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

RED-6945: Generate simplified text with linebreaks

Browse Source
This commit is contained in:
deiflaender 2023-06-20 16:09:37 +02:00
parent ada5df0048
commit c3b2f9d39f
3 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/SearchableText.java
View File

@ -259,7 +259,7 @@ public class SearchableText {
sb.append(word.toString());
previous = word;
}
return sb.append("\n").toString();
return TextNormalizationUtilities.removeHyphenLineBreaks(sb.toString());
}
}

2
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/AnalyzeService.java
View File

@ -367,7 +367,7 @@ public class AnalyzeService {
if (text != null) {
numberOfPages = text.getNumberOfPages();
text.getSectionTexts().stream().filter(Objects::nonNull).forEach(st -> sectionTexts.add(new SimplifiedSectionText(st.getSectionNumber(), st.getText())));
text.getSectionTexts().stream().filter(Objects::nonNull).forEach(st -> sectionTexts.add(new SimplifiedSectionText(st.getSectionNumber(), st.getSearchableText().getAsStringWithLinebreaks())));
}
return SimplifiedText.builder().numberOfPages(numberOfPages).sectionTexts(sectionTexts).build();
}

2
redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java
View File

@ -364,7 +364,7 @@ public class RedactionIntegrationTest {
@Test
public void titleExtraction() throws IOException {
AnalyzeRequest request = prepareStorage("files/new/APN3_Clean_6.1 (6.4.3.01-02)_Apple_211029.pdf");
AnalyzeRequest request = prepareStorage("files/Metolachlor/S-Metolachlor_RAR_01_Volume_1_2018-09-06.pdf");
analyzeService.analyzeDocumentStructure(new StructureAnalyzeRequest(request.getDossierId(), request.getFileId()));
AnalyzeResult result = analyzeService.analyze(request);