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

RED-6126: performance-test

Browse Source 
*fixed NullPointerException
*fixed StackOverFlowError by ignoring very small images
This commit is contained in:
Kilian Schuettler 2023-02-09 17:01:57 +01:00
parent 7065d098f3
commit adf1a4ce15
4 changed files with 29 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
ocr-service-v1/ocr-service-server-v1/src/main/java/com/iqser/red/service/ocr/v1/server/service/ImagePositionRetrievalService.java
View File

@ -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;
@ -27,6 +28,9 @@ 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,
@ -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);
@ -77,7 +87,7 @@ public class ImagePositionRetrievalService {
@SneakyThrows
public RectCollection mergeOverlappingRects(RectCollection imagePositions) {
if (imagePositions.getNumRects() == 1) {
if (imagePositions.getNumRects() < 2) {
return imagePositions;
}

10
ocr-service-v1/ocr-service-server-v1/src/main/java/com/iqser/red/service/ocr/v1/server/service/InvisibleElementRemovalService.java
View File

@ -76,6 +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()))
@ -221,8 +223,14 @@ 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();

8
ocr-service-v1/ocr-service-server-v1/src/test/java/com/iqser/red/service/ocr/v1/server/service/ImagePositionRetrievalServiceTest.java
View File

@ -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 {

BIN
ocr-service-v1/ocr-service-server-v1/src/test/resources/files/everyPointInDashedLineIsImage.pdf Normal file
View File

Binary file not shown.