Marmelator/layout-parser
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 289 Wiki Activity

Merge branch 'RED-9194' into 'main'

Browse Source 
RED-9194: allow single digit headline identifiers

See merge request fforesight/layout-parser!168
This commit is contained in:
Kilian Schüttler 2024-06-07 09:09:25 +02:00
commit 79795e408a
1 changed files with 11 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/classification/DocuMineClassificationService.java
View File

@ -24,9 +24,9 @@ import lombok.extern.slf4j.Slf4j;
@RequiredArgsConstructor
public class DocuMineClassificationService {
private static final Pattern pattern = Pattern.compile("^([1-9]\\d?\\.){1,3}\\d{1,2}\\.?\\s[0-9A-Za-z \\[\\]]{2,50}", Pattern.CASE_INSENSITIVE);
private static final Pattern pattern2 = Pattern.compile("\\p{L}{3,}", Pattern.CASE_INSENSITIVE);
private static final Pattern pattern3 = Pattern.compile("^(\\d{1,1}\\.){1,3}\\d{1,2}\\.?\\s[a-z]{1,2}\\/[a-z]{1,2}.*");
private static final Pattern HEADLINE_WITH_IDENTIFER_PATTERN = Pattern.compile("^([1-9]\\d?\\.){0,3}\\d{1,2}\\.?\\s[0-9A-Za-z \\[\\]]{2,50}", Pattern.CASE_INSENSITIVE);
private static final Pattern AT_LEAST_3_PATTERN = Pattern.compile("\\p{L}{3,}", Pattern.CASE_INSENSITIVE);
private static final Pattern HEADLINE_PATTTERN_WITH_SLASHES = Pattern.compile("^(\\d{1,1}\\.){0,3}\\d{1,2}\\.?\\s[a-z]{1,2}\\/[a-z]{1,2}.*");
public void classifyDocument(ClassificationDocument document) {
@ -58,9 +58,9 @@ public class DocuMineClassificationService {
log.debug("headlineFontSizes: {}", headlineFontSizes);
var bodyTextFrame = page.getBodyTextFrame();
Matcher matcher = pattern.matcher(textBlock.toString());
Matcher matcher2 = pattern2.matcher(textBlock.toString());
Matcher matcher3 = pattern3.matcher(textBlock.toString());
Matcher headlineWithIdentifierMatcher = HEADLINE_WITH_IDENTIFER_PATTERN.matcher(textBlock.toString());
Matcher atLeast3Matcher = AT_LEAST_3_PATTERN.matcher(textBlock.toString());
Matcher headlineWithSlashesMatcher = HEADLINE_PATTTERN_WITH_SLASHES.matcher(textBlock.toString());
if (textBlock.getClassification() != null && textBlock.getClassification().isHeadline()) {
headlineClassificationService.setLastHeadlineFromOutline(textBlock);
@ -101,19 +101,20 @@ public class DocuMineClassificationService {
|| textBlock.getMostPopularWordFontSize() > document.getFontSizeCounter().getMostPopular())
&& PositionUtils.getApproxLineCount(textBlock) < 5.9
&& (textBlock.getMostPopularWordStyle().contains("bold") && Character.isDigit(textBlock.toString().charAt(0)) && matcher2.reset().find() && !textBlock.toString()
&& (textBlock.getMostPopularWordStyle().contains("bold") && Character.isDigit(textBlock.toString().charAt(0)) && atLeast3Matcher.reset().find() && !textBlock.toString()
.contains(":")
|| textBlock.toString().equals(textBlock.toString().toUpperCase(Locale.ROOT)) && matcher2.reset().find() && !textBlock.toString().contains(":")
|| textBlock.toString().equals(textBlock.toString().toUpperCase(Locale.ROOT)) && atLeast3Matcher.reset().find() && !textBlock.toString().contains(":")
|| textBlock.toString().startsWith("APPENDIX")
|| textBlock.toString().startsWith("FIGURE")
|| textBlock.toString().startsWith("Continued TABLE")
|| textBlock.toString().startsWith("TABLE"))
&& !textBlock.toString().endsWith(":")
&& matcher2.reset().find()) {
&& atLeast3Matcher.reset().find()) {
PageBlockType headlineType = PageBlockType.getHeadlineType(1);
headlineClassificationService.classifyHeadline(textBlock, headlineType);
document.setHeadlines(true);
} else if (matcher.reset().find() && PositionUtils.getApproxLineCount(textBlock) < 2.9 && matcher2.reset().find() && !matcher3.reset().matches()) {
} else if (headlineWithIdentifierMatcher.reset().find() && PositionUtils.getApproxLineCount(textBlock) < 2.9 && atLeast3Matcher.reset().find() && !headlineWithSlashesMatcher.reset().matches()) {
PageBlockType headlineType = PageBlockType.getHeadlineType(2);
headlineClassificationService.classifyHeadline(textBlock, headlineType);
document.setHeadlines(true);