Marmelator/layout-parser
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 289 Wiki Activity

RED-7074: Design Subsection section tree structure algorithm

Browse Source 
* added abstract class SectionNode
* both Section and SuperSection extend the SectionNode class, so that there is no inheritance between Section and SuperSection as well as no field duplication
This commit is contained in:
maverickstuder 2024-05-22 13:02:16 +02:00
parent b08ed2037e
commit 0c8b2e6d44
10 changed files with 144 additions and 80 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/data/redaction/NodeType.java
View File

@ -6,6 +6,7 @@ import java.util.Locale;
public enum NodeType implements Serializable {
DOCUMENT,
SECTION,
SUPER_SECTION,
HEADLINE,
PARAGRAPH,
TABLE,

70
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Section.java
View File

@ -21,81 +21,17 @@ import lombok.experimental.FieldDefaults;
import lombok.experimental.SuperBuilder;
import lombok.extern.slf4j.Slf4j;
@Slf4j
@Data
@SuperBuilder
@AllArgsConstructor
@FieldDefaults(level = AccessLevel.PRIVATE)
public class Section implements GenericSemanticNode {
@Builder.Default
Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
List<Integer> treeId;
TextBlock textBlock;
@EqualsAndHashCode.Exclude
DocumentTree documentTree;
@Builder.Default
@EqualsAndHashCode.Exclude
Set<RedactionEntity> entities = new HashSet<>();
@EqualsAndHashCode.Exclude
Map<Page, Rectangle2D> bBoxCache;
@Override
public NodeType getType() {
return NodeType.SECTION;
}
public boolean hasTables() {
return streamAllSubNodesOfType(NodeType.TABLE).findAny()
.isPresent();
}
public boolean isLeafSection() {
return streamAllSubNodesOfType(NodeType.SECTION).findAny()
.isEmpty();
}
@Override
public TextBlock getTextBlock() {
if (textBlock == null) {
textBlock = GenericSemanticNode.super.getTextBlock();
}
return textBlock;
}
@EqualsAndHashCode(callSuper = true)
public class Section extends SectionNode {
@Override
public String toString() {
return treeId.toString() + ": " + NodeType.SECTION + ": " + this.getTextBlock().buildSummary();
}
public Headline getHeadline() {
return streamChildrenOfType(NodeType.HEADLINE)//
.map(node -> (Headline) node)//
.findFirst()//
.orElseGet(() -> getParent().getHeadline());
}
@Override
public Map<Page, Rectangle2D> getBBox() {
if (bBoxCache == null) {
bBoxCache = GenericSemanticNode.super.getBBox();
}
return bBoxCache;
return super.toString();
}
}

103
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/SectionNode.java Normal file
View File

@ -0,0 +1,103 @@
package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
import java.awt.geom.Rectangle2D;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
import lombok.AccessLevel;
import lombok.AllArgsConstructor;
import lombok.Builder;
import lombok.Data;
import lombok.EqualsAndHashCode;
import lombok.NoArgsConstructor;
import lombok.experimental.FieldDefaults;
import lombok.experimental.SuperBuilder;
import lombok.extern.slf4j.Slf4j;
@Slf4j
@Data
@SuperBuilder
@AllArgsConstructor
@NoArgsConstructor
@FieldDefaults(level = AccessLevel.PRIVATE)
public abstract class SectionNode implements GenericSemanticNode {
@Builder.Default
Set<LayoutEngine> engines = new HashSet<>(Set.of(LayoutEngine.ALGORITHM));
List<Integer> treeId;
TextBlock textBlock;
@EqualsAndHashCode.Exclude
DocumentTree documentTree;
@Builder.Default
@EqualsAndHashCode.Exclude
Set<RedactionEntity> entities = new HashSet<>();
@EqualsAndHashCode.Exclude
Map<Page, Rectangle2D> bBoxCache;
@Override
public NodeType getType() {
return NodeType.SECTION;
}
public boolean hasTables() {
return streamAllSubNodesOfType(NodeType.TABLE).findAny()
.isPresent();
}
public boolean isLeafSection() {
return streamAllSubNodesOfType(NodeType.SECTION).findAny()
.isEmpty();
}
@Override
public TextBlock getTextBlock() {
if (textBlock == null) {
textBlock = GenericSemanticNode.super.getTextBlock();
}
return textBlock;
}
@Override
public String toString() {
return treeId.toString() + ": " + NodeType.SECTION + ": " + this.getTextBlock().buildSummary();
}
public Headline getHeadline() {
return streamChildrenOfType(NodeType.HEADLINE)//
.map(node -> (Headline) node)//
.findFirst()//
.orElseGet(() -> getParent().getHeadline());
}
@Override
public Map<Page, Rectangle2D> getBBox() {
if (bBoxCache == null) {
bBoxCache = GenericSemanticNode.super.getBBox();
}
return bBoxCache;
}
}

23
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/SuperSection.java
View File

@ -1,14 +1,35 @@
package com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes;
import java.util.List;
import java.util.Set;
import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.LayoutEngine;
import com.knecon.fforesight.service.layoutparser.internal.api.data.redaction.NodeType;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.DocumentTree;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.entity.RedactionEntity;
import lombok.AccessLevel;
import lombok.AllArgsConstructor;
import lombok.Builder;
import lombok.Data;
import lombok.EqualsAndHashCode;
import lombok.ToString;
import lombok.experimental.FieldDefaults;
import lombok.experimental.SuperBuilder;
@Data
@SuperBuilder
@AllArgsConstructor
@FieldDefaults(level = AccessLevel.PRIVATE)
@EqualsAndHashCode(callSuper = true)
public class SuperSection extends Section {
public class SuperSection extends SectionNode {
@Override
public NodeType getType() {
return NodeType.SUPER_SECTION;
}
@Override
public String toString() {

3
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/TableOfContentItem.java
View File

@ -6,6 +6,7 @@ import java.util.stream.Collectors;
import com.knecon.fforesight.service.layoutparser.processor.model.AbstractPageBlock;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Section;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SectionNode;
import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
@ -24,7 +25,7 @@ public class TableOfContentItem {
private List<AbstractPageBlock> sectionBlocks = new ArrayList<>();
private List<ClassifiedImage> images = new ArrayList<>();
private Section section;
private SectionNode section;
public TableOfContentItem(TextPageBlock headline) {

5
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/factory/DocumentGraphFactory.java
View File

@ -32,6 +32,7 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Im
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Paragraph;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Section;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SectionNode;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.AtomicTextBlock;
import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
import com.knecon.fforesight.service.layoutparser.processor.model.outline.TableOfContentItem;
@ -74,7 +75,7 @@ public class DocumentGraphFactory {
for (TableOfContentItem tocItem : classificationDocument.getTableOfContents()) {
var parent = tocItem.getParent() == null ? null : tocItem.getParent().getSection();
Optional<Section> section = SectionNodeFactory.addSection(layoutParsingType,
Optional<SectionNode> section = SectionNodeFactory.addSection(layoutParsingType,
parent,
tocItem.getChildren().isEmpty(),
tocItem.getNonEmptySectionBlocks(),
@ -239,7 +240,7 @@ public class DocumentGraphFactory {
DocumentTree documentTree;
Map<Page, Integer> pages;
List<Section> sections;
List<SectionNode> sections;
List<ClassifiedImage> images;
TextBlockFactory textBlockFactory;

13
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/factory/SectionNodeFactory.java
View File

@ -18,6 +18,7 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Do
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.GenericSemanticNode;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Page;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Section;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SectionNode;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SuperSection;
import com.knecon.fforesight.service.layoutparser.processor.model.image.ClassifiedImage;
import com.knecon.fforesight.service.layoutparser.processor.model.table.TablePageBlock;
@ -29,7 +30,7 @@ import lombok.experimental.UtilityClass;
@UtilityClass
public class SectionNodeFactory {
public Optional<Section> addSection(LayoutParsingType layoutParsingType,
public Optional<SectionNode> addSection(LayoutParsingType layoutParsingType,
GenericSemanticNode parentNode,
boolean isLeaf,
List<AbstractPageBlock> pageBlocks,
@ -53,7 +54,7 @@ public class SectionNodeFactory {
Map<Integer, List<AbstractPageBlock>> blocksPerPage = pageBlocks.stream()
.collect(groupingBy(AbstractPageBlock::getPage));
Section section;
SectionNode section;
if (isLeaf) {
section = Section.builder().documentTree(context.getDocumentTree()).build();
} else {
@ -98,7 +99,7 @@ public class SectionNodeFactory {
}
private List<Integer> getTreeId(GenericSemanticNode parentNode, DocumentGraphFactory.Context context, Section section) {
private List<Integer> getTreeId(GenericSemanticNode parentNode, DocumentGraphFactory.Context context, SectionNode section) {
if (parentNode == null) {
return context.getDocumentTree().createNewMainEntryAndReturnId(section);
@ -111,7 +112,7 @@ public class SectionNodeFactory {
private void addFirstHeadlineDirectlyToSection(LayoutParsingType layoutParsingType,
List<AbstractPageBlock> pageBlocks,
DocumentGraphFactory.Context context,
Section section,
SectionNode section,
Document document) {
if (pageBlocks.get(0).isHeadline()) {
@ -124,7 +125,7 @@ public class SectionNodeFactory {
private void addTablesAndParagraphsAndHeadlinesToSection(LayoutParsingType layoutParsingType,
List<AbstractPageBlock> pageBlocks,
DocumentGraphFactory.Context context,
Section section,
SectionNode section,
Document document) {
Set<AbstractPageBlock> alreadyMerged = new HashSet<>();
@ -249,7 +250,7 @@ public class SectionNodeFactory {
}
private void addSectionNodeToPageNode(DocumentGraphFactory.Context context, Section section, Integer pageNumber) {
private void addSectionNodeToPageNode(DocumentGraphFactory.Context context, SectionNode section, Integer pageNumber) {
Page page = context.getPage(pageNumber);
page.getMainBody().add(section);

2
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/mapper/DocumentGraphMapper.java
View File

@ -61,7 +61,7 @@ public class DocumentGraphMapper {
List<Page> pages = Arrays.stream(entryData.getPageNumbers()).map(pageNumber -> getPage(pageNumber, context)).toList();
SemanticNode node = switch (entryData.getType()) {
case SECTION -> buildSection(context);
case SECTION, SUPER_SECTION -> buildSection(context);
case PARAGRAPH -> buildParagraph(context, entryData.getProperties());
case HEADLINE -> buildHeadline(context);
case HEADER -> buildHeader(context);

2
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/utils/PdfVisualisationUtility.java
View File

@ -112,8 +112,8 @@ public class PdfVisualisationUtility {
case DOCUMENT -> Color.LIGHT_GRAY;
case HEADER, FOOTER -> Color.GREEN;
case PARAGRAPH -> Color.BLUE;
case SUPER_SECTION, SECTION -> Color.BLACK;
case HEADLINE -> Color.RED;
case SECTION -> Color.BLACK;
case TABLE -> Color.ORANGE;
case TABLE_CELL -> Color.GRAY;
case IMAGE -> Color.MAGENTA;

2
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/utils/visualizations/PdfDraw.java
View File

@ -229,7 +229,7 @@ public class PdfDraw {
case HEADER, FOOTER -> Color.GREEN;
case PARAGRAPH -> Color.BLUE;
case HEADLINE -> Color.RED;
case SECTION -> Color.BLACK;
case SECTION, SUPER_SECTION -> Color.BLACK;
case TABLE -> Color.ORANGE;
case TABLE_CELL -> Color.GRAY;
case IMAGE -> Color.MAGENTA;