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

RSS-83: Test new rules for headline detection 4

Browse Source
This commit is contained in:
deiflaender 2022-10-04 09:19:05 +02:00
parent 5503c7929b
commit 65daab4372
2 changed files with 23 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/ClassificationService.java
View File

@ -67,15 +67,7 @@ public class ClassificationService {
// textBlock.setClassification("Other");
// return;
// }
if (PositionUtils.isOverBodyTextFrame(bodyTextFrame, textBlock, page.isRotated()) && (document.getFontSizeCounter()
.getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter()
.getMostPopular())) {
textBlock.setClassification("Header");
} else if (PositionUtils.isUnderBodyTextFrame(bodyTextFrame, textBlock) && (document.getFontSizeCounter()
.getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter()
.getMostPopular())) {
textBlock.setClassification("Footer");
// } else if (page.getPageNumber() == 1 && (!PositionUtils.isTouchingUnderBodyTextFrame(bodyTextFrame, textBlock) && PositionUtils
// .getHeightDifferenceBetweenChunkWordAndDocumentWord(textBlock, document.getTextHeightCounter()
// .getMostPopular()) > 2.5 && textBlock.getHighestFontSize() > document.getFontSizeCounter()
@ -83,11 +75,13 @@ public class ClassificationService {
// if (!Pattern.matches("[0-9]+", textBlock.toString())) {
// textBlock.setClassification("Title");
// }
} else
if (textBlock.getText().length() > 5 && textBlock.getMostPopularWordHeight() > document.getTextHeightCounter().getMostPopular() && PositionUtils.getApproxLineCount(textBlock) < 5.9
&& !textBlock.getOrientation().equals(Orientation.RIGHT)
&& ((textBlock.getMostPopularWordStyle().contains("bold") && Character.isDigit(textBlock.toString().charAt(0))) || textBlock.toString().equals(textBlock.toString().toUpperCase(Locale.ROOT)))) {
&& ((textBlock.getMostPopularWordStyle().contains("bold") && Character.isDigit(textBlock.toString().charAt(0)))
|| textBlock.toString().equals(textBlock.toString().toUpperCase(Locale.ROOT))
|| textBlock.toString().startsWith("APPENDIX") || textBlock.toString().startsWith("TABLE"))) {
textBlock.setClassification("H 1");
document.setHeadlines(true);
@ -97,6 +91,18 @@ public class ClassificationService {
textBlock.setClassification("H 2");
document.setHeadlines(true);
}
else if (PositionUtils.isOverBodyTextFrame(bodyTextFrame, textBlock, page.isRotated()) && (document.getFontSizeCounter()
.getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter()
.getMostPopular())) {
textBlock.setClassification("Header");
} else if (PositionUtils.isUnderBodyTextFrame(bodyTextFrame, textBlock) && (document.getFontSizeCounter()
.getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter()
.getMostPopular())) {
textBlock.setClassification("Footer");
}
// else if (textBlock.getMostPopularWordStyle().equals("bold") && !document.getFontStyleCounter()
// .getMostPopular()
// .equals("bold") && PositionUtils.getApproxLineCount(textBlock) < 2.9 && textBlock.getSequences().get(0).getTextPositions().get(0).getFontSizeInPt() >= textBlock.getMostPopularWordFontSize()) {
@ -121,10 +127,11 @@ public class ClassificationService {
// .getMostPopular()
// .equals("italic") && PositionUtils.getApproxLineCount(textBlock) < 2.9) {
// textBlock.setClassification("TextBlock Italic");
// } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock)) {
// textBlock.setClassification("TextBlock Unknown");
else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock)) {
textBlock.setClassification("TextBlock");
}
else {
textBlock.setClassification("TextBlock");
textBlock.setClassification("Other");
}
}

3
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/PdfSegmentationService.java
View File

@ -96,6 +96,9 @@ public class PdfSegmentationService {
boolean isLandscape = pdr.getWidth() > pdr.getHeight();
int rotation = pdPage.getRotation();
System.out.println("PageNr: " + pageNumber + " rotation: "+rotation);
boolean isRotated = rotation != 0 && rotation != 360;
CleanRulings cleanRulings = rulingCleaningService.getCleanRulings(stripper.getRulings(), stripper.getMinCharWidth(), stripper