Marmelator/ocr-service
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 64 Wiki Activity

RED-6126: performance-test

Browse Source 
*added PDFNet terminate
This commit is contained in:
Kilian Schuettler 2023-02-09 14:37:09 +01:00
parent 8db0b712f7
commit e2c0be95d5
2 changed files with 5 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ocr-service-v1/ocr-service-server-v1/src/main/java/com/iqser/red/service/ocr/v1/server/service/OCRService.java
View File

@ -72,7 +72,7 @@ public class OCRService {
invisibleElementRemovalService.removeInvisibleElements(fileStream, transferOutputStream, false);
}
long end = System.currentTimeMillis();
log.info("Invisible element removal successful for file with dossierId {} and fileId {}, took {}s", dossierId, fileId, format("%.1f", (float) ((end - start) / 1000)));
log.info("Invisible element removal successful for file with dossierId {} and fileId {}, took {}s", dossierId, fileId, format("%.1f", (end - start) / 1000.0));
try (InputStream transferInputStream = new ByteArrayInputStream(transferOutputStream.toByteArray())) {
runOcr(transferInputStream, out, fileId);
}
@ -93,12 +93,12 @@ public class OCRService {
Map<Integer, RectCollection> pageIdToRectCollection = imagePositionRetrievalService.getImagePositionPerPage(pdfDoc, true);
int numProcessedPages = 0;
// optimization: only scanning pages that contain images
for (Integer pageId : pageIdToRectCollection.keySet()) {
try {
// optimization: creating a new document is faster than reusing the same and adding/removing pages one by one
OCROptions options = new OCROptions();
PDFDoc ocrPageDoc = new PDFDoc();
// optimization: only scanning pages that contain images
Page pdfPage = pdfDoc.getPage(pageId);
// optimization: this line ensures the ocr text is placed correctly by PDFTron
pdfPage.setMediaBox(pdfPage.getCropBox());

4
ocr-service-v1/ocr-service-server-v1/src/main/java/com/iqser/red/service/ocr/v1/server/service/OcrMessageReceiver.java
View File

@ -15,6 +15,7 @@ import com.fasterxml.jackson.databind.ObjectMapper;
import com.iqser.red.service.ocr.v1.api.model.DocumentRequest;
import com.iqser.red.service.ocr.v1.server.client.FileStatusProcessingUpdateClient;
import com.iqser.red.service.ocr.v1.server.configuration.MessagingConfiguration;
import com.pdftron.pdf.PDFNet;
import feign.FeignException;
import lombok.RequiredArgsConstructor;
@ -57,11 +58,12 @@ public class OcrMessageReceiver {
throw new RuntimeException(e);
}
long end = System.currentTimeMillis();
log.info("Successfully processed ocr for file with dossierId {} and fileId {}, took {}", ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId(), end - start);
fileStatusProcessingUpdateClient.ocrSuccessful(ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId());
PDFNet.terminate();
}