Fix3

Merge in RED/redaction-service from RED-2223 to master

* commit '6005c476962dcdfd1d4e4fabf10fa840377b78db':
  RED-2223: Fixed adding annotations to rotated words on non rotated pages

redaction-service-v1/pom.xml

@@ -32,7 +32,7 @@
             <dependency>
                 <groupId>com.iqser.red</groupId>
                 <artifactId>platform-commons-dependency</artifactId>
-                <version>1.6.0</version>
+                <version>1.7.0</version>
                 <scope>import</scope>
                 <type>pom</type>
             </dependency>

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/controller/RedactionController.java

@@ -1,5 +1,6 @@
 package com.iqser.red.service.redaction.v1.server.controller;
 
+import com.amazonaws.services.kms.model.NotFoundException;
 import com.iqser.red.service.persistence.service.v1.api.model.dossiertemplate.dossier.file.FileType;
 import com.iqser.red.service.redaction.v1.model.*;
 import com.iqser.red.service.redaction.v1.resources.RedactionResource;
@@ -152,6 +153,10 @@ public class RedactionController implements RedactionResource {
 
         var redactionLog = redactionStorageService.getRedactionLog(redactionRequest.getDossierId(), redactionRequest.getFileId());
 
+        if (redactionLog == null) {
+            throw new NotFoundException("RedactionLog not present");
+        }
+
         log.info("Loaded redaction log with computationalVersion: {}", redactionLog.getAnalysisVersion());
         if (redactionLog.getAnalysisVersion() == 0) {
             // old redaction logs are returned directly

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/parsing/model/RedTextPosition.java

@@ -21,6 +21,7 @@ public class RedTextPosition {
     private float YDirAdj;
     private float width;
     private float heightDir;
+    private float widthDirAdj;
 
     // not used in reanalysis
     @JsonIgnore

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/parsing/model/TextPositionSequence.java

@@ -1,5 +1,7 @@
 package com.iqser.red.service.redaction.v1.server.parsing.model;
 
+import static java.util.stream.Collectors.toSet;
+
 import com.fasterxml.jackson.annotation.JsonIgnore;
 import com.fasterxml.jackson.annotation.JsonIgnoreProperties;
 import com.iqser.red.service.redaction.v1.model.Point;
@@ -244,7 +246,15 @@ public class TextPositionSequence implements CharSequence {
             posXInit = textPositions.get(0).getPageWidth() - textPositions.get(0).getYDirAdj() - 2;
             posXEnd = textPositions.get(0).getPageWidth() - textPositions.get(textPositions.size() - 1)
                     .getYDirAdj() + height;
-        } else {
+        } else if(textPositions.get(0).getRotation() == 0 && textPositions.stream().map(t -> t.getY()).collect(toSet()).size() > 1) {
+            posYInit = textPositions.get(0).getPageHeight() - getX1();
+            posYEnd = textPositions.get(0).getPageHeight() - getX2() - textPositions.get(0)
+                    .getWidthDirAdj() - textPositions.get(textPositions.size() - 1).getWidthDirAdj() - 3;
+            posXInit = textPositions.get(0).getPageWidth() - textPositions.get(0).getYDirAdj() - 2;
+            posXEnd = textPositions.get(0).getPageWidth() - textPositions.get(textPositions.size() - 1)
+                    .getYDirAdj() + height;
+        }
+        else {
             posXEnd = textPositions.get(textPositions.size() - 1)
                     .getXDirAdj() + textPositions.get(textPositions.size() - 1).getWidth() + 1;
             posYInit = textPositions.get(0).getPageHeight() - textPositions.get(0).getYDirAdj() - 2;

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/Dictionary.java

@@ -65,13 +65,13 @@ public class Dictionary {
     public boolean containsValue(String type, String value) {
 
         return localAccessMap.containsKey(type) && localAccessMap.get(type)
-                .getEntries().stream().map(DictionaryEntry::getValue).collect(toSet())
+                .getValues(false)
                 .contains(value) || localAccessMap.containsKey(type) && localAccessMap.get(type)
-                .getLocalEntries()
+                .getValues(true)
                 .contains(value) || localAccessMap.containsKey(RECOMMENDATION_PREFIX + type) && localAccessMap.get(RECOMMENDATION_PREFIX + type)
-                .getEntries().stream().map(DictionaryEntry::getValue).collect(toSet())
+                .getValues(false)
                 .contains(value) || localAccessMap.containsKey(RECOMMENDATION_PREFIX + type) && localAccessMap.get(RECOMMENDATION_PREFIX + type)
-                .getLocalEntries()
+                .getValues(true)
                 .contains(value);
     }
 

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/EntityRedactionService.java

@@ -209,18 +209,18 @@ public class EntityRedactionService {
         for (DictionaryModel model : dictionary.getDictionaryModels()) {
             if (model.isCaseInsensitive()) {
                 EntitySearchUtils.addOrAddEngine(found, EntitySearchUtils.find(lowercaseInputString, model.getValues(local), model
-                        .getType(), headline, sectionNumber, !local, model.isDossierDictionary(), Engine.DICTIONARY));
+                        .getType(), headline, sectionNumber, !local, model.isDossierDictionary(), local ? Engine.RULE : Engine.DICTIONARY));
             } else {
                 EntitySearchUtils.addOrAddEngine(found, EntitySearchUtils.find(searchableString, model.getValues(local), model
-                        .getType(), headline, sectionNumber, !local, model.isDossierDictionary(), Engine.DICTIONARY));
+                        .getType(), headline, sectionNumber, !local, model.isDossierDictionary(), local ? Engine.RULE : Engine.DICTIONARY));
             }
         }
 
         if (!local) {
             Map<String, Set<String>> nerValuesPerType = getNerValues(sectionNumber, nerEntities, cellstarts);
             nerValuesPerType.entrySet().forEach(entry -> {
-                EntitySearchUtils.addOrAddEngine(found, EntitySearchUtils.find(searchableString, entry.getValue(), entry
-                        .getKey(), headline, sectionNumber, false, false, Engine.NER));
+                EntitySearchUtils.addEntitiesWithHigherRank(found, EntitySearchUtils.find(searchableString, entry.getValue(), entry
+                        .getKey(), headline, sectionNumber, false, false, Engine.NER), dictionary);
             });
         }
 

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/NerAnalyserService.java

@@ -1,19 +1,17 @@
 package com.iqser.red.service.redaction.v1.server.redaction.service;
 
-import java.util.stream.Collectors;
-
 import com.iqser.red.service.persistence.service.v1.api.model.dossiertemplate.dossier.file.FileType;
-import org.apache.commons.codec.binary.Base64;
-import org.springframework.stereotype.Service;
-
 import com.iqser.red.service.redaction.v1.server.client.EntityRecognitionClient;
 import com.iqser.red.service.redaction.v1.server.client.model.EntityRecognitionRequest;
 import com.iqser.red.service.redaction.v1.server.client.model.EntityRecognitionSection;
 import com.iqser.red.service.redaction.v1.server.settings.RedactionServiceSettings;
 import com.iqser.red.service.redaction.v1.server.storage.RedactionStorageService;
-
 import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
+import org.apache.commons.codec.binary.Base64;
+import org.springframework.stereotype.Service;
+
+import java.util.stream.Collectors;
 
 @Slf4j
 @Service
@@ -31,20 +29,24 @@ public class NerAnalyserService {
 
             long start = System.currentTimeMillis();
 
-            var nerRequest = EntityRecognitionRequest.builder()
-                    .data(text.getSectionTexts()
-                            .stream()
-                            .map(sectionText -> new EntityRecognitionSection(sectionText.getSectionNumber(), new String(Base64
-                                    .encodeBase64(sectionText
-                                    .getText().getBytes()))))
-                            .collect(Collectors.toList()))
-                    .build();
+            if (text != null) {
+                var nerRequest = EntityRecognitionRequest.builder()
+                        .data(text.getSectionTexts()
+                                .stream()
+                                .map(sectionText -> new EntityRecognitionSection(sectionText.getSectionNumber(), new String(Base64
+                                        .encodeBase64(sectionText
+                                                .getText().getBytes()))))
+                                .collect(Collectors.toList()))
+                        .build();
 
-            var nerResponse = entityRecognitionClient.findAuthors(nerRequest);
+                var nerResponse = entityRecognitionClient.findAuthors(nerRequest);
 
-            log.info("Computing NER entities took: {} ms for dossierId {} and fileId {}", System.currentTimeMillis() - start, dossierId, fileId);
+                log.info("Computing NER entities took: {} ms for dossierId {} and fileId {}", System.currentTimeMillis() - start, dossierId, fileId);
 
-            redactionStorageService.storeObject(dossierId, fileId, FileType.NER_ENTITIES, nerResponse);
+                redactionStorageService.storeObject(dossierId, fileId, FileType.NER_ENTITIES, nerResponse);
+            } else {
+                log.warn("Warning, text for file: {} in dossier: {} is null", fileId, dossierId);
+            }
         }
     }
 

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/utils/EntitySearchUtils.java

@@ -146,11 +146,14 @@ public class EntitySearchUtils {
             }
             var existing = existingOptional.get();
 
-            if (dictionary.getDictionaryRank(existing.getType()) <= dictionary.getDictionaryRank(found.getType())) {
+            log.info("existing: {}", existing);
+            log.info("found: {} ", found);
+            if(existing.getType().equals(found.getType())){
+                existing.getEngines().addAll(found.getEngines());
+            } else if (dictionary.getDictionaryRank(existing.getType()) <= dictionary.getDictionaryRank(found.getType())){
+//                found.getEngines().addAll(existing.getEngines());
                 entities.remove(found);
                 entities.add(found);
-            } else {
-                existing.getEngines().addAll(found.getEngines());
             }
         } else {
             entities.add(found);

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/utils/Patterns.java

@@ -11,8 +11,7 @@ public class Patterns {
 
     public static Map<String, Pattern> patternCache = new HashMap<>();
 
-    public static Pattern AUTHOR_TABLE_SPITTER = Pattern.compile("((((di)|(van)) )|[A-Z]’)?[A-ZÄÖÜ][\\wäöüéèê]{2,}( ?[A-ZÄÖÜ]{1,2}\\.)+|((((di)|(van)) )|[A-Z]’)?[A-ZÄÖÜ][\\wäöüéèê]{2,}( ?[A-ZÄÖÜ]{1,2} )+");
-
+    public static Pattern AUTHOR_TABLE_SPITTER = Pattern.compile("(((((di)|(van)) )|[A-Z]’)?[A-ZÄÖÜ][\\wäöüéèê]{2,500}( ?[A-ZÄÖÜ]{1,2}\\.){1,3})|(((((di)|(van)) )|[A-Z]’)?[A-ZÄÖÜ][\\wäöüéèê]{2,500}( ?[A-ZÄÖÜ]{1,2} ){1,3})");
 
     public Pattern getCompiledPattern(String pattern, boolean caseInsensitive) {