth() < 1.1 || cell.getHeight() < 1.1);
+
+        List<List<TableCell>> rowsOfCells = calculateStructure(cells);
+
+        for (int i = 0; i < rowsOfCells.size(); i++) {
+            for (int j = 0; j < rowsOfCells.get(i).size(); j++) {
+                add(rowsOfCells.get(i).get(j), i, j);
+            }
+        }
+
+    }
+
+
+    /**
+     * Calculates the structure of the table. For spanning rows and columns multiple cells with the same values will be inserted.
+     *
+     * @param cells The found cells
+     * @return Table Structure
+     */
+    private List<List<TableCell>> calculateStructure(List<TableCell> cells) {
+
+        List<List<TableCell>> matrix = new ArrayList<>();
+
+        if (cells.isEmpty()) {
+            return matrix;
+        }
+
+        Set<Float> uniqueX = new HashSet<>();
+        Set<Float> 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());
+        });
+
+        var sortedUniqueX = uniqueX.stream().sorted().collect(Collectors.toList());
+        var sortedUniqueY = uniqueY.stream().sorted().collect(Collectors.toList());
+
+        Float prevY = null;
+        for (Float y : sortedUniqueY) {
+
+            List<TableCell> row = new ArrayList<>();
+
+            Float prevX = null;
+            for (Float x : sortedUniqueX) {
+
+                if (prevY != null && prevX != null) {
+                    var cell = new TableCell(new Point2D.Float(prevX, prevY), new Point2D.Float(x, y));
+
+                    var intersectionCell = cells.stream().filter(c -> cell.intersects(c) && cell.overlapRatio(c) > 0.1f).findFirst();
+                    if (intersectionCell.isPresent()) {
+                        cell.getTextBlocks().addAll(intersectionCell.get().getTextBlocks());
+                    }
+                    row.add(cell);
+                }
+                prevX = x;
+            }
+
+            if (prevY != null && prevX != null) {
+                matrix.add(row);
+            }
+            prevY = y;
+        }
+
+        Collections.reverse(matrix);
+
+        return matrix;
+    }
+
+
+    @Override
+    public String getText() {
+
+        StringBuilder sb = new StringBuilder();
+        List<List<TableCell>> rows = getRows();
+
+        int i = 0;
+        for (List<TableCell> row : rows) {
+            if (i != 0) {
+                sb.append("\n");
+            }
+            if (!row.isEmpty()) {
+                boolean firstColumn = true;
+                for (TableCell column : row) {
+                    if (!firstColumn) {
+                        sb.append(",");
+                    }
+                    if (column != null && column.getTextBlocks() != null) {
+                        boolean first = true;
+                        for (ClassificationTextBlock textBlock : column.getTextBlocks()) {
+                            if (!first) {
+                                sb.append("\n");
+                            }
+                            sb.append('\"').append(textBlock.getText().replaceAll("\"", "\\\"")).append('\"');
+                            first = false;
+                        }
+                    }
+                    firstColumn = false;
+                }
+            }
+            i++;
+        }
+
+        return sb.toString();
+    }
+
+
+    public String getTextAsHtml() {
+
+        StringBuilder sb = new StringBuilder();
+        List<List<TableCell>> rows = getRows();
+
+        sb.append("<table border=\"1\">");
+        int i = 0;
+        for (List<TableCell> row : rows) {
+            sb.append("\n<tr>");
+            if (!row.isEmpty()) {
+                for (TableCell column : row) {
+                    sb.append(i == 0 ? "\n<th>" : "\n<td>");
+                    if (column != null && column.getTextBlocks() != null) {
+                        boolean first = true;
+                        for (ClassificationTextBlock textBlock : column.getTextBlocks()) {
+                            if (!first) {
+                                sb.append("<br />");
+                            }
+                            sb.append(textBlock.getText().replaceAll("\\n", "<br />"));
+                            first = false;
+                        }
+                    }
+                    sb.append(i == 0 ? "</th>" : "</td>");
+                }
+            }
+            sb.append("</tr>");
+            i++;
+        }
+        sb.append("</table>");
+
+        return sb.toString();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/table/TableCell.java

@@ -0,0 +1,38 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.table;
+
+import java.awt.geom.Point2D;
+import java.util.ArrayList;
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.Rectangle;
+
+import lombok.Data;
+import lombok.EqualsAndHashCode;
+import lombok.NoArgsConstructor;
+
+@SuppressWarnings("serial")
+@Data
+@EqualsAndHashCode(callSuper = true)
+@NoArgsConstructor
+public class TableCell extends Rectangle {
+
+    private List<ClassificationTextBlock> textBlocks = new ArrayList<>();
+
+    private List<TableCell> headerCells = new ArrayList<>();
+
+    private boolean isHeaderCell;
+
+
+    public TableCell(Point2D topLeft, Point2D bottomRight) {
+
+        super((float) topLeft.getY(), (float) topLeft.getX(), (float) (bottomRight.getX() - topLeft.getX()), (float) (bottomRight.getY() - topLeft.getY()));
+    }
+
+
+    public void addTextBlock(ClassificationTextBlock textBlock) {
+
+        textBlocks.add(textBlock);
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/table/TableCellPosition.java

@@ -0,0 +1,22 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.table;
+
+import lombok.RequiredArgsConstructor;
+import lombok.Value;
+
+@Value
+@RequiredArgsConstructor
+public class TableCellPosition implements Comparable<TableCellPosition> {
+
+    int row;
+
+    int col;
+
+
+    @Override
+    public int compareTo(TableCellPosition other) {
+
+        int rowDiff = row - other.row;
+        return rowDiff != 0 ? rowDiff : col - other.col;
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/text/ClassificationTextBlock.java

@@ -0,0 +1,286 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.text;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.utils.TextNormalizationUtilities;
+
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.Data;
+import lombok.EqualsAndHashCode;
+import lombok.NoArgsConstructor;
+
+@EqualsAndHashCode(callSuper = true)
+@AllArgsConstructor
+@Builder
+@Data
+@NoArgsConstructor
+public class ClassificationTextBlock extends AbstractTextContainer {
+
+    @Builder.Default
+    private List<TextPositionSequence> sequences = new ArrayList<>();
+
+    private int rotation;
+
+    private int indexOnPage;
+
+    private String mostPopularWordFont;
+
+    private String mostPopularWordStyle;
+
+    private float mostPopularWordFontSize;
+
+    private float mostPopularWordHeight;
+
+    private float mostPopularWordSpaceWidth;
+
+    private float highestFontSize;
+
+    private String classification;
+
+
+    public TextDirection getDir() {
+
+        return sequences.get(0).getDir();
+    }
+
+    private float getPageHeight() {
+
+        return sequences.get(0).getPageHeight();
+    }
+
+
+    private float getPageWidth() {
+
+        return sequences.get(0).getPageWidth();
+    }
+
+
+    /**
+     * 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
+     */
+    public float getPdfMinX() {
+
+        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;
+        }
+    }
+
+    /**
+     * 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
+     */
+    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
+     */
+    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
+     */
+    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 ClassificationTextBlock(float minX, float maxX, float minY, float maxY, List<TextPositionSequence> sequences, int rotation, int indexOnPage) {
+        super();
+        this.indexOnPage = indexOnPage;
+        super.minX = minX;
+        super.maxX = maxX;
+        super.minY = minY;
+        super.maxY = maxY;
+        this.sequences = sequences;
+        this.rotation = rotation;
+    }
+
+
+    public ClassificationTextBlock union(TextPositionSequence r) {
+
+        ClassificationTextBlock union = this.copy();
+        union.add(r);
+        return union;
+    }
+
+
+    public ClassificationTextBlock union(ClassificationTextBlock r) {
+
+        ClassificationTextBlock union = this.copy();
+        union.add(r);
+        return union;
+    }
+
+
+    public void add(ClassificationTextBlock r) {
+
+        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());
+    }
+
+
+    public void add(TextPositionSequence r) {
+
+        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();
+        }
+    }
+
+
+    public ClassificationTextBlock copy() {
+
+        return new ClassificationTextBlock(minX, maxX, minY, maxY, sequences, rotation, indexOnPage);
+    }
+
+
+    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);
+    }
+
+
+    @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();
+
+    }
+
+
+    @Override
+    public String getText() {
+
+        StringBuilder sb = new StringBuilder();
+
+        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(' ');
+                }
+            }
+            sb.append(word.toString());
+            previous = word;
+        }
+
+        return TextNormalizationUtilities.removeHyphenLineBreaks(sb.toString());
+
+    }
+}

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

@@ -0,0 +1,106 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.text;
+
+import org.apache.pdfbox.text.TextPosition;
+import org.springframework.beans.BeanUtils;
+
+import com.dslplatform.json.CompiledJson;
+import com.dslplatform.json.JsonAttribute;
+import com.fasterxml.jackson.annotation.JsonIgnore;
+
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+import lombok.SneakyThrows;
+
+@Data
+@Builder
+@NoArgsConstructor
+@AllArgsConstructor
+@CompiledJson
+public class RedTextPosition {
+
+    private String textMatrix;
+    private float[] position;
+
+    @JsonIgnore
+    private int rotation;
+
+    @JsonIgnore
+    private float pageHeight;
+
+    @JsonIgnore
+    private float pageWidth;
+
+    private String unicode;
+
+    @JsonIgnore
+    private float dir;
+
+    // not used in reanalysis
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    private float widthOfSpace;
+
+    // not used in reanalysis
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    private float fontSizeInPt;
+
+    // not used in reanalysis
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    private String fontName;
+
+
+    @SneakyThrows
+    public static RedTextPosition fromTextPosition(TextPosition textPosition) {
+
+        var pos = new RedTextPosition();
+        BeanUtils.copyProperties(textPosition, pos);
+        pos.setFontName(textPosition.getFont().getName());
+
+        pos.setFontSizeInPt(textPosition.getFontSizeInPt());
+
+        pos.setTextMatrix(textPosition.getTextMatrix().toString());
+
+        var position = new float[4];
+
+        position[0] = textPosition.getXDirAdj();
+        position[1] = textPosition.getYDirAdj();
+        position[2] = textPosition.getWidthDirAdj();
+        position[3] = textPosition.getHeightDir();
+
+        pos.setPosition(position);
+        return pos;
+    }
+
+
+    @JsonIgnore
+    public float getXDirAdj() {
+
+        return position[0];
+    }
+
+
+    @JsonIgnore
+    public float getYDirAdj() {
+
+        return position[1];
+    }
+
+
+    @JsonIgnore
+    public float getWidthDirAdj() {
+
+        return position[2];
+    }
+
+
+    @JsonIgnore
+    public float getHeightDir() {
+
+        return position[3];
+    }
+
+}

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

@@ -0,0 +1,47 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.text;
+
+import java.util.HashMap;
+import java.util.Map;
+
+import lombok.Getter;
+
+public class StringFrequencyCounter {
+
+    @Getter
+    private final Map<String, Integer> countPerValue = new HashMap<>();
+
+
+    public void add(String value) {
+
+        if (!countPerValue.containsKey(value)) {
+            countPerValue.put(value, 1);
+        } else {
+            countPerValue.put(value, countPerValue.get(value) + 1);
+        }
+    }
+
+
+    public void addAll(Map<String, Integer> otherCounter) {
+
+        for (Map.Entry<String, Integer> entry : otherCounter.entrySet()) {
+            if (countPerValue.containsKey(entry.getKey())) {
+                countPerValue.put(entry.getKey(), countPerValue.get(entry.getKey()) + entry.getValue());
+            } else {
+                countPerValue.put(entry.getKey(), entry.getValue());
+            }
+        }
+    }
+
+
+    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;
+            }
+        }
+        return mostPopular != null ? mostPopular.getKey() : null;
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/dto/text/TextBlockOrientation.java

@@ -0,0 +1,8 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.text;
+
+public enum TextBlockOrientation {
+
+    NONE,
+    LEFT,
+    RIGHT
+}

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

@@ -0,0 +1,54 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.text;
+
+import com.fasterxml.jackson.annotation.JsonCreator;
+import com.fasterxml.jackson.annotation.JsonValue;
+
+import lombok.Getter;
+
+@Getter
+public enum TextDirection {
+    ZERO(0f),
+    QUARTER_CIRCLE(90f),
+    HALF_CIRCLE(180f),
+    THREE_QUARTER_CIRCLE(270f);
+
+    public static final String VALUE_STRING_SUFFIX = "°";
+
+    @JsonValue
+    private final float degrees;
+    private final float radians;
+
+
+    TextDirection(float degreeValue) {
+
+        degrees = degreeValue;
+        radians = (float) Math.toRadians(degreeValue);
+    }
+
+
+    @Override
+    public String toString() {
+
+        return degrees + VALUE_STRING_SUFFIX;
+    }
+
+
+    @com.dslplatform.json.JsonValue
+    public float jsonValue() {
+
+        return getDegrees();
+    }
+
+
+    @JsonCreator(mode = JsonCreator.Mode.DELEGATING)
+    public static TextDirection fromDegrees(float degrees) {
+
+        for (var dir : TextDirection.values()) {
+            if (degrees == dir.degrees) {
+                return dir;
+            }
+        }
+
+        throw new IllegalArgumentException(String.format("A value of %f is not supported by TextDirection", degrees));
+    }
+}

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

@@ -0,0 +1,298 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.text;
+
+import java.awt.geom.AffineTransform;
+import java.awt.geom.Point2D;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.stream.Collectors;
+
+import org.apache.pdfbox.text.TextPosition;
+
+import com.dslplatform.json.JsonAttribute;
+import com.fasterxml.jackson.annotation.JsonIgnore;
+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;
+
+
+    public TextPositionSequence(List<TextPosition> textPositions, int page) {
+
+        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();
+    }
+
+
+    @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);
+    }
+
+
+    @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
+     */
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    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
+     */
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    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.
+     */
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    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.
+     */
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    public float getMaxYDirAdj() {
+
+        return textPositions.get(0).getYDirAdj();
+
+    }
+
+
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    public float getTextHeight() {
+
+        return textPositions.get(0).getHeightDir() + HEIGHT_PADDING;
+    }
+
+
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    public float getHeight() {
+
+        return getMaxYDirAdj() - getMinYDirAdj();
+    }
+
+
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    public float getWidth() {
+
+        return getMaxXDirAdj() - getMinXDirAdj();
+    }
+
+
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    public String getFont() {
+
+        return textPositions.get(0).getFontName().toLowerCase().replaceAll(",bold", "").replaceAll(",italic", "");
+    }
+
+
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    public String getFontStyle() {
+
+        String lowercaseFontName = textPositions.get(0).getFontName().toLowerCase();
+
+        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";
+        }
+
+    }
+
+
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    public float getFontSize() {
+
+        return textPositions.get(0).getFontSizeInPt();
+    }
+
+
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    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
+     */
+    @JsonIgnore
+    @JsonAttribute(ignore = true)
+    @SneakyThrows
+    public Rectangle getRectangle() {
+
+        log.debug("ClassificationPage: '{}', 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/classification/dto/text/UnclassifiedText.java

@@ -0,0 +1,14 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.dto.text;
+
+import java.util.List;
+
+import lombok.AllArgsConstructor;
+import lombok.Data;
+
+@Data
+@AllArgsConstructor
+public class UnclassifiedText {
+
+    private List<ClassificationTextBlock> textBlocks;
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/parsing/LegacyPDFStreamEngine.java

@@ -0,0 +1,384 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package com.knecon.fforesight.service.layoutparser.processor.classification.parsing;
+
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.Map;
+import java.util.WeakHashMap;
+
+import org.apache.commons.logging.Log;
+import org.apache.commons.logging.LogFactory;
+import org.apache.fontbox.ttf.TrueTypeFont;
+import org.apache.fontbox.util.BoundingBox;
+import org.apache.pdfbox.contentstream.PDFStreamEngine;
+import org.apache.pdfbox.contentstream.operator.DrawObject;
+import org.apache.pdfbox.contentstream.operator.state.Concatenate;
+import org.apache.pdfbox.contentstream.operator.state.Restore;
+import org.apache.pdfbox.contentstream.operator.state.Save;
+import org.apache.pdfbox.contentstream.operator.state.SetGraphicsStateParameters;
+import org.apache.pdfbox.contentstream.operator.state.SetMatrix;
+import org.apache.pdfbox.contentstream.operator.text.BeginText;
+import org.apache.pdfbox.contentstream.operator.text.EndText;
+import org.apache.pdfbox.contentstream.operator.text.MoveText;
+import org.apache.pdfbox.contentstream.operator.text.MoveTextSetLeading;
+import org.apache.pdfbox.contentstream.operator.text.NextLine;
+import org.apache.pdfbox.contentstream.operator.text.SetCharSpacing;
+import org.apache.pdfbox.contentstream.operator.text.SetFontAndSize;
+import org.apache.pdfbox.contentstream.operator.text.SetTextHorizontalScaling;
+import org.apache.pdfbox.contentstream.operator.text.SetTextLeading;
+import org.apache.pdfbox.contentstream.operator.text.SetTextRenderingMode;
+import org.apache.pdfbox.contentstream.operator.text.SetTextRise;
+import org.apache.pdfbox.contentstream.operator.text.SetWordSpacing;
+import org.apache.pdfbox.contentstream.operator.text.ShowText;
+import org.apache.pdfbox.contentstream.operator.text.ShowTextAdjusted;
+import org.apache.pdfbox.contentstream.operator.text.ShowTextLine;
+import org.apache.pdfbox.contentstream.operator.text.ShowTextLineAndSpace;
+import org.apache.pdfbox.cos.COSDictionary;
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.pdmodel.common.PDRectangle;
+import org.apache.pdfbox.pdmodel.font.PDCIDFont;
+import org.apache.pdfbox.pdmodel.font.PDCIDFontType2;
+import org.apache.pdfbox.pdmodel.font.PDFont;
+import org.apache.pdfbox.pdmodel.font.PDFontDescriptor;
+import org.apache.pdfbox.pdmodel.font.PDSimpleFont;
+import org.apache.pdfbox.pdmodel.font.PDTrueTypeFont;
+import org.apache.pdfbox.pdmodel.font.PDType0Font;
+import org.apache.pdfbox.pdmodel.font.PDType3Font;
+import org.apache.pdfbox.pdmodel.font.encoding.GlyphList;
+import org.apache.pdfbox.pdmodel.graphics.state.PDGraphicsState;
+import org.apache.pdfbox.text.TextPosition;
+import org.apache.pdfbox.util.Matrix;
+import org.apache.pdfbox.util.Vector;
+
+/**
+ * LEGACY text calculations which are known to be incorrect but are depended on by PDFTextStripper.
+ * <p>
+ * This class exists only so that we don't break the code of users who have their own subclasses of
+ * PDFTextStripper. It replaces the mostly empty implementation of showGlyph() in PDFStreamEngine
+ * with a heuristic implementation which is backwards compatible.
+ * <p>
+ * DO NOT USE THIS CODE UNLESS YOU ARE WORKING WITH PDFTextStripper.
+ * THIS CODE IS DELIBERATELY INCORRECT, USE PDFStreamEngine INSTEAD.
+ */
+@SuppressWarnings({"PMD", "checkstyle:all"})
+class LegacyPDFStreamEngine extends PDFStreamEngine {
+
+    private static final Log LOG = LogFactory.getLog(LegacyPDFStreamEngine.class);
+
+    private int pageRotation;
+    private PDRectangle pageSize;
+    private Matrix translateMatrix;
+    private final GlyphList glyphList;
+    private final Map<COSDictionary, Float> fontHeightMap = new WeakHashMap<COSDictionary, Float>();
+
+
+    /**
+     * Constructor.
+     */
+    LegacyPDFStreamEngine() throws IOException {
+
+        addOperator(new BeginText());
+        addOperator(new Concatenate());
+        addOperator(new DrawObject()); // special text version
+        addOperator(new EndText());
+        addOperator(new SetGraphicsStateParameters());
+        addOperator(new Save());
+        addOperator(new Restore());
+        addOperator(new NextLine());
+        addOperator(new SetCharSpacing());
+        addOperator(new MoveText());
+        addOperator(new MoveTextSetLeading());
+        addOperator(new SetFontAndSize());
+        addOperator(new ShowText());
+        addOperator(new ShowTextAdjusted());
+        addOperator(new SetTextLeading());
+        addOperator(new SetMatrix());
+        addOperator(new SetTextRenderingMode());
+        addOperator(new SetTextRise());
+        addOperator(new SetWordSpacing());
+        addOperator(new SetTextHorizontalScaling());
+        addOperator(new ShowTextLine());
+        addOperator(new ShowTextLineAndSpace());
+
+        // load additional glyph list for Unicode mapping
+        String path = "/org/apache/pdfbox/resources/glyphlist/additional.txt";
+        InputStream input = GlyphList.class.getResourceAsStream(path);
+        glyphList = new GlyphList(GlyphList.getAdobeGlyphList(), input);
+    }
+
+
+    /**
+     * This will initialize and process the contents of the stream.
+     *
+     * @param page the page to process
+     * @throws IOException if there is an error accessing the stream.
+     */
+    @Override
+    public void processPage(PDPage page) throws IOException {
+
+        this.pageRotation = page.getRotation();
+        this.pageSize = page.getCropBox();
+
+        if (pageSize.getLowerLeftX() == 0 && pageSize.getLowerLeftY() == 0) {
+            translateMatrix = null;
+        } else {
+            // translation matrix for cropbox
+            translateMatrix = Matrix.getTranslateInstance(-pageSize.getLowerLeftX(), -pageSize.getLowerLeftY());
+        }
+        super.processPage(page);
+    }
+
+
+    /**
+     * Called when a glyph is to be processed. The heuristic calculations here were originally
+     * written by Ben Litchfield for PDFStreamEngine.
+     */
+    @Override
+    protected void showGlyph(Matrix textRenderingMatrix, PDFont font, int code,Vector displacement) throws IOException {
+        //
+        // legacy calculations which were previously in PDFStreamEngine
+        //
+        //  DO NOT USE THIS CODE UNLESS YOU ARE WORKING WITH PDFTextStripper.
+        //  THIS CODE IS DELIBERATELY INCORRECT
+        //
+
+        PDGraphicsState state = getGraphicsState();
+        Matrix ctm = state.getCurrentTransformationMatrix();
+        float fontSize = state.getTextState().getFontSize();
+        float horizontalScaling = state.getTextState().getHorizontalScaling() / 100f;
+        Matrix textMatrix = getTextMatrix();
+
+        float displacementX = displacement.getX();
+        // the sorting algorithm is based on the width of the character. As the displacement
+        // for vertical characters doesn't provide any suitable value for it, we have to 
+        // calculate our own
+        if (font.isVertical()) {
+            displacementX = font.getWidth(code) / 1000;
+            // there may be an additional scaling factor for true type fonts
+            TrueTypeFont ttf = null;
+            if (font instanceof PDTrueTypeFont) {
+                ttf = ((PDTrueTypeFont) font).getTrueTypeFont();
+            } else if (font instanceof PDType0Font) {
+                PDCIDFont cidFont = ((PDType0Font) font).getDescendantFont();
+                if (cidFont instanceof PDCIDFontType2) {
+                    ttf = ((PDCIDFontType2) cidFont).getTrueTypeFont();
+                }
+            }
+            if (ttf != null && ttf.getUnitsPerEm() != 1000) {
+                displacementX *= 1000f / ttf.getUnitsPerEm();
+            }
+        }
+
+        //
+        // legacy calculations which were previously in PDFStreamEngine
+        //
+        //  DO NOT USE THIS CODE UNLESS YOU ARE WORKING WITH PDFTextStripper.
+        //  THIS CODE IS DELIBERATELY INCORRECT
+        //
+
+        // (modified) combined displacement, this is calculated *without* taking the character
+        // spacing and word spacing into account, due to legacy code in TextStripper
+        float tx = displacementX * fontSize * horizontalScaling;
+        float ty = displacement.getY() * fontSize;
+
+        // (modified) combined displacement matrix
+        Matrix td = Matrix.getTranslateInstance(tx, ty);
+
+        // (modified) text rendering matrix
+        Matrix nextTextRenderingMatrix = td.multiply(textMatrix).multiply(ctm); // text space -> device space
+        float nextX = nextTextRenderingMatrix.getTranslateX();
+        float nextY = nextTextRenderingMatrix.getTranslateY();
+
+        // (modified) width and height calculations
+        float dxDisplay = nextX - textRenderingMatrix.getTranslateX();
+        Float fontHeight = fontHeightMap.get(font.getCOSObject());
+        if (fontHeight == null) {
+            fontHeight = computeFontHeight(font);
+            fontHeightMap.put(font.getCOSObject(), fontHeight);
+        }
+        float dyDisplay = fontHeight * textRenderingMatrix.getScalingFactorY();
+
+        //
+        // start of the original method
+        //
+
+        // Note on variable names. There are three different units being used in this code.
+        // Character sizes are given in glyph units, text locations are initially given in text
+        // units, and we want to save the data in display units. The variable names should end with
+        // Text or Disp to represent if the values are in text or disp units (no glyph units are
+        // saved).
+
+        float glyphSpaceToTextSpaceFactor = 1 / 1000f;
+        if (font instanceof PDType3Font) {
+            glyphSpaceToTextSpaceFactor = font.getFontMatrix().getScaleX();
+        }
+
+        float spaceWidthText = 0;
+        try {
+            // to avoid crash as described in PDFBOX-614, see what the space displacement should be
+            spaceWidthText = font.getSpaceWidth() * glyphSpaceToTextSpaceFactor;
+        } catch (Throwable exception) {
+            LOG.warn(exception, exception);
+        }
+
+        if (spaceWidthText == 0) {
+            spaceWidthText = font.getAverageFontWidth() * glyphSpaceToTextSpaceFactor;
+            // the average space width appears to be higher than necessary so make it smaller
+            spaceWidthText *= .80f;
+        }
+        if (spaceWidthText == 0) {
+            spaceWidthText = 1.0f; // if could not find font, use a generic value
+        }
+
+        // the space width has to be transformed into display units
+        float spaceWidthDisplay = spaceWidthText * textRenderingMatrix.getScalingFactorX();
+
+        // use our additional glyph list for Unicode mapping
+        String unicodeMapping = font.toUnicode(code, glyphList);
+
+        // when there is no Unicode mapping available, Acrobat simply coerces the character code
+        // into Unicode, so we do the same. Subclasses of PDFStreamEngine don't necessarily want
+        // this, which is why we leave it until this point in PDFTextStreamEngine.
+        if (unicodeMapping == null) {
+            if (font instanceof PDSimpleFont) {
+                char c = (char) code;
+                unicodeMapping = new String(new char[]{c});
+            } else {
+                // Acrobat doesn't seem to coerce composite font's character codes, instead it
+                // skips them. See the "allah2.pdf" TestTextStripper file.
+                return;
+            }
+        }
+
+        // adjust for cropbox if needed
+        Matrix translatedTextRenderingMatrix;
+        if (translateMatrix == null) {
+            translatedTextRenderingMatrix = textRenderingMatrix;
+        } else {
+            translatedTextRenderingMatrix = Matrix.concatenate(translateMatrix, textRenderingMatrix);
+            nextX -= pageSize.getLowerLeftX();
+            nextY -= pageSize.getLowerLeftY();
+        }
+
+        // This is a hack for unicode letter with 2 chars e.g. RA see unicodeProblem.pdf
+        if (unicodeMapping.length() == 2) {
+            processTextPosition(new TextPosition(pageRotation,
+                    pageSize.getWidth(),
+                    pageSize.getHeight(),
+                    translatedTextRenderingMatrix,
+                    nextX,
+                    nextY,
+                    Math.abs(dyDisplay),
+                    dxDisplay,
+                    Math.abs(spaceWidthDisplay),
+                    Character.toString(unicodeMapping.charAt(0)),
+                    new int[]{code},
+                    font,
+                    fontSize,
+                    (int) (fontSize * textMatrix.getScalingFactorX())));
+            processTextPosition(new TextPosition(pageRotation,
+                    pageSize.getWidth(),
+                    pageSize.getHeight(),
+                    translatedTextRenderingMatrix,
+                    nextX,
+                    nextY,
+                    Math.abs(dyDisplay),
+                    dxDisplay,
+                    Math.abs(spaceWidthDisplay),
+                    Character.toString(unicodeMapping.charAt(1)),
+                    new int[]{code},
+                    font,
+                    fontSize,
+                    (int) (fontSize * textMatrix.getScalingFactorX())));
+        } else {
+
+            processTextPosition(new TextPosition(pageRotation,
+                    pageSize.getWidth(),
+                    pageSize.getHeight(),
+                    translatedTextRenderingMatrix,
+                    nextX,
+                    nextY,
+                    Math.abs(dyDisplay),
+                    dxDisplay,
+                    Math.abs(spaceWidthDisplay),
+                    unicodeMapping,
+                    new int[]{code},
+                    font,
+                    fontSize,
+                    (int) (fontSize * textMatrix.getScalingFactorX())));
+        }
+    }
+
+
+    /**
+     * Compute the font height. Override this if you want to use own calculations.
+     *
+     * @param font the font.
+     * @return the font height.
+     * @throws IOException if there is an error while getting the font bounding box.
+     */
+    protected float computeFontHeight(PDFont font) throws IOException {
+
+        BoundingBox bbox = font.getBoundingBox();
+        if (bbox.getLowerLeftY() < Short.MIN_VALUE) {
+            // PDFBOX-2158 and PDFBOX-3130
+            // files by Salmat eSolutions / ClibPDF Library
+            bbox.setLowerLeftY(-(bbox.getLowerLeftY() + 65536));
+        }
+        // 1/2 the bbox is used as the height todo: why?
+        float glyphHeight = bbox.getHeight() / 2;
+
+        // sometimes the bbox has very high values, but CapHeight is OK
+        PDFontDescriptor fontDescriptor = font.getFontDescriptor();
+        if (fontDescriptor != null) {
+            float capHeight = fontDescriptor.getCapHeight();
+            if (Float.compare(capHeight, 0) != 0 && (capHeight < glyphHeight || Float.compare(glyphHeight, 0) == 0)) {
+                glyphHeight = capHeight;
+            }
+            // PDFBOX-3464, PDFBOX-4480, PDFBOX-4553:
+            // sometimes even CapHeight has very high value, but Ascent and Descent are ok
+            float ascent = fontDescriptor.getAscent();
+            float descent = fontDescriptor.getDescent();
+            if (capHeight > ascent && ascent > 0 && descent < 0 && ((ascent - descent) / 2 < glyphHeight || Float.compare(glyphHeight, 0) == 0)) {
+                glyphHeight = (ascent - descent) / 2;
+            }
+        }
+
+        // transformPoint from glyph space -> text space
+        float height;
+        if (font instanceof PDType3Font) {
+            height = font.getFontMatrix().transformPoint(0, glyphHeight).y;
+        } else {
+            height = glyphHeight / 1000;
+        }
+
+        return height;
+    }
+
+
+    /**
+     * A method provided as an event interface to allow a subclass to perform some specific
+     * functionality when text needs to be processed.
+     *
+     * @param text The text to be processed.
+     */
+    protected void processTextPosition(TextPosition text) {
+        // subclasses can override to provide specific functionality
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/parsing/PDFAreaTextStripper.java

@@ -0,0 +1,82 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.parsing;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+
+import org.apache.pdfbox.text.PDFTextStripperByArea;
+import org.apache.pdfbox.text.TextPosition;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.TextPositionSequence;
+
+import lombok.Getter;
+import lombok.Setter;
+
+public class PDFAreaTextStripper extends PDFTextStripperByArea {
+
+    @Getter
+    private List<TextPositionSequence> textPositionSequences = new ArrayList<>();
+
+    @Setter
+    private int pageNumber;
+
+
+    public PDFAreaTextStripper() throws IOException {
+
+    }
+
+
+    @Override
+    public void writeString(String text, List<TextPosition> textPositions) throws IOException {
+
+        int startIndex = 0;
+        for (int i = 0; i <= textPositions.size() - 1; i++) {
+
+            if (i == 0 && (textPositions.get(i).getUnicode().equals(" ") || textPositions.get(i).getUnicode().equals("\u00A0"))) {
+                startIndex++;
+                continue;
+            }
+
+            // Strange but sometimes this is happening, for example: Metolachlor2.pdf
+            if (i > 0 && textPositions.get(i).getX() < textPositions.get(i - 1).getX()) {
+                List<TextPosition> sublist = textPositions.subList(startIndex, i);
+                if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0).getUnicode().equals(" ") || sublist.get(0).getUnicode().equals("\u00A0")))) {
+                    textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
+                }
+                startIndex = i;
+            }
+
+            if (textPositions.get(i).getRotation() == 0 && i > 0 && textPositions.get(i).getX() > textPositions.get(i - 1).getEndX() + 1) {
+                List<TextPosition> sublist = textPositions.subList(startIndex, i);
+                if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0).getUnicode().equals(" ") || sublist.get(0).getUnicode().equals("\u00A0")))) {
+                    textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
+                }
+                startIndex = i;
+            }
+
+            if (i > 0 && (textPositions.get(i).getUnicode().equals(" ") || textPositions.get(i).getUnicode().equals("\u00A0")) && i <= textPositions.size() - 2) {
+                List<TextPosition> sublist = textPositions.subList(startIndex, i);
+                if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0).getUnicode().equals(" ") || sublist.get(0).getUnicode().equals("\u00A0")))) {
+                    textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
+                }
+                startIndex = i + 1;
+            }
+        }
+
+        List<TextPosition> sublist = textPositions.subList(startIndex, textPositions.size());
+        if (!sublist.isEmpty() && (sublist.get(sublist.size() - 1).getUnicode().equals(" ") || sublist.get(sublist.size() - 1).getUnicode().equals("\u00A0"))) {
+            sublist = sublist.subList(0, sublist.size() - 1);
+        }
+        if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0).getUnicode().equals(" ") || sublist.get(0).getUnicode().equals("\u00A0")))) {
+            textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
+        }
+        super.writeString(text);
+    }
+
+
+    public void clearPositions() {
+
+        textPositionSequences = new ArrayList<>();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/parsing/PDFLinesTextStripper.java

@@ -0,0 +1,335 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.parsing;
+
+import java.awt.geom.Point2D;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.Comparator;
+import java.util.List;
+
+import org.apache.pdfbox.contentstream.operator.Operator;
+import org.apache.pdfbox.contentstream.operator.OperatorName;
+import org.apache.pdfbox.contentstream.operator.color.SetNonStrokingColor;
+import org.apache.pdfbox.contentstream.operator.color.SetNonStrokingColorN;
+import org.apache.pdfbox.contentstream.operator.color.SetNonStrokingColorSpace;
+import org.apache.pdfbox.contentstream.operator.color.SetNonStrokingDeviceCMYKColor;
+import org.apache.pdfbox.contentstream.operator.color.SetNonStrokingDeviceGrayColor;
+import org.apache.pdfbox.contentstream.operator.color.SetNonStrokingDeviceRGBColor;
+import org.apache.pdfbox.contentstream.operator.color.SetStrokingColor;
+import org.apache.pdfbox.contentstream.operator.color.SetStrokingColorN;
+import org.apache.pdfbox.contentstream.operator.color.SetStrokingColorSpace;
+import org.apache.pdfbox.contentstream.operator.color.SetStrokingDeviceCMYKColor;
+import org.apache.pdfbox.contentstream.operator.color.SetStrokingDeviceGrayColor;
+import org.apache.pdfbox.contentstream.operator.color.SetStrokingDeviceRGBColor;
+import org.apache.pdfbox.contentstream.operator.state.SetFlatness;
+import org.apache.pdfbox.contentstream.operator.state.SetLineCapStyle;
+import org.apache.pdfbox.contentstream.operator.state.SetLineDashPattern;
+import org.apache.pdfbox.contentstream.operator.state.SetLineJoinStyle;
+import org.apache.pdfbox.contentstream.operator.state.SetLineMiterLimit;
+import org.apache.pdfbox.contentstream.operator.state.SetLineWidth;
+import org.apache.pdfbox.contentstream.operator.state.SetRenderingIntent;
+import org.apache.pdfbox.contentstream.operator.text.SetFontAndSize;
+import org.apache.pdfbox.cos.COSBase;
+import org.apache.pdfbox.cos.COSNumber;
+import org.apache.pdfbox.pdmodel.PDDocument;
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.text.TextPosition;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.RedTextPosition;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Ruling;
+
+import lombok.Getter;
+import lombok.Setter;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+public class PDFLinesTextStripper extends PDFTextStripper {
+
+    @Getter
+    private final List<TextPositionSequence> textPositionSequences = new ArrayList<>();
+    @Getter
+    private final List<Ruling> rulings = new ArrayList<>();
+    private final List<Ruling> graphicsPath = new ArrayList<>();
+    @Setter
+    protected PDPage pdpage;
+    @Getter
+    private int minCharWidth;
+    @Getter
+    private int maxCharWidth;
+    @Getter
+    private int minCharHeight;
+    @Getter
+    private int maxCharHeight;
+
+    private float path_x;
+    private float path_y;
+
+    @Setter
+    private int pageNumber;
+
+
+    public PDFLinesTextStripper() throws IOException {
+
+        super();
+        this.addOperator(new SetStrokingColorSpace());
+        this.addOperator(new SetNonStrokingColorSpace());
+        this.addOperator(new SetLineDashPattern());
+        this.addOperator(new SetStrokingDeviceGrayColor());
+        this.addOperator(new SetNonStrokingDeviceGrayColor());
+        this.addOperator(new SetFlatness());
+        this.addOperator(new SetLineJoinStyle());
+        this.addOperator(new SetLineCapStyle());
+        this.addOperator(new SetStrokingDeviceCMYKColor());
+        this.addOperator(new SetNonStrokingDeviceCMYKColor());
+        this.addOperator(new SetLineMiterLimit());
+        this.addOperator(new SetStrokingDeviceRGBColor());
+        this.addOperator(new SetNonStrokingDeviceRGBColor());
+        this.addOperator(new SetRenderingIntent());
+        this.addOperator(new SetStrokingColor());
+        this.addOperator(new SetNonStrokingColor());
+        this.addOperator(new SetStrokingColorN());
+        this.addOperator(new SetNonStrokingColorN());
+        this.addOperator(new SetFontAndSize());
+        this.addOperator(new SetLineWidth());
+    }
+
+
+    @Override
+    protected void processOperator(Operator operator, List<COSBase> arguments) throws IOException {
+
+        String operation = operator.getName();
+
+        //move
+        switch (operation) {
+            case OperatorName.MOVE_TO:
+                if (arguments.size() == 2) {
+                    Point2D.Float pos = transformPosition(floatValue(arguments.get(0)), floatValue(arguments.get(1)));
+                    path_x = (float) pos.getX();
+                    path_y = (float) pos.getY();
+                }
+                break;
+
+            //line
+            case OperatorName.LINE_TO:
+                if (arguments.size() == 2) {
+                    Point2D.Float pos = transformPosition(floatValue(arguments.get(0)), floatValue(arguments.get(1)));
+
+                    // The direction of vertical lines must always be from bottom to top for the table extraction algorithm.
+                    if (pos.getY() > path_y) {
+                        graphicsPath.add(new Ruling(new Point2D.Float(path_x, path_y), new Point2D.Float((float) pos.getX(), (float) pos.getY())));
+                    } else {
+                        graphicsPath.add(new Ruling(new Point2D.Float(path_x, (float) pos.getY()), new Point2D.Float((float) pos.getX(), path_y)));
+                    }
+
+                    path_x = (float) pos.getX();
+                    path_y = (float) pos.getY();
+                }
+                break;
+
+            //rectangle
+            case OperatorName.APPEND_RECT:
+
+                if (arguments.size() == 4) {
+                    float x = floatValue(arguments.get(0));
+                    float y = floatValue(arguments.get(1));
+                    float width = floatValue(arguments.get(2));
+                    float height = floatValue(arguments.get(3));
+
+                    Point2D p1 = transformPosition(x, y);
+                    Point2D p2 = transformPosition(x + width, y + height);
+
+                    // Horizontal lines
+                    graphicsPath.add(new Ruling(new Point2D.Float((float) p1.getX(), (float) p1.getY()), new Point2D.Float((float) p2.getX(), (float) p1.getY())));
+                    graphicsPath.add(new Ruling(new Point2D.Float((float) p1.getX(), (float) p2.getY()), new Point2D.Float((float) p2.getX(), (float) p2.getY())));
+
+                    // Vertical lines, direction must always be from bottom to top for the table extraction algorithm.
+                    if (p2.getY() > p1.getY()) {
+                        graphicsPath.add(new Ruling(new Point2D.Float((float) p2.getX(), (float) p1.getY()), new Point2D.Float((float) p2.getX(), (float) p2.getY())));
+                    } else {
+                        graphicsPath.add(new Ruling(new Point2D.Float((float) p2.getX(), (float) p2.getY()), new Point2D.Float((float) p2.getX(), (float) p1.getY())));
+                    }
+                    if (p2.getY() > p1.getY()) {
+                        graphicsPath.add(new Ruling(new Point2D.Float((float) p1.getX(), (float) p1.getY()), new Point2D.Float((float) p1.getX(), (float) p2.getY())));
+                    } else {
+                        graphicsPath.add(new Ruling(new Point2D.Float((float) p1.getX(), (float) p2.getY()), new Point2D.Float((float) p1.getX(), (float) p1.getY())));
+                    }
+                }
+                break;
+
+            //fill
+            case OperatorName.FILL_NON_ZERO:
+            case OperatorName.LEGACY_FILL_NON_ZERO:
+            case OperatorName.FILL_EVEN_ODD:
+                addVisibleRulings(graphicsPath, false);
+                graphicsPath.clear();
+                break;
+
+            //stroke
+            case OperatorName.STROKE_PATH:
+                addVisibleRulings(graphicsPath, true);
+                graphicsPath.clear();
+                break;
+
+            //cancel path
+            case OperatorName.ENDPATH:
+                graphicsPath.clear();
+                break;
+
+        }
+
+        super.processOperator(operator, arguments);
+    }
+
+
+    private float floatValue(COSBase value) {
+
+        if (value instanceof COSNumber) {
+            return ((COSNumber) value).floatValue();
+        } else {
+            return 0;
+        }
+    }
+
+
+    private Point2D.Float transformPosition(float x, float y) {
+
+        return super.transformedPoint(x, y);
+    }
+
+
+    private void addVisibleRulings(List<Ruling> path, boolean stroke) throws IOException {
+
+        try {
+            if (stroke && !getGraphicsState().getStrokingColor().isPattern() && getGraphicsState().getStrokingColor()
+                    .toRGB() == 0 || !stroke && !getGraphicsState().getNonStrokingColor().isPattern() && getGraphicsState().getNonStrokingColor().toRGB() == 0) {
+                rulings.addAll(path);
+            }
+        } catch (UnsupportedOperationException e) {
+            log.debug("UnsupportedOperationException: " + getGraphicsState().getStrokingColor().getColorSpace().getName() + " or " + getGraphicsState().getNonStrokingColor()
+                    .getColorSpace()
+                    .getName() + " does not support toRGB");
+        }
+    }
+
+
+    @Override
+    public void writeString(String text, List<TextPosition> textPositions) throws IOException {
+
+        int startIndex = 0;
+        RedTextPosition previous = null;
+
+        textPositions.sort(Comparator.comparing(TextPosition::getXDirAdj));
+
+        for (int i = 0; i <= textPositions.size() - 1; i++) {
+
+            if (!textPositionSequences.isEmpty()) {
+                previous = textPositionSequences.get(textPositionSequences.size() - 1)
+                        .getTextPositions()
+                        .get(textPositionSequences.get(textPositionSequences.size() - 1).getTextPositions().size() - 1);
+            }
+
+            int charWidth = (int) textPositions.get(i).getWidthDirAdj();
+            if (charWidth < minCharWidth) {
+                minCharWidth = charWidth;
+            }
+            if (charWidth > maxCharWidth) {
+                maxCharWidth = charWidth;
+            }
+
+            int charHeight = (int) textPositions.get(i).getHeightDir();
+            if (charHeight < minCharHeight) {
+                minCharHeight = charHeight;
+            }
+            if (charWidth > maxCharHeight) {
+                maxCharHeight = charHeight;
+            }
+
+            if (i == 0 && (textPositions.get(i).getUnicode().equals(" ") || textPositions.get(i).getUnicode().equals("\u00A0") || textPositions.get(i).getUnicode().equals("\t"))) {
+                startIndex++;
+                continue;
+            }
+
+            // Strange but sometimes this is happening, for example: Metolachlor2.pdf
+            if (i > 0 && textPositions.get(i).getXDirAdj() < textPositions.get(i - 1).getXDirAdj()) {
+                List<TextPosition> sublist = textPositions.subList(startIndex, i);
+                if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0).getUnicode().equals(" ") || sublist.get(0).getUnicode().equals("\u00A0") || sublist.get(0)
+                        .getUnicode()
+                        .equals("\t")))) {
+                    textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
+                }
+                startIndex = i;
+            }
+
+            if (textPositions.get(i).getRotation() == 0 && i > 0 && textPositions.get(i).getX() > textPositions.get(i - 1).getEndX() + 1) {
+                List<TextPosition> sublist = textPositions.subList(startIndex, i);
+                if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0).getUnicode().equals(" ") || sublist.get(0).getUnicode().equals("\u00A0") || sublist.get(0)
+                        .getUnicode()
+                        .equals("\t")))) {
+                    textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
+                }
+                startIndex = i;
+            }
+
+            if (i > 0 && (textPositions.get(i).getUnicode().equals(" ") || textPositions.get(i).getUnicode().equals("\u00A0") || textPositions.get(i)
+                    .getUnicode()
+                    .equals("\t")) && i <= textPositions.size() - 2) {
+                List<TextPosition> sublist = textPositions.subList(startIndex, i);
+                if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0).getUnicode().equals(" ") || sublist.get(0).getUnicode().equals("\u00A0") || sublist.get(0)
+                        .getUnicode()
+                        .equals("\t")))) {
+
+                    // Remove false sequence ends (whitespaces)
+                    if (previous != null && sublist.get(0).getYDirAdj() == previous.getYDirAdj() && sublist.get(0)
+                            .getXDirAdj() - (previous.getXDirAdj() + previous.getWidthDirAdj()) < 0.01) {
+                        for (TextPosition textPosition : sublist) {
+                            textPositionSequences.get(textPositionSequences.size() - 1).add(textPosition);
+                        }
+                    } else {
+                        textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
+                    }
+                }
+                startIndex = i + 1;
+            }
+        }
+
+        List<TextPosition> sublist = textPositions.subList(startIndex, textPositions.size());
+        if (!sublist.isEmpty() && (sublist.get(sublist.size() - 1).getUnicode().equals(" ") || sublist.get(sublist.size() - 1)
+                .getUnicode()
+                .equals("\u00A0") || sublist.get(sublist.size() - 1).getUnicode().equals("\t"))) {
+            sublist = sublist.subList(0, sublist.size() - 1);
+        }
+
+        if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0).getUnicode().equals(" ") || sublist.get(0).getUnicode().equals("\u00A0") || sublist.get(0)
+                .getUnicode()
+                .equals("\t")))) {
+            if (previous != null && sublist.get(0).getYDirAdj() == previous.getYDirAdj() && sublist.get(0)
+                    .getXDirAdj() - (previous.getXDirAdj() + previous.getWidthDirAdj()) < 0.01) {
+                for (TextPosition t : sublist) {
+                    textPositionSequences.get(textPositionSequences.size() - 1).add(t);
+                }
+            } else {
+                textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
+            }
+        }
+        super.writeString(text);
+    }
+
+
+    @Override
+    public String getText(PDDocument doc) throws IOException {
+
+        minCharWidth = Integer.MAX_VALUE;
+        maxCharWidth = 0;
+        minCharHeight = Integer.MAX_VALUE;
+        maxCharHeight = 0;
+        textPositionSequences.clear();
+        rulings.clear();
+        graphicsPath.clear();
+        path_x = 0.0f;
+        path_y = 0.0f;
+
+        return super.getText(doc);
+
+    }
+
+}

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

@@ -0,0 +1,279 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.service;
+
+import static java.util.stream.Collectors.toSet;
+
+import java.util.ArrayList;
+import java.util.Comparator;
+import java.util.Iterator;
+import java.util.List;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.FloatFrequencyCounter;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.TextBlockOrientation;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.StringFrequencyCounter;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Ruling;
+import com.knecon.fforesight.service.layoutparser.processor.classification.utils.RulingTextDirAdjustUtil;
+
+@Service
+@SuppressWarnings("all")
+public class BlockificationService {
+
+    static final float THRESHOLD = 1f;
+
+
+    /**
+     * This method is building blocks by expanding the minX/maxX and minY/maxY value on each word that is not split by the conditions.
+     * This method must use text direction adjusted postions (DirAdj). Where {0,0} is on the upper left. Never try to change this!
+     * Rulings (Table lines) must be adjusted to the text directions as well, when checking if a block is split by a ruling.
+     *
+     * @param textPositions         The words of a page.
+     * @param horizontalRulingLines Horizontal table lines.
+     * @param verticalRulingLines   Vertical table lines.
+     * @return ClassificationPage object that contains the Textblock and text statistics.
+     */
+    public ClassificationPage blockify(List<TextPositionSequence> textPositions, List<Ruling> horizontalRulingLines, List<Ruling> verticalRulingLines) {
+
+        int indexOnPage = 0;
+        List<TextPositionSequence> chunkWords = new ArrayList<>();
+        List<AbstractTextContainer> chunkBlockList1 = new ArrayList<>();
+
+        float minX = 1000, maxX = 0, minY = 1000, maxY = 0;
+        TextPositionSequence prev = null;
+
+        boolean wasSplitted = false;
+        Float splitX1 = null;
+        for (TextPositionSequence word : textPositions) {
+
+            boolean lineSeparation = word.getMinYDirAdj() - maxY > word.getHeight() * 1.25;
+            boolean startFromTop = prev != null && word.getMinYDirAdj() < prev.getMinYDirAdj() - prev.getTextHeight();
+            boolean splitByX = prev != null && maxX + 50 < word.getMinXDirAdj() && prev.getMinYDirAdj() == word.getMinYDirAdj();
+            boolean xIsBeforeFirstX = prev != null && word.getMinXDirAdj() < minX;
+            boolean newLineAfterSplit = prev != null && word.getMinYDirAdj() != prev.getMinYDirAdj() && wasSplitted && splitX1 != word.getMinXDirAdj();
+            boolean isSplitByRuling = isSplitByRuling(minX, minY, maxX, maxY, word, horizontalRulingLines, verticalRulingLines);
+            boolean splitByDir = prev != null && !prev.getDir().equals(word.getDir());
+
+            if (prev != null && (lineSeparation || startFromTop || splitByX || splitByDir || isSplitByRuling)) {
+
+                TextBlockOrientation prevOrientation = null;
+                if (!chunkBlockList1.isEmpty()) {
+                    prevOrientation = chunkBlockList1.get(chunkBlockList1.size() - 1).getOrientation();
+                }
+
+                ClassificationTextBlock cb1 = buildTextBlock(chunkWords, indexOnPage);
+                indexOnPage++;
+
+                chunkBlockList1.add(cb1);
+                chunkWords = new ArrayList<>();
+
+                if (splitByX && !isSplitByRuling) {
+                    wasSplitted = true;
+                    cb1.setOrientation(TextBlockOrientation.LEFT);
+                    splitX1 = word.getMinXDirAdj();
+                } else if (newLineAfterSplit && !isSplitByRuling) {
+                    wasSplitted = false;
+                    cb1.setOrientation(TextBlockOrientation.RIGHT);
+                    splitX1 = null;
+                } else if (prevOrientation != null && prevOrientation.equals(TextBlockOrientation.RIGHT) && (lineSeparation || !startFromTop || !splitByX || !newLineAfterSplit || !isSplitByRuling)) {
+                    cb1.setOrientation(TextBlockOrientation.LEFT);
+                }
+
+                minX = 1000;
+                maxX = 0;
+                minY = 1000;
+                maxY = 0;
+                prev = null;
+            }
+
+            chunkWords.add(word);
+
+            prev = word;
+            if (word.getMinXDirAdj() < minX) {
+                minX = word.getMinXDirAdj();
+            }
+            if (word.getMaxXDirAdj() > maxX) {
+                maxX = word.getMaxXDirAdj();
+            }
+            if (word.getMinYDirAdj() < minY) {
+                minY = word.getMinYDirAdj();
+            }
+            if (word.getMaxYDirAdj() > maxY) {
+                maxY = word.getMaxYDirAdj();
+            }
+        }
+
+        ClassificationTextBlock cb1 = buildTextBlock(chunkWords, indexOnPage);
+        if (cb1 != null) {
+            chunkBlockList1.add(cb1);
+        }
+
+        Iterator<AbstractTextContainer> itty = chunkBlockList1.iterator();
+
+        ClassificationTextBlock previousLeft = null;
+        ClassificationTextBlock previousRight = null;
+        while (itty.hasNext()) {
+            ClassificationTextBlock block = (ClassificationTextBlock) itty.next();
+
+            if (previousLeft != null && block.getOrientation().equals(TextBlockOrientation.LEFT)) {
+                if (previousLeft.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousLeft.getMinY()) {
+                    previousLeft.add(block);
+                    itty.remove();
+                    continue;
+                }
+            }
+
+            if (previousRight != null && block.getOrientation().equals(TextBlockOrientation.RIGHT)) {
+                if (previousRight.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousRight.getMinY()) {
+                    previousRight.add(block);
+                    itty.remove();
+                    continue;
+                }
+            }
+
+            if (block.getOrientation().equals(TextBlockOrientation.LEFT)) {
+                previousLeft = block;
+            } else if (block.getOrientation().equals(TextBlockOrientation.RIGHT)) {
+                previousRight = block;
+            }
+        }
+
+        itty = chunkBlockList1.iterator();
+        ClassificationTextBlock previous = null;
+        while (itty.hasNext()) {
+            ClassificationTextBlock block = (ClassificationTextBlock) itty.next();
+
+            if (previous != null && previous.getOrientation().equals(TextBlockOrientation.LEFT) && block.getOrientation().equals(TextBlockOrientation.LEFT) && equalsWithThreshold(block.getMaxY(),
+                    previous.getMaxY()) || previous != null && previous.getOrientation().equals(TextBlockOrientation.LEFT) && block.getOrientation()
+                    .equals(TextBlockOrientation.RIGHT) && equalsWithThreshold(block.getMaxY(), previous.getMaxY())) {
+                previous.add(block);
+                itty.remove();
+                continue;
+            }
+
+            previous = block;
+        }
+
+        return new ClassificationPage(chunkBlockList1);
+    }
+
+
+    private boolean equalsWithThreshold(float f1, float f2) {
+
+        return Math.abs(f1 - f2) < THRESHOLD;
+    }
+
+
+    private ClassificationTextBlock buildTextBlock(List<TextPositionSequence> wordBlockList, int indexOnPage) {
+
+        ClassificationTextBlock 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 ClassificationTextBlock(wordBlock.getMinXDirAdj(),
+                        wordBlock.getMaxXDirAdj(),
+                        wordBlock.getMinYDirAdj(),
+                        wordBlock.getMaxYDirAdj(),
+                        wordBlockList,
+                        wordBlock.getRotation(),
+                        indexOnPage);
+            } else {
+                ClassificationTextBlock 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 -> round(t.getMinYDirAdj(), 3)).collect(toSet()).size() == 1) {
+            textBlock.getSequences().sort(Comparator.comparing(TextPositionSequence::getMinXDirAdj));
+        }
+        return textBlock;
+    }
+
+
+    private boolean isSplitByRuling(float minX,
+                                    float minY,
+                                    float maxX,
+                                    float maxY,
+                                    TextPositionSequence word,
+                                    List<Ruling> horizontalRulingLines,
+                                    List<Ruling> verticalRulingLines) {
+
+        return isSplitByRuling(maxX,
+                minY,
+                word.getMinXDirAdj(),
+                word.getMinYDirAdj(),
+                verticalRulingLines,
+                word.getDir().getDegrees(),
+                word.getPageWidth(),
+                word.getPageHeight()) //
+                || isSplitByRuling(minX,
+                minY,
+                word.getMinXDirAdj(),
+                word.getMaxYDirAdj(),
+                horizontalRulingLines,
+                word.getDir().getDegrees(),
+                word.getPageWidth(),
+                word.getPageHeight()) //
+                || isSplitByRuling(maxX,
+                minY,
+                word.getMinXDirAdj(),
+                word.getMinYDirAdj(),
+                horizontalRulingLines,
+                word.getDir().getDegrees(),
+                word.getPageWidth(),
+                word.getPageHeight()) //
+                || isSplitByRuling(minX,
+                minY,
+                word.getMinXDirAdj(),
+                word.getMaxYDirAdj(),
+                verticalRulingLines,
+                word.getDir().getDegrees(),
+                word.getPageWidth(),
+                word.getPageHeight()); //
+    }
+
+
+    private boolean isSplitByRuling(float previousX2, float previousY1, float currentX1, float currentY1, List<Ruling> rulingLines, float dir, float pageWidth, float pageHeight) {
+
+        for (Ruling ruling : rulingLines) {
+            var line = RulingTextDirAdjustUtil.convertToDirAdj(ruling, dir, pageWidth, pageHeight);
+            if (line.intersectsLine(previousX2, previousY1, currentX1, currentY1)) {
+                return true;
+            }
+        }
+        return false;
+    }
+
+
+    private double round(float value, int decimalPoints) {
+
+        var d = Math.pow(10, decimalPoints);
+        return Math.round(value * d) / d;
+    }
+
+}

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

@@ -0,0 +1,160 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.service;
+import java.util.List;
+
+import org.springframework.stereotype.Service;
+
+import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
+import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Point;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.FloatFrequencyCounter;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.TableCell;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Table;
+import com.knecon.fforesight.service.layoutparser.processor.classification.utils.PositionUtils;
+
+@Service
+public class BodyTextFrameService {
+
+    /**
+     * Adjusts and sets the body text frame to a classificationPage.
+     * Note: This needs to use Pdf Coordinate System where {0,0} rotated with the classificationPage rotation.
+     * 0 -> LowerLeft
+     * 90 -> UpperLeft
+     * 180 -> UpperRight
+     * 270 -> LowerRight
+     * The aspect ratio of the classificationPage is also regarded.
+     *
+     * @param classificationPage                   The classificationPage
+     * @param bodyTextFrame          frame that contains the main text on portrait pages
+     * @param landscapeBodyTextFrame frame that contains the main text on landscape pages
+     */
+    public void setBodyTextFrameAdjustedToPage(ClassificationPage classificationPage, Rectangle bodyTextFrame, Rectangle landscapeBodyTextFrame) {
+
+        Rectangle textFrame = classificationPage.isLandscape() ? landscapeBodyTextFrame : bodyTextFrame;
+
+        if (classificationPage.getPageWidth() > classificationPage.getPageHeight() && classificationPage.getRotation() == 270) {
+            textFrame = new Rectangle(new Point(textFrame.getTopLeft().getY(), classificationPage.getPageHeight() - textFrame.getTopLeft().getX() - textFrame.getWidth()),
+                    textFrame.getHeight(),
+                    textFrame.getWidth(),
+                    0);
+        } else if (classificationPage.getPageWidth() > classificationPage.getPageHeight() && classificationPage.getRotation() != 0) {
+            textFrame = new Rectangle(new Point(textFrame.getTopLeft().getY(), textFrame.getTopLeft().getX()), textFrame.getHeight(), textFrame.getWidth(), classificationPage.getPageNumber());
+        } else if (classificationPage.getRotation() == 180) {
+            textFrame = new Rectangle(new Point(textFrame.getTopLeft().getX(), classificationPage.getPageHeight() - textFrame.getTopLeft().getY() - textFrame.getHeight()),
+                    textFrame.getWidth(),
+                    textFrame.getHeight(),
+                    0);
+        }
+        classificationPage.setBodyTextFrame(textFrame);
+    }
+
+
+    /**
+     * Calculates the frame that contains the main text, text outside the frame will be e.g. headers or footers.
+     * Note: This needs to use Pdf Coordinate System where {0,0} rotated with the page rotation.
+     * 0 -> LowerLeft
+     * 90 -> UpperLeft
+     * 180 -> UpperRight
+     * 270 -> LowerRight
+     * The aspect ratio of the page is also regarded.
+     *
+     * @param classificationPages                   List of all classificationPages
+     * @param documentFontSizeCounter Statistics of the document
+     * @param landscape               Calculate for landscape or portrait
+     * @return Rectangle of the text frame
+     */
+    public Rectangle calculateBodyTextFrame(List<ClassificationPage> classificationPages, FloatFrequencyCounter documentFontSizeCounter, boolean landscape) {
+
+        BodyTextFrameExpansionsRectangle expansionsRectangle = new BodyTextFrameExpansionsRectangle();
+
+        for (ClassificationPage classificationPage : classificationPages) {
+
+            if (classificationPage.getTextBlocks().isEmpty() || landscape != classificationPage.isLandscape()) {
+                continue;
+            }
+
+            for (AbstractTextContainer container : classificationPage.getTextBlocks()) {
+
+                if (container instanceof ClassificationTextBlock) {
+                    ClassificationTextBlock textBlock = (ClassificationTextBlock) container;
+                    if (textBlock.getMostPopularWordFont() == null || textBlock.getMostPopularWordStyle() == null) {
+                        continue;
+                    }
+
+                    float approxLineCount = PositionUtils.getApproxLineCount(textBlock);
+                    if (approxLineCount < 2.9f) {
+                        continue;
+                    }
+
+                    if (documentFontSizeCounter.getMostPopular() != null && textBlock.getMostPopularWordFontSize() >= documentFontSizeCounter.getMostPopular()) {
+
+                        expandRectangle(textBlock, classificationPage, expansionsRectangle);
+                    }
+                }
+
+                if (container instanceof Table) {
+                    Table table = (Table) container;
+                    for (List<TableCell> row : table.getRows()) {
+                        for (TableCell cell : row) {
+
+                            if (cell == null || cell.getTextBlocks() == null) {
+                                continue;
+                            }
+                            for (ClassificationTextBlock textBlock : cell.getTextBlocks()) {
+                                expandRectangle(textBlock, classificationPage, expansionsRectangle);
+                            }
+                        }
+                    }
+                }
+            }
+        }
+        return new Rectangle(new Point(expansionsRectangle.minX, expansionsRectangle.minY),
+                expansionsRectangle.maxX - expansionsRectangle.minX,
+                expansionsRectangle.maxY - expansionsRectangle.minY,
+                0);
+    }
+
+
+    private void expandRectangle(ClassificationTextBlock textBlock, ClassificationPage classificationPage, BodyTextFrameExpansionsRectangle expansionsRectangle) {
+
+        if (classificationPage.getPageWidth() > classificationPage.getPageHeight() && classificationPage.getRotation() != 0) {
+            if (textBlock.getPdfMinY() < expansionsRectangle.minX) {
+                expansionsRectangle.minX = textBlock.getPdfMinY();
+            }
+            if (textBlock.getPdfMaxY() > expansionsRectangle.maxX) {
+                expansionsRectangle.maxX = textBlock.getPdfMaxY();
+            }
+            if (textBlock.getPdfMinX() < expansionsRectangle.minY) {
+                expansionsRectangle.minY = textBlock.getPdfMinX();
+            }
+            if (textBlock.getPdfMaxX() > expansionsRectangle.maxY) {
+                expansionsRectangle.maxY = textBlock.getPdfMaxX();
+            }
+        } else {
+            if (textBlock.getPdfMinX() < expansionsRectangle.minX) {
+                expansionsRectangle.minX = textBlock.getPdfMinX();
+            }
+            if (textBlock.getPdfMaxX() > expansionsRectangle.maxX) {
+                expansionsRectangle.maxX = textBlock.getPdfMaxX();
+            }
+            if (textBlock.getPdfMinY() < expansionsRectangle.minY) {
+                expansionsRectangle.minY = textBlock.getPdfMinY();
+            }
+            if (textBlock.getPdfMaxY() > expansionsRectangle.maxY) {
+                expansionsRectangle.maxY = textBlock.getPdfMaxY();
+            }
+        }
+    }
+
+
+    private class BodyTextFrameExpansionsRectangle {
+
+        float minX = 10000;
+        float maxX = -100;
+        float minY = 10000;
+        float maxY = -100;
+
+    }
+
+}

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

@@ -0,0 +1,116 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.service;
+
+import java.util.List;
+import java.util.regex.Pattern;
+
+import org.springframework.stereotype.Service;
+
+import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationDocument;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.utils.PositionUtils;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class ClassificationService {
+
+    private final BodyTextFrameService bodyTextFrameService;
+
+
+    public void classifyDocument(ClassificationDocument document) {
+
+        Rectangle bodyTextFrame = bodyTextFrameService.calculateBodyTextFrame(document.getPages(), document.getFontSizeCounter(), false);
+        Rectangle landscapeBodyTextFrame = bodyTextFrameService.calculateBodyTextFrame(document.getPages(), document.getFontSizeCounter(), true);
+        List<Float> headlineFontSizes = document.getFontSizeCounter().getHighterThanMostPopular();
+
+        log.debug("Document FontSize counters are: {}", document.getFontSizeCounter().getCountPerValue());
+
+        for (ClassificationPage classificationPage : document.getPages()) {
+            bodyTextFrameService.setBodyTextFrameAdjustedToPage(classificationPage, bodyTextFrame, landscapeBodyTextFrame);
+            classifyPage(classificationPage, document, headlineFontSizes);
+        }
+    }
+
+
+    public void classifyPage(ClassificationPage classificationPage, ClassificationDocument document, List<Float> headlineFontSizes) {
+
+        for (AbstractTextContainer textBlock : classificationPage.getTextBlocks()) {
+            if (textBlock instanceof ClassificationTextBlock) {
+                classifyBlock((ClassificationTextBlock) textBlock, classificationPage, document, headlineFontSizes);
+            }
+        }
+    }
+
+
+    public void classifyBlock(ClassificationTextBlock textBlock, ClassificationPage page, ClassificationDocument document, List<Float> headlineFontSizes) {
+
+        var bodyTextFrame = page.getBodyTextFrame();
+
+        if (document.getFontSizeCounter().getMostPopular() == null) {
+            textBlock.setClassification("Other");
+            return;
+        }
+        if (PositionUtils.isOverBodyTextFrame(bodyTextFrame, textBlock, page.getRotation()) && (document.getFontSizeCounter()
+                .getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter().getMostPopular())) {
+            textBlock.setClassification("Header");
+
+        } else if (PositionUtils.isUnderBodyTextFrame(bodyTextFrame, textBlock, page.getRotation()) && (document.getFontSizeCounter()
+                .getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter().getMostPopular())) {
+            textBlock.setClassification("Footer");
+        } else if (page.getPageNumber() == 1 && (PositionUtils.getHeightDifferenceBetweenChunkWordAndDocumentWord(textBlock,
+                document.getTextHeightCounter().getMostPopular()) > 2.5 && textBlock.getHighestFontSize() > document.getFontSizeCounter().getMostPopular() || page.getTextBlocks()
+                .size() == 1)) {
+            if (!Pattern.matches("[0-9]+", textBlock.toString())) {
+                textBlock.setClassification("Title");
+            }
+        } else if (textBlock.getMostPopularWordFontSize() > document.getFontSizeCounter()
+                .getMostPopular() && PositionUtils.getApproxLineCount(textBlock) < 4.9 && (textBlock.getMostPopularWordStyle().equals("bold") || !document.getFontStyleCounter()
+                .getCountPerValue()
+                .containsKey("bold") && textBlock.getMostPopularWordFontSize() > document.getFontSizeCounter().getMostPopular() + 1) && textBlock.getSequences()
+                .get(0)
+                .getTextPositions()
+                .get(0)
+                .getFontSizeInPt() >= textBlock.getMostPopularWordFontSize()) {
+
+            for (int i = 1; i <= headlineFontSizes.size(); i++) {
+                if (textBlock.getMostPopularWordFontSize() == headlineFontSizes.get(i - 1)) {
+                    textBlock.setClassification("H " + i);
+                    document.setHeadlines(true);
+                }
+            }
+        } else if (!textBlock.getText().startsWith("Table ") && !textBlock.getText().startsWith("Figure ") && PositionUtils.isWithinBodyTextFrame(bodyTextFrame,
+                textBlock) && textBlock.getMostPopularWordStyle().equals("bold") && !document.getFontStyleCounter()
+                .getMostPopular()
+                .equals("bold") && PositionUtils.getApproxLineCount(textBlock) < 2.9 && textBlock.getSequences()
+                .get(0)
+                .getTextPositions()
+                .get(0)
+                .getFontSizeInPt() >= textBlock.getMostPopularWordFontSize()) {
+            textBlock.setClassification("H " + (headlineFontSizes.size() + 1));
+            document.setHeadlines(true);
+        } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFontSize() == document.getFontSizeCounter()
+                .getMostPopular() && textBlock.getMostPopularWordStyle().equals("bold") && !document.getFontStyleCounter().getMostPopular().equals("bold")) {
+            textBlock.setClassification("TextBlock Bold");
+        } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFont()
+                .equals(document.getFontCounter().getMostPopular()) && textBlock.getMostPopularWordStyle()
+                .equals(document.getFontStyleCounter().getMostPopular()) && textBlock.getMostPopularWordFontSize() == document.getFontSizeCounter().getMostPopular()) {
+            textBlock.setClassification("TextBlock");
+        } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFontSize() == document.getFontSizeCounter()
+                .getMostPopular() && textBlock.getMostPopularWordStyle().equals("italic") && !document.getFontStyleCounter()
+                .getMostPopular()
+                .equals("italic") && PositionUtils.getApproxLineCount(textBlock) < 2.9) {
+            textBlock.setClassification("TextBlock Italic");
+        } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock)) {
+            textBlock.setClassification("TextBlock Unknown");
+        } else {
+            textBlock.setClassification("Other");
+        }
+    }
+
+}

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

@@ -0,0 +1,134 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.service;
+
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Map;
+
+import org.apache.pdfbox.pdmodel.PDDocument;
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.pdmodel.common.PDRectangle;
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationDocument;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.TextPositionSequence;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.adapter.ImageServiceResponseAdapter;
+import com.knecon.fforesight.service.layoutparser.processor.classification.parsing.PDFLinesTextStripper;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CleanRulings;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CvParsedTableCell;
+
+import lombok.RequiredArgsConstructor;
+import lombok.SneakyThrows;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class PdfParsingService {
+
+    private final RulingCleaningService rulingCleaningService;
+    private final TableExtractionService tableExtractionService;
+    private final BlockificationService blockificationService;
+    private final ImageServiceResponseAdapter imageServiceResponseAdapter;
+
+
+    public ClassificationDocument parseDocument(PDDocument originDocument, Map<Integer, List<CvParsedTableCell>> pdfTableCells, Map<Integer, List<ClassifiedImage>> pdfImages) {
+
+        ClassificationDocument document = new ClassificationDocument();
+        List<ClassificationPage> classificationPages = new ArrayList<>();
+
+        originDocument.setAllSecurityToBeRemoved(true);
+        long pageCount = originDocument.getNumberOfPages();
+
+        for (int pageNumber = 1; pageNumber <= pageCount; pageNumber++) {
+            parsePage(pdfImages, originDocument, pdfTableCells, document, classificationPages, pageNumber);
+        }
+
+        document.setPages(classificationPages);
+
+        return document;
+    }
+
+
+    @SneakyThrows
+    private void parsePage(Map<Integer, List<ClassifiedImage>> pdfImages,
+                           PDDocument pdDocument,
+                           Map<Integer, List<CvParsedTableCell>> pdfTableCells,
+                           ClassificationDocument document,
+                           List<ClassificationPage> classificationPages,
+                           int pageNumber) {
+
+        PDFLinesTextStripper stripper = new PDFLinesTextStripper();
+        PDPage pdPage = pdDocument.getPage(pageNumber - 1);
+        stripper.setPageNumber(pageNumber);
+        stripper.setStartPage(pageNumber);
+        stripper.setEndPage(pageNumber);
+        stripper.setPdpage(pdPage);
+        stripper.getText(pdDocument);
+
+        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);
+
+        PDRectangle cropbox = pdPage.getCropBox();
+        CleanRulings cleanRulings = rulingCleaningService.getCleanRulings(pdfTableCells.get(pageNumber),
+                stripper.getRulings(),
+                stripper.getMinCharWidth(),
+                stripper.getMaxCharHeight());
+        ClassificationPage classificationPage = blockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
+
+        classificationPage.setRotation(rotation);
+        classificationPage.setLandscape(isLandscape);
+        classificationPage.setPageNumber(pageNumber);
+        classificationPage.setPageWidth(cropbox.getWidth());
+        classificationPage.setPageHeight(cropbox.getHeight());
+
+        // 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)) {
+            classificationPage.setImages(pdfImages.get(pageNumber));
+            imageServiceResponseAdapter.findOcr(classificationPage);
+        }
+
+        tableExtractionService.removeRedundantTableCells(cleanRulings, classificationPage);
+        buildPageStatistics(classificationPage);
+        increaseDocumentStatistics(classificationPage, document);
+
+        classificationPages.add(classificationPage);
+    }
+
+
+    private void increaseDocumentStatistics(ClassificationPage classificationPage, ClassificationDocument document) {
+
+        if (!classificationPage.isLandscape()) {
+            document.getFontSizeCounter().addAll(classificationPage.getFontSizeCounter().getCountPerValue());
+        }
+        document.getFontCounter().addAll(classificationPage.getFontCounter().getCountPerValue());
+        document.getTextHeightCounter().addAll(classificationPage.getTextHeightCounter().getCountPerValue());
+        document.getFontStyleCounter().addAll(classificationPage.getFontStyleCounter().getCountPerValue());
+    }
+
+
+    private void buildPageStatistics(ClassificationPage classificationPage) {
+
+        // Collect all statistics for the classificationPage, except from blocks inside tables, as tables will always be added to BodyTextFrame.
+        for (AbstractTextContainer textBlock : classificationPage.getTextBlocks()) {
+            if (textBlock instanceof ClassificationTextBlock) {
+                if (((ClassificationTextBlock) textBlock).getSequences() == null) {
+                    continue;
+                }
+                for (TextPositionSequence word : ((ClassificationTextBlock) textBlock).getSequences()) {
+                    classificationPage.getTextHeightCounter().add(word.getTextHeight());
+                    classificationPage.getFontCounter().add(word.getFont());
+                    classificationPage.getFontSizeCounter().add(word.getFontSize());
+                    classificationPage.getFontStyleCounter().add(word.getFontStyle());
+                }
+            }
+        }
+
+    }
+
+}

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

@@ -0,0 +1,231 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.service;
+
+import java.awt.geom.Line2D;
+import java.awt.geom.Point2D;
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.Collections;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CleanRulings;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CvParsedTableCell;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Ruling;
+import com.knecon.fforesight.service.layoutparser.processor.classification.utils.DoubleComparisons;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class RulingCleaningService {
+
+    public CleanRulings getCleanRulings(List<CvParsedTableCell> cvParsedTableCells, List<Ruling> rulings, float minCharWidth, float maxCharHeight) {
+
+        if (!rulings.isEmpty()) {
+            snapPoints(rulings, minCharWidth, maxCharHeight);
+        }
+
+        List<Ruling> vrs = new ArrayList<>();
+        for (Ruling vr : rulings) {
+            if (vr.vertical()) {
+                vrs.add(vr);
+            }
+        }
+        if (vrs.isEmpty()) {
+            vrs.addAll(extractVerticalRulings(cvParsedTableCells));
+        }
+        List<Ruling> verticalRulingLines = collapseOrientedRulings(vrs);
+
+        List<Ruling> hrs = new ArrayList<>();
+        for (Ruling hr : rulings) {
+            if (hr.horizontal()) {
+                hrs.add(hr);
+            }
+        }
+        if (hrs.isEmpty()) {
+            hrs.addAll(extractHorizontalRulings(cvParsedTableCells));
+        }
+        List<Ruling> horizontalRulingLines = collapseOrientedRulings(hrs);
+
+        return CleanRulings.builder().vertical(verticalRulingLines).horizontal(horizontalRulingLines).build();
+    }
+
+
+    public void snapPoints(List<? extends Line2D.Float> rulings, float xThreshold, float yThreshold) {
+
+        // collect points and keep a Line -> p1,p2 map
+        Map<Line2D.Float, Point2D[]> linesToPoints = new HashMap<>();
+        List<Point2D> points = new ArrayList<>();
+        for (Line2D.Float r : rulings) {
+            Point2D p1 = r.getP1();
+            Point2D p2 = r.getP2();
+            linesToPoints.put(r, new Point2D[]{p1, p2});
+            points.add(p1);
+            points.add(p2);
+        }
+
+        // snap by X
+        points.sort(Comparator.comparingDouble(Point2D::getX));
+
+        List<List<Point2D>> groupedPoints = new ArrayList<>();
+        groupedPoints.add(new ArrayList<>(Collections.singletonList(points.get(0))));
+
+        for (Point2D p : points.subList(1, points.size() - 1)) {
+            List<Point2D> last = groupedPoints.get(groupedPoints.size() - 1);
+            if (Math.abs(p.getX() - last.get(0).getX()) < xThreshold) {
+                groupedPoints.get(groupedPoints.size() - 1).add(p);
+            } else {
+                groupedPoints.add(new ArrayList<>(Collections.singletonList(p)));
+            }
+        }
+
+        for (List<Point2D> group : groupedPoints) {
+            float avgLoc = 0;
+            for (Point2D p : group) {
+                avgLoc += p.getX();
+            }
+            avgLoc /= group.size();
+            for (Point2D p : group) {
+                p.setLocation(avgLoc, p.getY());
+            }
+        }
+        // ---
+
+        // snap by Y
+        points.sort(Comparator.comparingDouble(Point2D::getY));
+
+        groupedPoints = new ArrayList<>();
+        groupedPoints.add(new ArrayList<>(Collections.singletonList(points.get(0))));
+
+        for (Point2D p : points.subList(1, points.size() - 1)) {
+            List<Point2D> last = groupedPoints.get(groupedPoints.size() - 1);
+            if (Math.abs(p.getY() - last.get(0).getY()) < yThreshold) {
+                groupedPoints.get(groupedPoints.size() - 1).add(p);
+            } else {
+                groupedPoints.add(new ArrayList<>(Collections.singletonList(p)));
+            }
+        }
+
+        for (List<Point2D> group : groupedPoints) {
+            float avgLoc = 0;
+            for (Point2D p : group) {
+                avgLoc += p.getY();
+            }
+            avgLoc /= group.size();
+            for (Point2D p : group) {
+                p.setLocation(p.getX(), avgLoc);
+            }
+        }
+        // ---
+
+        // finally, modify lines
+        for (Map.Entry<Line2D.Float, Point2D[]> ltp : linesToPoints.entrySet()) {
+            Point2D[] p = ltp.getValue();
+            ltp.getKey().setLine(p[0], p[1]);
+        }
+    }
+
+
+    private Collection<? extends Ruling> extractVerticalRulings(List<CvParsedTableCell> cvParsedTableCells) {
+
+        List<Ruling> vrs = new ArrayList<>();
+
+        if (cvParsedTableCells != null) {
+            for (CvParsedTableCell cvParsedTableCell : cvParsedTableCells) {
+                Ruling leftLine = createRuling(cvParsedTableCell.getX0(), cvParsedTableCell.getX0(), cvParsedTableCell.getY0(), cvParsedTableCell.getY1());
+                Ruling rightLine = createRuling(cvParsedTableCell.getX1(), cvParsedTableCell.getX1(), cvParsedTableCell.getY0(), cvParsedTableCell.getY1());
+                vrs.add(leftLine);
+                vrs.add(rightLine);
+            }
+        }
+        return vrs;
+    }
+
+
+    private Collection<? extends Ruling> extractHorizontalRulings(List<CvParsedTableCell> cvParsedTableCells) {
+
+        List<Ruling> hrs = new ArrayList<>();
+
+        if (cvParsedTableCells != null) {
+            for (CvParsedTableCell cvParsedTableCell : cvParsedTableCells) {
+                Ruling topLine = createRuling(cvParsedTableCell.getX0(), cvParsedTableCell.getX1(), cvParsedTableCell.getY1(), cvParsedTableCell.getY1());
+                Ruling baseLine = createRuling(cvParsedTableCell.getX0(), cvParsedTableCell.getX1(), cvParsedTableCell.getY0(), cvParsedTableCell.getY0());
+                hrs.add(topLine);
+                hrs.add(baseLine);
+            }
+        }
+
+        return hrs;
+    }
+
+
+    private Ruling createRuling(float tableCellX0, float tableCellX1, float tableCellY0, float tableCellY1) {
+
+        float x0 = tableCellX0;
+        float x1 = tableCellX1;
+        float y0 = tableCellY0;
+        float y1 = tableCellY1;
+
+        if (x1 < x0) {
+            x0 = tableCellX1;
+            x1 = tableCellX0;
+        }
+
+        if (y1 < y0) {
+            y0 = tableCellY1;
+            y1 = tableCellY0;
+        }
+
+        return new Ruling(new Point2D.Float(x0, y0), new Point2D.Float(x1, y1));
+    }
+
+
+    private List<Ruling> collapseOrientedRulings(List<Ruling> lines) {
+
+        int COLINEAR_OR_PARALLEL_PIXEL_EXPAND_AMOUNT = 1;
+        return collapseOrientedRulings(lines, COLINEAR_OR_PARALLEL_PIXEL_EXPAND_AMOUNT);
+    }
+
+
+    private List<Ruling> collapseOrientedRulings(List<Ruling> lines, int expandAmount) {
+
+        ArrayList<Ruling> rv = new ArrayList<>();
+        lines.sort((a, b) -> {
+            final float diff = a.getPosition() - b.getPosition();
+            return Float.compare(diff == 0 ? a.getStart() - b.getStart() : diff, 0f);
+        });
+
+        for (Ruling next_line : lines) {
+            Ruling last = rv.isEmpty() ? null : rv.get(rv.size() - 1);
+            // if current line colinear with next, and are "close enough": expand current line
+            if (last != null && DoubleComparisons.feq(next_line.getPosition(), last.getPosition()) && last.nearlyIntersects(next_line, expandAmount)) {
+                final float lastStart = last.getStart();
+                final float lastEnd = last.getEnd();
+
+                final boolean lastFlipped = lastStart > lastEnd;
+                final boolean nextFlipped = next_line.getStart() > next_line.getEnd();
+
+                boolean differentDirections = nextFlipped != lastFlipped;
+                float nextS = differentDirections ? next_line.getEnd() : next_line.getStart();
+                float nextE = differentDirections ? next_line.getStart() : next_line.getEnd();
+
+                final float newStart = lastFlipped ? Math.max(nextS, lastStart) : Math.min(nextS, lastStart);
+                final float newEnd = lastFlipped ? Math.min(nextE, lastEnd) : Math.max(nextE, lastEnd);
+                last.setStartEnd(newStart, newEnd);
+                assert !last.oblique();
+            } else if (next_line.length() == 0) {
+                continue;
+            } else {
+                rv.add(next_line);
+            }
+        }
+        return rv;
+    }
+
+}

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

@@ -0,0 +1,303 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.service;
+
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.stream.Collectors;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationDocument;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationFooter;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationSection;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationHeader;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.UnclassifiedText;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.TableCell;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Table;
+
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+public class SectionsBuilderService {
+
+    public void buildSections(ClassificationDocument document) {
+
+        List<AbstractTextContainer> chunkWords = new ArrayList<>();
+        List<ClassificationSection> chunkBlockList = new ArrayList<>();
+        List<ClassificationHeader> headers = new ArrayList<>();
+        List<ClassificationFooter> footers = new ArrayList<>();
+        List<UnclassifiedText> unclassifiedTexts = new ArrayList<>();
+
+        AbstractTextContainer prev = null;
+
+        String lastHeadline = "";
+        Table previousTable = null;
+        for (ClassificationPage classificationPage : document.getPages()) {
+            List<ClassificationTextBlock> header = new ArrayList<>();
+            List<ClassificationTextBlock> footer = new ArrayList<>();
+            List<ClassificationTextBlock> unclassifiedText = new ArrayList<>();
+            for (AbstractTextContainer current : classificationPage.getTextBlocks()) {
+
+                if (current.getClassification() == null) {
+                    continue;
+                }
+
+                current.setPage(classificationPage.getPageNumber());
+
+                if (current.getClassification().equals("Header")) {
+                    header.add((ClassificationTextBlock) current);
+                    continue;
+                }
+
+                if (current.getClassification().equals("Footer")) {
+                    footer.add((ClassificationTextBlock) current);
+                    continue;
+                }
+
+                if (current.getClassification().equals("Other")) {
+                    unclassifiedText.add((ClassificationTextBlock) current);
+                    continue;
+                }
+
+                if (prev != null && current.getClassification().startsWith("H ") && !prev.getClassification().startsWith("H ") || !document.isHeadlines()) {
+                    ClassificationSection chunkBlock = buildTextBlock(chunkWords, lastHeadline);
+                    chunkBlock.setHeadline(lastHeadline);
+                    if (document.isHeadlines()) {
+                        lastHeadline = current.getText();
+                    }
+                    chunkBlockList.add(chunkBlock);
+                    chunkWords = new ArrayList<>();
+                    if (!chunkBlock.getTables().isEmpty()) {
+                        previousTable = chunkBlock.getTables().get(chunkBlock.getTables().size() - 1);
+                    }
+                }
+                if (current instanceof Table table) {
+                    // Distribute header information for subsequent tables
+                    mergeTableMetadata(table, previousTable);
+                    previousTable = table;
+                }
+                chunkWords.add(current);
+                prev = current;
+            }
+
+            if (!header.isEmpty()) {
+                headers.add(new ClassificationHeader(header));
+            }
+            if (!footer.isEmpty()) {
+                footers.add(new ClassificationFooter(footer));
+            }
+            if (!unclassifiedText.isEmpty()) {
+                unclassifiedTexts.add(new UnclassifiedText(unclassifiedText));
+            }
+        }
+
+        ClassificationSection chunkBlock = buildTextBlock(chunkWords, lastHeadline);
+        chunkBlock.setHeadline(lastHeadline);
+        chunkBlockList.add(chunkBlock);
+
+        document.setSections(chunkBlockList);
+        document.setHeaders(headers);
+        document.setFooters(footers);
+        document.setUnclassifiedTexts(unclassifiedTexts);
+        addImagesToSections(document);
+    }
+
+
+    private void addImagesToSections(ClassificationDocument document) {
+
+        Map<Integer, List<ClassificationSection>> sectionMap = new HashMap<>();
+        for (ClassificationSection section : document.getSections()) {
+            for (AbstractTextContainer container : section.getPageBlocks()) {
+
+                List<ClassificationSection> sectionsOnPage = sectionMap.computeIfAbsent(container.getPage(), c -> new ArrayList<>());
+                if (sectionsOnPage.contains(section)) {
+                    continue;
+                }
+                sectionsOnPage.add(section);
+            }
+        }
+
+        if (sectionMap.isEmpty()) {
+            ClassificationSection section = new ClassificationSection();
+            document.getSections().add(section);
+            sectionMap.computeIfAbsent(1, x -> new ArrayList<>()).add(section);
+        }
+
+        // first page is always a paragraph, else we can't process pages 1..N,
+        // where N is the first found page with a paragraph
+        if (sectionMap.get(1) == null) {
+            ClassificationSection section = new ClassificationSection();
+            document.getSections().add(section);
+            sectionMap.computeIfAbsent(1, x -> new ArrayList<>()).add(section);
+        }
+
+        for (ClassificationPage classificationPage : document.getPages()) {
+            for (ClassifiedImage image : classificationPage.getImages()) {
+                List<ClassificationSection> sectionsOnPage = sectionMap.get(classificationPage.getPageNumber());
+                if (sectionsOnPage == null) {
+                    int i = classificationPage.getPageNumber();
+                    while (sectionsOnPage == null) {
+                        sectionsOnPage = sectionMap.get(i);
+                        i--;
+                    }
+                }
+                for (ClassificationSection section : sectionsOnPage) {
+                    Float xMin = null;
+                    Float yMin = null;
+                    Float xMax = null;
+                    Float yMax = null;
+
+                    for (AbstractTextContainer abs : section.getPageBlocks()) {
+                        if (abs.getPage() != classificationPage.getPageNumber()) {
+                            continue;
+                        }
+
+                        if (abs.getMinX() < abs.getMaxX()) {
+                            if (xMin == null || abs.getMinX() < xMin) {
+                                xMin = abs.getMinX();
+                            }
+                            if (xMax == null || abs.getMaxX() > xMax) {
+                                xMax = abs.getMaxX();
+                            }
+                        } else {
+                            if (xMin == null || abs.getMaxX() < xMin) {
+                                xMin = abs.getMaxX();
+                            }
+                            if (xMax == null || abs.getMinX() > xMax) {
+                                xMax = abs.getMinX();
+                            }
+                        }
+
+                        if (abs.getMinY() < abs.getMaxY()) {
+                            if (yMin == null || abs.getMinY() < yMin) {
+                                yMin = abs.getMinY();
+                            }
+                            if (yMax == null || abs.getMaxY() > yMax) {
+                                yMax = abs.getMaxY();
+                            }
+                        } else {
+                            if (yMin == null || abs.getMaxY() < yMin) {
+                                yMin = abs.getMaxY();
+                            }
+                            if (yMax == null || abs.getMinY() > yMax) {
+                                yMax = abs.getMinY();
+                            }
+                        }
+
+                    }
+
+                    log.debug("Image position x: {}, y: {}", image.getPosition().getX(), image.getPosition().getY());
+                    log.debug("Paragraph position xMin: {}, xMax: {}, yMin: {}, yMax: {}", xMin, xMax, yMin, yMax);
+
+                    if (xMin != null && xMax != null && yMin != null && yMax != null && image.getPosition().getX() >= xMin && image.getPosition()
+                            .getX() <= xMax && image.getPosition().getY() >= yMin && image.getPosition().getY() <= yMax) {
+                        section.getImages().add(image);
+                        image.setAppendedToSection(true);
+                    }
+                }
+                if (!image.isAppendedToSection()) {
+                    log.debug("Image uses first paragraph");
+                    sectionsOnPage.get(0).getImages().add(image);
+                    image.setAppendedToSection(true);
+                }
+            }
+        }
+    }
+
+
+    private void mergeTableMetadata(Table currentTable, Table previousTable) {
+
+        // Distribute header information for subsequent tables
+        if (previousTable != null && hasInvalidHeaderInformation(currentTable) && hasValidHeaderInformation(previousTable)) {
+            List<TableCell> previousTableNonHeaderRow = getRowWithNonHeaderCells(previousTable);
+            List<TableCell> 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 -> {
+                    TableCell fakeCell = new TableCell(cell.getPoints()[0], cell.getPoints()[2]);
+                    fakeCell.setHeaderCells(Collections.singletonList(cell));
+                    return fakeCell;
+                }).collect(Collectors.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<TableCell> 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 ClassificationSection buildTextBlock(List<AbstractTextContainer> wordBlockList, String lastHeadline) {
+
+        ClassificationSection section = new ClassificationSection();
+
+        for (AbstractTextContainer container : wordBlockList) {
+            if (container instanceof Table table) {
+
+                if (lastHeadline == null || lastHeadline.isEmpty()) {
+                    table.setHeadline("Text in table");
+                } else {
+                    table.setHeadline("Table in: " + lastHeadline);
+                }
+
+                section.getPageBlocks().add(table);
+                continue;
+            }
+
+            ClassificationTextBlock wordBlock = (ClassificationTextBlock) container;
+            section.getPageBlocks().add(wordBlock);
+        }
+        return section;
+    }
+
+
+    private boolean hasValidHeaderInformation(Table table) {
+
+        return !hasInvalidHeaderInformation(table);
+    }
+
+
+    private boolean hasInvalidHeaderInformation(Table table) {
+
+        return table.getRows().stream().flatMap(row -> row.stream().filter(cell -> !cell.getHeaderCells().isEmpty())).findAny().isEmpty();
+
+    }
+
+
+    private List<TableCell> getRowWithNonHeaderCells(Table table) {
+
+        for (int i = table.getRowCount() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of table
+            List<TableCell> row = table.getRows().get(i);
+            if (row.size() == 1) {
+                continue;
+            }
+            boolean allNonHeader = true;
+            for (TableCell 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/classification/service/TableExtractionService.java

@@ -0,0 +1,338 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.service;
+
+import java.awt.geom.Point2D;
+import java.util.ArrayList;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.Iterator;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
+import java.util.stream.Collectors;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.TableCell;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.CleanRulings;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Ruling;
+import com.knecon.fforesight.service.layoutparser.processor.classification.utils.QuickSort;
+import com.knecon.fforesight.service.layoutparser.processor.classification.utils.DoubleComparisons;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.Rectangle;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Table;
+
+@Service
+public class TableExtractionService {
+
+    private static final Comparator<Point2D> X_FIRST_POINT_COMPARATOR = (arg0, arg1) -> {
+
+        int rv = 0;
+        float arg0X = DoubleComparisons.round(arg0.getX(), 2);
+        float arg0Y = DoubleComparisons.round(arg0.getY(), 2);
+        float arg1X = DoubleComparisons.round(arg1.getX(), 2);
+        float arg1Y = DoubleComparisons.round(arg1.getY(), 2);
+
+        if (arg0X > arg1X) {
+            rv = 1;
+        } else if (arg0X < arg1X) {
+            rv = -1;
+        } else if (arg0Y > arg1Y) {
+            rv = 1;
+        } else if (arg0Y < arg1Y) {
+            rv = -1;
+        }
+        return rv;
+    };
+    private static final Comparator<Point2D> POINT_COMPARATOR = (arg0, arg1) -> {
+
+        int rv = 0;
+        float arg0X = DoubleComparisons.round(arg0.getX(), 2);
+        float arg0Y = DoubleComparisons.round(arg0.getY(), 2);
+        float arg1X = DoubleComparisons.round(arg1.getX(), 2);
+        float arg1Y = DoubleComparisons.round(arg1.getY(), 2);
+
+        if (arg0Y > arg1Y) {
+            rv = 1;
+        } else if (arg0Y < arg1Y) {
+            rv = -1;
+        } else if (arg0X > arg1X) {
+            rv = 1;
+        } else if (arg0X < arg1X) {
+            rv = -1;
+        }
+        return rv;
+    };
+
+
+    /**
+     * Finds tables on a classificationPage and moves textblocks into cells of the found tables.
+     * Note: This algorithm uses Pdf Coordinate System where {0,0} rotated with the classificationPage rotation.
+     * 0 -> LowerLeft
+     * 90 -> UpperLeft
+     * 180 -> UpperRight
+     * 270 -> LowerRight
+     *
+     * DirAdj (Text direction adjusted) values can not be used here.
+     *
+     * @param cleanRulings The lines used to build the table.
+     * @param classificationPage ClassificationPage object that contains textblocks and statistics.
+     */
+    public void removeRedundantTableCells(CleanRulings cleanRulings, ClassificationPage classificationPage) {
+
+        List<TableCell> cells = findCells(cleanRulings.getHorizontal(), cleanRulings.getVertical());
+
+        List<ClassificationTextBlock> toBeRemoved = new ArrayList<>();
+
+        for (AbstractTextContainer abstractTextContainer : classificationPage.getTextBlocks()) {
+            ClassificationTextBlock textBlock = (ClassificationTextBlock) abstractTextContainer;
+            for (TableCell cell : cells) {
+                if (cell.intersects(textBlock.getPdfMinX(),
+                        textBlock.getPdfMinY(),
+                        textBlock.getPdfMaxX() - textBlock.getPdfMinX(),
+                        textBlock.getPdfMaxY() - textBlock.getPdfMinY())) {
+                    cell.addTextBlock(textBlock);
+                    toBeRemoved.add(textBlock);
+                    break;
+                }
+            }
+        }
+
+        cells = new ArrayList<>(new HashSet<>(cells));
+        QuickSort.sort(cells, Rectangle.ILL_DEFINED_ORDER);
+
+        List<Rectangle> spreadsheetAreas = findSpreadsheetsFromCells(cells).stream().filter(r -> r.getWidth() > 0f && r.getHeight() > 0f).collect(Collectors.toList());
+
+        List<Table> tables = new ArrayList<>();
+        for (Rectangle area : spreadsheetAreas) {
+
+            List<TableCell> overlappingCells = new ArrayList<>();
+            for (TableCell c : cells) {
+                if (c.intersects(area)) {
+                    overlappingCells.add(c);
+                }
+            }
+            tables.add(new Table(overlappingCells, area, classificationPage.getRotation()));
+        }
+
+        for (Table table : tables) {
+            int position = -1;
+
+            Iterator<AbstractTextContainer> itty = classificationPage.getTextBlocks().iterator();
+            while (itty.hasNext()) {
+                AbstractTextContainer textBlock = itty.next();
+                if (textBlock instanceof ClassificationTextBlock ? table.containsBlock((ClassificationTextBlock) textBlock) : table.contains(textBlock) && position == -1) {
+                    position = classificationPage.getTextBlocks().indexOf(textBlock);
+                }
+            }
+            if (position != -1) {
+                classificationPage.getTextBlocks().add(position, table);
+            }
+        }
+
+        classificationPage.getTextBlocks().removeAll(toBeRemoved);
+    }
+
+
+    public List<TableCell> findCells(List<Ruling> horizontalRulingLines, List<Ruling> verticalRulingLines) {
+
+        List<TableCell> cellsFound = new ArrayList<>();
+        Map<Point2D, Ruling[]> intersectionPoints = Ruling.findIntersections(horizontalRulingLines, verticalRulingLines);
+        List<Point2D> intersectionPointsList = new ArrayList<>(intersectionPoints.keySet());
+        intersectionPointsList.sort(POINT_COMPARATOR);
+
+        for (int i = 0; i < intersectionPointsList.size(); i++) {
+            Point2D topLeft = intersectionPointsList.get(i);
+            Ruling[] hv = intersectionPoints.get(topLeft);
+
+            // CrossingPointsDirectlyBelow( topLeft );
+            List<Point2D> xPoints = new ArrayList<>();
+            // CrossingPointsDirectlyToTheRight( topLeft );
+            List<Point2D> yPoints = new ArrayList<>();
+
+            for (Point2D p : intersectionPointsList.subList(i, intersectionPointsList.size())) {
+                if (p.getX() == topLeft.getX() && p.getY() > topLeft.getY()) {
+                    xPoints.add(p);
+                }
+                if (p.getY() == topLeft.getY() && p.getX() > topLeft.getX()) {
+                    yPoints.add(p);
+                }
+            }
+            outer:
+            for (Point2D xPoint : xPoints) {
+                // is there a vertical edge b/w topLeft and xPoint?
+                if (!hv[1].equals(intersectionPoints.get(xPoint)[1])) {
+                    continue;
+                }
+                for (Point2D yPoint : yPoints) {
+                    // is there an horizontal edge b/w topLeft and yPoint ?
+                    if (!hv[0].equals(intersectionPoints.get(yPoint)[0])) {
+                        continue;
+                    }
+                    Point2D btmRight = new Point2D.Float((float) yPoint.getX(), (float) xPoint.getY());
+                    if (intersectionPoints.containsKey(btmRight) && intersectionPoints.get(btmRight)[0].equals(intersectionPoints.get(xPoint)[0]) && intersectionPoints.get(btmRight)[1].equals(
+                            intersectionPoints.get(yPoint)[1])) {
+                        cellsFound.add(new TableCell(topLeft, btmRight));
+                        break outer;
+                    }
+                }
+            }
+        }
+
+        // TODO create cells for vertical ruling lines with aligned endpoints at the top/bottom of a grid
+        // that aren't connected with an horizontal ruler?
+        // see: https://github.com/jazzido/tabula-extractor/issues/78#issuecomment-41481207
+
+        return cellsFound;
+    }
+
+
+    private List<Rectangle> findSpreadsheetsFromCells(List<? extends Rectangle> cells) {
+        // via: http://stackoverflow.com/questions/13746284/merging-multiple-adjacent-rectangles-into-one-polygon
+        List<Rectangle> rectangles = new ArrayList<>();
+        Set<Point2D> pointSet = new HashSet<>();
+        Map<Point2D, Point2D> edgesH = new HashMap<>();
+        Map<Point2D, Point2D> edgesV = new HashMap<>();
+        int i = 0;
+
+        for (Rectangle cell : cells) {
+            for (Point2D pt : cell.getPoints()) {
+                if (pointSet.contains(pt)) { // shared vertex, remove it
+                    pointSet.remove(pt);
+                } else {
+                    pointSet.add(pt);
+                }
+            }
+        }
+
+        // X first sort
+        List<Point2D> pointsSortX = new ArrayList<>(pointSet);
+        pointsSortX.sort(X_FIRST_POINT_COMPARATOR);
+        // Y first sort
+        List<Point2D> pointsSortY = new ArrayList<>(pointSet);
+        pointsSortY.sort(POINT_COMPARATOR);
+
+        while (i < pointSet.size()) {
+            float currY = (float) pointsSortY.get(i).getY();
+            while (i < pointSet.size() && DoubleComparisons.feq(pointsSortY.get(i).getY(), currY)) {
+                edgesH.put(pointsSortY.get(i), pointsSortY.get(i + 1));
+                edgesH.put(pointsSortY.get(i + 1), pointsSortY.get(i));
+                i += 2;
+            }
+        }
+
+        i = 0;
+        while (i < pointSet.size()) {
+            float currX = (float) pointsSortX.get(i).getX();
+            while (i < pointSet.size() && DoubleComparisons.feq(pointsSortX.get(i).getX(), currX)) {
+                edgesV.put(pointsSortX.get(i), pointsSortX.get(i + 1));
+                edgesV.put(pointsSortX.get(i + 1), pointsSortX.get(i));
+                i += 2;
+            }
+        }
+
+        // Get all the polygons
+        List<List<PolygonVertex>> polygons = new ArrayList<>();
+        Point2D nextVertex;
+        while (!edgesH.isEmpty()) {
+            ArrayList<PolygonVertex> polygon = new ArrayList<>();
+            Point2D first = edgesH.keySet().iterator().next();
+            polygon.add(new PolygonVertex(first, Direction.HORIZONTAL));
+            edgesH.remove(first);
+
+            while (true) {
+                PolygonVertex curr = polygon.get(polygon.size() - 1);
+                PolygonVertex lastAddedVertex;
+                if (curr.direction == Direction.HORIZONTAL) {
+                    nextVertex = edgesV.get(curr.point);
+                    edgesV.remove(curr.point);
+                    lastAddedVertex = new PolygonVertex(nextVertex, Direction.VERTICAL);
+                } else {
+                    nextVertex = edgesH.get(curr.point);
+                    edgesH.remove(curr.point);
+                    lastAddedVertex = new PolygonVertex(nextVertex, Direction.HORIZONTAL);
+                }
+                polygon.add(lastAddedVertex);
+
+                if (lastAddedVertex.equals(polygon.get(0))) {
+                    // closed polygon
+                    polygon.remove(polygon.size() - 1);
+                    break;
+                }
+            }
+
+            for (PolygonVertex vertex : polygon) {
+                edgesH.remove(vertex.point);
+                edgesV.remove(vertex.point);
+            }
+            polygons.add(polygon);
+        }
+
+        // calculate grid-aligned minimum area rectangles for each found polygon
+        for (List<PolygonVertex> poly : polygons) {
+            float top = Float.MAX_VALUE;
+            float left = Float.MAX_VALUE;
+            float bottom = Float.MIN_VALUE;
+            float right = Float.MIN_VALUE;
+            for (PolygonVertex pt : poly) {
+                top = (float) Math.min(top, pt.point.getY());
+                left = (float) Math.min(left, pt.point.getX());
+                bottom = (float) Math.max(bottom, pt.point.getY());
+                right = (float) Math.max(right, pt.point.getX());
+            }
+            rectangles.add(new Rectangle(top, left, right - left, bottom - top));
+        }
+
+        return rectangles;
+    }
+
+
+    private enum Direction {
+        HORIZONTAL,
+        VERTICAL
+    }
+
+    static class PolygonVertex {
+
+        Point2D point;
+        Direction direction;
+
+
+        PolygonVertex(Point2D point, Direction direction) {
+
+            this.direction = direction;
+            this.point = point;
+        }
+
+
+        @Override
+        public boolean equals(Object other) {
+
+            if (this == other) {
+                return true;
+            }
+            if (!(other instanceof PolygonVertex)) {
+                return false;
+            }
+            return this.point.equals(((PolygonVertex) other).point);
+        }
+
+
+        @Override
+        public int hashCode() {
+
+            return this.point.hashCode();
+        }
+
+
+        @Override
+        public String toString() {
+
+            return String.format("%s[point=%s,direction=%s]", this.getClass().getName(), this.point.toString(), this.direction.toString());
+        }
+
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/utils/CohenSutherlandClipping.java

@@ -0,0 +1,142 @@
+/*
+ * CohenSutherland.java
+ * --------------------
+ * (c) 2007 by Intevation GmbH
+ *
+ * @author Sascha L. Teichmann (teichmann@intevation.de)
+ * @author Ludwig Reiter       (ludwig@intevation.de)
+ *
+ * This program is free software under the LGPL (>=v2.1)
+ * Read the file LICENSE.txt coming with the sources for details.
+ */
+package com.knecon.fforesight.service.layoutparser.processor.classification.utils;
+
+import java.awt.geom.Line2D;
+import java.awt.geom.Rectangle2D;
+
+/**
+ * Implements the well known Cohen Sutherland line
+ * clipping algorithm (line against clip rectangle).
+ */
+@SuppressWarnings("all")
+public final class CohenSutherlandClipping {
+
+    private static final int INSIDE = 0;
+    private static final int LEFT = 1;
+    private static final int RIGHT = 2;
+    private static final int BOTTOM = 4;
+    private static final int TOP = 8;
+    private double xMin;
+    private double yMin;
+    private double xMax;
+    private double yMax;
+
+
+    /**
+     * Creates a Cohen Sutherland clipper with clip rect (0, 0, 0, 0).
+     */
+    public CohenSutherlandClipping() {
+
+    }
+
+
+    /**
+     * Creates a Cohen Sutherland clipper with the given clip rectangle.
+     *
+     * @param clip the clip rectangle to use
+     */
+    public CohenSutherlandClipping(Rectangle2D clip) {
+
+        setClip(clip);
+    }
+
+
+    /**
+     * Sets the clip rectangle.
+     *
+     * @param clip the clip rectangle
+     */
+    public void setClip(Rectangle2D clip) {
+
+        xMin = clip.getX();
+        xMax = xMin + clip.getWidth();
+        yMin = clip.getY();
+        yMax = yMin + clip.getHeight();
+    }
+
+
+    private final int regionCode(double x, double y) {
+
+        int code = x < xMin ? LEFT : x > xMax ? RIGHT : INSIDE;
+        if (y < yMin) {
+            code |= BOTTOM;
+        } else if (y > yMax) {
+            code |= TOP;
+        }
+        return code;
+    }
+
+
+    /**
+     * Clips a given line against the clip rectangle.
+     * The modification (if needed) is done in place.
+     *
+     * @param line the line to clip
+     * @return true if line is clipped, false if line is
+     * totally outside the clip rect.
+     */
+    public boolean clip(Line2D.Float line) {
+
+        double p1x = line.getX1();
+        double p1y = line.getY1();
+        double p2x = line.getX2();
+        double p2y = line.getY2();
+
+        double qx = 0d;
+        double qy = 0d;
+
+        boolean vertical = p1x == p2x;
+
+        double slope = vertical ? 0d : (p2y - p1y) / (p2x - p1x);
+
+        int c1 = regionCode(p1x, p1y);
+        int c2 = regionCode(p2x, p2y);
+
+        while (c1 != INSIDE || c2 != INSIDE) {
+
+            if ((c1 & c2) != INSIDE) {
+                return false;
+            }
+
+            int c = c1 == INSIDE ? c2 : c1;
+
+            if ((c & LEFT) != INSIDE) {
+                qx = xMin;
+                qy = (DoubleComparisons.feq(qx, p1x) ? 0 : qx - p1x) * slope + p1y;
+            } else if ((c & RIGHT) != INSIDE) {
+                qx = xMax;
+                qy = (DoubleComparisons.feq(qx, p1x) ? 0 : qx - p1x) * slope + p1y;
+            } else if ((c & BOTTOM) != INSIDE) {
+                qy = yMin;
+                qx = vertical ? p1x : (DoubleComparisons.feq(qy, p1y) ? 0 : qy - p1y) / slope + p1x;
+            } else if ((c & TOP) != INSIDE) {
+                qy = yMax;
+                qx = vertical ? p1x : (DoubleComparisons.feq(qy, p1y) ? 0 : qy - p1y) / slope + p1x;
+            }
+
+            if (c == c1) {
+                p1x = qx;
+                p1y = qy;
+                c1 = regionCode(p1x, p1y);
+            } else {
+                p2x = qx;
+                p2y = qy;
+                c2 = regionCode(p2x, p2y);
+            }
+        }
+        line.setLine(p1x, p1y, p2x, p2y);
+        return true;
+    }
+
+}
+// end of file

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/utils/DoubleComparisons.java

@@ -0,0 +1,30 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.utils;
+
+import java.math.BigDecimal;
+
+import lombok.experimental.UtilityClass;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@UtilityClass
+@SuppressWarnings("all")
+public final class DoubleComparisons {
+
+    private final static float EPSILON = 0.1f;
+
+
+    public static boolean feq(double f1, double f2) {
+
+        return (Math.abs(f1 - f2) < EPSILON);
+    }
+
+
+    public static float round(double d, int decimalPlace) {
+        BigDecimal bd = BigDecimal.valueOf(d);
+        bd = bd.setScale(decimalPlace, BigDecimal.ROUND_HALF_UP);
+        return bd.floatValue();
+    }
+
+
+}
+

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/utils/PositionUtils.java

@@ -0,0 +1,119 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.utils;
+
+import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlog.Rectangle;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+
+import lombok.experimental.UtilityClass;
+
+@UtilityClass
+@SuppressWarnings("all")
+public final class PositionUtils {
+
+    // TODO This currently uses pdf coord system. In the futher this should use java coord system.
+    // Note: DirAdj (TextDirection Adjusted) can not be user for this.
+    public boolean isWithinBodyTextFrame(Rectangle btf, ClassificationTextBlock textBlock) {
+
+        if (btf == null || textBlock == null) {
+            return false;
+        }
+
+        double threshold = textBlock.getMostPopularWordHeight() * 3;
+
+        if (textBlock.getPdfMinX() + threshold > btf.getTopLeft().getX() && textBlock.getPdfMaxX() - threshold < btf.getTopLeft()
+                .getX() + btf.getWidth() && textBlock.getPdfMinY() + threshold > btf.getTopLeft().getY() && textBlock.getPdfMaxY() - threshold < btf.getTopLeft()
+                .getY() + btf.getHeight()) {
+            return true;
+        } else {
+            return false;
+        }
+
+    }
+
+
+    // TODO This currently uses pdf coord system. In the futher this should use java coord system.
+    // Note: DirAdj (TextDirection Adjusted) can not be user for this.
+    public boolean isOverBodyTextFrame(Rectangle btf, ClassificationTextBlock textBlock, int rotation) {
+
+        if (btf == null || textBlock == null) {
+            return false;
+        }
+
+        if (rotation == 90 && textBlock.getPdfMaxX() < btf.getTopLeft().getX()) {
+            return true;
+        }
+
+        if (rotation == 180 && textBlock.getPdfMaxY() < btf.getTopLeft().getY()) {
+            return true;
+        }
+
+        if (rotation == 270 && textBlock.getPdfMinX() > btf.getTopLeft().getX() + btf.getWidth()) {
+            return true;
+        }
+
+        if (rotation == 0 && textBlock.getPdfMinY() > btf.getTopLeft().getY() + btf.getHeight()) {
+            return true;
+        } else {
+            return false;
+        }
+
+    }
+
+    // TODO This currently uses pdf coord system. In the futher this should use java coord system.
+    // Note: DirAdj (TextDirection Adjusted) can not be user for this.
+    public boolean isUnderBodyTextFrame(Rectangle btf, ClassificationTextBlock textBlock, int rotation) {
+
+        if (btf == null || textBlock == null) {
+            return false;
+        }
+
+        if (rotation == 90 && textBlock.getPdfMinX() > btf.getTopLeft().getX() + btf.getWidth()) {
+            return true;
+        }
+
+        if (rotation == 180 && textBlock.getPdfMinY() > btf.getTopLeft().getY() + btf.getHeight()) {
+            return true;
+        }
+
+        if (rotation == 270 && textBlock.getPdfMaxX() < btf.getTopLeft().getX()) {
+            return true;
+        }
+
+        if (rotation == 0 && textBlock.getPdfMaxY() < btf.getTopLeft().getY()) {
+            return true;
+        } else {
+            return false;
+        }
+
+    }
+
+    // TODO This currently uses pdf coord system. In the futher this should use java coord system.
+    // Note: DirAdj (TextDirection Adjusted) can not be user for this.
+    public boolean isTouchingUnderBodyTextFrame(Rectangle btf, ClassificationTextBlock textBlock) {
+
+        //TODO Currently this is not working for rotated pages.
+
+        if (btf == null || textBlock == null) {
+            return false;
+        }
+
+        if (textBlock.getMinY() < btf.getTopLeft().getY()) {
+            return true;
+        } else {
+            return false;
+        }
+
+    }
+
+
+    public float getHeightDifferenceBetweenChunkWordAndDocumentWord(ClassificationTextBlock textBlock, Float documentMostPopularWordHeight) {
+
+        return textBlock.getMostPopularWordHeight() - documentMostPopularWordHeight;
+    }
+
+
+    public Float getApproxLineCount(ClassificationTextBlock textBlock) {
+
+        return textBlock.getHeight() / textBlock.getMostPopularWordHeight();
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/utils/QuickSort.java

@@ -0,0 +1,109 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.utils;
+
+import java.util.ArrayDeque;
+import java.util.Comparator;
+import java.util.Deque;
+import java.util.List;
+
+import lombok.experimental.UtilityClass;
+
+/**
+ * Copied and minimal modified from PDFBox.
+ */
+@UtilityClass
+public final class QuickSort {
+
+    private static final Comparator<? extends Comparable> OBJCOMP = new Comparator<Comparable>() {
+        @Override
+        public int compare(Comparable object1, Comparable object2) {
+
+            return object1.compareTo(object2);
+        }
+    };
+
+
+    /**
+     * Sorts the given list using the given comparator.
+     *
+     * @param <T>  type of the objects to be sorted.
+     * @param list list to be sorted
+     * @param cmp  comparator used to compare the objects within the list
+     */
+    public static <T> void sort(List<T> list, Comparator<? super T> cmp) {
+
+        int size = list.size();
+        if (size < 2) {
+            return;
+        }
+        quicksort(list, cmp);
+    }
+
+
+    /**
+     * Sorts the given list using compareTo as comparator.
+     *
+     * @param <T>  type of the objects to be sorted.
+     * @param list list to be sorted
+     */
+    public static <T extends Comparable> void sort(List<T> list) {
+
+        sort(list, (Comparator<T>) OBJCOMP);
+    }
+
+
+    private static <T> void quicksort(List<T> list, Comparator<? super T> cmp) {
+
+        Deque<Integer> stack = new ArrayDeque<Integer>();
+        stack.push(0);
+        stack.push(list.size());
+        while (!stack.isEmpty()) {
+            int right = stack.pop();
+            int left = stack.pop();
+            if (right - left < 2) {
+                continue;
+            }
+            int p = left + ((right - left) / 2);
+            p = partition(list, cmp, p, left, right);
+
+            stack.push(p + 1);
+            stack.push(right);
+
+            stack.push(left);
+            stack.push(p);
+        }
+    }
+
+
+    private static <T> int partition(List<T> list, Comparator<? super T> cmp, int p, int start, int end) {
+
+        int l = start;
+        int h = end - 2;
+        T piv = list.get(p);
+        swap(list, p, end - 1);
+
+        while (l < h) {
+            if (cmp.compare(list.get(l), piv) <= 0) {
+                l++;
+            } else if (cmp.compare(piv, list.get(h)) <= 0) {
+                h--;
+            } else {
+                swap(list, l, h);
+            }
+        }
+        int idx = h;
+        if (cmp.compare(list.get(h), piv) < 0) {
+            idx++;
+        }
+        swap(list, end - 1, idx);
+        return idx;
+    }
+
+
+    private static <T> void swap(List<T> list, int i, int j) {
+
+        T tmp = list.get(i);
+        list.set(i, list.get(j));
+        list.set(j, tmp);
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/utils/RulingTextDirAdjustUtil.java

@@ -0,0 +1,64 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.utils;
+
+import java.awt.geom.Line2D;
+import java.awt.geom.Point2D;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Ruling;
+
+import lombok.experimental.UtilityClass;
+
+@UtilityClass
+public final class RulingTextDirAdjustUtil {
+
+    /**
+     * Converts a ruling (line of a table) the same way TextPositions are converted in PDFBox.
+     * This will get the y position of the text, adjusted so that 0,0 is upper left and it is adjusted based on the text direction.
+     *
+     * See org.apache.pdfbox.text.TextPosition
+     */
+    public Line2D.Float convertToDirAdj(Ruling ruling, float dir, float pageWidth, float pageHeight) {
+
+        return new Line2D.Float(convertPoint(ruling.x1, ruling.y1, dir, pageWidth, pageHeight), convertPoint(ruling.x2, ruling.y2, dir, pageWidth, pageHeight));
+    }
+
+
+    private Point2D convertPoint(float x, float y, float dir, float pageWidth, float pageHeight) {
+
+        var xAdj = getXRot(x, y, dir, pageWidth, pageHeight);
+        var yLowerLeftRot = getYLowerLeftRot(x, y, dir, pageWidth, pageHeight);
+        var yAdj = dir == 0 || dir == 180 ? pageHeight - yLowerLeftRot : pageWidth - yLowerLeftRot;
+        return new Point2D.Float(xAdj, yAdj);
+    }
+
+
+    @SuppressWarnings("SuspiciousNameCombination")
+    private float getXRot(float x, float y, float dir, float pageWidth, float pageHeight) {
+
+        if (dir == 0) {
+            return x;
+        } else if (dir == 90) {
+            return y;
+        } else if (dir == 180) {
+            return pageWidth - x;
+        } else if (dir == 270) {
+            return pageHeight - y;
+        }
+        return 0;
+    }
+
+
+    private float getYLowerLeftRot(float x, float y, float dir, float pageWidth, float pageHeight) {
+
+        if (dir == 0) {
+            return y;
+        } else if (dir == 90) {
+            return pageWidth - x;
+        } else if (dir == 180) {
+            return pageHeight - y;
+        } else if (dir == 270) {
+            return x;
+        }
+        return 0;
+    }
+
+}

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/classification/utils/TextNormalizationUtilities.java

@@ -0,0 +1,19 @@
+package com.knecon.fforesight.service.layoutparser.processor.classification.utils;
+
+import lombok.experimental.UtilityClass;
+
+@UtilityClass
+public final class TextNormalizationUtilities {
+
+    /**
+     * Revert hyphenation due to line breaks.
+     *
+     * @param text Text to be processed.
+     * @return Text without line-break hyphenation.
+     */
+    public static String removeHyphenLineBreaks(String text) {
+
+        return text.replaceAll("([^\\s\\d\\-]{2,500})[\\-\\u00AD]\\R", "$1");
+    }
+
+}

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

@@ -0,0 +1,386 @@
+package com.knecon.fforesight.service.layoutparser.processor.factory;
+
+import static java.lang.String.format;
+import static java.util.stream.Collectors.groupingBy;
+import static java.util.stream.Collectors.toList;
+
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.Map;
+import java.util.NoSuchElementException;
+import java.util.Set;
+import java.util.concurrent.atomic.AtomicInteger;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.DocumentGraph;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.TableOfContents;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.FooterNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.HeaderNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.HeadlineNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.ImageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.NodeType;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.PageNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.ParagraphNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SectionNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.SemanticNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.TableCellNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.nodes.TableNode;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.AtomicTextBlock;
+import com.knecon.fforesight.service.layoutparser.internal.api.graph.textblock.TextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationDocument;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationFooter;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationHeader;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationPage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Table;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.TableCell;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.TextPositionSequence;
+
+
+@Service
+public class DocumentGraphFactory {
+
+    public static final double TABLE_CELL_MERGE_CONTENTS_SIZE_THRESHOLD = 0.05;
+
+
+    public DocumentGraph buildDocumentGraph(ClassificationDocument document) {
+
+        TextBlockFactory textBlockFactory = new TextBlockFactory();
+        Context context = new Context(new TableOfContents(), new HashMap<>(), new LinkedList<>(), new LinkedList<>(), textBlockFactory);
+
+        document.getPages().stream().map(this::buildPage).forEach(page -> context.pages().put(page, new AtomicInteger(1)));
+        document.getSections().stream().flatMap(section -> section.getImages().stream()).forEach(image -> context.images().add(image));
+        addSections(document, context);
+        addHeaderAndFooterToEachPage(document, context);
+
+        DocumentGraph documentGraph = DocumentGraph.builder().numberOfPages(context.pages.size()).pages(context.pages.keySet()).tableOfContents(context.tableOfContents).build();
+
+        documentGraph.setTextBlock(documentGraph.buildTextBlock());
+        return documentGraph;
+    }
+
+
+    private void addSections(ClassificationDocument document, Context context) {
+
+        document.getSections().forEach(section -> addSection(null, section.getPageBlocks(), section.getImages(), context));
+    }
+
+
+    private void addSection(SemanticNode parentNode, List<AbstractTextContainer> pageBlocks, List<ClassifiedImage> images, Context context) {
+
+        Map<Integer, List<AbstractTextContainer>> blocksPerPage = pageBlocks.stream().collect(groupingBy(AbstractTextContainer::getPage));
+        SectionNode sectionNode = SectionNode.builder().entities(new HashSet<>()).tableOfContents(context.tableOfContents()).build();
+
+        context.sections().add(sectionNode);
+        blocksPerPage.keySet().forEach(pageNumber -> addSectionNodeToPageNode(context, sectionNode, pageNumber));
+
+        List<Integer> tocId;
+        if (parentNode == null) {
+            tocId = context.tableOfContents.createNewEntryAndReturnId(NodeType.SECTION, sectionNode);
+        } else {
+            tocId = context.tableOfContents.createNewChildEntryAndReturnId(parentNode.getTocId(), NodeType.SECTION, sectionNode);
+        }
+        sectionNode.setTocId(tocId);
+        Set<AbstractTextContainer> alreadyMerged = new HashSet<>();
+        for (AbstractTextContainer abstractTextContainer : pageBlocks) {
+
+            if (alreadyMerged.contains(abstractTextContainer)) {
+                continue;
+            }
+
+            if (abstractTextContainer instanceof ClassificationTextBlock) {
+                List<ClassificationTextBlock> textBlocks = findTextBlocksWithSameClassificationAndAlignsY(abstractTextContainer, pageBlocks);
+                alreadyMerged.addAll(textBlocks);
+                addParagraphOrHeadline(sectionNode, (ClassificationTextBlock) abstractTextContainer, context, textBlocks);
+            }
+            if (abstractTextContainer instanceof Table) {
+                addTable(sectionNode, (Table) abstractTextContainer, context);
+            }
+        }
+        for (ClassifiedImage image : images) {
+
+            addImage(sectionNode, image, context);
+        }
+    }
+
+
+    private static List<ClassificationTextBlock> findTextBlocksWithSameClassificationAndAlignsY(AbstractTextContainer atc, List<AbstractTextContainer> pageBlocks) {
+
+        return pageBlocks.stream()
+                .filter(abstractTextContainer -> !abstractTextContainer.equals(atc))
+                .filter(abstractTextContainer -> abstractTextContainer.getPage() == atc.getPage())
+                .filter(abstractTextContainer -> abstractTextContainer instanceof ClassificationTextBlock)
+                .filter(abstractTextContainer -> abstractTextContainer.intersectsY(atc))
+                .map(abstractTextContainer -> (ClassificationTextBlock) abstractTextContainer)
+                .toList();
+    }
+
+
+    private void addSectionNodeToPageNode(Context context, SectionNode sectionNode, Integer pageNumber) {
+
+        PageNode page = getPage(pageNumber, context);
+        page.getMainBody().add(sectionNode);
+    }
+
+
+    private void addTable(SemanticNode parentNode, Table table, Context context) {
+
+        PageNode page = getPage(table.getPage(), context);
+        TableNode tableNode = TableNode.builder().tableOfContents(context.tableOfContents()).numberOfCols(table.getColCount()).numberOfRows(table.getRowCount()).build();
+
+        if (!page.getMainBody().contains(parentNode)) {
+            parentNode.getPages().add(page);
+        }
+
+        page.getMainBody().add(tableNode);
+
+        List<Integer> tocId = context.tableOfContents().createNewChildEntryAndReturnId(parentNode.getTocId(), NodeType.TABLE, tableNode);
+        tableNode.setTocId(tocId);
+
+        addTableCells(table.getRows(), tableNode, context, table.getPage());
+    }
+
+
+    private void addTableCells(List<List<TableCell>> rows, SemanticNode parentNode, Context context, int pageNumber) {
+
+        for (int rowIndex = 0; rowIndex < rows.size(); rowIndex++) {
+            for (int colIndex = 0; colIndex < rows.get(rowIndex).size(); colIndex++) {
+                addTableCell(rows.get(rowIndex).get(colIndex), rowIndex, colIndex, parentNode, pageNumber, context);
+            }
+        }
+    }
+
+
+    private void addTableCell(TableCell cell, int rowIndex, int colIndex, SemanticNode parentNode, int pageNumber, Context context) {
+
+        PageNode page = getPage(pageNumber, context);
+        cell.getTextBlocks().stream().filter(tb -> tb.getPage() == 0).forEach(tb -> tb.setPage(pageNumber));
+
+        TableCellNode tableCellNode = TableCellNode.builder()
+                .tableOfContents(context.tableOfContents())
+                .row(rowIndex)
+                .col(colIndex)
+                .header(cell.isHeaderCell())
+                .bBox(cell.getBounds2D())
+                .build();
+        page.getMainBody().add(tableCellNode);
+
+        TextBlock textBlock;
+
+        List<Integer> tocId = context.tableOfContents().createNewChildEntryAndReturnId(parentNode.getTocId(), NodeType.TABLE_CELL, tableCellNode);
+        tableCellNode.setTocId(tocId);
+
+        if (cell.getTextBlocks().isEmpty()) {
+            tableCellNode.setTerminalTextBlock(context.textBlockFactory.emptyTextBlock(parentNode, context, page));
+            tableCellNode.setTerminal(true);
+
+        } else if (cell.getTextBlocks().size() == 1) {
+            textBlock = context.textBlockFactory().buildAtomicTextBlock(cell.getTextBlocks().get(0).getSequences(), tableCellNode, context, page);
+            tableCellNode.setTerminalTextBlock(textBlock);
+            tableCellNode.setTerminal(true);
+
+        } else if (firstTextBlockIsHeadline(cell)) {
+            addSection(tableCellNode, cell.getTextBlocks().stream().map(tb -> (AbstractTextContainer) tb).toList(), Collections.emptyList(), context);
+            tableCellNode.setTerminal(false);
+
+        } else if (cellAreaIsSmallerThanPageAreaTimesThreshold(cell, page)) {
+            List<TextPositionSequence> sequences = TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(cell.getTextBlocks());
+            textBlock = context.textBlockFactory().buildAtomicTextBlock(sequences, tableCellNode, context, page);
+            tableCellNode.setTerminalTextBlock(textBlock);
+            tableCellNode.setTerminal(true);
+
+        } else {
+            cell.getTextBlocks().forEach(tb -> addParagraphOrHeadline(tableCellNode, tb, context));
+            tableCellNode.setTerminal(false);
+        }
+
+    }
+
+
+    private static boolean cellAreaIsSmallerThanPageAreaTimesThreshold(TableCell cell, PageNode page) {
+
+        return cell.getArea() < TABLE_CELL_MERGE_CONTENTS_SIZE_THRESHOLD * page.getHeight() * page.getWidth();
+    }
+
+
+    private static boolean firstTextBlockIsHeadline(TableCell cell) {
+
+        String classification = cell.getTextBlocks().get(0).getClassification();
+        return classification != null && classification.startsWith("H");
+    }
+
+
+    private void addParagraphOrHeadline(SemanticNode parentNode, ClassificationTextBlock originalTextBlock, Context context) {
+
+        addParagraphOrHeadline(parentNode, originalTextBlock, context, Collections.emptyList());
+    }
+
+
+    private void addParagraphOrHeadline(SemanticNode parentNode, ClassificationTextBlock originalTextBlock, Context context, List<ClassificationTextBlock> textBlocksToMerge) {
+
+        PageNode page = getPage(originalTextBlock.getPage(), context);
+
+        SemanticNode node;
+        if (originalTextBlock.getClassification() != null && originalTextBlock.getClassification().startsWith("H")) {
+            node = HeadlineNode.builder().tableOfContents(context.tableOfContents()).build();
+        } else {
+            node = ParagraphNode.builder().tableOfContents(context.tableOfContents()).build();
+        }
+
+        page.getMainBody().add(node);
+
+        List<ClassificationTextBlock> textBlocks = new LinkedList<>(textBlocksToMerge);
+        textBlocks.add(originalTextBlock);
+        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(textBlocks), node, context, page);
+
+        if (node instanceof HeadlineNode headlineNode) {
+            List<Integer> tocId = context.tableOfContents.createNewChildEntryAndReturnId(parentNode.getTocId(), NodeType.HEADLINE, node);
+            headlineNode.setTerminalTextBlock(textBlock);
+            headlineNode.setTocId(tocId);
+        }
+        if (node instanceof ParagraphNode paragraphNode) {
+            List<Integer> tocId = context.tableOfContents.createNewChildEntryAndReturnId(parentNode.getTocId(), NodeType.PARAGRAPH, node);
+            paragraphNode.setTerminalTextBlock(textBlock);
+            paragraphNode.setTocId(tocId);
+        }
+    }
+
+
+    private void addImage(SectionNode sectionNode, ClassifiedImage image, Context context) {
+
+        PageNode page = getPage(image.getPage(), context);
+        ImageNode imageNode = ImageNode.builder()
+                .imageType(image.getImageType())
+                .position(image.getPosition())
+                .transparency(image.isHasTransparency())
+                .page(page)
+                .tableOfContents(context.tableOfContents())
+                .build();
+        page.getMainBody().add(imageNode);
+
+        List<Integer> tocId = context.tableOfContents().createNewChildEntryAndReturnId(sectionNode.getTocId(), NodeType.IMAGE, imageNode);
+        imageNode.setTocId(tocId);
+    }
+
+
+    private void addHeaderAndFooterToEachPage(ClassificationDocument document, Context context) {
+
+        Map<Integer, List<ClassificationTextBlock>> headers = document.getHeaders()
+                .stream()
+                .map(ClassificationHeader::getTextBlocks)
+                .flatMap(List::stream)
+                .collect(groupingBy(AbstractTextContainer::getPage, toList()));
+
+        Map<Integer, List<ClassificationTextBlock>> footers = document.getFooters()
+                .stream()
+                .map(ClassificationFooter::getTextBlocks)
+                .flatMap(List::stream)
+                .collect(groupingBy(AbstractTextContainer::getPage, toList()));
+
+        for (int pageIndex = 1; pageIndex <= document.getPages().size(); pageIndex++) {
+            if (headers.containsKey(pageIndex)) {
+                addHeader(headers.get(pageIndex), context);
+            } else {
+                addEmptyHeader(pageIndex, context);
+            }
+        }
+
+        for (int pageIndex = 1; pageIndex <= document.getPages().size(); pageIndex++) {
+            if (footers.containsKey(pageIndex)) {
+                addFooter(footers.get(pageIndex), context);
+            } else {
+                addEmptyFooter(pageIndex, context);
+            }
+        }
+    }
+
+
+    private void addFooter(List<ClassificationTextBlock> textBlocks, Context context) {
+
+        PageNode page = getPage(textBlocks.get(0).getPage(), context);
+        FooterNode footer = FooterNode.builder().tableOfContents(context.tableOfContents()).build();
+        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(textBlocks),
+                footer,
+                context,
+                page);
+        List<Integer> tocId = context.tableOfContents().createNewEntryAndReturnId(NodeType.FOOTER, footer);
+        footer.setTocId(tocId);
+        footer.setTerminalTextBlock(textBlock);
+        page.setFooter(footer);
+    }
+
+
+    public void addHeader(List<ClassificationTextBlock> textBlocks, Context context) {
+
+        PageNode page = getPage(textBlocks.get(0).getPage(), context);
+        HeaderNode header = HeaderNode.builder().tableOfContents(context.tableOfContents()).build();
+        AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(textBlocks),
+                header,
+                context,
+                0,
+                page);
+        List<Integer> tocId = context.tableOfContents().createNewEntryAndReturnId(NodeType.HEADER, header);
+        header.setTocId(tocId);
+        header.setTerminalTextBlock(textBlock);
+        page.setHeader(header);
+    }
+
+
+    private void addEmptyFooter(int pageIndex, Context context) {
+
+        PageNode page = getPage(pageIndex, context);
+        FooterNode footer = FooterNode.builder().tableOfContents(context.tableOfContents()).build();
+        AtomicTextBlock textBlock = context.textBlockFactory.emptyTextBlock(footer, context, page);
+        List<Integer> tocId = context.tableOfContents().createNewEntryAndReturnId(NodeType.FOOTER, footer);
+        footer.setTocId(tocId);
+        footer.setTerminalTextBlock(textBlock);
+        page.setFooter(footer);
+    }
+
+
+    private void addEmptyHeader(int pageIndex, Context context) {
+
+        PageNode page = getPage(pageIndex, context);
+        HeaderNode header = HeaderNode.builder().tableOfContents(context.tableOfContents()).build();
+        AtomicTextBlock textBlock = context.textBlockFactory.emptyTextBlock(header, 0, page);
+        List<Integer> tocId = context.tableOfContents().createNewEntryAndReturnId(NodeType.HEADER, header);
+        header.setTocId(tocId);
+        header.setTerminalTextBlock(textBlock);
+        page.setHeader(header);
+    }
+
+
+    private PageNode buildPage(ClassificationPage p) {
+
+        return PageNode.builder()
+                .height((int) p.getPageHeight())
+                .width((int) p.getPageWidth())
+                .number(p.getPageNumber())
+                .rotation(p.getRotation())
+                .mainBody(new LinkedList<>())
+                .build();
+    }
+
+
+    private PageNode getPage(int pageIndex, Context context) {
+
+        return context.pages.keySet()
+                .stream()
+                .filter(page -> page.getNumber() == pageIndex)
+                .findFirst()
+                .orElseThrow(() -> new NoSuchElementException(format("ClassificationPage with number %d not found", pageIndex)));
+    }
+
+
+    record Context(
+            TableOfContents tableOfContents, Map<PageNode, AtomicInteger> pages, List<SectionNode> sections, List<ClassifiedImage> images, TextBlockFactory textBlockFactory) {
+
+    }
+
+}

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

@@ -0,0 +1,132 @@
+package com.knecon.fforesight.service.layoutparser.processor.factory;
+
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.Optional;
+import java.util.stream.Collectors;
+
+import org.springframework.stereotype.Service;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationDocument;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.ClassificationSection;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.image.ClassifiedImage;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.Table;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.table.TableCell;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.ClassificationTextBlock;
+
+@Service
+public class ImageSortService {
+
+    public SortedImages sortImagesIntoStructure(ClassificationDocument document) {
+
+        SortedImages sortedImages = new SortedImages(new HashMap<>(), new HashMap<>(), new HashMap<>(), new HashMap<>(), new HashMap<>());
+
+        Map<Integer, List<ClassifiedImage>> imagesByPage = document.getSections()
+                .stream()
+                .flatMap(section -> section.getImages().stream())
+                .distinct()
+                .collect(Collectors.groupingBy(ClassifiedImage::getPage));
+
+        for (int pageNumber : imagesByPage.keySet()) {
+            List<AbstractTextContainer> textContainersOnPage = document.getSections()
+                    .stream()
+                    .flatMap(section -> section.getPageBlocks().stream())
+                    .filter(abstractTextContainer -> abstractTextContainer.getPage() == pageNumber)
+                    .toList();
+
+            List<ClassificationSection> sectionsOnPage = document.getSections()
+                    .stream()
+                    .filter(section -> section.getPageBlocks().stream().anyMatch(block -> block.getPage() == pageNumber))
+                    .toList();
+
+            for (ClassifiedImage image : imagesByPage.get(pageNumber)) {
+                sortImage(textContainersOnPage, sectionsOnPage, image, sortedImages);
+            }
+        }
+        return sortedImages;
+    }
+
+
+    private void sortImage(List<AbstractTextContainer> textContainersOnPage, List<ClassificationSection> sectionsOnPage, ClassifiedImage image, SortedImages sortedImages) {
+
+        Optional<AbstractTextContainer> containingTextContainer = getContainingTextContainer(image, textContainersOnPage);
+        Optional<ClassificationSection> sectionContainingTextContainer = getContainingSection(image, sectionsOnPage);
+        List<AbstractTextContainer> containedTextContainers = getContainedTextContainers(image, textContainersOnPage);
+        List<ClassificationSection> containedSections = getContainedSections(image, sectionsOnPage);
+        if (containingTextContainer.isPresent()) {
+            if (sortImageIntoTextContainerOrCell(image, sortedImages, containingTextContainer.get())) {
+                return;
+            }
+        }
+    }
+
+
+    private static boolean sortImageIntoTextContainerOrCell(ClassifiedImage image, SortedImages sortedImages, AbstractTextContainer containingTextContainer) {
+
+        if (containingTextContainer instanceof ClassificationTextBlock) {
+            sortedImages.containedInTextContainer().computeIfAbsent(containingTextContainer, sortedImage -> new ArrayList<>()).add(image);
+            return true;
+        }
+        if (containingTextContainer instanceof Table) {
+            Optional<TableCell> containingCell = getContainingCell((Table) containingTextContainer, image);
+            if (containingCell.isPresent()) {
+                sortedImages.containedInCell().computeIfAbsent(containingCell.get(), sortedImage -> new ArrayList<>()).add(image);
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+
+    private static Optional<TableCell> getContainingCell(Table table, ClassifiedImage image) {
+
+        return table.getRows().stream().flatMap(List::stream).filter(cell -> cell.contains(image.getPosition())).findFirst();
+    }
+
+
+    private List<ClassificationSection> getContainedSections(ClassifiedImage image, List<ClassificationSection> sectionsOnPage) {
+
+        return sectionsOnPage.stream()
+                .filter(section -> image.getPosition().contains(RectangleTransformations.bBoxUnionAbstractTextContainer(section.getPageBlocks()
+                        .stream()
+                        .filter(block -> block.getPage() == image.getPage())
+                        .toList())))
+                .toList();
+    }
+
+
+    private List<AbstractTextContainer> getContainedTextContainers(ClassifiedImage image, List<AbstractTextContainer> textContainersOnPage) {
+
+        return textContainersOnPage.stream().filter(textContainer -> image.getPosition().contains(RectangleTransformations.toRectangle2D(textContainer))).toList();
+    }
+
+
+    private Optional<ClassificationSection> getContainingSection(ClassifiedImage image, List<ClassificationSection> sectionsOnPage) {
+
+        return sectionsOnPage.stream()//
+                .filter(section -> //
+                        RectangleTransformations.bBoxUnionAbstractTextContainer(section.getPageBlocks().stream().filter(block -> block.getPage() == image.getPage()).toList())//
+                                .contains(image.getPosition())).findFirst();
+    }
+
+
+    private Optional<AbstractTextContainer> getContainingTextContainer(ClassifiedImage image, List<AbstractTextContainer> textContainersOnPage) {
+
+        return textContainersOnPage.stream().filter(textContainer -> RectangleTransformations.toRectangle2D(textContainer).contains(image.getPosition())).findFirst();
+    }
+
+
+    public record SortedImages(
+            Map<TableCell, List<ClassifiedImage>> containedInCell,
+            Map<AbstractTextContainer, List<ClassifiedImage>> containedInTextContainer,
+            Map<ClassificationSection, List<ClassifiedImage>> containedInSection,
+            Map<ClassifiedImage, List<AbstractTextContainer>> containedByImage,
+            Map<ClassifiedImage, List<ClassificationSection>> sectionContainedByImage) {
+
+    }
+
+}

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

@@ -0,0 +1,105 @@
+package com.knecon.fforesight.service.layoutparser.processor.factory;
+
+import static java.lang.String.format;
+
+import java.awt.geom.Area;
+import java.awt.geom.Rectangle2D;
+import java.util.Arrays;
+import java.util.List;
+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 org.apache.pdfbox.pdmodel.common.PDRectangle;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.AbstractTextContainer;
+
+public class RectangleTransformations {
+
+    public static Rectangle2D pad(Rectangle2D rectangle2D, int deltaX, int deltaY) {
+
+        return new Rectangle2D.Double(rectangle2D.getMinX() - deltaX, rectangle2D.getMinY() - deltaY, rectangle2D.getWidth() + 2 * deltaX, rectangle2D.getHeight() + 2 * deltaY);
+    }
+
+
+    public static Rectangle2D bBoxUnionAbstractTextContainer(List<AbstractTextContainer> abstractTextContainers) {
+
+        return abstractTextContainers.stream().map(RectangleTransformations::toRectangle2D).collect(new Rectangle2DUnion());
+    }
+
+    public static Rectangle2D rectangleUnion(List<Rectangle2D> rectangle2DList) {
+
+        return rectangle2DList.stream().collect(new Rectangle2DUnion());
+    }
+
+
+    public static Rectangle2D toRectangle2D(AbstractTextContainer abstractTextContainer) {
+
+        return new Rectangle2D.Float(abstractTextContainer.getMinX(), abstractTextContainer.getMinY(), abstractTextContainer.getWidth(), abstractTextContainer.getHeight());
+    }
+
+
+    public static Rectangle2D toRectangle2D(PDRectangle rectangle) {
+
+        return new Rectangle2D.Double(rectangle.getLowerLeftX(), rectangle.getLowerLeftY(), rectangle.getWidth(), rectangle.getHeight());
+    }
+
+
+    public static String toString(Rectangle2D rectangle2D) {
+
+        return format("%f,%f,%f,%f", rectangle2D.getX(), rectangle2D.getY(), rectangle2D.getWidth(), rectangle2D.getHeight());
+    }
+
+
+    public static Rectangle2D parseRectangle2D(String bBox) {
+
+        List<Float> floats = Arrays.stream(bBox.split(",")).map(Float::parseFloat).toList();
+        return new Rectangle2D.Float(floats.get(0), floats.get(1), floats.get(2), floats.get(3));
+    }
+
+
+    private static class Rectangle2DUnion implements Collector<Rectangle2D, Area, Rectangle2D> {
+
+        @Override
+        public Supplier<Area> supplier() {
+
+            return Area::new;
+        }
+
+
+        @Override
+        public BiConsumer<Area, Rectangle2D> accumulator() {
+
+            return (area, rectangle2D) -> area.add(new Area(rectangle2D));
+        }
+
+
+        @Override
+        public BinaryOperator<Area> combiner() {
+
+            return (area1, area2) -> {
+                area1.add(area2);
+                return area1;
+            };
+        }
+
+
+        @Override
+        public Function<Area, Rectangle2D> finisher() {
+
+            return Area::getBounds2D;
+        }
+
+
+        @Override
+        public Set<Characteristics> characteristics() {
+
+            return Set.of(Characteristics.CONCURRENT, Characteristics.UNORDERED);
+        }
+
+    }
+
+}

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

@@ -0,0 +1,156 @@
+package com.knecon.fforesight.service.layoutparser.processor.factory;
+
+import java.awt.geom.AffineTransform;
+import java.awt.geom.Rectangle2D;
+import java.util.Collections;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.Objects;
+
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.RedTextPosition;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.TextDirection;
+import com.knecon.fforesight.service.layoutparser.processor.classification.dto.text.TextPositionSequence;
+
+public class SearchTextWithTextPositionFactory {
+
+    public static final int HEIGHT_PADDING = 2;
+
+
+    public static SearchTextWithTextPositionModel buildSearchTextToTextPositionModel(List<TextPositionSequence> sequences) {
+
+        if (sequences.isEmpty() || sequences.stream().allMatch(sequence -> sequence.getTextPositions().isEmpty())) {
+            return SearchTextWithTextPositionModel.builder()
+                    .searchText("")
+                    .lineBreaks(Collections.emptyList())
+                    .positions(Collections.emptyList())
+                    .stringCoordsToPositionCoords(Collections.emptyList())
+                    .build();
+        }
+
+        List<Integer> stringIdxToPositionIdx = new LinkedList<>();
+        List<Integer> lineBreaksStringIdx = new LinkedList<>();
+        StringBuilder sb = new StringBuilder();
+
+        int stringIdx = 0;
+        int positionIdx = 0;
+        int lastHyphenIdx = -3;
+
+        RedTextPosition currentTextPosition = sequences.get(0).getTextPositions().get(0);
+        RedTextPosition previousTextPosition = RedTextPosition.builder().unicode(" ").position(currentTextPosition.getPosition()).build();
+
+        for (TextPositionSequence word : sequences) {
+            for (int i = 0; i < word.getTextPositions().size(); ++i) {
+
+                currentTextPosition = word.getTextPositions().get(i);
+
+                if (isLineBreak(currentTextPosition, previousTextPosition)) {
+
+                    if (stringIdx - lastHyphenIdx < 3) {
+                        sb.delete(lastHyphenIdx, sb.length());
+                        stringIdxToPositionIdx = stringIdxToPositionIdx.subList(0, lastHyphenIdx);
+                        stringIdx = lastHyphenIdx;
+                        lastHyphenIdx = -3;
+                    }
+                    lineBreaksStringIdx.add(stringIdx);
+                }
+                if (!isRepeatedWhitespace(currentTextPosition.getUnicode(), previousTextPosition.getUnicode())) {
+
+                    if (isHyphen(currentTextPosition.getUnicode())) {
+                        lastHyphenIdx = stringIdx;
+                    }
+                    sb.append(currentTextPosition.getUnicode());
+                    stringIdxToPositionIdx.add(positionIdx);
+                    ++stringIdx;
+                }
+
+                previousTextPosition = currentTextPosition;
+
+                ++positionIdx;
+            }
+
+            previousTextPosition = RedTextPosition.builder().unicode(" ").position(previousTextPosition.getPosition()).build();
+            sb.append(previousTextPosition.getUnicode());
+            stringIdxToPositionIdx.add(positionIdx);
+            ++stringIdx;
+        }
+
+        assert sb.length() == stringIdxToPositionIdx.size();
+
+        List<Rectangle2D> positions = sequences.stream()
+                .flatMap(sequence -> sequence.getTextPositions().stream().map(textPosition -> mapRedTextPositionToInitialUserSpace(textPosition, sequence)))
+                .toList();
+
+        return SearchTextWithTextPositionModel.builder()
+                .searchText(sb.toString())
+                .lineBreaks(lineBreaksStringIdx)
+                .stringCoordsToPositionCoords(stringIdxToPositionIdx)
+                .positions(positions)
+                .build();
+    }
+
+
+    private static boolean isLineBreak(RedTextPosition currentTextPosition, RedTextPosition previousTextPosition) {
+
+        return Objects.equals(currentTextPosition.getUnicode(), "\n") || isDeltaYLargerThanTextHeight(currentTextPosition, previousTextPosition);
+    }
+
+
+    private static boolean isDeltaYLargerThanTextHeight(RedTextPosition currentPosition, RedTextPosition previousPosition) {
+
+        if (previousPosition == null) {
+            return false;
+        }
+
+        float deltaY = Math.abs(currentPosition.getYDirAdj() - previousPosition.getYDirAdj());
+        return deltaY >= currentPosition.getHeightDir();
+    }
+
+
+    private static boolean isRepeatedWhitespace(String currentUnicode, String previousUnicode) {
+
+        return Objects.equals(previousUnicode, " ") && Objects.equals(currentUnicode, " ");
+    }
+
+
+    private static boolean isHyphen(String unicodeCharacter) {
+
+        return Objects.equals(unicodeCharacter, "-") || //
+                Objects.equals(unicodeCharacter, "~") || //
+                Objects.equals(unicodeCharacter, "‐") || //
+                Objects.equals(unicodeCharacter, "‒") || //
+                Objects.equals(unicodeCharacter, "⁻") || //
+                Objects.equals(unicodeCharacter, "−") || //
+                Objects.equals(unicodeCharacter, "﹣") || //
+                Objects.equals(unicodeCharacter, "゠") || //
+                Objects.equals(unicodeCharacter, "⁓") || //
+                Objects.equals(unicodeCharacter, "‑") || //
+                Objects.equals(unicodeCharacter, "\u00AD");
+    }
+
+
+    private static Rectangle2D mapRedTextPositionToInitialUserSpace(RedTextPosition textPosition, TextPositionSequence sequence) {
+
+        float textHeight = sequence.getTextHeight() + HEIGHT_PADDING;
+        Rectangle2D rectangle2D = new Rectangle2D.Double(textPosition.getXDirAdj(),
+                textPosition.getYDirAdj() - textHeight,
+                textPosition.getWidthDirAdj(),
+                textHeight + HEIGHT_PADDING);
+
+        AffineTransform transform = new AffineTransform();
+
+        if (sequence.getDir() == TextDirection.ZERO || sequence.getDir() == TextDirection.HALF_CIRCLE) {
+            transform.rotate(sequence.getDir().getRadians(), sequence.getPageWidth() / 2f, sequence.getPageHeight() / 2f);
+            transform.translate(0f, sequence.getPageHeight());
+        } else if (sequence.getDir() == TextDirection.QUARTER_CIRCLE) {
+            transform.rotate(sequence.getDir().getRadians(), sequence.getPageWidth() / 2f, sequence.getPageWidth() / 2f);
+            transform.translate(0f, sequence.getPageWidth());
+        } else {
+            transform.rotate(sequence.getDir().getRadians(), sequence.getPageHeight() / 2f, sequence.getPageHeight() / 2f);
+            transform.translate(0f, sequence.getPageWidth());
+        }
+        transform.scale(1., -1.);
+
+        return transform.createTransformedShape(rectangle2D).getBounds2D();
+    }
+
+}