Compare commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
546341ee75 | ||
|
|
0ed1481517 | ||
|
|
b2a47f66ae | ||
|
|
3835d03036 | ||
|
|
b867deb9f9 | ||
|
|
8648ed0952 |
+10
-5
@@ -105,16 +105,21 @@ public class LayoutParsingPipeline {
|
||||
log.info("Starting layout parsing for {}", layoutParsingRequest.identifier());
|
||||
|
||||
File originFile = layoutParsingStorageService.getOriginFile(layoutParsingRequest.originFileStorageId());
|
||||
File viewerDocumentFile = layoutParsingStorageService.getViewerDocFile(layoutParsingRequest.viewerDocumentStorageId()).orElse(originFile);
|
||||
// File viewerDocumentFile = layoutParsingStorageService.getViewerDocFile(layoutParsingRequest.viewerDocumentStorageId()).orElse(originFile);
|
||||
File viewerDocumentFile = originFile;
|
||||
|
||||
VisualLayoutParsingResponse visualLayoutParsingResponse = new VisualLayoutParsingResponse();
|
||||
if (layoutParsingRequest.visualLayoutParsingFileId().isPresent()) {
|
||||
visualLayoutParsingResponse = layoutParsingStorageService.getVisualLayoutParsingFile(layoutParsingRequest.visualLayoutParsingFileId().get());
|
||||
if (layoutParsingRequest.visualLayoutParsingFileId()
|
||||
.isPresent()) {
|
||||
visualLayoutParsingResponse = layoutParsingStorageService.getVisualLayoutParsingFile(layoutParsingRequest.visualLayoutParsingFileId()
|
||||
.get());
|
||||
}
|
||||
|
||||
ImageServiceResponse imageServiceResponse = new ImageServiceResponse();
|
||||
if (layoutParsingRequest.imagesFileStorageId().isPresent()) {
|
||||
imageServiceResponse = layoutParsingStorageService.getImagesFile(layoutParsingRequest.imagesFileStorageId().get());
|
||||
if (layoutParsingRequest.imagesFileStorageId()
|
||||
.isPresent()) {
|
||||
imageServiceResponse = layoutParsingStorageService.getImagesFile(layoutParsingRequest.imagesFileStorageId()
|
||||
.get());
|
||||
}
|
||||
|
||||
TableServiceResponse tableServiceResponse = new TableServiceResponse();
|
||||
|
||||
+1
-1
@@ -63,7 +63,7 @@ public class DocstrumBlockificationService {
|
||||
}
|
||||
|
||||
if (layoutParsingType == LayoutParsingType.CLARIFYND) {
|
||||
mergeIntersectingBlocks(classificationPage, usedRulings, 0, 6.5f);
|
||||
mergeIntersectingBlocks(classificationPage, usedRulings, 0, 0);
|
||||
}
|
||||
|
||||
return classificationPage;
|
||||
|
||||
+1
-2
@@ -71,8 +71,7 @@ public class DocuMineClassificationService {
|
||||
== null
|
||||
|| textBlock.getHighestFontSize()
|
||||
<= document.getFontSizeCounter()
|
||||
.getMostPopular()))
|
||||
|| HeaderFooterDetection.isLikelyHeader(textBlock, document, page)) {
|
||||
.getMostPopular()))) {
|
||||
textBlock.setClassification(PageBlockType.HEADER);
|
||||
|
||||
} else if (MarkedContentUtils.intersects(textBlock, page.getMarkedContentBboxPerType(), MarkedContentUtils.FOOTER)
|
||||
|
||||
+2
-3
@@ -52,17 +52,16 @@ public class ContentStreamUtility {
|
||||
|
||||
public static Optional<COSName> findLayoutGridOCGName(PDPage page) {
|
||||
|
||||
Optional<COSName> layoutGridOCGName = Optional.empty();
|
||||
var resourceIterator = page.getResources().getPropertiesNames();
|
||||
for (COSName cosName : resourceIterator) {
|
||||
COSBase cosBase = page.getResources().getProperties(cosName).getCOSObject().getDictionaryObject(COSName.NAME);
|
||||
if (cosBase instanceof COSString string) {
|
||||
if (ContentStreams.KNECON_LAYOUT.name().equals(string.getString())) {
|
||||
layoutGridOCGName = Optional.of(cosName);
|
||||
return Optional.of(cosName);
|
||||
}
|
||||
}
|
||||
}
|
||||
return layoutGridOCGName;
|
||||
return Optional.empty();
|
||||
}
|
||||
|
||||
|
||||
|
||||
Reference in New Issue
Block a user