diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/SearchableText.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/SearchableText.java index 82debd10..061b2fee 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/SearchableText.java +++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/SearchableText.java @@ -259,7 +259,7 @@ public class SearchableText { sb.append(word.toString()); previous = word; } - return sb.append("\n").toString(); + return TextNormalizationUtilities.removeHyphenLineBreaks(sb.toString()); } } diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/AnalyzeService.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/AnalyzeService.java index f8cfe19a..85e9ec2c 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/AnalyzeService.java +++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/AnalyzeService.java @@ -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(); } diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java b/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java index d4972a98..f8bef256 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java +++ b/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java @@ -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);