Marmelator/redaction-service
1
0
Fork 0
Code Issues Pull Requests 14 Actions Packages Projects Releases 622 Wiki Activity

RED-6009: Document Tree Structure

Browse Source 
* extracted classification to enum
* extracted approxLineCount as constant
* removed NodeType from Entry in DocumentTree
* renamed a variable
This commit is contained in:
Kilian Schuettler 2023-05-17 13:29:41 +02:00
parent 1dd55d9af6
commit c8ee4444c2
14 changed files with 109 additions and 69 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/classification/model/AbstractPageBlock.java
View File

@ -22,7 +22,7 @@ public abstract class AbstractPageBlock {
@JsonIgnore
protected float maxY;
@JsonIgnore
protected String classification;
protected PageBlockType classification;
@JsonIgnore
protected int page;
@ -35,7 +35,7 @@ public abstract class AbstractPageBlock {
public boolean isHeadline() {
return this instanceof ClassificationTextBlock && this.getClassification() != null && this.getClassification().startsWith("H");
return this instanceof ClassificationTextBlock && this.getClassification() != null && this.getClassification().isHeadline();
}

34
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/classification/model/PageBlockType.java Normal file
View File

@ -0,0 +1,34 @@
package com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model;
public enum PageBlockType {
H1,
H2,
H3,
H4,
HEADER,
FOOTER,
TITLE,
PARAGRAPH,
PARAGRAPH_BOLD,
PARAGRAPH_ITALIC,
PARAGRAPH_UNKNOWN,
OTHER,
TABLE;
public static PageBlockType getHeadlineType(int i) {
return switch (i) {
case 1 -> PageBlockType.H1;
case 2 -> PageBlockType.H2;
case 3 -> PageBlockType.H3;
default -> PageBlockType.H4;
};
}
public boolean isHeadline() {
return this.equals(H1) || this.equals(H2) || this.equals(H3) || this.equals(H4);
}
}

3
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/classification/model/table/TablePageBlock.java
View File

@ -12,6 +12,7 @@ import java.util.stream.Collectors;
import org.apache.commons.collections4.CollectionUtils;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.AbstractPageBlock;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.PageBlockType;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.text.ClassificationTextBlock;
import lombok.Getter;
@ -39,7 +40,7 @@ public class TablePageBlock extends AbstractPageBlock {
minY = area.getBottom();
maxX = area.getRight();
maxY = area.getTop();
classification = "TablePageBlock";
classification = PageBlockType.TABLE;
this.rotation = rotation;
}

3
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/classification/model/text/ClassificationTextBlock.java
View File

@ -5,6 +5,7 @@ import java.util.List;
import com.fasterxml.jackson.annotation.JsonIgnore;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.AbstractPageBlock;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.PageBlockType;
import com.iqser.red.service.redaction.v1.server.redaction.utils.TextNormalizationUtilities;
import lombok.AllArgsConstructor;
@ -45,7 +46,7 @@ public class ClassificationTextBlock extends AbstractPageBlock {
private float highestFontSize;
@JsonIgnore
private String classification;
private PageBlockType classification;
@JsonIgnore

18
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/classification/service/BlockificationService.java
View File

@ -40,7 +40,7 @@ public class BlockificationService {
int indexOnPage = 0;
List<TextPositionSequence> chunkWords = new ArrayList<>();
List<AbstractPageBlock> chunkBlockList1 = new ArrayList<>();
List<AbstractPageBlock> chunkBlockList = new ArrayList<>();
float minX = 1000, maxX = 0, minY = 1000, maxY = 0;
TextPositionSequence prev = null;
@ -60,14 +60,14 @@ public class BlockificationService {
if (prev != null && (lineSeparation || startFromTop || splitByX || splitByDir || isSplitByRuling)) {
Orientation prevOrientation = null;
if (!chunkBlockList1.isEmpty()) {
prevOrientation = chunkBlockList1.get(chunkBlockList1.size() - 1).getOrientation();
if (!chunkBlockList.isEmpty()) {
prevOrientation = chunkBlockList.get(chunkBlockList.size() - 1).getOrientation();
}
ClassificationTextBlock cb1 = buildTextBlock(chunkWords, indexOnPage);
indexOnPage++;
chunkBlockList1.add(cb1);
chunkBlockList.add(cb1);
chunkWords = new ArrayList<>();
if (splitByX && !isSplitByRuling) {
@ -108,10 +108,10 @@ public class BlockificationService {
ClassificationTextBlock cb1 = buildTextBlock(chunkWords, indexOnPage);
if (cb1 != null) {
chunkBlockList1.add(cb1);
chunkBlockList.add(cb1);
}
Iterator<AbstractPageBlock> itty = chunkBlockList1.iterator();
Iterator<AbstractPageBlock> itty = chunkBlockList.iterator();
ClassificationTextBlock previousLeft = null;
ClassificationTextBlock previousRight = null;
@ -141,7 +141,7 @@ public class BlockificationService {
}
}
itty = chunkBlockList1.iterator();
itty = chunkBlockList.iterator();
ClassificationTextBlock previous = null;
while (itty.hasNext()) {
ClassificationTextBlock block = (ClassificationTextBlock) itty.next();
@ -157,7 +157,7 @@ public class BlockificationService {
previous = block;
}
return new ClassificationPage(chunkBlockList1);
return new ClassificationPage(chunkBlockList);
}
@ -254,7 +254,7 @@ public class BlockificationService {
verticalRulingLines,
word.getDir().getDegrees(),
word.getPageWidth(),
word.getPageHeight()); //
word.getPageHeight());
}

5
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/classification/service/BodyTextFrameService.java
View File

@ -17,6 +17,9 @@ import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.ut
@Service
public class BodyTextFrameService {
private static final float APPROXIMATE_HEADER_LINE_COUNT = 2.9f;
/**
* Adjusts and sets the body text frame to a page.
* Note: This needs to use Pdf Coordinate System where {0,0} rotated with the page rotation.
@ -84,7 +87,7 @@ public class BodyTextFrameService {
}
float approxLineCount = PositionUtils.getApproxLineCount(textBlock);
if (approxLineCount < 2.9f) {
if (approxLineCount < APPROXIMATE_HEADER_LINE_COUNT) {
continue;
}

29
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/classification/service/ClassificationService.java
View File

@ -9,6 +9,7 @@ import com.iqser.red.service.persistence.service.v1.api.shared.model.redactionlo
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.AbstractPageBlock;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.ClassificationDocument;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.ClassificationPage;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.PageBlockType;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.text.ClassificationTextBlock;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.utils.PositionUtils;
@ -53,21 +54,21 @@ public class ClassificationService {
var bodyTextFrame = page.getBodyTextFrame();
if (document.getFontSizeCounter().getMostPopular() == null) {
textBlock.setClassification("Other");
textBlock.setClassification(PageBlockType.OTHER);
return;
}
if (PositionUtils.isOverBodyTextFrame(bodyTextFrame, textBlock, page.getRotation()) && (document.getFontSizeCounter()
.getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter().getMostPopular())) {
textBlock.setClassification("Header");
textBlock.setClassification(PageBlockType.HEADER);
} else if (PositionUtils.isUnderBodyTextFrame(bodyTextFrame, textBlock, page.getRotation()) && (document.getFontSizeCounter()
.getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter().getMostPopular())) {
textBlock.setClassification("Footer");
textBlock.setClassification(PageBlockType.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");
textBlock.setClassification(PageBlockType.TITLE);
}
} else if (textBlock.getMostPopularWordFontSize() > document.getFontSizeCounter()
.getMostPopular() && PositionUtils.getApproxLineCount(textBlock) < 4.9 && (textBlock.getMostPopularWordStyle().equals("bold") || !document.getFontStyleCounter()
@ -80,36 +81,34 @@ public class ClassificationService {
for (int i = 1; i <= headlineFontSizes.size(); i++) {
if (textBlock.getMostPopularWordFontSize() == headlineFontSizes.get(i - 1)) {
textBlock.setClassification("H " + i);
textBlock.setClassification(PageBlockType.getHeadlineType(i));
document.setHeadlines(true);
}
}
} else if (!textBlock.getText().startsWith("TablePageBlock ") && !textBlock.getText().startsWith("Figure ") && PositionUtils.isWithinBodyTextFrame(bodyTextFrame,
textBlock) && textBlock.getMostPopularWordStyle().equals("bold") && !document.getFontStyleCounter()
.getMostPopular()
.equals("bold") && PositionUtils.getApproxLineCount(textBlock) < 2.9 && textBlock.getSequences()
} else if (!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));
textBlock.setClassification(PageBlockType.getHeadlineType(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");
textBlock.setClassification(PageBlockType.PARAGRAPH_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");
textBlock.setClassification(PageBlockType.PARAGRAPH);
} 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");
textBlock.setClassification(PageBlockType.PARAGRAPH_ITALIC);
} else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock)) {
textBlock.setClassification("TextBlock Unknown");
textBlock.setClassification(PageBlockType.PARAGRAPH_UNKNOWN);
} else {
textBlock.setClassification("Other");
textBlock.setClassification(PageBlockType.OTHER);
}
}

9
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/classification/service/SectionsBuilderService.java
View File

@ -16,6 +16,7 @@ import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.mo
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.ClassificationHeader;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.ClassificationPage;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.ClassificationSection;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.PageBlockType;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.image.ClassifiedImage;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.table.Cell;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.table.TablePageBlock;
@ -52,22 +53,22 @@ public class SectionsBuilderService {
current.setPage(page.getPageNumber());
if (current.getClassification().equals("Header")) {
if (current.getClassification().equals(PageBlockType.HEADER)) {
header.add((ClassificationTextBlock) current);
continue;
}
if (current.getClassification().equals("Footer")) {
if (current.getClassification().equals(PageBlockType.FOOTER)) {
footer.add((ClassificationTextBlock) current);
continue;
}
if (current.getClassification().equals("Other")) {
if (current.getClassification().equals(PageBlockType.OTHER)) {
unclassifiedText.add((ClassificationTextBlock) current);
continue;
}
if (prev != null && current.getClassification().startsWith("H ") && !prev.getClassification().startsWith("H ") || !document.isHeadlines()) {
if (prev != null && current.getClassification().isHeadline() && !prev.getClassification().isHeadline() || !document.isHeadlines()) {
ClassificationSection chunkBlock = buildTextBlock(chunkWords, lastHeadline);
chunkBlock.setHeadline(lastHeadline);
if (document.isHeadlines()) {

2
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/data/mapper/DocumentGraphMapper.java
View File

@ -93,7 +93,7 @@ public class DocumentGraphMapper {
} else {
pages.forEach(page -> page.getMainBody().add(node));
}
newEntries.add(DocumentTree.Entry.builder().treeId(treeId).type(entryData.getType()).children(buildEntries(entryData.getSubEntries(), context)).node(node).build());
newEntries.add(DocumentTree.Entry.builder().treeId(treeId).children(buildEntries(entryData.getSubEntries(), context)).node(node).build());
}
return newEntries;
}

13
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/factory/DocumentGraphFactory.java
View File

@ -26,7 +26,6 @@ import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.no
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Header;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Headline;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Image;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.NodeType;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Page;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Paragraph;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Section;
@ -87,7 +86,7 @@ public class DocumentGraphFactory {
List<ClassificationTextBlock> textBlocks = new LinkedList<>(textBlocksToMerge);
textBlocks.add(originalTextBlock);
AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(textBlocks), node, context, page);
List<Integer> treeId = context.documentTree.createNewChildEntryAndReturnId(parentNode, node.getType(), node);
List<Integer> treeId = context.documentTree.createNewChildEntryAndReturnId(parentNode, node);
node.setLeafTextBlock(textBlock);
node.setTreeId(treeId);
}
@ -107,7 +106,7 @@ public class DocumentGraphFactory {
.build();
page.getMainBody().add(imageNode);
List<Integer> tocId = context.getDocumentTree().createNewChildEntryAndReturnId(section, NodeType.IMAGE, imageNode);
List<Integer> tocId = context.getDocumentTree().createNewChildEntryAndReturnId(section, imageNode);
imageNode.setTreeId(tocId);
}
@ -152,7 +151,7 @@ public class DocumentGraphFactory {
footer,
context,
page);
List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(NodeType.FOOTER, footer);
List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(footer);
footer.setTreeId(tocId);
footer.setLeafTextBlock(textBlock);
page.setFooter(footer);
@ -164,7 +163,7 @@ public class DocumentGraphFactory {
Page page = context.getPage(textBlocks.get(0).getPage());
Header header = Header.builder().documentTree(context.getDocumentTree()).build();
AtomicTextBlock textBlock = context.textBlockFactory.buildAtomicTextBlock(TextPositionOperations.mergeAndSortTextPositionSequenceByYThenX(textBlocks), header, 0, page);
List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(NodeType.HEADER, header);
List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(header);
header.setTreeId(tocId);
header.setLeafTextBlock(textBlock);
page.setHeader(header);
@ -176,7 +175,7 @@ public class DocumentGraphFactory {
Page page = context.getPage(pageIndex);
Footer footer = Footer.builder().documentTree(context.getDocumentTree()).build();
AtomicTextBlock textBlock = context.textBlockFactory.emptyTextBlock(footer, context, page);
List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(NodeType.FOOTER, footer);
List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(footer);
footer.setTreeId(tocId);
footer.setLeafTextBlock(textBlock);
page.setFooter(footer);
@ -188,7 +187,7 @@ public class DocumentGraphFactory {
Page page = context.getPage(pageIndex);
Header header = Header.builder().documentTree(context.getDocumentTree()).build();
AtomicTextBlock textBlock = context.textBlockFactory.emptyTextBlock(header, 0, page);
List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(NodeType.HEADER, header);
List<Integer> tocId = context.getDocumentTree().createNewMainEntryAndReturnId(header);
header.setTreeId(tocId);
header.setLeafTextBlock(textBlock);
page.setHeader(header);

14
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/factory/SearchTextWithTextPositionFactory.java
View File

@ -44,19 +44,19 @@ public class SearchTextWithTextPositionFactory {
}
previousTextPosition = RedTextPosition.builder().unicode(" ").position(previousTextPosition.getPosition()).build();
context.sb.append(" ");
context.stringBuilder.append(" ");
context.stringIdxToPositionIdx.add(context.positionIdx);
++context.stringIdx;
}
assert context.sb.length() == context.stringIdxToPositionIdx.size();
assert context.stringBuilder.length() == context.stringIdxToPositionIdx.size();
List<Rectangle2D> positions = sequences.stream()
.flatMap(sequence -> sequence.getTextPositions().stream().map(textPosition -> mapRedTextPositionToInitialUserSpace(textPosition, sequence)))
.toList();
return SearchTextWithTextPositionDto.builder()
.searchText(context.sb.toString())
.searchText(context.stringBuilder.toString())
.lineBreaks(context.lineBreaksStringIdx)
.stringCoordsToPositionCoords(context.stringIdxToPositionIdx)
.positions(positions)
@ -66,7 +66,7 @@ public class SearchTextWithTextPositionFactory {
private static void appendCurrentTextPosition(Context context, RedTextPosition currentTextPosition) {
context.sb.append(currentTextPosition.getUnicode());
context.stringBuilder.append(currentTextPosition.getUnicode());
// unicode characters with more than 16-bit encoding have a length > 1 in java strings
for (int j = 0; j < currentTextPosition.getUnicode().length(); j++) {
@ -81,7 +81,7 @@ public class SearchTextWithTextPositionFactory {
if (isLineBreak(currentTextPosition, previousTextPosition)) {
if (lastHyphenDirectlyBeforeLineBreak(context)) {
context.sb.delete(context.lastHyphenIdx, context.sb.length());
context.stringBuilder.delete(context.lastHyphenIdx, context.stringBuilder.length());
context.stringIdxToPositionIdx = context.stringIdxToPositionIdx.subList(0, context.lastHyphenIdx);
context.stringIdx = context.lastHyphenIdx;
context.lastHyphenIdx = -3;
@ -166,14 +166,14 @@ public class SearchTextWithTextPositionFactory {
List<Integer> stringIdxToPositionIdx = new LinkedList<>();
List<Integer> lineBreaksStringIdx = new LinkedList<>();
StringBuilder sb = new StringBuilder();
StringBuilder stringBuilder = new StringBuilder();
int stringIdx;
int positionIdx;
// when checking for a hyphen linebreak, we need to check after a linebreak if the last hyphen was less than three symbols away.
// We detect a linebreak as either a "\n" character or if two adjacent symbol's position differ in y-coordinates by at least one character height.
// If there is a hyphen linebreak, the hyphen will be 1 position in front of a "\n" or 2 positions in front of the character which has a lower y-coordinate
// This is why, we need to initialize this to <= -3, otherwise, if the very first symbol is a \n we would detect a hyphen linebreak that isn't there.
// This is why, we need to initialize this to < -2, otherwise, if the very first symbol is a \n we would detect a hyphen linebreak that isn't there.
// Also, Integer.MIN_VALUE is a bad idea due to potential overflow during arithmetic operations. This is why the default should be -3.
int lastHyphenIdx = -3;

5
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/factory/SectionNodeFactory.java
View File

@ -14,7 +14,6 @@ import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.mo
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.table.TablePageBlock;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.text.ClassificationTextBlock;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.GenericSemanticNode;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.NodeType;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Page;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Section;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.utils.TableMergingUtility;
@ -54,9 +53,9 @@ public class SectionNodeFactory {
private static List<Integer> getTreeId(GenericSemanticNode parentNode, DocumentGraphFactory.Context context, Section section) {
if (parentNode == null) {
return context.getDocumentTree().createNewMainEntryAndReturnId(NodeType.SECTION, section);
return context.getDocumentTree().createNewMainEntryAndReturnId(section);
} else {
return context.getDocumentTree().createNewChildEntryAndReturnId(parentNode, NodeType.SECTION, section);
return context.getDocumentTree().createNewChildEntryAndReturnId(parentNode, section);
}
}

8
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/factory/TableNodeFactory.java
View File

@ -12,7 +12,6 @@ import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.mo
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.table.TablePageBlock;
import com.iqser.red.service.redaction.v1.server.layoutparsing.classification.model.text.TextPositionSequence;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.GenericSemanticNode;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.NodeType;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Page;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.SemanticNode;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.Table;
@ -37,7 +36,7 @@ public class TableNodeFactory {
pages.forEach(page -> addTableToPage(page, parentNode, table));
List<Integer> tocId = context.getDocumentTree().createNewChildEntryAndReturnId(parentNode, NodeType.TABLE, table);
List<Integer> tocId = context.getDocumentTree().createNewChildEntryAndReturnId(parentNode, table);
table.setTreeId(tocId);
addTableCells(mergedRows, table, context);
@ -104,7 +103,7 @@ public class TableNodeFactory {
TextBlock textBlock;
List<Integer> tocId = context.getDocumentTree().createNewTableChildEntryAndReturnId(tableNode, NodeType.TABLE_CELL, tableCell);
List<Integer> tocId = context.getDocumentTree().createNewTableChildEntryAndReturnId(tableNode, tableCell);
tableCell.setTreeId(tocId);
if (cell.getTextBlocks().isEmpty()) {
@ -142,8 +141,7 @@ public class TableNodeFactory {
private boolean firstTextBlockIsHeadline(Cell cell) {
String classification = cell.getTextBlocks().get(0).getClassification();
return classification != null && classification.startsWith("H");
return cell.getTextBlocks().get(0).isHeadline();
}
}

31
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/graph/DocumentTree.java
View File

@ -33,7 +33,7 @@ public class DocumentTree {
public DocumentTree(Document document) {
root = Entry.builder().treeId(Collections.emptyList()).type(NodeType.DOCUMENT).children(new LinkedList<>()).node(document).build();
root = Entry.builder().treeId(Collections.emptyList()).children(new LinkedList<>()).node(document).build();
}
@ -43,32 +43,32 @@ public class DocumentTree {
}
public List<Integer> createNewMainEntryAndReturnId(NodeType nodeType, GenericSemanticNode node) {
public List<Integer> createNewMainEntryAndReturnId(GenericSemanticNode node) {
return createNewChildEntryAndReturnIdImpl(Collections.emptyList(), nodeType, node);
return createNewChildEntryAndReturnIdImpl(Collections.emptyList(), node);
}
public List<Integer> createNewChildEntryAndReturnId(GenericSemanticNode parentNode, NodeType nodeType, GenericSemanticNode node) {
public List<Integer> createNewChildEntryAndReturnId(GenericSemanticNode parentNode, GenericSemanticNode node) {
return createNewChildEntryAndReturnIdImpl(parentNode.getTreeId(), nodeType, node);
return createNewChildEntryAndReturnIdImpl(parentNode.getTreeId(), node);
}
public List<Integer> createNewChildEntryAndReturnId(GenericSemanticNode parentNode, NodeType nodeType, Table node) {
public List<Integer> createNewChildEntryAndReturnId(GenericSemanticNode parentNode, Table node) {
return createNewChildEntryAndReturnIdImpl(parentNode.getTreeId(), nodeType, node);
return createNewChildEntryAndReturnIdImpl(parentNode.getTreeId(), node);
}
public List<Integer> createNewTableChildEntryAndReturnId(Table parentTable, NodeType nodeType, TableCell tableCell) {
public List<Integer> createNewTableChildEntryAndReturnId(Table parentTable, TableCell tableCell) {
return createNewChildEntryAndReturnIdImpl(parentTable.getTreeId(), nodeType, tableCell);
return createNewChildEntryAndReturnIdImpl(parentTable.getTreeId(), tableCell);
}
@SuppressWarnings("PMD.UnusedPrivateMethod") // PMD actually flags this wrong
private List<Integer> createNewChildEntryAndReturnIdImpl(List<Integer> parentId, NodeType nodeType, SemanticNode node) {
private List<Integer> createNewChildEntryAndReturnIdImpl(List<Integer> parentId, SemanticNode node) {
if (!entryExists(parentId)) {
throw new UnsupportedOperationException(format("parentId %s does not exist!", parentId));
@ -77,7 +77,7 @@ public class DocumentTree {
Entry parent = getEntryById(parentId);
List<Integer> newId = new LinkedList<>(parentId);
newId.add(parent.children.size());
parent.children.add(Entry.builder().treeId(newId).node(node).type(nodeType).children(new LinkedList<>()).build());
parent.children.add(Entry.builder().treeId(newId).node(node).children(new LinkedList<>()).build());
return newId;
}
@ -119,7 +119,7 @@ public class DocumentTree {
public Stream<SemanticNode> streamChildNodesOfType(List<Integer> treeId, NodeType nodeType) {
return getEntryById(treeId).children.stream().filter(entry -> entry.type.equals(nodeType)).map(Entry::getNode);
return getEntryById(treeId).children.stream().filter(entry -> entry.node.getType().equals(nodeType)).map(Entry::getNode);
}
@ -195,7 +195,6 @@ public class DocumentTree {
public static class Entry {
List<Integer> treeId;
NodeType type;
SemanticNode node;
List<Entry> children;
@ -207,6 +206,12 @@ public class DocumentTree {
}
public NodeType getType() {
return node.getType();
}
@Override
public int hashCode() {