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

RED-8481: Use visual layout parsing to detect signatures

Browse Source 
addressed review comments
This commit is contained in:
yhampe 2024-02-15 12:12:23 +01:00
parent b4a225144d
commit bdf1161c91
11 changed files with 20 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
layoutparser-service/layoutparser-service-internal-api/src/main/java/com/knecon/fforesight/service/layoutparser/internal/api/queue/LayoutParsingRequest.java
View File

@ -25,7 +25,7 @@ public record LayoutParsingRequest(
@Schema(description = "Optional Path to the image classification file.")//
Optional<String> imagesFileStorageId,//
@Schema(description = "Optional Path to the the visual layout parsing service file") Optional<String> visualLayoutParsingFileId,
@Schema(description = "Optional Path to the the visual layout parsing service file") Optional<String> visualLayoutParsingFileId,//
@Schema(description = "Path where the Document Structure File will be stored.")//
@NonNull String structureFileStorageId,//

12
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/LayoutParsingPipeline.java
View File

@ -38,7 +38,6 @@ import com.knecon.fforesight.service.layoutparser.processor.python_api.model.ima
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableCells;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResult;
import com.knecon.fforesight.service.layoutparser.processor.services.BodyTextFrameService;
import com.knecon.fforesight.service.layoutparser.processor.services.RulingCleaningService;
import com.knecon.fforesight.service.layoutparser.processor.services.SectionsBuilderService;
@ -101,8 +100,8 @@ public class LayoutParsingPipeline {
File viewerDocumentFile = layoutParsingStorageService.getViewerDocFile(layoutParsingRequest.viewerDocumentStorageId()).orElse(originFile);
VisualLayoutParsingResponse visualLayoutParsingResponse = new VisualLayoutParsingResponse();
if (layoutParsingRequest.visualLayoutParsingFileId() != null && layoutParsingRequest.visualLayoutParsingFileId().isPresent()) {
visualLayoutParsingResponse = layoutParsingStorageService.getExtractedTablesFile(layoutParsingRequest.visualLayoutParsingFileId().get());
if (layoutParsingRequest.visualLayoutParsingFileId().isPresent()) {
visualLayoutParsingResponse = layoutParsingStorageService.getVisualLayoutParsingFile(layoutParsingRequest.visualLayoutParsingFileId().get());
}
ImageServiceResponse imageServiceResponse = new ImageServiceResponse();
@ -215,10 +214,9 @@ public class LayoutParsingPipeline {
addNumberOfPagesToTrace(originDocument.getNumberOfPages(), Files.size(originFile.toPath()));
Map<Integer, List<TableCells>> pdfTableCells = cvTableParsingAdapter.buildCvParsedTablesPerPage(tableServiceResponse);
Map<Integer, List<ClassifiedImage>> pdfImages = imageServiceResponseAdapter.buildClassifiedImagesPerPage(imageServiceResponse);
Map<Integer, List<ClassifiedImage>> signatures = null;
if(visualLayoutParsingResponse != null) {
log.info("response: {}",visualLayoutParsingResponse);
signatures = visualLayoutParsingAdapter.buildExtractedSignaturesPerPage(visualLayoutParsingResponse);
Map<Integer, List<ClassifiedImage>> signatures = new HashMap<>();
if(signatures.size() > 0) {
visualLayoutParsingAdapter.buildExtractedSignaturesPerPage(visualLayoutParsingResponse);
}
ClassificationDocument classificationDocument = new ClassificationDocument();

5
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/LayoutParsingStorageService.java
View File

@ -10,8 +10,6 @@ import java.nio.file.Paths;
import java.nio.file.StandardOpenOption;
import java.util.Optional;
import org.apache.pdfbox.Loader;
import org.apache.pdfbox.pdmodel.PDDocument;
import org.springframework.stereotype.Service;
import com.fasterxml.jackson.databind.ObjectMapper;
@ -80,11 +78,10 @@ public class LayoutParsingStorageService {
}
}
public VisualLayoutParsingResponse getExtractedTablesFile(String storageId) throws IOException {
public VisualLayoutParsingResponse getVisualLayoutParsingFile(String storageId) throws IOException {
try (InputStream inputStream = getObject(storageId)) {
VisualLayoutParsingResponse visualLayoutParsingResponse = objectMapper.readValue(inputStream, VisualLayoutParsingResponse.class);
inputStream.close();
return visualLayoutParsingResponse;
}
}

4
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/python_api/model/table/VisualLayoutParsingBox.java
View File

@ -1,7 +1,5 @@
package com.knecon.fforesight.service.layoutparser.processor.python_api.model.table;
import java.util.List;
import lombok.AllArgsConstructor;
import lombok.Builder;
import lombok.Data;
@ -13,7 +11,7 @@ import lombok.NoArgsConstructor;
@AllArgsConstructor
public class VisualLayoutParsingBox {
private com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingBoxValue box;
private VisualLayoutParsingBoxValue box;
private String label;
private float probability;

3
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/BdrJsonBuildTest.java
View File

@ -27,6 +27,7 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Do
import com.knecon.fforesight.service.layoutparser.processor.model.graph.textblock.TextBlock;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.image.ImageServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
import com.knecon.fforesight.service.layoutparser.processor.services.factory.DocumentGraphFactory;
import com.knecon.fforesight.service.layoutparser.processor.services.mapper.TaasDocumentDataMapper;
import com.knecon.fforesight.service.layoutparser.server.utils.AbstractTest;
@ -50,7 +51,7 @@ public class BdrJsonBuildTest extends AbstractTest {
file,
new ImageServiceResponse(),
new TableServiceResponse(),
null,
new VisualLayoutParsingResponse(),
file.toString()));
}

3
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/HeadlinesGoldStandardIntegrationTest.java
View File

@ -33,6 +33,7 @@ import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Do
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.SemanticNode;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.image.ImageServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
import com.knecon.fforesight.service.layoutparser.processor.services.factory.DocumentGraphFactory;
import com.knecon.fforesight.tenantcommons.TenantsClient;
@ -98,7 +99,7 @@ public class HeadlinesGoldStandardIntegrationTest {
pdfFileResource.getFile(),
new ImageServiceResponse(),
new TableServiceResponse(),
null,
new VisualLayoutParsingResponse(),
filePath));
var foundHeadlines = documentGraph.streamAllSubNodes()

5
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/VisualLayoutParsingServiceTests.java
View File

@ -1,5 +0,0 @@
package com.knecon.fforesight.service.layoutparser.server;
public class VisualLayoutParsingServiceTests {
}

3
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/graph/DocumentGraphJsonWritingTest.java
View File

@ -16,6 +16,7 @@ import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsi
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.image.ImageServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
import com.knecon.fforesight.service.layoutparser.processor.services.factory.DocumentGraphFactory;
import com.knecon.fforesight.service.layoutparser.processor.services.mapper.DocumentDataMapper;
import com.knecon.fforesight.service.layoutparser.server.utils.BuildDocumentTest;
@ -58,7 +59,7 @@ public class DocumentGraphJsonWritingTest extends BuildDocumentTest {
filename.toFile(),
new ImageServiceResponse(),
new TableServiceResponse(),
null,
new VisualLayoutParsingResponse(),
filename.toFile().toString()));
DocumentData documentData = DocumentDataMapper.toDocumentData(documentGraph);

3
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/graph/ViewerDocumentTest.java
View File

@ -12,6 +12,7 @@ import com.knecon.fforesight.service.layoutparser.internal.api.queue.LayoutParsi
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.image.ImageServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
import com.knecon.fforesight.service.layoutparser.processor.services.factory.DocumentGraphFactory;
import com.knecon.fforesight.service.layoutparser.processor.services.visualization.LayoutGridService;
import com.knecon.fforesight.service.layoutparser.server.utils.BuildDocumentTest;
@ -51,7 +52,7 @@ public class ViewerDocumentTest extends BuildDocumentTest {
var tableResponse = mapper.readValue(new ClassPathResource(tableFileName).getInputStream(), TableServiceResponse.class);
var documentFile = new ClassPathResource(fileName).getFile();
var classificationDocument = layoutParsingPipeline.parseLayout(LayoutParsingType.DOCUMINE, documentFile, new ImageServiceResponse(), tableResponse, null,Path.of(fileName).getFileName().toFile().toString());
var classificationDocument = layoutParsingPipeline.parseLayout(LayoutParsingType.DOCUMINE, documentFile, new ImageServiceResponse(), tableResponse, new VisualLayoutParsingResponse(),Path.of(fileName).getFileName().toFile().toString());
ViewerDocumentService viewerDocumentService = new ViewerDocumentService(null);
LayoutGridService layoutGridService = new LayoutGridService(viewerDocumentService);
Document document = DocumentGraphFactory.buildDocumentGraph(classificationDocument);

3
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/segmentation/PdfSegmentationServiceTest.java
View File

@ -33,6 +33,7 @@ import com.knecon.fforesight.service.layoutparser.processor.python_api.adapter.C
import com.knecon.fforesight.service.layoutparser.processor.python_api.adapter.ImageServiceResponseAdapter;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.image.ImageServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
import com.knecon.fforesight.service.layoutparser.processor.services.SectionsBuilderService;
import com.knecon.fforesight.service.layoutparser.processor.services.classification.RedactManagerClassificationService;
import com.knecon.fforesight.service.layoutparser.server.utils.AbstractTest;
@ -67,7 +68,7 @@ public class PdfSegmentationServiceTest extends AbstractTest {
originDocument,
new ImageServiceResponse(),
tableServiceResponse,
null,
new VisualLayoutParsingResponse(),
"document");
redactManagerClassificationService.classifyDocument(classificationDocument);

3
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/utils/BuildDocumentTest.java
View File

@ -10,6 +10,7 @@ import com.knecon.fforesight.service.layoutparser.processor.LayoutParsingPipelin
import com.knecon.fforesight.service.layoutparser.processor.model.ClassificationDocument;
import com.knecon.fforesight.service.layoutparser.processor.model.graph.nodes.Document;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.VisualLayoutParsingResponse;
import com.knecon.fforesight.service.layoutparser.processor.services.factory.DocumentGraphFactory;
import lombok.SneakyThrows;
@ -25,7 +26,7 @@ public abstract class BuildDocumentTest extends AbstractTest {
File fileResource = new ClassPathResource(filename).getFile();
prepareStorage(filename);
return layoutParsingPipeline.parseLayout(layoutParsingType, fileResource, layoutParsingStorageService.getImagesFile(IMAGE_FILE_ID), new TableServiceResponse(), null,filename);
return layoutParsingPipeline.parseLayout(layoutParsingType, fileResource, layoutParsingStorageService.getImagesFile(IMAGE_FILE_ID), new TableServiceResponse(), new VisualLayoutParsingResponse(),filename);
}