Pull request #174 : RED-1326: Enabled to recategorize images

Merge in RED/redaction-service from RED-1326 to master * commit '8ff3e463eae960a35e11fce7589e24131ccb51f2': RED-1326: Enabled to recategorize images
RED-1326: Enabled to recategorize images
2021-06-18 11:53:21 +02:00 · 2021-06-18 11:29:43 +02:00 · 2021-06-17 12:21:01 +02:00 · 2021-06-17 12:16:07 +02:00 · 2021-06-16 11:20:40 +02:00 · 2021-06-16 10:54:51 +02:00
13 changed files with 320 additions and 180 deletions
@@ -6,6 +6,7 @@ import lombok.Data;
 import lombok.NoArgsConstructor;

 import java.time.OffsetDateTime;
+import java.util.Set;

@Data
@Builder
@@ -19,6 +20,7 @@ public class AnalyzeRequest {
    private boolean reanalyseOnlyIfPossible;
    private ManualRedactions manualRedactions;
    private OffsetDateTime lastProcessed;
+    private Set<Integer> excludedPages;

 }

@@ -0,0 +1,21 @@
+package com.iqser.red.service.redaction.v1.model;
+
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+@Data
+@Builder
+@AllArgsConstructor
+@NoArgsConstructor
+public class ManualImageRecategorization {
+
+    private String id;
+    private String user;
+    private Status status;
+    private String type;
+    private String legalBasis;
+    private boolean redacted;
+
+}
@@ -1,5 +1,5 @@
 package com.iqser.red.service.redaction.v1.model;

 public enum ManualRedactionType {
-    ADD, REMOVE, FORCE_REDACT
+    ADD, REMOVE, FORCE_REDACT, RECATEGORIZE
 }
@@ -26,6 +26,9 @@ public class ManualRedactions {
    @Builder.Default
    private Set<ManualRedactionEntry> entriesToAdd = new HashSet<>();

+    @Builder.Default
+    private Set<ManualImageRecategorization> imageRecategorizations = new HashSet<>();
+
    @Builder.Default
    private Map<String, List<Comment>> comments = new HashMap<>();

@@ -44,4 +44,6 @@ public class RedactionChangeLogEntry {

    private boolean isDossierDictionaryEntry;

+    private boolean excluded;
+
 }
@@ -49,4 +49,8 @@ public class RedactionLogEntry {

    private boolean isDossierDictionaryEntry;

+    private boolean excluded;
+
+    private String recategorizationType;
+
 }
@@ -14,16 +14,19 @@ public class AnalyzeResponseService {

        boolean hasRequests = redactionLog.getRedactionLogEntry()
                .stream()
+                .filter(entry -> !entry.isExcluded())
                .anyMatch(entry -> entry.isManual() && entry.getStatus()
                        .equals(com.iqser.red.service.redaction.v1.model.Status.REQUESTED));

        boolean hasRedactions = redactionLog.getRedactionLogEntry()
                .stream()
+                .filter(entry -> !entry.isExcluded())
                .anyMatch(entry -> entry.isRedacted() && !entry.isManual() || entry.isManual() && entry.getStatus()
                        .equals(com.iqser.red.service.redaction.v1.model.Status.APPROVED));

        boolean hasImages = redactionLog.getRedactionLogEntry()
                .stream()
+                .filter(entry -> !entry.isExcluded())
                .anyMatch(entry -> entry.isHint() && entry.getType().equals("image"));

        boolean hasUpdates = redactionChangeLog != null && redactionChangeLog.getRedactionLogEntry() != null && !redactionChangeLog
@@ -1,14 +1,44 @@
 package com.iqser.red.service.redaction.v1.server.redaction.service;

+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
+import java.util.stream.Collectors;
+import java.util.stream.Stream;
+
+import org.kie.api.runtime.KieContainer;
+import org.springframework.stereotype.Service;
+import org.springframework.web.bind.annotation.RequestBody;
+
 import com.iqser.red.service.file.management.v1.api.model.FileType;
-import com.iqser.red.service.redaction.v1.model.*;
+import com.iqser.red.service.redaction.v1.model.AnalyzeRequest;
+import com.iqser.red.service.redaction.v1.model.AnalyzeResult;
+import com.iqser.red.service.redaction.v1.model.Comment;
+import com.iqser.red.service.redaction.v1.model.IdRemoval;
+import com.iqser.red.service.redaction.v1.model.ManualForceRedact;
+import com.iqser.red.service.redaction.v1.model.ManualRedactionEntry;
+import com.iqser.red.service.redaction.v1.model.ManualRedactions;
+import com.iqser.red.service.redaction.v1.model.Rectangle;
+import com.iqser.red.service.redaction.v1.model.RedactionLog;
+import com.iqser.red.service.redaction.v1.model.RedactionLogEntry;
+import com.iqser.red.service.redaction.v1.model.SectionArea;
 import com.iqser.red.service.redaction.v1.server.classification.model.Document;
 import com.iqser.red.service.redaction.v1.server.classification.model.SectionText;
 import com.iqser.red.service.redaction.v1.server.classification.model.Text;
 import com.iqser.red.service.redaction.v1.server.client.LegalBasisClient;
 import com.iqser.red.service.redaction.v1.server.exception.RedactionException;
 import com.iqser.red.service.redaction.v1.server.redaction.model.Dictionary;
-import com.iqser.red.service.redaction.v1.server.redaction.model.*;
+import com.iqser.red.service.redaction.v1.server.redaction.model.DictionaryIncrement;
+import com.iqser.red.service.redaction.v1.server.redaction.model.DictionaryVersion;
+import com.iqser.red.service.redaction.v1.server.redaction.model.Entity;
+import com.iqser.red.service.redaction.v1.server.redaction.model.EntityPositionSequence;
+import com.iqser.red.service.redaction.v1.server.redaction.model.Image;
+import com.iqser.red.service.redaction.v1.server.redaction.model.RedRectangle2D;
+import com.iqser.red.service.redaction.v1.server.redaction.model.Section;
+import com.iqser.red.service.redaction.v1.server.redaction.model.SectionSearchableTextPair;
 import com.iqser.red.service.redaction.v1.server.redaction.utils.EntitySearchUtils;
 import com.iqser.red.service.redaction.v1.server.segmentation.PdfSegmentationService;
 import com.iqser.red.service.redaction.v1.server.storage.RedactionStorageService;
@@ -17,14 +47,6 @@ import lombok.RequiredArgsConstructor;
 import lombok.SneakyThrows;
 import lombok.extern.slf4j.Slf4j;

-import org.kie.api.runtime.KieContainer;
-import org.springframework.stereotype.Service;
-import org.springframework.web.bind.annotation.RequestBody;
-
-import java.util.*;
-import java.util.stream.Collectors;
-import java.util.stream.Stream;
-
@Slf4j
@Service
@RequiredArgsConstructor
@@ -41,6 +63,7 @@ public class ReanalyzeService {
    private final AnalyzeResponseService analyzeResponseService;
    private final LegalBasisClient legalBasisClient;

+
    public AnalyzeResult analyze(AnalyzeRequest analyzeRequest) {

        long startTime = System.currentTimeMillis();
@@ -66,11 +89,11 @@ public class ReanalyzeService {
        log.info("Redaction analysis successful...");

        var legalBasis = legalBasisClient.getLegalBasisMapping(analyzeRequest.getDossierTemplateId());
-        var redactionLog = new RedactionLog(classifiedDoc.getRedactionLogEntities(),legalBasis,
-                classifiedDoc.getDictionaryVersion().getDossierTemplateVersion(),
-                classifiedDoc.getDictionaryVersion().getDossierVersion(),
-                classifiedDoc.getRulesVersion(),
-                legalBasisClient.getVersion(analyzeRequest.getDossierTemplateId()));
+        var redactionLog = new RedactionLog(classifiedDoc.getRedactionLogEntities(), legalBasis, classifiedDoc.getDictionaryVersion()
+                .getDossierTemplateVersion(), classifiedDoc.getDictionaryVersion()
+                .getDossierVersion(), classifiedDoc.getRulesVersion(), legalBasisClient.getVersion(analyzeRequest.getDossierTemplateId()));
+
+        excludeExcludedPages(redactionLog, analyzeRequest.getExcludedPages());

        log.info("Analyzed with rules {} and dictionary {} for dossierTemplate: {}", classifiedDoc.getRulesVersion(), classifiedDoc
                .getDictionaryVersion(), analyzeRequest.getDossierTemplateId());
@@ -165,14 +188,15 @@ public class ReanalyzeService {

        KieContainer kieContainer = droolsExecutionService.updateRules(analyzeRequest.getDossierTemplateId());

-        Dictionary dictionary = dictionaryService.getDeepCopyDictionary(analyzeRequest.getDossierTemplateId(), analyzeRequest.getDossierId());
+        Dictionary dictionary = dictionaryService.getDeepCopyDictionary(analyzeRequest.getDossierTemplateId(), analyzeRequest
+                .getDossierId());

        List<SectionSearchableTextPair> sectionSearchableTextPairs = new ArrayList<>();
        for (SectionText reanalysisSection : reanalysisSections) {

            Set<Entity> entities = entityRedactionService.findEntities(reanalysisSection.getSearchableText(), reanalysisSection
                    .getHeadline(), reanalysisSection.getSectionNumber(), dictionary, false);
-            if (reanalysisSection.getCellStarts() != null) {
+            if (reanalysisSection.getCellStarts() != null && !reanalysisSection.getCellStarts().isEmpty()) {
                surroundingWordsService.addSurroundingText(entities, reanalysisSection.getSearchableText(), dictionary, reanalysisSection
                        .getCellStarts());
            } else {
@@ -240,8 +264,7 @@ public class ReanalyzeService {
                    .getDossierTemplateId()));
        }

-        redactionLog.getRedactionLogEntry()
-                .removeIf(entry -> sectionsToReanalyse.contains(entry.getSectionNumber()));
+        redactionLog.getRedactionLogEntry().removeIf(entry -> sectionsToReanalyse.contains(entry.getSectionNumber()));
        redactionLog.getRedactionLogEntry().addAll(newRedactionLogEntries);
        return finalizeAnalysis(analyzeRequest, startTime, redactionLog, text, dictionaryIncrement);

@@ -255,6 +278,8 @@ public class ReanalyzeService {
        redactionLog.setDictionaryVersion(dictionaryIncrement.getDictionaryVersion().getDossierTemplateVersion());
        redactionLog.setDossierDictionaryVersion(dictionaryIncrement.getDictionaryVersion().getDossierVersion());

+        excludeExcludedPages(redactionLog, analyzeRequest.getExcludedPages());
+
        var changeLog = redactionChangeLogService.createAndStoreChangeLog(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), redactionLog);
        redactionStorageService.storeObject(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), FileType.REDACTION_LOG, redactionLog);

@@ -292,4 +317,18 @@ public class ReanalyzeService {
                .build();
    }

+
+    private void excludeExcludedPages(RedactionLog redactionLog, Set<Integer> excludedPages) {
+
+        redactionLog.getRedactionLogEntry().forEach(entry -> {
+            entry.getPositions().forEach(pos -> {
+                if (excludedPages != null && excludedPages.contains(pos.getPage())) {
+                    entry.setExcluded(true);
+                } else {
+                    entry.setExcluded(false);
+                }
+            });
+        });
+    }
+
 }
@@ -90,6 +90,7 @@ public class RedactionChangeLogService {
                .comments(entry.getComments())
                .changeType(changeType)
                .isDossierDictionaryEntry(entry.isDossierDictionaryEntry())
+                .excluded(entry.isExcluded())
                .build();
    }

@@ -13,7 +13,9 @@ import com.iqser.red.service.redaction.v1.server.redaction.utils.IdBuilder;
 import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractTextContainer;
 import com.iqser.red.service.redaction.v1.server.tableextraction.model.Cell;
 import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table;
+
 import lombok.RequiredArgsConstructor;
+
 import org.apache.commons.collections4.CollectionUtils;
 import org.springframework.stereotype.Service;

@@ -87,6 +89,36 @@ public class RedactionLogCreatorService {
                    .section(image.getSection())
                    .build();

+            if (manualRedactions != null && !manualRedactions.getImageRecategorizations().isEmpty()) {
+                for (ManualImageRecategorization recategorization : manualRedactions.getImageRecategorizations()) {
+                    if (recategorization.getId().equals(id)) {
+                        String manualOverrideReason = null;
+                        if (recategorization.getStatus().equals(Status.APPROVED)) {
+                            image.setType(recategorization.getType());
+                            image.setRedaction(recategorization.isRedacted());
+                            image.setLegalBasis(recategorization.getLegalBasis());
+                            redactionLogEntry.setRedacted(recategorization.isRedacted());
+                            redactionLogEntry.setStatus(Status.APPROVED);
+                            redactionLogEntry.setLegalBasis(recategorization.getLegalBasis());
+                            manualOverrideReason = image.getRedactionReason() + ", recategorized by manual override";
+                            redactionLogEntry.setColor(getColorForImage(image, dossierTemplateId, false));
+                        } else if (recategorization.getStatus().equals(Status.REQUESTED)) {
+                            manualOverrideReason = image.getRedactionReason() + ", requested to recategorize";
+                            redactionLogEntry.setStatus(Status.REQUESTED);
+                            redactionLogEntry.setColor(getColorForImage(image, dossierTemplateId, true));
+                            redactionLogEntry.setRecategorizationType(recategorization.getType());
+                        } else {
+                            redactionLogEntry.setStatus(Status.DECLINED);
+                        }
+
+                        image.setRedactionReason(manualOverrideReason != null ? manualOverrideReason : image.getRedactionReason());
+                        redactionLogEntry.setReason(manualOverrideReason);
+                        redactionLogEntry.setManual(true);
+                        redactionLogEntry.setManualRedactionType(ManualRedactionType.RECATEGORIZE);
+                    }
+                }
+            }
+
            if (manualRedactions != null && !manualRedactions.getIdsToRemove().isEmpty()) {
                for (IdRemoval manualRemoval : manualRedactions.getIdsToRemove()) {
                    if (manualRemoval.getId().equals(id)) {
@@ -276,20 +308,22 @@ public class RedactionLogCreatorService {

        List<Rectangle> rectangles = new ArrayList<>();
        if (textPositions.size() == 1) {
-            rectangles.add( TextPositionSequence.fromData(textPositions, page).getRectangle());
+            rectangles.add(TextPositionSequence.fromData(textPositions, page).getRectangle());
        } else {
            float y = textPositions.get(0).getYDirAdj();
            int startIndex = 0;
            for (int i = 1; i < textPositions.size(); i++) {
                float yDirAdj = textPositions.get(i).getYDirAdj();
                if (yDirAdj != y) {
-                    rectangles.add( TextPositionSequence.fromData(textPositions.subList(startIndex, i), page).getRectangle());
+                    rectangles.add(TextPositionSequence.fromData(textPositions.subList(startIndex, i), page)
+                            .getRectangle());
                    y = yDirAdj;
                    startIndex = i;
                }
            }
            if (startIndex != textPositions.size()) {
-                rectangles.add( TextPositionSequence.fromData(textPositions.subList(startIndex, textPositions.size()), page).getRectangle());
+                rectangles.add(TextPositionSequence.fromData(textPositions.subList(startIndex, textPositions.size()), page)
+                        .getRectangle());
            }
        }

@@ -0,0 +1,165 @@
+package com.iqser.red.service.redaction.v1.server.segmentation;
+
+import java.awt.Graphics;
+import java.awt.geom.Rectangle2D;
+import java.awt.image.BufferedImage;
+import java.util.ArrayList;
+import java.util.List;
+
+import org.springframework.stereotype.Service;
+
+import com.iqser.red.service.redaction.v1.server.redaction.model.PdfImage;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class ImageMergeService {
+
+
+    public List<PdfImage> mergeImages(List<PdfImage> images, int rotation){
+
+        List<PdfImage> mergedList = processImages(images, rotation);
+
+        List<PdfImage> imagesInImage = new ArrayList<>();
+        for(PdfImage image: mergedList){
+            for (PdfImage inner: mergedList){
+                if(image != inner && image.getPosition().contains(inner.getPosition().getX(), inner.getPosition().getY(), inner.getPosition().getWidth(), inner.getPosition().getHeight())){
+                    imagesInImage.add(inner);
+                }
+            }
+        }
+        mergedList.removeAll(imagesInImage);
+
+        return mergedList;
+    }
+
+
+    //merge images, if they are separated during pdf import, return new list of Pdfimages
+    private List<PdfImage> processImages(List<PdfImage> imageList, int rotation) {
+        if (imageList.size() > 1) {
+            List<PdfImage> mergedList = new ArrayList<>();
+            int countElementsInList = 0;
+            boolean beginImage = true;
+
+            // a List of Boolean, true = candidate for merging, false = no merging
+            List<Boolean> candidatesList = getCandidatesList(imageList, rotation);
+
+            // loop through list, if there are candidates for merging (true), merge images and add it to mergedList
+            for (int i = 0; i < candidatesList.size(); i++) {
+                if (candidatesList.get(i)) {
+                    if (beginImage) {
+                        //begin of image, merge two parts of imageList
+                        PdfImage mergedImage = mergeTwoImages(imageList.get(i), imageList.get(i + 1), rotation);
+                        // image merge successful
+                        if (mergedImage != null) {
+                            mergedList.add(mergedImage);
+                            countElementsInList++;
+                        }
+                    } else {
+                        //middle of an image, merge current piece auf mergedList with image of imageList
+                        PdfImage mergedImage = mergeTwoImages(mergedList.get(countElementsInList - 1), imageList.get(i + 1), rotation);
+                        // image merge successful
+                        if (mergedImage != null) {
+                            mergedList.set(countElementsInList - 1, mergedImage);
+                        }
+                    }
+                    beginImage = false;
+                } else {
+                    // if the last candidate is false, then both images i and i+1 must be added
+                    if (i == candidatesList.size() - 1) {
+                        if (countElementsInList > 0 && mergedList.get(countElementsInList - 1) == imageList.get(i)) {
+                            mergedList.add(imageList.get(i + 1));
+                        } else {
+                            mergedList.add(imageList.get(i));
+                            mergedList.add(imageList.get(i + 1));
+                        }
+                    } else {
+                        //first image is not splitted, add i to resultlist
+                        if (beginImage) {
+                            mergedList.add(imageList.get(i));
+                            countElementsInList++;
+                        } else {
+                            // i is the end of an image, add begin of new image
+                            mergedList.add(imageList.get(i + 1));
+                            countElementsInList++;
+                            beginImage = false;
+                        }
+                    }
+                }
+            }
+            return mergedList;
+        } else {
+            return imageList;
+        }
+    }
+
+    private PdfImage mergeTwoImages(PdfImage image1, PdfImage image2, int rotation) {
+
+        // diese Angaben von getPosition scheinen nicht richtig zu sein, damit werden teile des Bildes abgeschnitten
+        double width = image1.getPosition().getWidth();
+        double width2 = image2.getPosition().getWidth();
+        double height1 = image1.getPosition().getHeight();
+        double height2 = image2.getPosition().getHeight();
+        // mit  den Werten, die unter Image gespeichert sind, funktioniert es
+        double img1height = image1.getImage().getHeight();
+        double img1width = image1.getImage().getWidth();
+        double img2height = image2.getImage().getHeight();
+
+        BufferedImage mergedImage = new BufferedImage((int) img1width, (int) (img1height + img2height), BufferedImage.TYPE_INT_RGB);
+        Graphics mergedImageGraphics = mergedImage.getGraphics();
+        try {
+            mergedImageGraphics.drawImage(image1.getImage(), 0, 0, null);
+            mergedImageGraphics.drawImage(image2.getImage(), 0, (int) (img1height), null);
+
+            // set Image, Position and type for merged Image
+            //set position for merged image with values of image1 and the height of both
+            Rectangle2D pos = new Rectangle2D.Float();
+            pos.setRect(image1.getPosition().getX(), image2.getPosition().getY(), rotation == 90 ? width + width2: width, rotation == 90 ? height1 : height1 + height2);
+            PdfImage newPdfImage = new PdfImage(mergedImage, pos, image1.getPage());
+            // Graphics need to be disposed
+
+            image1.getImage().flush();
+            image2.getImage().flush();
+
+            mergedImage.flush();
+            mergedImageGraphics.dispose();
+
+            return newPdfImage;
+        } catch (Exception e) {
+            // failed to merge image
+            log.error("Failed to merge image", e);
+            return null;
+        }
+
+
+    }
+
+    //make a list of true and false, if the image is a candidate for merging
+    private List<Boolean> getCandidatesList(List<PdfImage> imageList, int rotation) {
+        List<Boolean> candidatesList = new ArrayList<>();
+        for (int i = 0; i < imageList.size(); i++) {
+            if (i >= 1) {
+                candidatesList.add(isCandidateForMerging(imageList.get(i - 1), imageList.get(i), rotation));
+            }
+        }
+        return candidatesList;
+    }
+
+    // evaluate if two images are candidates for merging, depending on their coordinates, width and height
+    private boolean isCandidateForMerging(PdfImage image1, PdfImage image2, int rotation) {
+        double x1 = rotation == 90 ? image1.getPosition().getY() : image1.getPosition().getX();
+        double y1 = rotation == 90 ? image1.getPosition().getX() : image1.getPosition().getY();
+        double width1 = rotation == 90 ? image1.getPosition().getHeight() : image1.getPosition().getWidth();
+        double x2 = rotation == 90 ? image2.getPosition().getY() : image2.getPosition().getX();
+        double y2 = rotation == 90 ? image2.getPosition().getX() : image2.getPosition().getY();
+        double width2 = rotation == 90 ? image2.getPosition().getHeight() : image2.getPosition().getWidth();
+        double height2 = rotation == 90 ? image2.getPosition().getWidth() : image2.getPosition().getHeight();
+        //if the x-coordinates and widths of images are equal and the height is equal to difference between y-coordinates,
+        // then it is the same picture and has to be merged -> return true
+        return x1 == x2 && width1 == width2 && Math.ceil(height2) == Math.ceil(rotation == 90 ? y2 - y1 : y1 - y2) && width2 > (height2 / 6);
+    }
+
+}
@@ -1,6 +1,19 @@
 package com.iqser.red.service.redaction.v1.server.segmentation;

-import com.iqser.red.service.redaction.v1.model.Rectangle;
+import java.io.File;
+import java.io.FileOutputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.ArrayList;
+import java.util.List;
+
+import org.apache.commons.io.IOUtils;
+import org.apache.pdfbox.io.MemoryUsageSetting;
+import org.apache.pdfbox.pdmodel.PDDocument;
+import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.pdmodel.common.PDRectangle;
+import org.springframework.stereotype.Service;
+
 import com.iqser.red.service.redaction.v1.server.classification.model.Document;
 import com.iqser.red.service.redaction.v1.server.classification.model.Page;
 import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
@@ -15,24 +28,9 @@ import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractT
 import com.iqser.red.service.redaction.v1.server.tableextraction.model.CleanRulings;
 import com.iqser.red.service.redaction.v1.server.tableextraction.service.RulingCleaningService;
 import com.iqser.red.service.redaction.v1.server.tableextraction.service.TableExtractionService;
+
 import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
-import org.apache.commons.io.IOUtils;
-import org.apache.pdfbox.io.MemoryUsageSetting;
-import org.apache.pdfbox.pdmodel.PDDocument;
-import org.apache.pdfbox.pdmodel.PDPage;
-import org.apache.pdfbox.pdmodel.common.PDRectangle;
-import org.springframework.stereotype.Service;
-
-import java.awt.Graphics;
-import java.awt.geom.Rectangle2D;
-import java.awt.image.BufferedImage;
-import java.io.File;
-import java.io.FileOutputStream;
-import java.io.IOException;
-import java.io.InputStream;
-import java.util.ArrayList;
-import java.util.List;

@Slf4j
@Service
@@ -47,13 +45,17 @@ public class PdfSegmentationService {
    private final ClassificationService classificationService;
    private final SectionsBuilderService sectionsBuilderService;
    private final ImageClassificationService imageClassificationService;
+    private final ImageMergeService imageMergeService;


    public Document parseDocument(InputStream documentInputStream) throws IOException {
+
        return parseDocument(documentInputStream, false);
    }

+
    public Document parseDocument(InputStream documentInputStream, boolean ignoreImages) throws IOException {
+
        PDDocument pdDocument = null;
        try {
            //create tempFile
@@ -64,7 +66,6 @@ public class PdfSegmentationService {
            Document document = new Document();
            List<Page> pages = new ArrayList<>();

-
            pdDocument = reinitializePDDocument(tempFile, null);
            long pageCount = pdDocument.getNumberOfPages();

@@ -101,32 +102,19 @@ public class PdfSegmentationService {
                page.setRotation(rotation);
                page.setLandscape(isLandscape || isRotated);
                page.setPageNumber(pageNumber);
-                List<PdfImage> mergedList = processImages(stripper.getImages());
-
-                List<PdfImage> imagesInImage = new ArrayList<>();
-                for(PdfImage image: mergedList){
-                    for (PdfImage inner: mergedList){
-                        if(image != inner && image.getPosition().contains(inner.getPosition().getX(), inner.getPosition().getY(), inner.getPosition().getWidth(), inner.getPosition().getHeight())){
-                            imagesInImage.add(inner);
-                        }
-                    }
-                }
-                mergedList.removeAll(imagesInImage);

+                List<PdfImage> mergedList = imageMergeService.mergeImages(stripper.getImages(), rotation);
                page.setImages(mergedList);

                tableExtractionService.extractTables(cleanRulings, page);
                buildPageStatistics(page);
                increaseDocumentStatistics(page, document);

-
                if (!ignoreImages) {
                    imageClassificationService.classifyImages(page);
                }

                pages.add(page);
-
-
            }

            document.setPages(pages);
@@ -149,7 +137,9 @@ public class PdfSegmentationService {
        }
    }

+
    private PDDocument reinitializePDDocument(File tempFile, PDDocument pdDocument) throws IOException {
+
        if (pdDocument != null) {
            pdDocument.close();
        }
@@ -164,130 +154,6 @@ public class PdfSegmentationService {
        return newPDDocument;
    }

-    //merge images, if they are separated during pdf import, return new list of Pdfimages
-    private List<PdfImage> processImages(List<PdfImage> imageList) {
-        if (imageList.size() > 1) {
-            List<PdfImage> mergedList = new ArrayList<>();
-            int countElementsInList = 0;
-            boolean beginImage = true;
-
-            // a List of Boolean, true = candidate for merging, false = no merging
-            List<Boolean> candidatesList = getCandidatesList(imageList);
-
-            // loop through list, if there are candidates for merging (true), merge images and add it to mergedList
-            for (int i = 0; i < candidatesList.size(); i++) {
-                if (candidatesList.get(i)) {
-                    if (beginImage) {
-                        //begin of image, merge two parts of imageList
-                        PdfImage mergedImage = mergeTwoImages(imageList.get(i), imageList.get(i + 1));
-                        // image merge successful
-                        if (mergedImage != null) {
-                            mergedList.add(mergedImage);
-                            countElementsInList++;
-                        }
-                    } else {
-                        //middle of an image, merge current piece auf mergedList with image of imageList
-                        PdfImage mergedImage = mergeTwoImages(mergedList.get(countElementsInList - 1), imageList.get(i + 1));
-                        // image merge successful
-                        if (mergedImage != null) {
-                            mergedList.set(countElementsInList - 1, mergedImage);
-                        }
-                    }
-                    beginImage = false;
-                } else {
-                    // if the last candidate is false, then both images i and i+1 must be added
-                    if (i == candidatesList.size() - 1) {
-                        if (countElementsInList > 0 && mergedList.get(countElementsInList - 1) == imageList.get(i)) {
-                            mergedList.add(imageList.get(i + 1));
-                        } else {
-                            mergedList.add(imageList.get(i));
-                            mergedList.add(imageList.get(i + 1));
-                        }
-                    } else {
-                        //first image is not splitted, add i to resultlist
-                        if (beginImage) {
-                            mergedList.add(imageList.get(i));
-                            countElementsInList++;
-                        } else {
-                            // i is the end of an image, add begin of new image
-                            mergedList.add(imageList.get(i + 1));
-                            countElementsInList++;
-                            beginImage = false;
-                        }
-                    }
-                }
-            }
-            return mergedList;
-        } else {
-            return imageList;
-        }
-    }
-
-    private PdfImage mergeTwoImages(PdfImage image1, PdfImage image2) {
-
-        // diese Angaben von getPosition scheinen nicht richtig zu sein, damit werden teile des Bildes abgeschnitten
-        double width = image1.getPosition().getWidth();
-        double height1 = image1.getPosition().getHeight();
-        double height2 = image2.getPosition().getHeight();
-        // mit  den Werten, die unter Image gespeichert sind, funktioniert es
-        double img1height = image1.getImage().getHeight();
-        double img1width = image1.getImage().getWidth();
-        double img2height = image2.getImage().getHeight();
-
-        BufferedImage mergedImage = new BufferedImage((int) img1width, (int) (img1height + img2height), BufferedImage.TYPE_INT_RGB);
-        Graphics mergedImageGraphics = mergedImage.getGraphics();
-        try {
-            mergedImageGraphics.drawImage(image1.getImage(), 0, 0, null);
-            mergedImageGraphics.drawImage(image2.getImage(), 0, (int) (img1height), null);
-
-            // set Image, Position and type for merged Image
-            //set position for merged image with values of image1 and the height of both
-            Rectangle2D pos = new Rectangle2D.Float();
-            pos.setRect(image1.getPosition().getX(), image2.getPosition().getY(), width, height1 + height2);
-            PdfImage newPdfImage = new PdfImage(mergedImage, pos, image1.getPage());
-            // Graphics need to be disposed
-
-            image1.getImage().flush();
-            image2.getImage().flush();
-
-            mergedImage.flush();
-            mergedImageGraphics.dispose();
-
-            return newPdfImage;
-        } catch (Exception e) {
-            // failed to merge image
-            log.error("Failed to merge image", e);
-            return null;
-        }
-
-
-    }
-
-    //make a list of true and false, if the image is a candidate for merging
-    private List<Boolean> getCandidatesList(List<PdfImage> imageList) {
-        List<Boolean> candidatesList = new ArrayList<>();
-        for (int i = 0; i < imageList.size(); i++) {
-            if (i >= 1) {
-                candidatesList.add(isCandidateForMerging(imageList.get(i - 1), imageList.get(i)));
-            }
-        }
-        return candidatesList;
-    }
-
-    // evaluate if two images are candidates for merging, depending on their coordinates, width and height
-    private boolean isCandidateForMerging(PdfImage image1, PdfImage image2) {
-        double x1 = image1.getPosition().getX();
-        double y1 = image1.getPosition().getY();
-        double width1 = image1.getPosition().getWidth();
-        double x2 = image2.getPosition().getX();
-        double y2 = image2.getPosition().getY();
-        double width2 = image2.getPosition().getWidth();
-        double height2 = image2.getPosition().getHeight();
-        //if the x-coordinates and widths of images are equal and the height is equal to difference between y-coordinates,
-        // then it is the same picture and has to be merged -> return true
-        return x1 == x2 && width1 == width2 && Math.ceil(height2) == Math.ceil(y1 - y2) && width2 > (height2 / 6);
-    }
-

    private void increaseDocumentStatistics(Page page, Document document) {

@@ -319,5 +185,4 @@ public class PdfSegmentationService {

    }

-
 }
@@ -633,6 +633,7 @@ public class RedactionIntegrationTest {
        long start = System.currentTimeMillis();
        ClassPathResource pdfFileResource = new ClassPathResource("files/new/Single Study - Oral (Gavage) Mouse.pdf");
        AnalyzeRequest request = prepareStorage(pdfFileResource.getInputStream());
+        request.setExcludedPages(Set.of(1));

        AnalyzeResult result = reanalyzeService.analyze(request);
Author	SHA1	Message	Date
Dominique Eiflaender	03b8a95c33	Pull request #174 : RED-1326: Enabled to recategorize images Merge in RED/redaction-service from RED-1326 to master * commit '8ff3e463eae960a35e11fce7589e24131ccb51f2': RED-1326: Enabled to recategorize images	2021-06-18 11:53:21 +02:00
Dominique Eifländer	8ff3e463ea	RED-1326: Enabled to recategorize images	2021-06-18 11:29:43 +02:00
Dominique Eiflaender	e8b7e1fe7c	Pull request #173 : RED-1212: Enabled to exclude pages Merge in RED/redaction-service from RED-1212 to master * commit 'b166ab29ee16ab1bec7a8c310c6bd1cda238ae3f': RED-1212: Enabled to exclude pages	2021-06-17 12:21:01 +02:00
Dominique Eifländer	b166ab29ee	RED-1212: Enabled to exclude pages	2021-06-17 12:16:07 +02:00
Dominique Eiflaender	3a4b8833c8	Pull request #172 : RED-1539: Fixed missing textBefore & textAfter on reanalyis that leads to wrong changelog Merge in RED/redaction-service from RED-1539 to master * commit '143877ff2d6003597cd2a8fc0fe8f835f9d96de5': RED-1539: Fixed missing textBefore & textAfter on reanalyis that leads to wrong changelog	2021-06-16 11:20:40 +02:00
Dominique Eifländer	143877ff2d	RED-1539: Fixed missing textBefore & textAfter on reanalyis that leads to wrong changelog	2021-06-16 10:54:51 +02:00
Dominique Eiflaender	925b2a274c	Pull request #171 : RED-1472: Fixed image merging in rotated pages Merge in RED/redaction-service from RED-1472 to master * commit 'b8dc0e448d9101db8099aaf638ac929a16c87c3a': RED-1472: Fixed image merging in rotated pages	2021-06-15 13:11:25 +02:00
Dominique Eifländer	b8dc0e448d	RED-1472: Fixed image merging in rotated pages	2021-06-15 12:29:46 +02:00
Dominique Eiflaender	00af39bb6d	Pull request #169 : Fixed endless processing Merge in RED/redaction-service from FixEndlessProcessingOnError to master * commit '6bd8c2cf0f61eba94717d5478cdefd87f1cb4dbb': Fixed endless processing	2021-06-02 13:52:05 +02:00
Dominique Eiflaender	1f288d04ac	Pull request #167 : Ignore failing Ruling expansion Merge in RED/redaction-service from IgnoreFailedRulingExpansion to master * commit '418faac923e79c17f2d2b88a02920ab459fc36bf': Ignore failing Ruling expansion	2021-06-02 12:58:28 +02:00