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

Pull request #267: Backport fixes

Browse Source 
Merge in RED/redaction-service from backport-fixes to release/2.52.x

* commit 'c49f85916b1503fe08abcacb970a2efdaa165090':
  RED-2624: Fixed footnote is recognized as headline
  RED-2623: Fixed headline handline on strange ocr document
This commit is contained in:
Dominique Eiflaender 2021-11-03 12:41:18 +01:00
commit 1b925b3afa
1 changed files with 5 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/ClassificationService.java
View File

@ -73,8 +73,10 @@ public class ClassificationService {
} }
} else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFontSize() > document } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFontSize() > document
.getFontSizeCounter() .getFontSizeCounter()
.getMostPopular() && PositionUtils.getApproxLineCount(textBlock) < 4.9 && textBlock.getMostPopularWordStyle() .getMostPopular() && PositionUtils.getApproxLineCount(textBlock) < 4.9 && (textBlock.getMostPopularWordStyle()
.equals("bold")) { .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++) { for (int i = 1; i <= headlineFontSizes.size(); i++) {
if (textBlock.getMostPopularWordFontSize() == headlineFontSizes.get(i - 1)) { if (textBlock.getMostPopularWordFontSize() == headlineFontSizes.get(i - 1)) {
@ -86,7 +88,7 @@ public class ClassificationService {
.startsWith("Figure ") && PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordStyle() .startsWith("Figure ") && PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordStyle()
.equals("bold") && !document.getFontStyleCounter() .equals("bold") && !document.getFontStyleCounter()
.getMostPopular() .getMostPopular()
.equals("bold") && PositionUtils.getApproxLineCount(textBlock) < 2.9) { .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("H " + (headlineFontSizes.size() + 1));
document.setHeadlines(true); document.setHeadlines(true);
} else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFontSize() == document } else if (PositionUtils.isWithinBodyTextFrame(bodyTextFrame, textBlock) && textBlock.getMostPopularWordFontSize() == document