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

Pull request #116: Fixed table extraction problems

Browse Source 
Merge in RED/redaction-service from TableExtractionProbemFix to master

* commit 'fc2ac03691871a8a1f3c8a29756e3a60e63e16bf':
  Fixed table extraction problems
This commit is contained in:
Dominique Eiflaender 2021-02-03 14:49:01 +01:00
commit 76ecfdabd3
4 changed files with 9 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/parsing/PDFLinesTextStripper.java
View File

@ -54,10 +54,7 @@ public class PDFLinesTextStripper extends PDFTextStripper {
protected PDPage pdpage;
@Getter
private int maxCharWidths;
@Getter
private int maxCharHeight;
private int minCharWidths;
@Getter
private final List<TextPositionSequence> textPositionSequences = new ArrayList<>();
@ -282,14 +279,9 @@ public class PDFLinesTextStripper extends PDFTextStripper {
int startIndex = 0;
for (int i = 0; i <= textPositions.size() - 1; i++) {
int charHeight = (int) textPositions.get(i).getHeightDir();
if (charHeight > maxCharHeight) {
maxCharHeight = charHeight;
}
int charWidth = (int) textPositions.get(i).getWidthDirAdj();
if (charWidth > maxCharWidths) {
maxCharWidths = charWidth;
if (charWidth < minCharWidths) {
minCharWidths = charWidth;
}
if (i == 0 && (textPositions.get(i).getUnicode().equals(" ") || textPositions.get(i)
@ -341,8 +333,7 @@ public class PDFLinesTextStripper extends PDFTextStripper {
@Override
public String getText(PDDocument doc) throws IOException {
maxCharWidths = 0;
maxCharWidths = 0;
minCharWidths = Integer.MAX_VALUE;
textPositionSequences.clear();
imageBounds = new ArrayList<>();
rulings.clear();

3
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/parsing/model/ParsedElements.java
View File

@ -19,6 +19,5 @@ public class ParsedElements {
private boolean landscape;
private boolean rotated;
private float maxCharWidth;
private float maxCharHeight;
private float minCharWidth;
}

5
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/PdfSegmentationService.java
View File

@ -61,14 +61,13 @@ public class PdfSegmentationService {
.rulings(stripper.getRulings())
.sequences(stripper.getTextPositionSequences())
.imageBounds(stripper.getImageBounds())
.maxCharWidth(stripper.getMaxCharWidths())
.maxCharHeight(stripper.getMaxCharWidths())
.minCharWidth(stripper.getMinCharWidths())
.landscape(isLandscape)
.rotated(isRotated)
.build();
CleanRulings cleanRulings = rulingCleaningService.getCleanRulings(parsedElements.getRulings(), parsedElements
.getMaxCharWidth(), parsedElements.getMaxCharHeight());
.getMinCharWidth());
Page page = blockificationService.blockify(parsedElements.getSequences(), cleanRulings.getHorizontal(), cleanRulings
.getVertical());

4
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/tableextraction/service/RulingCleaningService.java
View File

@ -18,9 +18,9 @@ import com.iqser.red.service.redaction.v1.server.tableextraction.utils.Utils;
@Service
public class RulingCleaningService {
public CleanRulings getCleanRulings(List<Ruling> rulings, float maxCharWidth, float maxCharHeight){
public CleanRulings getCleanRulings(List<Ruling> rulings, float minCharWidth){
if (!rulings.isEmpty()) {
snapPoints(rulings, maxCharWidth , maxCharHeight);
snapPoints(rulings, minCharWidth , minCharWidth);
}
List<Ruling> vrs = new ArrayList<>();