Pull request #7: RED-6126 performance test
Merge in RED/ocr-service from RED-6126-performance-test to master * commit '37f1e03ebcd5356e0f0b403a5c0cdd20fc133997': RED-6126: performance-test *refactor to improve cleanness *closed inputStream RED-6126: performance-test *fixed NullPointerException *fixed StackOverFlowError by ignoring very small images and moving to while loop instead of recursion RED-6126: performance-test *fixed time calculation RED-6126: performance-test *improved error logging RED-6126: performance-test *re-enabled overlap detection *re-creating helper document for every page instead of reusing and adding/removing pages RED-6126: Performance Tests *moved to streams for pdf file transfer *disabled overlap detection
This commit is contained in:
Kilian Schuettler
commit
001719a34c
@ -15,6 +15,7 @@ import com.pdftron.common.Matrix2D;
|
||||
import com.pdftron.common.PDFNetException;
|
||||
import com.pdftron.pdf.Element;
|
||||
import com.pdftron.pdf.ElementReader;
|
||||
import com.pdftron.pdf.Image;
|
||||
import com.pdftron.pdf.PDFDoc;
|
||||
import com.pdftron.pdf.Page;
|
||||
import com.pdftron.pdf.Rect;
|
||||
@ -27,9 +28,12 @@ public class ImagePositionRetrievalService {
|
||||
|
||||
private static final double TOLERANCE = 1e-1;
|
||||
|
||||
// any image with smaller height and width than this gets thrown out, see everyPointInDashedLineIsImage.pdf
|
||||
private static final int PIXEL_THRESHOLD = 10;
|
||||
|
||||
|
||||
/**
|
||||
* Iterates over all elements in a PDF Document and retrieves the bounding box for each image,
|
||||
* Iterates over all elements in a PDF Document and retrieves the bounding box for each image, that is larger than the pixel threshold of 10 in either dimension.
|
||||
* Then it adjusts the bounding boxes for the page rotation.
|
||||
* If the mirrorY flag is set, the Y Coordinates are mirrored and moved up by the page height. This is required for PDFTrons OCRModule.
|
||||
*
|
||||
@ -63,7 +67,13 @@ public class ImagePositionRetrievalService {
|
||||
Element element;
|
||||
while ((element = reader.next()) != null) {
|
||||
switch (element.getType()) {
|
||||
case Element.e_image, Element.e_inline_image -> imagePositions.addRect(toRotationAdjustedRect(element.getBBox(), currentPage, mirrorY));
|
||||
case Element.e_image, Element.e_inline_image -> {
|
||||
Image image = new Image(element.getXObject());
|
||||
// see everyPointInDashedLineIsImage.pdf TestFile
|
||||
if (image.getImageHeight() > PIXEL_THRESHOLD || image.getImageWidth() > PIXEL_THRESHOLD) {
|
||||
imagePositions.addRect(toRotationAdjustedRect(element.getBBox(), currentPage, mirrorY));
|
||||
}
|
||||
}
|
||||
case Element.e_form -> {
|
||||
reader.formBegin();
|
||||
findImagePositionsOnPage(reader, imagePositions, currentPage, mirrorY);
|
||||
@ -77,39 +87,49 @@ public class ImagePositionRetrievalService {
|
||||
@SneakyThrows
|
||||
public RectCollection mergeOverlappingRects(RectCollection imagePositions) {
|
||||
|
||||
if (imagePositions.getNumRects() == 1) {
|
||||
if (imagePositions.getNumRects() < 2) {
|
||||
return imagePositions;
|
||||
}
|
||||
|
||||
List<Rectangle2D> rectangleList = toSortedRectangleList(imagePositions);
|
||||
|
||||
rectangleList = mergeRectangleListRecursive(rectangleList, 0);
|
||||
mergeRectangleList(rectangleList);
|
||||
|
||||
return toRectCollection(rectangleList);
|
||||
}
|
||||
|
||||
|
||||
// Sometimes images are split up into stripes, here we merge the positions of aligned and intersecting rectangles into one larger rectangle
|
||||
private List<Rectangle2D> mergeRectangleListRecursive(List<Rectangle2D> rectangleList, int currentIdx) {
|
||||
private void mergeRectangleList(List<Rectangle2D> rectangleList) {
|
||||
|
||||
if (rectangleList.size() < currentIdx + 2) {
|
||||
return rectangleList;
|
||||
for (int idx = 0; rectangleList.size() >= idx + 2; ) {
|
||||
|
||||
var rect1 = rectangleList.get(idx);
|
||||
var rect2 = rectangleList.get(idx + 1);
|
||||
|
||||
if (intersects(rect1, rect2) && isAlignedXOrY(rect1, rect2)) {
|
||||
rectangleList.remove(idx + 1);
|
||||
rectangleList.remove(idx);
|
||||
rectangleList.add(idx, rect1.createUnion(rect2));
|
||||
} else {
|
||||
++idx;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
var rect1 = rectangleList.get(currentIdx);
|
||||
var rect2 = rectangleList.get(currentIdx + 1);
|
||||
|
||||
private boolean intersects(Rectangle2D rect1, Rectangle2D rect2) {
|
||||
|
||||
return rect1.intersects(rect2.getMinX() - TOLERANCE, rect2.getMinY() - TOLERANCE, rect2.getWidth() + (2 * TOLERANCE), rect2.getHeight() + (2 * TOLERANCE));
|
||||
}
|
||||
|
||||
|
||||
private boolean isAlignedXOrY(Rectangle2D rect1, Rectangle2D rect2) {
|
||||
|
||||
boolean isAlignedX = Math.abs(rect1.getMinX() - rect2.getMinX()) < TOLERANCE && Math.abs(rect1.getMaxX() - rect2.getMaxX()) < TOLERANCE;
|
||||
boolean isAlignedY = Math.abs(rect1.getMinY() - rect2.getMinY()) < TOLERANCE && Math.abs(rect1.getMaxY() - rect2.getMaxY()) < TOLERANCE;
|
||||
boolean intersects = rect1.intersects(rect2.getMinX() - TOLERANCE, rect2.getMinY() - TOLERANCE, rect2.getWidth() + (2 * TOLERANCE), rect2.getHeight() + (2 * TOLERANCE));
|
||||
|
||||
if (intersects && (isAlignedX || isAlignedY)) {
|
||||
rectangleList.remove(currentIdx + 1);
|
||||
rectangleList.remove(currentIdx);
|
||||
rectangleList.add(currentIdx, rect1.createUnion(rect2));
|
||||
return mergeRectangleListRecursive(rectangleList, currentIdx);
|
||||
} else {
|
||||
return mergeRectangleListRecursive(rectangleList, currentIdx + 1);
|
||||
}
|
||||
return isAlignedX || isAlignedY;
|
||||
}
|
||||
|
||||
|
||||
|
||||
@ -5,6 +5,7 @@ import java.awt.geom.AffineTransform;
|
||||
import java.awt.geom.GeneralPath;
|
||||
import java.awt.geom.Rectangle2D;
|
||||
import java.io.InputStream;
|
||||
import java.io.OutputStream;
|
||||
import java.util.ArrayList;
|
||||
import java.util.Iterator;
|
||||
import java.util.List;
|
||||
@ -59,10 +60,10 @@ public class InvisibleElementRemovalService {
|
||||
* @param pdfFile The PDF file to process
|
||||
* @param delta If this flag is set only the removed Elements will be written to the output file.
|
||||
* The Elements are red if they are removed by clipping path, blue for transparency, and a green bounding box for overlap.
|
||||
* @return The resulting PDF File as bytes.
|
||||
* @param out OutputStream to write the resulting file to
|
||||
**/
|
||||
@SneakyThrows
|
||||
public byte[] removeInvisibleElements(InputStream pdfFile, boolean delta) {
|
||||
public void removeInvisibleElements(InputStream pdfFile, OutputStream out, boolean delta) {
|
||||
|
||||
PDFDoc pdfDoc = new PDFDoc(pdfFile);
|
||||
|
||||
@ -75,6 +76,8 @@ public class InvisibleElementRemovalService {
|
||||
Page page = iterator.next();
|
||||
|
||||
visitedXObjIds.add(page.getSDFObj().getObjNum());
|
||||
|
||||
|
||||
InvisibleElementRemovalContext context = InvisibleElementRemovalContext.builder()
|
||||
.reader(reader)
|
||||
.clippingPathStack(new ClippingPathStack(page.getMediaBox()))
|
||||
@ -90,7 +93,7 @@ public class InvisibleElementRemovalService {
|
||||
|
||||
removeOverlappedElements(page, writer, context);
|
||||
}
|
||||
return pdfDoc.save(SDFDoc.SaveMode.REMOVE_UNUSED, null);
|
||||
pdfDoc.save(out, SDFDoc.SaveMode.REMOVE_UNUSED, null);
|
||||
}
|
||||
|
||||
|
||||
@ -220,12 +223,18 @@ public class InvisibleElementRemovalService {
|
||||
|
||||
|
||||
private void processPath(Element pathElement, ElementWriter writer, InvisibleElementRemovalContext context) throws PDFNetException {
|
||||
PathData pathData = pathElement.getPathData();
|
||||
|
||||
GeneralPath linePath = convertToGeneralPath(pathElement.getPathData());
|
||||
if (pathData.getOperators().length == 0 && pathData.getPoints().length == 0) {
|
||||
writer.writeGStateChanges(pathElement);
|
||||
return;
|
||||
}
|
||||
|
||||
GeneralPath linePath = convertToGeneralPath(pathData);
|
||||
|
||||
//transform path to initial user space
|
||||
var ctm = pathElement.getCTM();
|
||||
var affineTransform = getAffineTransform(ctm);
|
||||
var affineTransform = toAffineTransform(ctm);
|
||||
linePath.transform(affineTransform);
|
||||
|
||||
var rect = linePath.getBounds2D();
|
||||
@ -244,8 +253,13 @@ public class InvisibleElementRemovalService {
|
||||
writer.writeElement(pathElement);
|
||||
|
||||
} else {
|
||||
if (pathElement.isWindingFill()) {
|
||||
linePath.setWindingRule(GeneralPath.WIND_NON_ZERO);
|
||||
} else {
|
||||
linePath.setWindingRule(GeneralPath.WIND_EVEN_ODD);
|
||||
}
|
||||
|
||||
if (inClippingPath) {
|
||||
// TODO: WINDING RULE
|
||||
if (isFilledAndNonTransparent(pathElement)) {
|
||||
List<ElementFeatures> currentOverlappedElements = context.visibleElements()
|
||||
.stream()
|
||||
@ -270,12 +284,6 @@ public class InvisibleElementRemovalService {
|
||||
}
|
||||
|
||||
|
||||
private static AffineTransform getAffineTransform(Matrix2D ctm) throws PDFNetException {
|
||||
|
||||
return new AffineTransform(ctm.getA(), ctm.getB(), ctm.getC(), ctm.getD(), ctm.getH(), ctm.getV());
|
||||
}
|
||||
|
||||
|
||||
private void removeOverlappedElements(Page page, ElementWriter writer, InvisibleElementRemovalContext context) throws PDFNetException {
|
||||
|
||||
context.reader().begin(page);
|
||||
@ -422,6 +430,12 @@ public class InvisibleElementRemovalService {
|
||||
}
|
||||
|
||||
|
||||
private static AffineTransform toAffineTransform(Matrix2D ctm) throws PDFNetException {
|
||||
|
||||
return new AffineTransform(ctm.getA(), ctm.getB(), ctm.getC(), ctm.getD(), ctm.getH(), ctm.getV());
|
||||
}
|
||||
|
||||
|
||||
@Builder
|
||||
private record InvisibleElementRemovalContext(
|
||||
boolean delta,
|
||||
|
||||
@ -1,10 +1,14 @@
|
||||
package com.iqser.red.service.ocr.v1.server.service;
|
||||
|
||||
import static java.lang.String.format;
|
||||
|
||||
import java.io.ByteArrayInputStream;
|
||||
import java.io.ByteArrayOutputStream;
|
||||
import java.io.IOException;
|
||||
import java.io.InputStream;
|
||||
import java.io.OutputStream;
|
||||
import java.util.Map;
|
||||
|
||||
import io.micrometer.core.annotation.Timed;
|
||||
import org.springframework.amqp.rabbit.core.RabbitTemplate;
|
||||
import org.springframework.stereotype.Service;
|
||||
|
||||
@ -21,6 +25,7 @@ import com.pdftron.pdf.Page;
|
||||
import com.pdftron.pdf.RectCollection;
|
||||
import com.pdftron.sdf.SDFDoc;
|
||||
|
||||
import io.micrometer.core.annotation.Timed;
|
||||
import lombok.RequiredArgsConstructor;
|
||||
import lombok.SneakyThrows;
|
||||
import lombok.extern.slf4j.Slf4j;
|
||||
@ -55,58 +60,56 @@ public class OCRService {
|
||||
*
|
||||
* @param dossierId The dossier id
|
||||
* @param fileId The file id
|
||||
* @return the resulting PDF file as an InputStream
|
||||
* @param out OutputStream to write the file to
|
||||
*/
|
||||
@SneakyThrows
|
||||
@Timed("redactmanager_runOcrOnDocument")
|
||||
public InputStream runOcrOnDocument(String dossierId, String fileId) {
|
||||
|
||||
InputStream fileStream = fileStorageService.getOriginalFileAsStream(dossierId, fileId);
|
||||
|
||||
byte[] fileWithoutInvisibleTextBytes = invisibleElementRemovalService.removeInvisibleElements(fileStream, false);
|
||||
|
||||
byte[] ocrBytes = runOcr(fileWithoutInvisibleTextBytes, fileId);
|
||||
|
||||
return new ByteArrayInputStream(ocrBytes);
|
||||
public void runOcrOnDocument(String dossierId, String fileId, OutputStream out) throws IOException {
|
||||
|
||||
try (ByteArrayOutputStream transferOutputStream = new ByteArrayOutputStream()) {
|
||||
try (InputStream fileStream = fileStorageService.getOriginalFileAsStream(dossierId, fileId)) {
|
||||
long start = System.currentTimeMillis();
|
||||
log.debug("Start invisible element removal for file with dossierId {} and fileId {}", dossierId, fileId);
|
||||
invisibleElementRemovalService.removeInvisibleElements(fileStream, transferOutputStream, false);
|
||||
long end = System.currentTimeMillis();
|
||||
log.debug("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())) {
|
||||
long start = System.currentTimeMillis();
|
||||
runOcr(transferInputStream, out, fileId);
|
||||
long end = System.currentTimeMillis();
|
||||
log.info("ocr successful for file with dossierId {} and fileId {}, took {}s", dossierId, fileId, format("%.1f", (end - start) / 1000.0));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@SneakyThrows
|
||||
private byte[] runOcr(byte[] file, String fileId) {
|
||||
private void runOcr(InputStream fileStream, OutputStream out, String fileId) {
|
||||
|
||||
PDFDoc pdfDoc = new PDFDoc(file);
|
||||
PDFDoc pdfDoc = new PDFDoc(fileStream);
|
||||
|
||||
Map<Integer, RectCollection> pageIdToRectCollection = imagePositionRetrievalService.getImagePositionPerPage(pdfDoc, true);
|
||||
|
||||
OCROptions options = new OCROptions();
|
||||
PDFDoc ocrPageDoc = new PDFDoc();
|
||||
// Optimization:
|
||||
// When a page does not have a TextZone, PDFTron whites out the page. But, PDFTron scans it anyway, resulting in a longer runtime.
|
||||
// So, we need to remove pages without images.
|
||||
// Furthermore, creating a new document is *much* faster than reusing the same document and adding/removing pages one by one.
|
||||
// Therefore, we create a new Document with a single page for every page that contains text.
|
||||
int numProcessedPages = 0;
|
||||
for (Integer pageId : pageIdToRectCollection.keySet()) {
|
||||
try {
|
||||
// optimization by only scanning pages that contain images
|
||||
Page pdfPage = pdfDoc.getPage(pageId);
|
||||
pdfPage.setMediaBox(pdfPage.getCropBox()); // this line ensures the ocr text is placed correctly by PDFTron
|
||||
ocrPageDoc.pagePushBack(pdfPage);
|
||||
options.addTextZonesForPage(pageIdToRectCollection.get(pageId), 1);
|
||||
options.addLang(ENGLISH);
|
||||
options.addDPI(settings.getOcrDPI());
|
||||
|
||||
OCRModule.processPDF(ocrPageDoc, options);
|
||||
PDFDoc singlePagePdfDoc = extractSinglePagePdfDoc(pdfDoc, pageId);
|
||||
processOcr(pageIdToRectCollection, pageId, singlePagePdfDoc);
|
||||
++numProcessedPages;
|
||||
|
||||
StringBuilder zonesString = new StringBuilder();
|
||||
for (int j = 0; j < pageIdToRectCollection.get(pageId).getNumRects(); ++j) {
|
||||
var r = pageIdToRectCollection.get(pageId).getRectAt(j);
|
||||
zonesString.append(String.format("[lower left (%.1f|%.1f) upper right (%.1f|%.1f)]", r.getX1(), r.getY1(), r.getX2(), r.getY2()));
|
||||
}
|
||||
log.info("{}/{} Page {} done, OCR regions {}", numProcessedPages, pageIdToRectCollection.size(), pageId, zonesString);
|
||||
log.info("{}/{} Page {} done, OCR regions {}",
|
||||
numProcessedPages,
|
||||
pageIdToRectCollection.size(),
|
||||
pageId,
|
||||
getAllOcrTextZonesAsString(pageIdToRectCollection, pageId));
|
||||
|
||||
// re-adding OCR pages
|
||||
Page ocrPage = ocrPageDoc.getPage(1);
|
||||
pdfDoc.pageInsert(pdfDoc.getPageIterator(pageId), ocrPage);
|
||||
pdfDoc.pageRemove(pdfDoc.getPageIterator(pageId + 1));
|
||||
ocrPageDoc.pageRemove(ocrPageDoc.getPageIterator(1));
|
||||
replaceOriginalPageWithOcrPage(pdfDoc, pageId, singlePagePdfDoc);
|
||||
singlePagePdfDoc.close();
|
||||
|
||||
rabbitTemplate.convertAndSend(MessagingConfiguration.OCR_STATUS_UPDATE_RESPONSE_QUEUE,
|
||||
objectMapper.writeValueAsString(OCRStatusUpdateResponse.builder()
|
||||
@ -116,13 +119,11 @@ public class OCRService {
|
||||
.build()));
|
||||
|
||||
} catch (PDFNetException e) {
|
||||
log.error("failed to process page {}", pageId);
|
||||
log.error("Failed to process page {}", pageId);
|
||||
throw new RuntimeException(e);
|
||||
}
|
||||
}
|
||||
|
||||
ocrPageDoc.close();
|
||||
|
||||
rabbitTemplate.convertAndSend(MessagingConfiguration.OCR_STATUS_UPDATE_RESPONSE_QUEUE,
|
||||
objectMapper.writeValueAsString(OCRStatusUpdateResponse.builder()
|
||||
.fileId(fileId)
|
||||
@ -132,6 +133,52 @@ public class OCRService {
|
||||
.build()));
|
||||
|
||||
Optimizer.optimize(pdfDoc);
|
||||
return pdfDoc.save(SDFDoc.SaveMode.LINEARIZED, null);
|
||||
try {
|
||||
pdfDoc.save(out, SDFDoc.SaveMode.LINEARIZED, null);
|
||||
} catch (Exception e) {
|
||||
log.error("Processed File with fileId {} could not be saved", fileId);
|
||||
throw new RuntimeException(e);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
private void processOcr(Map<Integer, RectCollection> pageIdToRectCollection, Integer pageId, PDFDoc singlePagePdfDoc) throws PDFNetException {
|
||||
|
||||
OCROptions options = new OCROptions();
|
||||
options.addTextZonesForPage(pageIdToRectCollection.get(pageId), 1);
|
||||
options.addLang(ENGLISH);
|
||||
options.addDPI(settings.getOcrDPI());
|
||||
|
||||
OCRModule.processPDF(singlePagePdfDoc, options);
|
||||
}
|
||||
|
||||
|
||||
private static PDFDoc extractSinglePagePdfDoc(PDFDoc pdfDoc, Integer pageId) throws PDFNetException {
|
||||
|
||||
PDFDoc singlePagePdfDoc = new PDFDoc();
|
||||
Page page = pdfDoc.getPage(pageId);
|
||||
page.setMediaBox(page.getCropBox()); // this line ensures the ocr text is placed correctly by PDFTron, see TestFile MediaBoxBiggerThanCropBox.pdf
|
||||
singlePagePdfDoc.pagePushBack(page);
|
||||
return singlePagePdfDoc;
|
||||
}
|
||||
|
||||
|
||||
private static void replaceOriginalPageWithOcrPage(PDFDoc pdfDoc, Integer pageId, PDFDoc ocrPageDoc) throws PDFNetException {
|
||||
|
||||
Page ocrPage = ocrPageDoc.getPage(1);
|
||||
pdfDoc.pageInsert(pdfDoc.getPageIterator(pageId), ocrPage);
|
||||
pdfDoc.pageRemove(pdfDoc.getPageIterator(pageId + 1));
|
||||
}
|
||||
|
||||
|
||||
private static StringBuilder getAllOcrTextZonesAsString(Map<Integer, RectCollection> pageIdToRectCollection, Integer pageId) throws PDFNetException {
|
||||
|
||||
StringBuilder zonesString = new StringBuilder();
|
||||
for (int j = 0; j < pageIdToRectCollection.get(pageId).getNumRects(); ++j) {
|
||||
var r = pageIdToRectCollection.get(pageId).getRectAt(j);
|
||||
zonesString.append(format("[lower left (%.1f|%.1f) upper right (%.1f|%.1f)]", r.getX1(), r.getY1(), r.getX2(), r.getY2()));
|
||||
}
|
||||
return zonesString;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
@ -1,5 +1,9 @@
|
||||
package com.iqser.red.service.ocr.v1.server.service;
|
||||
|
||||
import java.io.ByteArrayInputStream;
|
||||
import java.io.ByteArrayOutputStream;
|
||||
import java.io.IOException;
|
||||
|
||||
import org.springframework.amqp.AmqpRejectAndDontRequeueException;
|
||||
import org.springframework.amqp.rabbit.annotation.RabbitHandler;
|
||||
import org.springframework.amqp.rabbit.annotation.RabbitListener;
|
||||
@ -8,9 +12,9 @@ import org.springframework.stereotype.Service;
|
||||
|
||||
import com.fasterxml.jackson.core.JsonProcessingException;
|
||||
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.iqser.red.service.ocr.v1.api.model.DocumentRequest;
|
||||
|
||||
import feign.FeignException;
|
||||
import lombok.RequiredArgsConstructor;
|
||||
@ -34,7 +38,6 @@ public class OcrMessageReceiver {
|
||||
|
||||
DocumentRequest ocrRequestMessage = objectMapper.readValue(in, DocumentRequest.class);
|
||||
|
||||
long start = System.currentTimeMillis();
|
||||
log.info("Start ocr for file with dossierId {} and fileId {}", ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId());
|
||||
|
||||
setStatusOcrProcessing(ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId());
|
||||
@ -44,14 +47,18 @@ public class OcrMessageReceiver {
|
||||
fileStorageService.storeUntouchedFile(ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId(), originalFile);
|
||||
}
|
||||
|
||||
var ocrResult = ocrService.runOcrOnDocument(ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId());
|
||||
|
||||
fileStorageService.storeOriginalFile(ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId(), ocrResult);
|
||||
|
||||
long end = System.currentTimeMillis();
|
||||
log.info("Successfully processed ocr for file with dossierId {} and fileId {}, took {}", ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId(), end - start);
|
||||
try (var transferStream = new ByteArrayOutputStream()) {
|
||||
ocrService.runOcrOnDocument(ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId(), transferStream);
|
||||
try (var inputStream = new ByteArrayInputStream(transferStream.toByteArray())) {
|
||||
fileStorageService.storeOriginalFile(ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId(), inputStream);
|
||||
}
|
||||
} catch (IOException e) {
|
||||
log.error("Failed to store file with dossierId {} and fileId {}", ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId());
|
||||
throw new RuntimeException(e);
|
||||
}
|
||||
|
||||
fileStatusProcessingUpdateClient.ocrSuccessful(ocrRequestMessage.getDossierId(), ocrRequestMessage.getFileId());
|
||||
|
||||
}
|
||||
|
||||
|
||||
|
||||
@ -3,14 +3,14 @@ package com.iqser.red.service.ocr.v1.server;
|
||||
import static com.iqser.red.service.ocr.v1.server.utils.OsUtils.getTemporaryDirectory;
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.io.FileInputStream;
|
||||
import java.io.FileOutputStream;
|
||||
import java.io.IOException;
|
||||
import java.io.InputStream;
|
||||
import java.util.ArrayList;
|
||||
import java.util.List;
|
||||
import java.util.concurrent.TimeUnit;
|
||||
|
||||
import io.micrometer.prometheus.PrometheusMeterRegistry;
|
||||
import io.micrometer.prometheus.PrometheusTimer;
|
||||
import org.junit.jupiter.api.AfterEach;
|
||||
import org.junit.jupiter.api.BeforeEach;
|
||||
import org.junit.jupiter.api.Disabled;
|
||||
@ -36,12 +36,15 @@ import com.iqser.red.service.ocr.v1.server.utils.FileSystemBackedStorageService;
|
||||
import com.iqser.red.service.persistence.service.v1.api.model.dossiertemplate.dossier.file.FileType;
|
||||
import com.iqser.red.storage.commons.StorageAutoConfiguration;
|
||||
import com.iqser.red.storage.commons.service.StorageService;
|
||||
import com.pdftron.common.PDFNetException;
|
||||
import com.pdftron.pdf.OCRModule;
|
||||
import com.pdftron.pdf.PDFDoc;
|
||||
import com.pdftron.pdf.Page;
|
||||
import com.pdftron.pdf.PageIterator;
|
||||
import com.pdftron.pdf.TextExtractor;
|
||||
|
||||
import io.micrometer.prometheus.PrometheusMeterRegistry;
|
||||
import io.micrometer.prometheus.PrometheusTimer;
|
||||
import lombok.SneakyThrows;
|
||||
|
||||
@ExtendWith(SpringExtension.class)
|
||||
@ -80,19 +83,20 @@ public class OcrServiceIntegrationTest {
|
||||
|
||||
@Test
|
||||
@Disabled // OCRModule is not available on build server. If you want to run the test set the property at the top.
|
||||
public void testOCRMetrics(){
|
||||
public void testOCRMetrics() {
|
||||
|
||||
testOCR("Watermark");
|
||||
testOCR("Watermark");
|
||||
testOCR("Watermark");
|
||||
|
||||
var ocrOnDocumentMeter = registry.getMeters().stream()
|
||||
.filter(m -> m.getId().getName().equalsIgnoreCase("redactmanager_runOcrOnDocument")).findAny();
|
||||
var ocrOnDocumentMeter = registry.getMeters().stream().filter(m -> m.getId().getName().equalsIgnoreCase("redactmanager_runOcrOnDocument")).findAny();
|
||||
assertThat(ocrOnDocumentMeter.isPresent()).isTrue();
|
||||
PrometheusTimer timer = (PrometheusTimer) ocrOnDocumentMeter.get();
|
||||
assertThat(timer.count()).isEqualTo(3);
|
||||
assertThat(timer.mean(TimeUnit.SECONDS)).isGreaterThan(0.1);
|
||||
}
|
||||
|
||||
|
||||
@Test
|
||||
@Disabled // OCRModule is not available on build server. If you want to run the test set the property at the top.
|
||||
public void testOcr() {
|
||||
@ -153,34 +157,42 @@ public class OcrServiceIntegrationTest {
|
||||
private String testOCR(String fileName) {
|
||||
|
||||
ClassPathResource pdfFileResource = new ClassPathResource("files/" + fileName + ".pdf");
|
||||
|
||||
var originId = FileStorageService.getStorageId("dossier", "file", FileType.ORIGIN);
|
||||
storageService.storeObject(originId, pdfFileResource.getInputStream());
|
||||
|
||||
try (InputStream ocrDocument = ocrService.runOcrOnDocument("dossier", "file")) {
|
||||
byte[] ocrDocumentBytes = ocrDocument.readAllBytes();
|
||||
try (var out = new FileOutputStream(getTemporaryDirectory() + "/" + fileName + ".pdf")) {
|
||||
out.write(ocrDocumentBytes);
|
||||
try (var fileStream = pdfFileResource.getInputStream()) {
|
||||
storageService.storeObject(originId, fileStream);
|
||||
}
|
||||
try (var out = new FileOutputStream(getTemporaryDirectory() + "/" + fileName + ".pdf")) {
|
||||
ocrService.runOcrOnDocument("dossier", "file", out);
|
||||
}
|
||||
|
||||
System.out.println("File:" + getTemporaryDirectory() + "/" + fileName + ".pdf");
|
||||
try (var fileStream = new FileInputStream(getTemporaryDirectory() + "/" + fileName + ".pdf")) {
|
||||
return extractAllTextFromDocument(fileStream);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
private static String extractAllTextFromDocument(InputStream fileStream) throws IOException, PDFNetException {
|
||||
|
||||
PDFDoc pdfDoc = new PDFDoc(fileStream);
|
||||
TextExtractor extractor = new TextExtractor();
|
||||
List<String> texts = new ArrayList<>();
|
||||
PDFDoc pdfDoc = new PDFDoc(ocrDocumentBytes);
|
||||
|
||||
PageIterator iterator = pdfDoc.getPageIterator();
|
||||
while (iterator.hasNext()) {
|
||||
Page page = iterator.next();
|
||||
extractor.begin(page);
|
||||
texts.add(extractor.getAsText());
|
||||
}
|
||||
System.out.println("File:" + getTemporaryDirectory() + "/" + fileName + ".pdf");
|
||||
|
||||
return String.join("\n", texts);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@SneakyThrows
|
||||
public void dummyTest() {
|
||||
|
||||
// Build needs one text to not fail.
|
||||
// Build needs one test to not fail.
|
||||
assertThat(1).isEqualTo(1);
|
||||
}
|
||||
|
||||
@ -200,7 +212,7 @@ public class OcrServiceIntegrationTest {
|
||||
|
||||
@Bean
|
||||
@Primary
|
||||
public StorageService inmemoryStorage() {
|
||||
public StorageService inMemoryStorage() {
|
||||
|
||||
return new FileSystemBackedStorageService();
|
||||
}
|
||||
|
||||
@ -122,6 +122,14 @@ class ImagePositionRetrievalServiceTest {
|
||||
assertThat(allRectCoords.size()).isEqualTo(48);
|
||||
}
|
||||
|
||||
@Test
|
||||
@SneakyThrows
|
||||
public void testEveryPointInDashedLineIsImage() {
|
||||
String fileName = "everyPointInDashedLineIsImage";
|
||||
List<int[]> allRectCoords = testImagePositionDetection(fileName);
|
||||
assertThat(allRectCoords.size()).isEqualTo(0);
|
||||
}
|
||||
|
||||
|
||||
private List<int[]> testImagePositionDetection(String fileName) throws IOException, PDFNetException {
|
||||
|
||||
|
||||
@ -3,9 +3,8 @@ package com.iqser.red.service.ocr.v1.server.service;
|
||||
import static com.iqser.red.service.ocr.v1.server.utils.OsUtils.getTemporaryDirectory;
|
||||
import static org.assertj.core.api.AssertionsForClassTypes.assertThat;
|
||||
|
||||
import java.io.FileInputStream;
|
||||
import java.io.FileOutputStream;
|
||||
import java.io.IOException;
|
||||
import java.nio.file.Files;
|
||||
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.junit.jupiter.api.extension.ExtendWith;
|
||||
@ -46,22 +45,23 @@ public class InvisibleElementRemovalServiceTest {
|
||||
|
||||
ClassPathResource pdfFileResource = new ClassPathResource("files/" + fileName + ".pdf");
|
||||
|
||||
var initialFileStream = Files.newInputStream(pdfFileResource.getFile().toPath());
|
||||
var fileWithoutInvisibleElements = invisibleElementRemovalService.removeInvisibleElements(initialFileStream, false);
|
||||
try (var initialFileStream = pdfFileResource.getInputStream(); var out = new FileOutputStream(getTemporaryDirectory() + "/" + fileName + ".pdf")) {
|
||||
invisibleElementRemovalService.removeInvisibleElements(initialFileStream, out,false);
|
||||
}
|
||||
|
||||
initialFileStream = Files.newInputStream(pdfFileResource.getFile().toPath());
|
||||
var deltaFile = invisibleElementRemovalService.removeInvisibleElements(initialFileStream, true);
|
||||
try (var initialFileStream = pdfFileResource.getInputStream(); var out = new FileOutputStream(getTemporaryDirectory() + "/" + fileName + "_delta.pdf")) {
|
||||
invisibleElementRemovalService.removeInvisibleElements(initialFileStream, out,true);
|
||||
}
|
||||
|
||||
String fileWithoutInvisibleTextLocation = getTemporaryDirectory() + "/" + fileName + ".pdf";
|
||||
String deltaFileLocation = getTemporaryDirectory() + "/" + fileName + "_delta.pdf";
|
||||
|
||||
saveToFile(fileWithoutInvisibleTextLocation, fileWithoutInvisibleElements);
|
||||
saveToFile(deltaFileLocation, deltaFile);
|
||||
|
||||
System.out.println("Output File without invisible elements: " + fileWithoutInvisibleTextLocation);
|
||||
System.out.println("Output Delta File: " + deltaFileLocation);
|
||||
System.out.println("Output File without invisible elements: files/" + fileName + ".pdf");
|
||||
System.out.println("Output Delta File without invisible elements: files/" + fileName + "_delta.pdf");
|
||||
TextExtractor extractor = new TextExtractor();
|
||||
PDFDoc pdfDoc = new PDFDoc(fileWithoutInvisibleElements);
|
||||
|
||||
PDFDoc pdfDoc;
|
||||
try(var fileStream = new FileInputStream(getTemporaryDirectory() + "/" + fileName + ".pdf")) {
|
||||
pdfDoc = new PDFDoc(fileStream);
|
||||
}
|
||||
|
||||
PageIterator iterator = pdfDoc.getPageIterator();
|
||||
while (iterator.hasNext()) {
|
||||
Page page = iterator.next();
|
||||
@ -70,16 +70,4 @@ public class InvisibleElementRemovalServiceTest {
|
||||
assertThat(text).contains("APPENDIX 16 Pathology Report", "Amendment 1", "Page 255 of 260");
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
private void saveToFile(String location, byte[] fileBytes) {
|
||||
|
||||
try (var f_out = new FileOutputStream(location)) {
|
||||
f_out.write(fileBytes);
|
||||
} catch (IOException e) {
|
||||
throw new RuntimeException("File location: " + location + "could not be openend, no file will be saved");
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
Binary file not shown.
@ -0,0 +1,16 @@
|
||||
<Configuration>
|
||||
|
||||
<Appenders>
|
||||
<Console name="CONSOLE" target="SYSTEM_OUT">
|
||||
<PatternLayout pattern="%d{HH:mm:ss.SSS} [%t] %-5level %logger{36} - %msg%n"/>
|
||||
</Console>
|
||||
</Appenders>
|
||||
|
||||
<Loggers>
|
||||
<Root level="warn">
|
||||
<AppenderRef ref="CONSOLE"/>
|
||||
</Root>
|
||||
<Logger name="com.iqser" level="info"/>
|
||||
</Loggers>
|
||||
|
||||
</Configuration>
|
||||
Loading…
x
Reference in New Issue
Block a user