Compare commits

...
Author SHA1 Message Date
Dominique Eifländer 6bd8c2cf0f Fixed endless processing 2021-06-02 13:47:48 +02:00
Dominique Eifländer 418faac923 Ignore failing Ruling expansion 2021-06-02 12:53:30 +02:00
Dominique Eiflaender 2ef4e7889e Pull request #165: Fixed text parsing orientation problem
Merge in RED/redaction-service from ParsingOrientationFix to master

* commit 'ab8a710318802d4a836c006638dddb987f5d29c2':
  Fixed text parsing orientation problem
2021-06-02 10:58:00 +02:00
Dominique Eifländer ab8a710318 Fixed text parsing orientation problem 2021-06-02 10:47:40 +02:00
Dominique Eiflaender 1556dbad10 Pull request #164: RED-1535: Improved document parsing, added orientation to textblock
Merge in RED/redaction-service from RED-1535 to master

* commit '53ce6cb47cd5806e1880dcfeaec173e2efed13ff':
  RED-1535: Improved document parsing, added orientation to textblock
2021-06-01 13:47:59 +02:00
Dominique Eifländer 53ce6cb47c RED-1535: Improved document parsing, added orientation to textblock 2021-06-01 13:41:22 +02:00
Timo Bejan 837f97c157 Pull request #163: rename project and ruleSet - finalization with dependencies
Merge in RED/redaction-service from RED-1329 to master

* commit 'f60b0f25e579da2270e6925cca78b51ccdc71b38':
  rename project and ruleSet - finalization with dependencies
2021-05-31 14:21:57 +02:00
Timo f60b0f25e5 rename project and ruleSet - finalization with dependencies 2021-05-31 15:04:15 +03:00
Timo Bejan 3995c9a3e7 Pull request #162: migrated ruleset and projectid to new naming
Merge in RED/redaction-service from RED-1329 to master

* commit 'ceaaed4fc721c9488582d5fce223a9ced928862a':
  migrated ruleset and projectid to new naming
  migrated ruleset and projectid to new naming
2021-05-31 10:44:17 +02:00
Timo ceaaed4fc7 migrated ruleset and projectid to new naming 2021-05-31 11:40:02 +03:00
Timo e01074ef27 migrated ruleset and projectid to new naming 2021-05-31 10:56:48 +03:00
Dominique Eiflaender b7d399c461 Pull request #161: Fixed missing whitespaces
Merge in RED/redaction-service from fixMissingSpaces to master

* commit '35dec94ccdfc569f24d6983c5b2810dfd0f80e7c':
  Fixed missing whitespaces
2021-05-28 15:04:47 +02:00
Dominique Eifländer 35dec94ccd Fixed missing whitespaces 2021-05-28 14:43:30 +02:00
Dominique Eiflaender 42f08194fd Pull request #160: RED-1536: Fixed changelog creation
Merge in RED/redaction-service from RED-1536 to master

* commit 'dc1dba00c11f4778dc5483f47fc147b258b830d2':
  RED-1536: Fixed changelog creation
2021-05-25 13:52:24 +02:00
Dominique Eifländer dc1dba00c1 RED-1536: Fixed changelog creation 2021-05-25 13:45:33 +02:00
Christoph  Schabert a900f5b617 Pull request #159: Swich to maven buld script
Merge in RED/redaction-service from RED-1533 to master

* commit 'd614fa6001dd46b82da63e219fe924d7152768e7':
  Swich to maven buld script
2021-05-25 08:49:15 +02:00
cschabert d614fa6001 Swich to maven buld script 2021-05-21 16:36:17 +02:00
Dominique Eiflaender e9dcd3535f Pull request #158: Ignore images that are contained in others
Merge in RED/redaction-service from ignoreInnerImages to master

* commit 'd87cbdeaeb87f9767fed876076fdec223824f3f6':
  Ignore images that are contained in others
2021-05-20 10:33:45 +02:00
Dominique Eifländer d87cbdeaeb Ignore images that are contained in others 2021-05-20 09:20:22 +02:00
Dominique Eiflaender 8ece616229 Pull request #157: RED-1390: Do not classify images that are nearly as big as the page
Merge in RED/redaction-service from RED-1390 to master

* commit '13974d13734cc9c43e44338bef885a1ace2e5f50':
  RED-1390: Do not classify images that are nearly as big as the page
2021-05-19 12:03:43 +02:00
Dominique Eifländer 13974d1373 RED-1390: Do not classify images that are nearly as big as the page 2021-05-19 11:46:08 +02:00
Dominique Eiflaender 63229316c8 Pull request #156: RED-1506: Fixed unpocess images in strange formats
Merge in RED/redaction-service from RED-1506 to master

* commit '816f44edf0e4a5fe9685c978850b08c09fd083ca':
  RED-1506: Fixed unpocess images in strange formats
2021-05-19 10:21:59 +02:00
Dominique Eifländer 816f44edf0 RED-1506: Fixed unpocess images in strange formats 2021-05-19 10:18:11 +02:00
Timo Bejan be92e3c036 Pull request #155: RED-785
Merge in RED/redaction-service from RED-785 to master

* commit '513501abf476dc7f08446bb3035da888d2fd30a2':
  save legal basis version on redaction log
  adjusted tests
  legal basis is now stored in redaction log
2021-05-14 14:07:56 +02:00
Timo 2334a6cb5b save legal basis version on redaction log 2021-05-14 15:07:07 +03:00
Timo 513501abf4 save legal basis version on redaction log 2021-05-14 15:00:22 +03:00
Timo 6a660c6a73 adjusted tests 2021-05-14 14:50:06 +03:00
Timo 8b898a9038 legal basis is now stored in redaction log 2021-05-14 14:49:21 +03:00
44 changed files with 620 additions and 376 deletions
+1 -1
View File
@@ -5,7 +5,7 @@
<parent>
<groupId>com.atlassian.bamboo</groupId>
<artifactId>bamboo-specs-parent</artifactId>
<version>7.1.2</version>
<version>7.2.2</version>
<relativePath/>
</parent>
@@ -21,6 +21,8 @@ import com.atlassian.bamboo.specs.builders.task.VcsTagTask;
import com.atlassian.bamboo.specs.builders.trigger.BitbucketServerTrigger;
import com.atlassian.bamboo.specs.model.task.InjectVariablesScope;
import com.atlassian.bamboo.specs.util.BambooServer;
import com.atlassian.bamboo.specs.builders.task.ScriptTask;
import com.atlassian.bamboo.specs.model.task.ScriptTaskProperties.Location;
import static com.atlassian.bamboo.specs.builders.task.TestParserTask.createJUnitParserTask;
@@ -84,23 +86,9 @@ public class PlanSpec {
.checkoutItems(new CheckoutItem().defaultRepository()),
new ScriptTask()
.description("Build")
.environmentVariables("MAVEN_OPTS="+JVM_ARGS)
.inlineBody("#!/bin/bash\n" +
"set -e\n" +
"export MAVEN_OPTS=\"$MAVEN_OPTS "+JVM_ARGS +"\"\n" +
"if [[ \"${bamboo.version_tag}\" != \"dev\" ]]; then ${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn --no-transfer-progress -f ${bamboo_build_working_directory}/" + SERVICE_NAME + "-v1/pom.xml versions:set -DnewVersion=${bamboo.version_tag}; fi\n" +
"if [[ \"${bamboo.version_tag}\" != \"dev\" ]]; then ${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn --no-transfer-progress -f ${bamboo_build_working_directory}/" + SERVICE_NAME + "-image-v1/pom.xml versions:set -DnewVersion=${bamboo.version_tag}; fi\n" +
"if [[ \"${bamboo.version_tag}\" = \"dev\" ]]; then ${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn -f ${bamboo_build_working_directory}/" + SERVICE_NAME + "-v1/pom.xml --no-transfer-progress clean install -Djava.security.egd=file:/dev/./urandom; fi\n" +
"if [[ \"${bamboo.version_tag}\" != \"dev\" ]]; then ${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn -f ${bamboo_build_working_directory}/" + SERVICE_NAME + "-v1/pom.xml --no-transfer-progress clean deploy -e -DdeployAtEnd=true -Dmaven.wagon.http.ssl.insecure=true -Dmaven.wagon.http.ssl.allowall=true -Dmaven.wagon.http.ssl.ignore.validity.dates=true -DaltDeploymentRepository=iqser_release::default::https://nexus.iqser.com/repository/red-platform-releases; fi\n" +
"${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn --no-transfer-progress -f ${bamboo_build_working_directory}/" + SERVICE_NAME + "-image-v1/pom.xml package\n" +
"${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn --no-transfer-progress -f ${bamboo_build_working_directory}/" + SERVICE_NAME + "-image-v1/pom.xml docker:push\n" +
"if [[ \"${bamboo.version_tag}\" = \"dev\" ]]; then echo \"gitTag=${bamboo.planRepository.1.branch}_${bamboo.buildNumber}\" > git.tag; fi\n" +
"if [[ \"${bamboo.version_tag}\" != \"dev\" ]]; then echo \"gitTag=${bamboo.version_tag}\" > git.tag; fi\n"),
.location(Location.FILE)
.fileFromPath("bamboo-specs/src/main/resources/scripts/build-java.sh")
.argument(SERVICE_NAME),
createJUnitParserTask()
.description("Resultparser")
.resultDirectories("**/test-reports/*.xml, **/target/surefire-reports/*.xml, **/target/failsafe-reports/*.xml")
+51
View File
@@ -0,0 +1,51 @@
#!/bin/bash
set -e
SERVICE_NAME=$1
if [[ "${bamboo_version_tag}" = "dev" ]]
then
${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn \
-f ${bamboo_build_working_directory}/$SERVICE_NAME-v1/pom.xml \
--no-transfer-progress \
clean install \
-Djava.security.egd=file:/dev/./urandomelse
else
${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn \
--no-transfer-progress \
-f ${bamboo_build_working_directory}/$SERVICE_NAME-v1/pom.xml \
versions:set \
-DnewVersion=${bamboo_version_tag}
${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn \
--no-transfer-progress \
-f ${bamboo_build_working_directory}/$SERVICE_NAME-image-v1/pom.xml \
versions:set \
-DnewVersion=${bamboo_version_tag}
${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn \
-f ${bamboo_build_working_directory}/$SERVICE_NAME-v1/pom.xml \
--no-transfer-progress \
clean deploy \
-e \
-DdeployAtEnd=true \
-Dmaven.wagon.http.ssl.insecure=true \
-Dmaven.wagon.http.ssl.allowall=true \
-Dmaven.wagon.http.ssl.ignore.validity.dates=true \
-DaltDeploymentRepository=iqser_release::default::https://nexus.iqser.com/repository/red-platform-releases
fi
${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn \
--no-transfer-progress \
-f ${bamboo_build_working_directory}/$SERVICE_NAME-image-v1/pom.xml \
package
${bamboo_capability_system_builder_mvn3_Maven_3}/bin/mvn \
--no-transfer-progress \
-f ${bamboo_build_working_directory}/$SERVICE_NAME-image-v1/pom.xml \
docker:push
if [[ "${bamboo_version_tag}" = "dev" ]]
then
echo "gitTag=${bamboo_planRepository_1_branch}_${bamboo_buildNumber}" > git.tag
else
echo "gitTag=${bamboo_version_tag}" > git.tag
fi
@@ -17,5 +17,16 @@
<artifactId>spring-web</artifactId>
<optional>true</optional>
</dependency>
<dependency>
<groupId>com.iqser.red.service</groupId>
<artifactId>configuration-service-api-v1</artifactId>
<version>2.11.0</version>
<exclusions>
<exclusion>
<groupId>com.iqser.red.service</groupId>
<artifactId>file-management-service-api-v1</artifactId>
</exclusion>
</exclusions>
</dependency>
</dependencies>
</project>
@@ -13,9 +13,9 @@ import java.time.OffsetDateTime;
@AllArgsConstructor
public class AnalyzeRequest {
private String projectId;
private String dossierId;
private String fileId;
private String ruleSetId;
private String dossierTemplateId;
private boolean reanalyseOnlyIfPossible;
private ManualRedactions manualRedactions;
private OffsetDateTime lastProcessed;
@@ -11,7 +11,7 @@ import lombok.NoArgsConstructor;
@AllArgsConstructor
public class AnalyzeResult {
private String projectId;
private String dossierId;
private String fileId;
private long duration;
private int numberOfPages;
@@ -23,6 +23,7 @@ public class AnalyzeResult {
private long dictionaryVersion;
private long dossierDictionaryVersion;
private long rulesVersion;
private long legalBasisVersion;
}
@@ -11,6 +11,7 @@ import lombok.NoArgsConstructor;
@AllArgsConstructor
public class AnnotateRequest {
private String projectId;
private String dossierId;
private String dossierTemplateId;
private String fileId;
}
@@ -15,8 +15,8 @@ public class RedactionChangeLog {
private List<RedactionChangeLogEntry> redactionLogEntry = new ArrayList<>();
private long dictionaryVersion = -1;
private long dossierDictionaryVersion = -1;
private long rulesVersion = -1;
private String ruleSetId;
private long legalBasisVersion = -1;
}
@@ -1,32 +1,21 @@
package com.iqser.red.service.redaction.v1.model;
import com.iqser.red.service.configuration.v1.api.model.LegalBasisMapping;
import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;
import java.util.List;
@Data
@NoArgsConstructor
@AllArgsConstructor
public class RedactionLog {
private List<RedactionLogEntry> redactionLogEntry;
private List<LegalBasisMapping> legalBasis;
private long dictionaryVersion = -1;
private long rulesVersion = -1;
private String ruleSetId;
private long dossierDictionaryVersion = -1;
public RedactionLog(List<RedactionLogEntry> redactionLogEntry, long dictionaryVersion, long rulesVersion, String ruleSetId, long dossierDictionaryVersion) {
this.redactionLogEntry = redactionLogEntry;
this.dictionaryVersion = dictionaryVersion;
this.rulesVersion = rulesVersion;
this.ruleSetId = ruleSetId;
this.dossierDictionaryVersion = dossierDictionaryVersion;
}
private long rulesVersion = -1;
private long legalBasisVersion = -1;
}
@@ -3,6 +3,7 @@ package com.iqser.red.service.redaction.v1.model;
import lombok.AllArgsConstructor;
import lombok.Builder;
import lombok.Data;
import lombok.EqualsAndHashCode;
import lombok.NoArgsConstructor;
import java.util.ArrayList;
@@ -12,6 +13,7 @@ import java.util.List;
@Builder
@NoArgsConstructor
@AllArgsConstructor
@EqualsAndHashCode
public class RedactionLogEntry {
private String id;
@@ -11,8 +11,8 @@ import lombok.NoArgsConstructor;
@AllArgsConstructor
public class RedactionRequest {
private String projectId;
private String dossierId;
private String fileId;
private String ruleSetId;
private String dossierTemplateId;
private ManualRedactions manualRedactions;
}
@@ -10,7 +10,7 @@ public interface RedactionResource {
String SERVICE_NAME = "redaction-service-v1";
String RULE_SET_PARAMETER_NAME = "ruleSetId";
String RULE_SET_PARAMETER_NAME = "dossierTemplateId";
String RULE_SET_PATH_VARIABLE = "/{" + RULE_SET_PARAMETER_NAME + "}";
@@ -27,7 +27,7 @@ public interface RedactionResource {
RedactionResult htmlTables(@RequestBody RedactionRequest redactionRequest);
@PostMapping(value = "/rules/update" + RULE_SET_PATH_VARIABLE, consumes = MediaType.APPLICATION_JSON_VALUE)
void updateRules(@PathVariable(RULE_SET_PARAMETER_NAME) String ruleSetId);
void updateRules(@PathVariable(RULE_SET_PARAMETER_NAME) String dossierTemplateId);
@PostMapping(value = "/rules/test", consumes = MediaType.APPLICATION_JSON_VALUE)
void testRules(@RequestBody String rules);
@@ -21,20 +21,19 @@
<artifactId>redaction-service-api-v1</artifactId>
<version>${project.version}</version>
</dependency>
<dependency>
<groupId>com.iqser.red.service</groupId>
<artifactId>configuration-service-api-v1</artifactId>
<version>2.5.6</version>
</dependency>
<dependency>
<groupId>com.iqser.red.service</groupId>
<artifactId>file-management-service-api-v1</artifactId>
<version>2.7.4</version>
<version>2.25.0</version>
<exclusions>
<exclusion>
<groupId>com.iqser.red.service</groupId>
<artifactId>redaction-service-api-v1</artifactId>
</exclusion>
<exclusion>
<groupId>com.iqser.red.service</groupId>
<artifactId>configuration-service-api-v1</artifactId>
</exclusion>
</exclusions>
</dependency>
<dependency>
@@ -57,6 +56,23 @@
<artifactId>guava</artifactId>
<version>29.0-jre</version>
</dependency>
<dependency>
<groupId>org.apache.pdfbox</groupId>
<artifactId>jbig2-imageio</artifactId>
<version>3.0.3</version>
</dependency>
<dependency>
<groupId>com.github.jai-imageio</groupId>
<artifactId>jai-imageio-core</artifactId>
<version>1.4.0</version>
</dependency>
<dependency>
<groupId>com.github.jai-imageio</groupId>
<artifactId>jai-imageio-jpeg2000</artifactId>
<version>1.4.0</version>
</dependency>
<!-- commons -->
<dependency>
<groupId>com.iqser.red.commons</groupId>
@@ -0,0 +1,6 @@
package com.iqser.red.service.redaction.v1.server.classification.model;
public enum Orientation {
NONE, LEFT, RIGHT
}
@@ -30,6 +30,8 @@ public class Page {
private StringFrequencyCounter fontCounter = new StringFrequencyCounter();
private StringFrequencyCounter fontStyleCounter = new StringFrequencyCounter();
private double cropBoxArea;
public boolean isRotated() {
@@ -32,6 +32,7 @@ public class TextBlock extends AbstractTextContainer {
private String classification;
public TextBlock(float minX, float maxX, float minY, float maxY, List<TextPositionSequence> sequences, int rotation) {
this.minX = minX;
this.maxX = maxX;
@@ -1,6 +1,7 @@
package com.iqser.red.service.redaction.v1.server.classification.service;
import com.iqser.red.service.redaction.v1.server.classification.model.FloatFrequencyCounter;
import com.iqser.red.service.redaction.v1.server.classification.model.Orientation;
import com.iqser.red.service.redaction.v1.server.classification.model.Page;
import com.iqser.red.service.redaction.v1.server.classification.model.StringFrequencyCounter;
import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
@@ -11,16 +12,21 @@ import com.iqser.red.service.redaction.v1.server.tableextraction.model.Cell;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Rectangle;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Ruling;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table;
import org.springframework.stereotype.Service;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.List;
@Service
@SuppressWarnings("all")
public class BlockificationService {
public Page blockify(List<TextPositionSequence> textPositions, List<Ruling> horizontalRulingLines, List<Ruling> verticalRulingLines) {
static final float THRESHOLD = 1f;
public Page blockify(List<TextPositionSequence> textPositions, List<Ruling> horizontalRulingLines,
List<Ruling> verticalRulingLines) {
List<TextPositionSequence> chunkWords = new ArrayList<>();
List<AbstractTextContainer> chunkBlockList1 = new ArrayList<>();
@@ -28,21 +34,46 @@ public class BlockificationService {
float minX = 1000, maxX = 0, minY = 1000, maxY = 0;
TextPositionSequence prev = null;
boolean wasSplitted = false;
Float splitX1 = null;
for (TextPositionSequence word : textPositions) {
boolean lineSeparation = minY - word.getY2() > word.getHeight() * 1.25;
boolean startFromTop = word.getY1() > maxY + word.getHeight();
boolean splitByX = prev != null && maxX + 50 < word.getX1() && prev.getY1() == word.getY1();
boolean newLineAfterSplit = prev != null && word.getY1() != prev.getY1() && wasSplitted && splitX1 != word.getX1();
boolean splittedByRuling = word.getRotation() == 0 && isSplittedByRuling(maxX, minY, word.getX1(), word.getY1(), verticalRulingLines) || word
.getRotation() == 0 && isSplittedByRuling(minX, minY, word.getX1(), word.getY2(), horizontalRulingLines) || word
.getRotation() == 90 && isSplittedByRuling(maxX, minY, word.getX1(), word.getY1(), horizontalRulingLines) || word
.getRotation() == 90 && isSplittedByRuling(minX, minY, word.getX1(), word.getY2(), verticalRulingLines);
if (prev != null && (lineSeparation || startFromTop || word.getRotation() == 0 && isSplittedByRuling(maxX, minY, word
.getX1(), word.getY1(), verticalRulingLines) || word.getRotation() == 0 && isSplittedByRuling(minX, minY, word
.getX1(), word.getY2(), horizontalRulingLines) || word.getRotation() == 90 && isSplittedByRuling(maxX, minY, word
.getX1(), word.getY1(), horizontalRulingLines) || word.getRotation() == 90 && isSplittedByRuling(minX, minY, word
.getX1(), word.getY2(), verticalRulingLines))) {
if (prev != null && (lineSeparation || startFromTop || splitByX || newLineAfterSplit || splittedByRuling)) {
Orientation prevOrientation = null;
if(!chunkBlockList1.isEmpty()) {
prevOrientation = chunkBlockList1.get(chunkBlockList1.size() - 1).getOrientation();
}
TextBlock cb1 = buildTextBlock(chunkWords);
chunkBlockList1.add(cb1);
chunkWords = new ArrayList<>();
if (splitByX && !splittedByRuling) {
wasSplitted = true;
cb1.setOrientation(Orientation.LEFT);
splitX1 = word.getX1();
} else
if (newLineAfterSplit && !splittedByRuling) {
wasSplitted = false;
cb1.setOrientation(Orientation.RIGHT);
splitX1 = null;
} else
if(prevOrientation != null && prevOrientation.equals(Orientation.RIGHT) && (lineSeparation || !startFromTop || !splitByX || !newLineAfterSplit || !splittedByRuling)){
cb1.setOrientation(Orientation.LEFT);
}
minX = 1000;
maxX = 0;
minY = 1000;
@@ -72,9 +103,62 @@ public class BlockificationService {
chunkBlockList1.add(cb1);
}
Iterator<AbstractTextContainer> itty = chunkBlockList1.iterator();
TextBlock previousLeft = null;
TextBlock previousRight = null;
while (itty.hasNext()) {
TextBlock block = (TextBlock) itty.next();
if(previousLeft != null && block.getOrientation().equals(Orientation.LEFT)){
if (previousLeft.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousLeft.getMinY()){
previousLeft.add(block);
itty.remove();
continue;
}
}
if(previousRight != null && block.getOrientation().equals(Orientation.RIGHT)){
if (previousRight.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousRight.getMinY()){
previousRight.add(block);
itty.remove();
continue;
}
}
if (block.getOrientation().equals(Orientation.LEFT)) {
previousLeft = block;
} else if (block.getOrientation().equals(Orientation.RIGHT)) {
previousRight = block;
}
}
itty = chunkBlockList1.iterator();
TextBlock previous = null;
while (itty.hasNext()) {
TextBlock block = (TextBlock) itty.next();
if(previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation().equals(Orientation.LEFT) && equalsWithThreshold(block.getMaxY(), previous
.getMaxY())||
previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation().equals(Orientation.RIGHT) && equalsWithThreshold(block.getMaxY(), previous
.getMaxY())){
previous.add(block);
itty.remove();
continue;
}
previous = block;
}
return new Page(chunkBlockList1);
}
private boolean equalsWithThreshold(float f1, float f2){
return Math.abs(f1 - f2) < THRESHOLD;
}
private TextBlock buildTextBlock(List<TextPositionSequence> wordBlockList) {
@@ -117,7 +201,8 @@ public class BlockificationService {
}
private boolean isSplittedByRuling(float previousX2, float previousY1, float currentX1, float currentY1, List<Ruling> rulingLines) {
private boolean isSplittedByRuling(float previousX2, float previousY1, float currentX1, float currentY1,
List<Ruling> rulingLines) {
for (Ruling ruling : rulingLines) {
if (ruling.intersectsLine(previousX2, previousY1, currentX1, currentY1)) {
@@ -128,7 +213,8 @@ public class BlockificationService {
}
public Rectangle calculateBodyTextFrame(List<Page> pages, FloatFrequencyCounter documentFontSizeCounter, boolean landscape) {
public Rectangle calculateBodyTextFrame(List<Page> pages, FloatFrequencyCounter documentFontSizeCounter,
boolean landscape) {
float minX = 10000;
float maxX = -100;
@@ -0,0 +1,8 @@
package com.iqser.red.service.redaction.v1.server.client;
import com.iqser.red.service.configuration.v1.api.resource.LegalBasisMappingResource;
import org.springframework.cloud.openfeign.FeignClient;
@FeignClient(name = "LegalBasisMappingResource", url = "${configuration-service.url}")
public interface LegalBasisClient extends LegalBasisMappingResource {
}
@@ -43,14 +43,14 @@ public class RedactionController implements RedactionResource {
public AnnotateResponse annotate(@RequestBody AnnotateRequest annotateRequest) {
var storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(annotateRequest.getProjectId(), annotateRequest.getFileId(), FileType.ORIGIN));
var redactionLog = redactionStorageService.getRedactionLog(annotateRequest.getProjectId(), annotateRequest.getFileId());
var sectionsGrid = redactionStorageService.getSectionGrid(annotateRequest.getProjectId(), annotateRequest.getFileId());
var storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(annotateRequest.getDossierId(), annotateRequest.getFileId(), FileType.ORIGIN));
var redactionLog = redactionStorageService.getRedactionLog(annotateRequest.getDossierId(), annotateRequest.getFileId());
var sectionsGrid = redactionStorageService.getSectionGrid(annotateRequest.getDossierId(), annotateRequest.getFileId());
try (PDDocument pdDocument = PDDocument.load(storedObjectStream, MemoryUsageSetting.setupTempFileOnly())) {
pdDocument.setAllSecurityToBeRemoved(true);
dictionaryService.updateDictionary(redactionLog.getRuleSetId(), annotateRequest.getProjectId());
dictionaryService.updateDictionary(annotateRequest.getDossierTemplateId(), annotateRequest.getDossierId());
annotationService.annotate(pdDocument, redactionLog, sectionsGrid);
try (ByteArrayOutputStream byteArrayOutputStream = new ByteArrayOutputStream()) {
@@ -66,11 +66,11 @@ public class RedactionController implements RedactionResource {
@Override
public RedactionResult classify(@RequestBody RedactionRequest redactionRequest) {
var storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getProjectId(), redactionRequest.getFileId(), FileType.ORIGIN));
var storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getDossierId(), redactionRequest.getFileId(), FileType.ORIGIN));
try {
Document classifiedDoc = pdfSegmentationService.parseDocument(storedObjectStream);
storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getProjectId(), redactionRequest.getFileId(), FileType.ORIGIN));
storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getDossierId(), redactionRequest.getFileId(), FileType.ORIGIN));
try (PDDocument pdDocument = PDDocument.load(storedObjectStream)) {
pdDocument.setAllSecurityToBeRemoved(true);
@@ -91,11 +91,11 @@ public class RedactionController implements RedactionResource {
@Override
public RedactionResult sections(@RequestBody RedactionRequest redactionRequest) {
var storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getProjectId(), redactionRequest.getFileId(), FileType.ORIGIN));
var storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getDossierId(), redactionRequest.getFileId(), FileType.ORIGIN));
try {
Document classifiedDoc = pdfSegmentationService.parseDocument(storedObjectStream);
storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getProjectId(), redactionRequest.getFileId(), FileType.ORIGIN));
storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getDossierId(), redactionRequest.getFileId(), FileType.ORIGIN));
try (PDDocument pdDocument = PDDocument.load(storedObjectStream)) {
pdDocument.setAllSecurityToBeRemoved(true);
@@ -120,7 +120,7 @@ public class RedactionController implements RedactionResource {
Document classifiedDoc;
try {
var storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getProjectId(), redactionRequest.getFileId(), FileType.ORIGIN));
var storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(redactionRequest.getDossierId(), redactionRequest.getFileId(), FileType.ORIGIN));
classifiedDoc = pdfSegmentationService.parseDocument(storedObjectStream, true);
} catch (Exception e) {
throw new RedactionException(e);
@@ -143,9 +143,9 @@ public class RedactionController implements RedactionResource {
@Override
public void updateRules(@PathVariable(RULE_SET_PARAMETER_NAME) String ruleSetId) {
public void updateRules(@PathVariable(RULE_SET_PARAMETER_NAME) String dossierTemplateId) {
droolsExecutionService.updateRules(ruleSetId);
droolsExecutionService.updateRules(dossierTemplateId);
}
@@ -46,6 +46,17 @@ public class PDFAreaTextStripper extends PDFTextStripperByArea {
startIndex = i;
}
if (textPositions.get(i).getRotation() == 0 && i > 0 && textPositions.get(i).getX() > textPositions.get(i - 1).getEndX() + 1) {
List<TextPosition> sublist = textPositions.subList(startIndex, i);
if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0)
.getUnicode()
.equals(" ") || sublist.get(0).getUnicode().equals("\u00A0")))) {
textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
}
startIndex = i;
}
if (i > 0 && (textPositions.get(i).getUnicode().equals(" ") || textPositions.get(i)
.getUnicode()
.equals("\u00A0")) && i <= textPositions.size() - 2) {
@@ -300,6 +300,18 @@ public class PDFLinesTextStripper extends PDFTextStripper {
startIndex = i;
}
if (textPositions.get(i).getRotation() == 0 && i > 0 && textPositions.get(i).getX() > textPositions.get(i - 1).getEndX() + 1) {
List<TextPosition> sublist = textPositions.subList(startIndex, i);
if (!(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0)
.getUnicode()
.equals(" ") || sublist.get(0).getUnicode().equals("\u00A0")))) {
textPositionSequences.add(new TextPositionSequence(sublist, pageNumber));
}
startIndex = i;
}
if (i > 0 && (textPositions.get(i).getUnicode().equals(" ") || textPositions.get(i)
.getUnicode()
.equals("\u00A0")) && i <= textPositions.size() - 2) {
@@ -21,7 +21,7 @@ public class MessagingConfiguration {
return QueueBuilder.durable(REDACTION_QUEUE)
.withArgument("x-dead-letter-exchange", "")
.withArgument("x-dead-letter-routing-key", REDACTION_QUEUE)
.withArgument("x-dead-letter-routing-key", REDACTION_DQL)
.maxPriority(2)
.build();
}
@@ -38,7 +38,7 @@ public class RedactionMessageReceiver {
}
log.info("Successfully analyzed {}", analyzeRequest);
fileStatusProcessingUpdateClient.analysisSuccessful(analyzeRequest.getProjectId(), analyzeRequest.getFileId(), result);
fileStatusProcessingUpdateClient.analysisSuccessful(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), result);
}
@RabbitHandler
@@ -48,7 +48,7 @@ public class RedactionMessageReceiver {
var analyzeRequest = objectMapper.readValue(in, AnalyzeRequest.class);
log.info("Failed to process analyze request: {}", analyzeRequest);
fileStatusProcessingUpdateClient.analysisFailed(analyzeRequest.getProjectId(), analyzeRequest.getFileId());
fileStatusProcessingUpdateClient.analysisFailed(analyzeRequest.getDossierId(), analyzeRequest.getFileId());
}
}
@@ -10,7 +10,7 @@ import java.util.Map;
@Data
public class DictionaryRepresentation {
private String ruleSetId;
private String dossierTemplateId;
private long dictionaryVersion = -1;
private List<DictionaryModel> dictionary = new ArrayList<>();
private float[] defaultColor;
@@ -11,6 +11,6 @@ import lombok.NoArgsConstructor;
@AllArgsConstructor
public class DictionaryVersion {
long rulesetVersion;
long dossierTemplateVersion;
long dossierVersion;
}
@@ -11,6 +11,8 @@ import lombok.NoArgsConstructor;
@AllArgsConstructor
public class RedRectangle2D {
public static final double THRESHOLD = 0.01;
private double x;
private double y;
private double width;
@@ -27,9 +29,14 @@ public class RedRectangle2D {
}
double x0 = getX();
double y0 = getY();
return x >= x0 &&
y >= y0 &&
(x + w) <= x0 + getWidth() &&
(y + h) <= y0 + getHeight();
return round(x) >= round(x0) &&
round(y) >= round(y0) &&
(x + w) - (x0 + getWidth()) <= THRESHOLD &&
(y + h) - (y0 + getHeight()) <= THRESHOLD;
}
private double round(double value) {
double d = Math.pow(10, 2);
return Math.round(value * d) / d;
}
}
@@ -9,7 +9,7 @@ import org.springframework.stereotype.Service;
@Service
public class AnalyzeResponseService {
public AnalyzeResult createAnalyzeResponse(String projectId, String fileId, long duration, int pageCount, RedactionLog redactionLog, RedactionChangeLog redactionChangeLog) {
public AnalyzeResult createAnalyzeResponse(String dossierId, String fileId, long duration, int pageCount, RedactionLog redactionLog, RedactionChangeLog redactionChangeLog) {
boolean hasHints = redactionLog.getRedactionLogEntry().stream().anyMatch(RedactionLogEntry::isHint);
boolean hasRequests = redactionLog.getRedactionLogEntry()
@@ -31,7 +31,7 @@ public class AnalyzeResponseService {
.isEmpty() && redactionChangeLog.getRedactionLogEntry().stream().anyMatch(entry -> !entry.getType().equals("false_positive"));
return AnalyzeResult.builder()
.projectId(projectId)
.dossierId(dossierId)
.fileId(fileId)
.duration(duration)
.numberOfPages(pageCount)
@@ -42,6 +42,7 @@ public class AnalyzeResponseService {
.hasUpdates(hasUpdates)
.rulesVersion(redactionLog.getRulesVersion())
.dictionaryVersion(redactionLog.getDictionaryVersion())
.legalBasisVersion(redactionLog.getLegalBasisVersion())
.dossierDictionaryVersion(redactionLog.getDossierDictionaryVersion())
.build();
}
@@ -1,19 +1,12 @@
package com.iqser.red.service.redaction.v1.server.redaction.service;
import static com.iqser.red.service.configuration.v1.api.resource.DictionaryResource.GLOBAL_DOSSIER;
import com.iqser.red.service.configuration.v1.api.model.Colors;
import com.iqser.red.service.configuration.v1.api.model.DictionaryEntry;
import com.iqser.red.service.configuration.v1.api.model.TypeResponse;
import com.iqser.red.service.configuration.v1.api.model.TypeResult;
import com.iqser.red.service.redaction.v1.server.client.DictionaryClient;
import com.iqser.red.service.redaction.v1.server.redaction.model.Dictionary;
import com.iqser.red.service.redaction.v1.server.redaction.model.DictionaryIncrement;
import com.iqser.red.service.redaction.v1.server.redaction.model.DictionaryIncrementValue;
import com.iqser.red.service.redaction.v1.server.redaction.model.DictionaryModel;
import com.iqser.red.service.redaction.v1.server.redaction.model.DictionaryRepresentation;
import com.iqser.red.service.redaction.v1.server.redaction.model.DictionaryVersion;
import com.iqser.red.service.redaction.v1.server.redaction.model.*;
import feign.FeignException;
import lombok.RequiredArgsConstructor;
import lombok.extern.slf4j.Slf4j;
@@ -25,6 +18,8 @@ import java.awt.Color;
import java.util.*;
import java.util.stream.Collectors;
import static com.iqser.red.service.configuration.v1.api.resource.DictionaryResource.GLOBAL_DOSSIER;
@Slf4j
@Service
@RequiredArgsConstructor
@@ -32,37 +27,37 @@ public class DictionaryService {
private final DictionaryClient dictionaryClient;
private final Map<String, DictionaryRepresentation> dictionariesByRuleSets = new HashMap<>();
private final Map<String, DictionaryRepresentation> dictionariesByDossierTemplate = new HashMap<>();
private final Map<String, DictionaryRepresentation> dictionariesByDossier = new HashMap<>();
public DictionaryVersion updateDictionary(String ruleSetId, String dossierId) {
public DictionaryVersion updateDictionary(String dossierTemplateId, String dossierId) {
long rulesetDictionaryVersion = dictionaryClient.getVersion(ruleSetId, GLOBAL_DOSSIER);
var rulesetDictionary = dictionariesByRuleSets.get(ruleSetId);
if (rulesetDictionary == null || rulesetDictionaryVersion > rulesetDictionary.getDictionaryVersion()) {
updateDictionaryEntry(ruleSetId, rulesetDictionaryVersion, GLOBAL_DOSSIER);
long dossierTemplateDictionaryVersion = dictionaryClient.getVersion(dossierTemplateId, GLOBAL_DOSSIER);
var dossierTemplateDictionary = dictionariesByDossierTemplate.get(dossierTemplateId);
if (dossierTemplateDictionary == null || dossierTemplateDictionaryVersion > dossierTemplateDictionary.getDictionaryVersion()) {
updateDictionaryEntry(dossierTemplateId, dossierTemplateDictionaryVersion, GLOBAL_DOSSIER);
}
long dossierDictionaryVersion = dictionaryClient.getVersion(ruleSetId, dossierId);
long dossierDictionaryVersion = dictionaryClient.getVersion(dossierTemplateId, dossierId);
var dossierDictionary = dictionariesByDossier.get(dossierId);
if (dossierDictionary == null || dossierDictionaryVersion > dossierDictionary.getDictionaryVersion()) {
updateDictionaryEntry(ruleSetId, dossierDictionaryVersion, dossierId);
updateDictionaryEntry(dossierTemplateId, dossierDictionaryVersion, dossierId);
}
return DictionaryVersion.builder().rulesetVersion(rulesetDictionaryVersion).dossierVersion(dossierDictionaryVersion).build();
return DictionaryVersion.builder().dossierTemplateVersion(dossierTemplateDictionaryVersion).dossierVersion(dossierDictionaryVersion).build();
}
public DictionaryIncrement getDictionaryIncrements(String ruleSetId, DictionaryVersion fromVersion, String dossierId) {
public DictionaryIncrement getDictionaryIncrements(String dossierTemplateId, DictionaryVersion fromVersion, String dossierId) {
DictionaryVersion version = updateDictionary(ruleSetId, dossierId);
DictionaryVersion version = updateDictionary(dossierTemplateId, dossierId);
Set<DictionaryIncrementValue> newValues = new HashSet<>();
List<DictionaryModel> dictionaryModels = dictionariesByRuleSets.get(ruleSetId).getDictionary();
List<DictionaryModel> dictionaryModels = dictionariesByDossierTemplate.get(dossierTemplateId).getDictionary();
dictionaryModels.forEach(dictionaryModel -> {
dictionaryModel.getEntries().forEach(dictionaryEntry -> {
if (dictionaryEntry.getVersion() > fromVersion.getRulesetVersion()) {
if (dictionaryEntry.getVersion() > fromVersion.getDossierTemplateVersion()) {
newValues.add(new DictionaryIncrementValue(dictionaryEntry.getValue(), dictionaryModel.isCaseInsensitive()));
}
});
@@ -83,35 +78,35 @@ public class DictionaryService {
}
private void updateDictionaryEntry(String ruleSetId, long version, String dossierId) {
private void updateDictionaryEntry(String dossierTemplateId, long version, String dossierId) {
try {
DictionaryRepresentation dictionaryRepresentation = new DictionaryRepresentation();
TypeResponse typeResponse = dictionaryClient.getAllTypes(ruleSetId, dossierId);
TypeResponse typeResponse = dictionaryClient.getAllTypes(dossierTemplateId, dossierId);
if (typeResponse != null && CollectionUtils.isNotEmpty(typeResponse.getTypes())) {
List<DictionaryModel> dictionary = typeResponse.getTypes()
.stream()
.map(t -> new DictionaryModel(t.getType(), t.getRank(), convertColor(t.getHexColor()), t.isCaseInsensitive(), t
.isHint(), t.isRecommendation(), convertEntries(t, dossierId), new HashSet<>(),dossierId.equals(GLOBAL_DOSSIER) ? false : true))
.isHint(), t.isRecommendation(), convertEntries(t, dossierId), new HashSet<>(), !dossierId.equals(GLOBAL_DOSSIER)))
.sorted(Comparator.comparingInt(DictionaryModel::getRank).reversed())
.collect(Collectors.toList());
dictionary.forEach(dm -> dictionaryRepresentation.getLocalAccessMap().put(dm.getType(), dm));
Colors colors = dictionaryClient.getColors(ruleSetId);
Colors colors = dictionaryClient.getColors(dossierTemplateId);
dictionaryRepresentation.setDefaultColor(convertColor(colors.getDefaultColor()));
dictionaryRepresentation.setRequestAddColor(convertColor(colors.getRequestAdd()));
dictionaryRepresentation.setRequestRemoveColor(convertColor(colors.getRequestRemove()));
dictionaryRepresentation.setNotRedactedColor(convertColor(colors.getNotRedacted()));
dictionaryRepresentation.setRuleSetId(ruleSetId);
dictionaryRepresentation.setDossierTemplateId(dossierTemplateId);
dictionaryRepresentation.setDictionaryVersion(version);
dictionaryRepresentation.setDictionary(dictionary);
if(dossierId.equals(GLOBAL_DOSSIER)) {
dictionariesByRuleSets.put(ruleSetId, dictionaryRepresentation);
dictionariesByDossierTemplate.put(dossierTemplateId, dictionaryRepresentation);
} else {
dictionariesByDossier.put(dossierId, dictionaryRepresentation);
}
@@ -123,14 +118,14 @@ public class DictionaryService {
}
public void updateExternalDictionary(Dictionary dictionary, String ruleSetId) {
public void updateExternalDictionary(Dictionary dictionary, String dossierTemplateId) {
dictionary.getDictionaryModels().forEach(dm -> {
if (dm.isRecommendation() && !dm.getLocalEntries().isEmpty()) {
dictionaryClient.addEntries(dm.getType(), ruleSetId, new ArrayList<>(dm.getLocalEntries()), false, GLOBAL_DOSSIER);
long externalVersion = dictionaryClient.getVersion(ruleSetId, GLOBAL_DOSSIER);
if (externalVersion == dictionary.getVersion().getRulesetVersion() + 1) {
dictionary.getVersion().setRulesetVersion(externalVersion);
dictionaryClient.addEntries(dm.getType(), dossierTemplateId, new ArrayList<>(dm.getLocalEntries()), false, GLOBAL_DOSSIER);
long externalVersion = dictionaryClient.getVersion(dossierTemplateId, GLOBAL_DOSSIER);
if (externalVersion == dictionary.getVersion().getDossierTemplateVersion() + 1) {
dictionary.getVersion().setDossierTemplateVersion(externalVersion);
}
}
});
@@ -139,7 +134,7 @@ public class DictionaryService {
private Set<DictionaryEntry> convertEntries(TypeResult t, String dossierId) {
Set<DictionaryEntry> entries = new HashSet<>(dictionaryClient.getDictionaryForType(t.getType(), t.getRuleSetId(), dossierId)
Set<DictionaryEntry> entries = new HashSet<>(dictionaryClient.getDictionaryForType(t.getType(), t.getDossierTemplateId(), dossierId)
.getEntries());
if (t.isCaseInsensitive()) {
@@ -156,9 +151,9 @@ public class DictionaryService {
}
public boolean isCaseInsensitiveDictionary(String type, String ruleSetId) {
public boolean isCaseInsensitiveDictionary(String type, String dossierTemplateId) {
DictionaryModel dictionaryModel = dictionariesByRuleSets.get(ruleSetId).getLocalAccessMap().get(type);
DictionaryModel dictionaryModel = dictionariesByDossierTemplate.get(dossierTemplateId).getLocalAccessMap().get(type);
if (dictionaryModel != null) {
return dictionaryModel.isCaseInsensitive();
}
@@ -166,19 +161,19 @@ public class DictionaryService {
}
public float[] getColor(String type, String ruleSetId) {
public float[] getColor(String type, String dossierTemplateId) {
DictionaryModel model = dictionariesByRuleSets.get(ruleSetId).getLocalAccessMap().get(type);
DictionaryModel model = dictionariesByDossierTemplate.get(dossierTemplateId).getLocalAccessMap().get(type);
if (model != null) {
return model.getColor();
}
return dictionariesByRuleSets.get(ruleSetId).getDefaultColor();
return dictionariesByDossierTemplate.get(dossierTemplateId).getDefaultColor();
}
public boolean isHint(String type, String ruleSetId) {
public boolean isHint(String type, String dossierTemplateId) {
DictionaryModel model = dictionariesByRuleSets.get(ruleSetId).getLocalAccessMap().get(type);
DictionaryModel model = dictionariesByDossierTemplate.get(dossierTemplateId).getLocalAccessMap().get(type);
if (model != null) {
return model.isHint();
}
@@ -186,9 +181,9 @@ public class DictionaryService {
}
public boolean isRecommendation(String type, String ruleSetId) {
public boolean isRecommendation(String type, String dossierTemplateId) {
DictionaryModel model = dictionariesByRuleSets.get(ruleSetId).getLocalAccessMap().get(type);
DictionaryModel model = dictionariesByDossierTemplate.get(dossierTemplateId).getLocalAccessMap().get(type);
if (model != null) {
return model.isRecommendation();
}
@@ -196,12 +191,12 @@ public class DictionaryService {
}
public Dictionary getDeepCopyDictionary(String ruleSetId, String dossierId) {
public Dictionary getDeepCopyDictionary(String dossierTemplateId, String dossierId) {
List<DictionaryModel> copy = new ArrayList<>();
var rulesetRepresentation = dictionariesByRuleSets.get(ruleSetId);
rulesetRepresentation.getDictionary().forEach(dm -> {
var dossierTemplateRepresentation = dictionariesByDossierTemplate.get(dossierTemplateId);
dossierTemplateRepresentation.getDictionary().forEach(dm -> {
copy.add(SerializationUtils.clone(dm));
});
@@ -215,25 +210,25 @@ public class DictionaryService {
dossierDictionaryVersion = dossierRepresentation.getDictionaryVersion();
}
return new Dictionary(copy, DictionaryVersion.builder().rulesetVersion(rulesetRepresentation.getDictionaryVersion()).dossierVersion(dossierDictionaryVersion).build());
return new Dictionary(copy, DictionaryVersion.builder().dossierTemplateVersion(dossierTemplateRepresentation.getDictionaryVersion()).dossierVersion(dossierDictionaryVersion).build());
}
public float[] getRequestRemoveColor(String ruleSetId) {
public float[] getRequestRemoveColor(String dossierTemplateId) {
return dictionariesByRuleSets.get(ruleSetId).getRequestAddColor();
return dictionariesByDossierTemplate.get(dossierTemplateId).getRequestAddColor();
}
public float[] getNotRedactedColor(String ruleSetId) {
public float[] getNotRedactedColor(String dossierTemplateId) {
return dictionariesByRuleSets.get(ruleSetId).getNotRedactedColor();
return dictionariesByDossierTemplate.get(dossierTemplateId).getNotRedactedColor();
}
public float[] getRequestAddColor(String ruleSetId) {
public float[] getRequestAddColor(String dossierTemplateId) {
return dictionariesByRuleSets.get(ruleSetId).getRequestAddColor();
return dictionariesByDossierTemplate.get(dossierTemplateId).getRequestAddColor();
}
}
@@ -28,14 +28,14 @@ public class DroolsExecutionService {
private final Map<String, KieContainer> kieContainers = new HashMap<>();
private final Map<String, Long> rulesVersionPerRuleSetId = new HashMap<>();
private final Map<String, Long> rulesVersionPerDossierTemplateId = new HashMap<>();
public KieContainer getKieContainer(String ruleSetId) {
public KieContainer getKieContainer(String dossierTemplateId) {
KieContainer container = kieContainers.get(ruleSetId);
KieContainer container = kieContainers.get(dossierTemplateId);
if (container == null) {
return createOrUpdateKieContainer(ruleSetId);
return createOrUpdateKieContainer(dossierTemplateId);
} else {
return container;
}
@@ -55,43 +55,43 @@ public class DroolsExecutionService {
}
public KieContainer updateRules(String ruleSetId) {
public KieContainer updateRules(String dossierTemplateId) {
long version = rulesClient.getVersion(ruleSetId);
Long rulesVersion = rulesVersionPerRuleSetId.get(ruleSetId);
long version = rulesClient.getVersion(dossierTemplateId);
Long rulesVersion = rulesVersionPerDossierTemplateId.get(dossierTemplateId);
if (rulesVersion == null) {
rulesVersion = -1L;
}
if (version > rulesVersion.longValue()) {
rulesVersionPerRuleSetId.put(ruleSetId, version);
return createOrUpdateKieContainer(ruleSetId);
rulesVersionPerDossierTemplateId.put(dossierTemplateId, version);
return createOrUpdateKieContainer(dossierTemplateId);
}
return getKieContainer(ruleSetId);
return getKieContainer(dossierTemplateId);
}
private KieContainer createOrUpdateKieContainer(String ruleSetId) {
private KieContainer createOrUpdateKieContainer(String dossierTemplateId) {
try {
RulesResponse rules = rulesClient.getRules(ruleSetId);
RulesResponse rules = rulesClient.getRules(dossierTemplateId);
if (rules == null || StringUtils.isEmpty(rules.getRules())) {
throw new RuntimeException("Rules cannot be empty.");
}
KieServices kieServices = KieServices.Factory.get();
KieModule kieModule = getKieModule(ruleSetId, rules.getRules(), kieServices);
KieModule kieModule = getKieModule(dossierTemplateId, rules.getRules(), kieServices);
var container = kieContainers.get(ruleSetId);
var container = kieContainers.get(dossierTemplateId);
if (container != null) {
container.updateToVersion(kieModule.getReleaseId());
return container;
}
container = kieServices.newKieContainer(kieModule.getReleaseId());
kieContainers.put(ruleSetId, container);
kieContainers.put(dossierTemplateId, container);
return container;
} catch (Exception e) {
throw new RulesValidationException("Could not update rules: " + e.getMessage(), e);
@@ -100,11 +100,11 @@ public class DroolsExecutionService {
}
private KieModule getKieModule(String ruleSetId, String rules, KieServices kieServices) {
private KieModule getKieModule(String dossierTemplateId, String rules, KieServices kieServices) {
KieFileSystem kieFileSystem = kieServices.newKieFileSystem();
InputStream input = new ByteArrayInputStream(rules.getBytes(StandardCharsets.UTF_8));
kieFileSystem.write("src/main/resources/drools/rules" + ruleSetId + ".drl", kieServices.getResources()
kieFileSystem.write("src/main/resources/drools/rules" + dossierTemplateId + ".drl", kieServices.getResources()
.newInputStreamResource(input));
KieBuilder kieBuilder = kieServices.newKieBuilder(kieFileSystem);
kieBuilder.buildAll();
@@ -122,9 +122,9 @@ public class DroolsExecutionService {
}
public long getRulesVersion(String ruleSetId) {
public long getRulesVersion(String dossierTemplateId) {
Long rulesVersion = rulesVersionPerRuleSetId.get(ruleSetId);
Long rulesVersion = rulesVersionPerDossierTemplateId.get(dossierTemplateId);
if (rulesVersion == null) {
return -1;
}
@@ -33,13 +33,13 @@ public class EntityRedactionService {
private final SurroundingWordsService surroundingWordsService;
public void processDocument(Document classifiedDoc, String ruleSetId, ManualRedactions manualRedactions, String dossierId) {
public void processDocument(Document classifiedDoc, String dossierTemplateId, ManualRedactions manualRedactions, String dossierId) {
dictionaryService.updateDictionary(ruleSetId, dossierId);
KieContainer container = droolsExecutionService.updateRules(ruleSetId);
long rulesVersion = droolsExecutionService.getRulesVersion(ruleSetId);
dictionaryService.updateDictionary(dossierTemplateId, dossierId);
KieContainer container = droolsExecutionService.updateRules(dossierTemplateId);
long rulesVersion = droolsExecutionService.getRulesVersion(dossierTemplateId);
Dictionary dictionary = dictionaryService.getDeepCopyDictionary(ruleSetId, dossierId);
Dictionary dictionary = dictionaryService.getDeepCopyDictionary(dossierTemplateId, dossierId);
Set<Entity> documentEntities = new HashSet<>(findEntities(classifiedDoc, container, manualRedactions, dictionary, false, null));
@@ -75,7 +75,7 @@ public class EntityRedactionService {
}
}
dictionaryService.updateExternalDictionary(dictionary, ruleSetId);
dictionaryService.updateExternalDictionary(dictionary, dossierTemplateId);
classifiedDoc.setDictionaryVersion(dictionary.getVersion());
classifiedDoc.setRulesVersion(rulesVersion);
@@ -5,6 +5,7 @@ import com.iqser.red.service.redaction.v1.server.client.ImageClassificationClien
import com.iqser.red.service.redaction.v1.server.client.ImageClassificationResponse;
import com.iqser.red.service.redaction.v1.server.client.MockMultipartFile;
import com.iqser.red.service.redaction.v1.server.redaction.model.ImageType;
import com.iqser.red.service.redaction.v1.server.redaction.model.PdfImage;
import com.iqser.red.service.redaction.v1.server.settings.RedactionServiceSettings;
import lombok.RequiredArgsConstructor;
import lombok.extern.slf4j.Slf4j;
@@ -26,7 +27,7 @@ public class ImageClassificationService {
page.getImages().forEach(image -> {
if (settings.isEnableImageClassification()) {
if (settings.isEnableImageClassification() && !isEntirePageImage(image, page)) {
long start = System.currentTimeMillis();
try (ByteArrayOutputStream baos = new ByteArrayOutputStream()) {
@@ -38,7 +39,6 @@ public class ImageClassificationService {
log.error("Could not classify image", e);
image.setImageType(ImageType.OTHER);
}
log.info("Image classification took: " + (System.currentTimeMillis() - start));
} else {
image.setImageType(ImageType.OTHER);
@@ -59,4 +59,13 @@ public class ImageClassificationService {
}
private boolean isEntirePageImage(PdfImage image, Page page){
double imageArea = image.getPosition().getHeight() * image.getPosition().getWidth();
if(imageArea / page.getCropBoxArea() >= settings.getMaxImageCropboxRatio()){
log.info("Skipping image classification because images is almost as large as the entire page");
return true;
}
return false;
}
}
@@ -5,6 +5,7 @@ import com.iqser.red.service.redaction.v1.model.*;
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.*;
@@ -38,7 +39,7 @@ public class ReanalyzeService {
private final PdfSegmentationService pdfSegmentationService;
private final RedactionChangeLogService redactionChangeLogService;
private final AnalyzeResponseService analyzeResponseService;
private final LegalBasisClient legalBasisClient;
public AnalyzeResult analyze(AnalyzeRequest analyzeRequest) {
@@ -49,7 +50,7 @@ public class ReanalyzeService {
try {
var storedObjectStream = redactionStorageService.getStoredObject(RedactionStorageService.StorageIdUtils.getStorageId(analyzeRequest
.getProjectId(), analyzeRequest.getFileId(), FileType.ORIGIN));
.getDossierId(), analyzeRequest.getFileId(), FileType.ORIGIN));
classifiedDoc = pdfSegmentationService.parseDocument(storedObjectStream);
pageCount = classifiedDoc.getPages().size();
} catch (Exception e) {
@@ -57,31 +58,34 @@ public class ReanalyzeService {
}
log.info("Document structure analysis successful, starting redaction analysis...");
entityRedactionService.processDocument(classifiedDoc, analyzeRequest.getRuleSetId(), analyzeRequest.getManualRedactions(), analyzeRequest
.getProjectId());
entityRedactionService.processDocument(classifiedDoc, analyzeRequest.getDossierTemplateId(), analyzeRequest.getManualRedactions(), analyzeRequest
.getDossierId());
redactionLogCreatorService.createRedactionLog(classifiedDoc, pageCount, analyzeRequest.getManualRedactions(), analyzeRequest
.getRuleSetId());
.getDossierTemplateId());
log.info("Redaction analysis successful...");
var redactionLog = new RedactionLog(classifiedDoc.getRedactionLogEntities(), classifiedDoc.getDictionaryVersion()
.getRulesetVersion(), classifiedDoc.getRulesVersion(), analyzeRequest.getRuleSetId(), classifiedDoc.getDictionaryVersion()
.getDossierVersion());
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()));
log.info("Analyzed with rules {} and dictionary {} for ruleSet: {}", classifiedDoc.getRulesVersion(), classifiedDoc
.getDictionaryVersion(), analyzeRequest.getRuleSetId());
log.info("Analyzed with rules {} and dictionary {} for dossierTemplate: {}", classifiedDoc.getRulesVersion(), classifiedDoc
.getDictionaryVersion(), analyzeRequest.getDossierTemplateId());
// first create changelog - this only happens when we migrate files analyzed via the old process and we don't want to loose changeLog data
var changeLog = redactionChangeLogService.createAndStoreChangeLog(analyzeRequest.getProjectId(), analyzeRequest.getFileId(), redactionLog);
var changeLog = redactionChangeLogService.createAndStoreChangeLog(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), redactionLog);
// store redactionLog
redactionStorageService.storeObject(analyzeRequest.getProjectId(), analyzeRequest.getFileId(), FileType.REDACTION_LOG, redactionLog);
redactionStorageService.storeObject(analyzeRequest.getProjectId(), analyzeRequest.getFileId(), FileType.TEXT, new Text(pageCount, classifiedDoc
redactionStorageService.storeObject(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), FileType.REDACTION_LOG, redactionLog);
redactionStorageService.storeObject(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), FileType.TEXT, new Text(pageCount, classifiedDoc
.getSectionText()));
redactionStorageService.storeObject(analyzeRequest.getProjectId(), analyzeRequest.getFileId(), FileType.SECTION_GRID, classifiedDoc
redactionStorageService.storeObject(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), FileType.SECTION_GRID, classifiedDoc
.getSectionGrid());
long duration = System.currentTimeMillis() - startTime;
return analyzeResponseService.createAnalyzeResponse(analyzeRequest.getProjectId(), analyzeRequest.getFileId(), duration, pageCount, redactionLog, changeLog);
return analyzeResponseService.createAnalyzeResponse(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), duration, pageCount, redactionLog, changeLog);
}
@@ -90,16 +94,16 @@ public class ReanalyzeService {
long startTime = System.currentTimeMillis();
var redactionLog = redactionStorageService.getRedactionLog(analyzeRequest.getProjectId(), analyzeRequest.getFileId());
var text = redactionStorageService.getText(analyzeRequest.getProjectId(), analyzeRequest.getFileId());
var redactionLog = redactionStorageService.getRedactionLog(analyzeRequest.getDossierId(), analyzeRequest.getFileId());
var text = redactionStorageService.getText(analyzeRequest.getDossierId(), analyzeRequest.getFileId());
// not yet ready for reanalysis
if (redactionLog == null || text == null || text.getNumberOfPages() == 0) {
return analyze(analyzeRequest);
}
DictionaryIncrement dictionaryIncrement = dictionaryService.getDictionaryIncrements(analyzeRequest.getRuleSetId(), new DictionaryVersion(redactionLog
.getDictionaryVersion(), redactionLog.getDossierDictionaryVersion()), analyzeRequest.getProjectId());
DictionaryIncrement dictionaryIncrement = dictionaryService.getDictionaryIncrements(analyzeRequest.getDossierTemplateId(), new DictionaryVersion(redactionLog
.getDictionaryVersion(), redactionLog.getDossierDictionaryVersion()), analyzeRequest.getDossierId());
Set<String> manualForceAndRemoveIds = getForceAndRemoveIds(analyzeRequest.getManualRedactions());
Map<String, List<Comment>> comments = null;
@@ -159,9 +163,9 @@ public class ReanalyzeService {
//--
KieContainer kieContainer = droolsExecutionService.updateRules(analyzeRequest.getRuleSetId());
KieContainer kieContainer = droolsExecutionService.updateRules(analyzeRequest.getDossierTemplateId());
Dictionary dictionary = dictionaryService.getDeepCopyDictionary(analyzeRequest.getRuleSetId(), analyzeRequest.getProjectId());
Dictionary dictionary = dictionaryService.getDeepCopyDictionary(analyzeRequest.getDossierTemplateId(), analyzeRequest.getDossierId());
List<SectionSearchableTextPair> sectionSearchableTextPairs = new ArrayList<>();
for (SectionText reanalysisSection : reanalysisSections) {
@@ -224,16 +228,16 @@ public class ReanalyzeService {
for (int page = 1; page <= text.getNumberOfPages(); page++) {
if (entitiesPerPage.get(page) != null) {
newRedactionLogEntries.addAll(redactionLogCreatorService.addEntries(entitiesPerPage, analyzeRequest.getManualRedactions(), page, analyzeRequest
.getRuleSetId()));
.getDossierTemplateId()));
}
if (imagesPerPage.get(page) != null) {
newRedactionLogEntries.addAll(redactionLogCreatorService.addImageEntries(imagesPerPage, analyzeRequest.getManualRedactions(), page, analyzeRequest
.getRuleSetId()));
.getDossierTemplateId()));
}
newRedactionLogEntries.addAll(redactionLogCreatorService.addManualAddEntries(manualAdds, comments, page, analyzeRequest
.getRuleSetId()));
.getDossierTemplateId()));
}
redactionLog.getRedactionLogEntry()
@@ -248,15 +252,15 @@ public class ReanalyzeService {
RedactionLog redactionLog, Text text,
DictionaryIncrement dictionaryIncrement) {
redactionLog.setDictionaryVersion(dictionaryIncrement.getDictionaryVersion().getRulesetVersion());
redactionLog.setDictionaryVersion(dictionaryIncrement.getDictionaryVersion().getDossierTemplateVersion());
redactionLog.setDossierDictionaryVersion(dictionaryIncrement.getDictionaryVersion().getDossierVersion());
var changeLog = redactionChangeLogService.createAndStoreChangeLog(analyzeRequest.getProjectId(), analyzeRequest.getFileId(), redactionLog);
redactionStorageService.storeObject(analyzeRequest.getProjectId(), analyzeRequest.getFileId(), FileType.REDACTION_LOG, redactionLog);
var changeLog = redactionChangeLogService.createAndStoreChangeLog(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), redactionLog);
redactionStorageService.storeObject(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), FileType.REDACTION_LOG, redactionLog);
long duration = System.currentTimeMillis() - startTime;
return analyzeResponseService.createAnalyzeResponse(analyzeRequest.getProjectId(), analyzeRequest.getFileId(), duration, text
return analyzeResponseService.createAnalyzeResponse(analyzeRequest.getDossierId(), analyzeRequest.getFileId(), duration, text
.getNumberOfPages(), redactionLog, changeLog);
}
@@ -22,12 +22,12 @@ public class RedactionChangeLogService {
private final RedactionStorageService redactionStorageService;
public RedactionChangeLog createAndStoreChangeLog(String projectId, String fileId, RedactionLog currentRedactionLog) {
public RedactionChangeLog createAndStoreChangeLog(String dossierId, String fileId, RedactionLog currentRedactionLog) {
try {
RedactionLog previousRedactionLog = redactionStorageService.getRedactionLog(projectId, fileId);
RedactionLog previousRedactionLog = redactionStorageService.getRedactionLog(dossierId, fileId);
var changeLog = createChangeLog(currentRedactionLog, previousRedactionLog);
redactionStorageService.storeObject(projectId, fileId, FileType.REDACTION_CHANGELOG, changeLog);
redactionStorageService.storeObject(dossierId, fileId, FileType.REDACTION_CHANGELOG, changeLog);
return changeLog;
} catch (Exception e) {
log.debug("Previous redaction log not available");
@@ -57,8 +57,11 @@ public class RedactionChangeLogService {
.map(entry -> convert(entry, ChangeType.REMOVED))
.collect(Collectors.toList()));
return new RedactionChangeLog(changeLogEntries, currentRedactionLog.getDictionaryVersion(), currentRedactionLog.getRulesVersion(), currentRedactionLog
.getRuleSetId());
return new RedactionChangeLog(changeLogEntries,
currentRedactionLog.getDictionaryVersion(),
currentRedactionLog.getDossierDictionaryVersion(),
currentRedactionLog.getRulesVersion(),
currentRedactionLog.getLegalBasisVersion());
}
@@ -32,7 +32,7 @@ public class RedactionLogCreatorService {
public void createRedactionLog(Document classifiedDoc, int numberOfPages, ManualRedactions manualRedactions,
String ruleSetId) {
String dossierTemplateId) {
Set<Integer> manualRedactionPages = getManualRedactionPages(manualRedactions);
@@ -42,24 +42,24 @@ public class RedactionLogCreatorService {
if (classifiedDoc.getEntities().get(page) != null) {
classifiedDoc.getRedactionLogEntities()
.addAll(addEntries(classifiedDoc.getEntities(), manualRedactions, page, ruleSetId));
.addAll(addEntries(classifiedDoc.getEntities(), manualRedactions, page, dossierTemplateId));
}
if (manualRedactionPages.contains(page)) {
classifiedDoc.getRedactionLogEntities()
.addAll(addManualAddEntries(manualRedactions.getEntriesToAdd(), manualRedactions.getComments(), page, ruleSetId));
.addAll(addManualAddEntries(manualRedactions.getEntriesToAdd(), manualRedactions.getComments(), page, dossierTemplateId));
}
if (classifiedDoc.getImages().get(page) != null && !classifiedDoc.getImages().get(page).isEmpty()) {
classifiedDoc.getRedactionLogEntities()
.addAll(addImageEntries(classifiedDoc.getImages(), manualRedactions, page, ruleSetId));
.addAll(addImageEntries(classifiedDoc.getImages(), manualRedactions, page, dossierTemplateId));
}
}
}
public List<RedactionLogEntry> addImageEntries(Map<Integer, Set<Image>> images, ManualRedactions manualRedactions,
int pageNumber, String ruleSetId) {
int pageNumber, String dossierTemplateId) {
List<RedactionLogEntry> redactionLogEntities = new ArrayList<>();
@@ -69,14 +69,14 @@ public class RedactionLogCreatorService {
RedactionLogEntry redactionLogEntry = RedactionLogEntry.builder()
.id(id)
.color(getColorForImage(image, ruleSetId, false))
.color(getColorForImage(image, dossierTemplateId, false))
.isImage(true)
.type(image.getType())
.redacted(image.isRedaction())
.reason(image.getRedactionReason())
.legalBasis(image.getLegalBasis())
.matchedRule(image.getMatchedRule())
.isHint(dictionaryService.isHint(image.getType(), ruleSetId))
.isHint(dictionaryService.isHint(image.getType(), dossierTemplateId))
.manual(false)
.isDictionaryEntry(false)
.isRecommendation(false)
@@ -96,11 +96,11 @@ public class RedactionLogCreatorService {
redactionLogEntry.setRedacted(false);
redactionLogEntry.setStatus(Status.APPROVED);
manualOverrideReason = image.getRedactionReason() + ", removed by manual override";
redactionLogEntry.setColor(getColorForImage(image, ruleSetId, false));
redactionLogEntry.setColor(getColorForImage(image, dossierTemplateId, false));
} else if (manualRemoval.getStatus().equals(Status.REQUESTED)) {
manualOverrideReason = image.getRedactionReason() + ", requested to remove";
redactionLogEntry.setStatus(Status.REQUESTED);
redactionLogEntry.setColor(getColorForImage(image, ruleSetId, true));
redactionLogEntry.setColor(getColorForImage(image, dossierTemplateId, true));
} else {
redactionLogEntry.setStatus(Status.DECLINED);
}
@@ -121,13 +121,13 @@ public class RedactionLogCreatorService {
image.setRedaction(true);
redactionLogEntry.setRedacted(true);
redactionLogEntry.setStatus(Status.APPROVED);
redactionLogEntry.setColor(getColorForImage(image, ruleSetId, false));
redactionLogEntry.setColor(getColorForImage(image, dossierTemplateId, false));
manualOverrideReason = image.getRedactionReason() + ", forced by manual override";
redactionLogEntry.setLegalBasis(manualForceRedact.getLegalBasis());
} else if (manualForceRedact.getStatus().equals(Status.REQUESTED)) {
manualOverrideReason = image.getRedactionReason() + ", requested to force redact";
redactionLogEntry.setStatus(Status.REQUESTED);
redactionLogEntry.setColor(getColorForImage(image, ruleSetId, true));
redactionLogEntry.setColor(getColorForImage(image, dossierTemplateId, true));
redactionLogEntry.setLegalBasis(manualForceRedact.getLegalBasis());
} else {
redactionLogEntry.setStatus(Status.DECLINED);
@@ -166,7 +166,7 @@ public class RedactionLogCreatorService {
public List<RedactionLogEntry> addEntries(Map<Integer, List<Entity>> entities, ManualRedactions manualRedactions,
int page, String ruleSetId) {
int page, String dossierTemplateId) {
List<RedactionLogEntry> redactionLogEntities = new ArrayList<>();
@@ -180,7 +180,7 @@ public class RedactionLogCreatorService {
for (EntityPositionSequence entityPositionSequence : entity.getPositionSequences()) {
RedactionLogEntry redactionLogEntry = createRedactionLogEntry(entity, ruleSetId);
RedactionLogEntry redactionLogEntry = createRedactionLogEntry(entity, dossierTemplateId);
if (processedIds.contains(entityPositionSequence.getId())) {
// TODO refactor this outer loop jump as soon as we have the time.
@@ -199,11 +199,11 @@ public class RedactionLogCreatorService {
redactionLogEntry.setRedacted(false);
redactionLogEntry.setStatus(Status.APPROVED);
manualOverrideReason = entity.getRedactionReason() + ", removed by manual override";
redactionLogEntry.setColor(getColor(entity, ruleSetId, false));
redactionLogEntry.setColor(getColor(entity, dossierTemplateId, false));
} else if (manualRemoval.getStatus().equals(Status.REQUESTED)) {
manualOverrideReason = entity.getRedactionReason() + ", requested to remove";
redactionLogEntry.setStatus(Status.REQUESTED);
redactionLogEntry.setColor(getColor(entity, ruleSetId, true));
redactionLogEntry.setColor(getColor(entity, dossierTemplateId, true));
} else {
redactionLogEntry.setStatus(Status.DECLINED);
}
@@ -224,13 +224,13 @@ public class RedactionLogCreatorService {
entity.setRedaction(true);
redactionLogEntry.setRedacted(true);
redactionLogEntry.setStatus(Status.APPROVED);
redactionLogEntry.setColor(getColor(entity, ruleSetId, false));
redactionLogEntry.setColor(getColor(entity, dossierTemplateId, false));
manualOverrideReason = entity.getRedactionReason() + ", forced by manual override";
redactionLogEntry.setLegalBasis(manualForceRedact.getLegalBasis());
} else if (manualForceRedact.getStatus().equals(Status.REQUESTED)) {
manualOverrideReason = entity.getRedactionReason() + ", requested to force redact";
redactionLogEntry.setStatus(Status.REQUESTED);
redactionLogEntry.setColor(getColor(entity, ruleSetId, true));
redactionLogEntry.setColor(getColor(entity, dossierTemplateId, true));
redactionLogEntry.setLegalBasis(manualForceRedact.getLegalBasis());
} else {
redactionLogEntry.setStatus(Status.DECLINED);
@@ -299,7 +299,7 @@ public class RedactionLogCreatorService {
public List<RedactionLogEntry> addManualAddEntries(Set<ManualRedactionEntry> manualAdds,
Map<String, List<Comment>> comments, int page,
String ruleSetId) {
String dossierTemplateId) {
List<RedactionLogEntry> redactionLogEntities = new ArrayList<>();
@@ -311,7 +311,7 @@ public class RedactionLogCreatorService {
String id = manualRedactionEntry.getId();
RedactionLogEntry redactionLogEntry = createRedactionLogEntry(manualRedactionEntry, id, ruleSetId);
RedactionLogEntry redactionLogEntry = createRedactionLogEntry(manualRedactionEntry, id, dossierTemplateId);
List<Rectangle> rectanglesOnPage = new ArrayList<>();
for (Rectangle rectangle : manualRedactionEntry.getPositions()) {
@@ -338,11 +338,11 @@ public class RedactionLogCreatorService {
private RedactionLogEntry createRedactionLogEntry(ManualRedactionEntry manualRedactionEntry, String id,
String ruleSetId) {
String dossierTemplateId) {
return RedactionLogEntry.builder()
.id(id)
.color(getColorForManualAdd(manualRedactionEntry.getType(), ruleSetId, manualRedactionEntry.getStatus()))
.color(getColorForManualAdd(manualRedactionEntry.getType(), dossierTemplateId, manualRedactionEntry.getStatus()))
.reason(manualRedactionEntry.getReason())
.legalBasis(manualRedactionEntry.getLegalBasis())
.value(manualRedactionEntry.getValue())
@@ -360,17 +360,17 @@ public class RedactionLogCreatorService {
}
private RedactionLogEntry createRedactionLogEntry(Entity entity, String ruleSetId) {
private RedactionLogEntry createRedactionLogEntry(Entity entity, String dossierTemplateId) {
return RedactionLogEntry.builder()
.color(getColor(entity, ruleSetId, false))
.color(getColor(entity, dossierTemplateId, false))
.reason(entity.getRedactionReason())
.legalBasis(entity.getLegalBasis())
.value(entity.getWord())
.type(entity.getType())
.redacted(entity.isRedaction())
.isHint(isHint(entity, ruleSetId))
.isRecommendation(isRecommendation(entity, ruleSetId))
.isHint(isHint(entity, dossierTemplateId))
.isRecommendation(isRecommendation(entity, dossierTemplateId))
.section(entity.getHeadline())
.sectionNumber(entity.getSectionNumber())
.matchedRule(entity.getMatchedRule())
@@ -384,56 +384,56 @@ public class RedactionLogCreatorService {
}
private float[] getColor(Entity entity, String ruleSetId, boolean requestedToRemove) {
private float[] getColor(Entity entity, String dossierTemplateId, boolean requestedToRemove) {
if (requestedToRemove) {
return dictionaryService.getRequestRemoveColor(ruleSetId);
return dictionaryService.getRequestRemoveColor(dossierTemplateId);
}
if (!entity.isRedaction() && !isHint(entity, ruleSetId)) {
return dictionaryService.getNotRedactedColor(ruleSetId);
if (!entity.isRedaction() && !isHint(entity, dossierTemplateId)) {
return dictionaryService.getNotRedactedColor(dossierTemplateId);
}
return dictionaryService.getColor(entity.getType(), ruleSetId);
return dictionaryService.getColor(entity.getType(), dossierTemplateId);
}
private float[] getColorForManualAdd(String type, String ruleSetId, Status status) {
private float[] getColorForManualAdd(String type, String dossierTemplateId, Status status) {
if (status.equals(Status.REQUESTED)) {
return dictionaryService.getRequestAddColor(ruleSetId);
return dictionaryService.getRequestAddColor(dossierTemplateId);
} else if (status.equals(Status.DECLINED)) {
return dictionaryService.getNotRedactedColor(ruleSetId);
return dictionaryService.getNotRedactedColor(dossierTemplateId);
}
return getColor(type, ruleSetId);
return getColor(type, dossierTemplateId);
}
private float[] getColor(String type, String ruleSetId) {
private float[] getColor(String type, String dossierTemplateId) {
return dictionaryService.getColor(type, ruleSetId);
return dictionaryService.getColor(type, dossierTemplateId);
}
private float[] getColorForImage(Image image, String ruleSetId, boolean requestedToRemove) {
private float[] getColorForImage(Image image, String dossierTemplateId, boolean requestedToRemove) {
if (requestedToRemove) {
return dictionaryService.getRequestRemoveColor(ruleSetId);
return dictionaryService.getRequestRemoveColor(dossierTemplateId);
}
if (!image.isRedaction() && !dictionaryService.isHint(image.getType(), ruleSetId)) {
return dictionaryService.getNotRedactedColor(ruleSetId);
if (!image.isRedaction() && !dictionaryService.isHint(image.getType(), dossierTemplateId)) {
return dictionaryService.getNotRedactedColor(dossierTemplateId);
}
return dictionaryService.getColor(image.getType(), ruleSetId);
return dictionaryService.getColor(image.getType(), dossierTemplateId);
}
private boolean isHint(Entity entity, String ruleSetId) {
private boolean isHint(Entity entity, String dossierTemplateId) {
return dictionaryService.isHint(entity.getType(), ruleSetId);
return dictionaryService.isHint(entity.getType(), dossierTemplateId);
}
private boolean isRecommendation(Entity entity, String ruleSetId) {
private boolean isRecommendation(Entity entity, String dossierTemplateId) {
return dictionaryService.isRecommendation(entity.getType(), ruleSetId);
return dictionaryService.isRecommendation(entity.getType(), dossierTemplateId);
}
@@ -1,5 +1,6 @@
package com.iqser.red.service.redaction.v1.server.segmentation;
import com.iqser.red.service.redaction.v1.model.Rectangle;
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;
@@ -93,10 +94,25 @@ public class PdfSegmentationService {
Page page = blockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings
.getVertical());
PDRectangle cropbox = pdPage.getCropBox();
float cropboxArea = cropbox.getHeight() * cropbox.getWidth();
page.setCropBoxArea(cropboxArea);
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);
page.setImages(mergedList);
tableExtractionService.extractTables(cleanRulings, page);
@@ -13,4 +13,6 @@ public class RedactionServiceSettings {
private boolean enableImageClassification = true;
private float maxImageCropboxRatio = 0.9f;
}
@@ -32,16 +32,16 @@ public class RedactionStorageService {
@SneakyThrows
public void storeObject(String projectId, String fileId, FileType fileType, Object any) {
storageService.storeObject(StorageIdUtils.getStorageId(projectId, fileId, fileType), objectMapper.writeValueAsBytes(any));
public void storeObject(String dossierId, String fileId, FileType fileType, Object any) {
storageService.storeObject(StorageIdUtils.getStorageId(dossierId, fileId, fileType), objectMapper.writeValueAsBytes(any));
}
public RedactionLog getRedactionLog(String projectId, String fileId) {
public RedactionLog getRedactionLog(String dossierId, String fileId) {
InputStreamResource inputStreamResource;
try {
inputStreamResource = storageService.getObject(StorageIdUtils.getStorageId(projectId, fileId, FileType.REDACTION_LOG));
inputStreamResource = storageService.getObject(StorageIdUtils.getStorageId(dossierId, fileId, FileType.REDACTION_LOG));
} catch (StorageObjectDoesNotExist e) {
log.debug("Text not available.");
return null;
@@ -55,11 +55,11 @@ public class RedactionStorageService {
}
public Text getText(String projectId, String fileId) {
public Text getText(String dossierId, String fileId) {
InputStreamResource inputStreamResource;
try {
inputStreamResource = storageService.getObject(StorageIdUtils.getStorageId(projectId, fileId, FileType.TEXT));
inputStreamResource = storageService.getObject(StorageIdUtils.getStorageId(dossierId, fileId, FileType.TEXT));
} catch (StorageObjectDoesNotExist e) {
log.debug("Text not available.");
return null;
@@ -73,9 +73,9 @@ public class RedactionStorageService {
}
public SectionGrid getSectionGrid(String projectId, String fileId) {
public SectionGrid getSectionGrid(String dossierId, String fileId) {
var sectionGrid = storageService.getObject(StorageIdUtils.getStorageId(projectId, fileId, FileType.SECTION_GRID));
var sectionGrid = storageService.getObject(StorageIdUtils.getStorageId(dossierId, fileId, FileType.SECTION_GRID));
try {
return objectMapper.readValue(sectionGrid.getInputStream(), SectionGrid.class);
} catch (IOException e) {
@@ -95,8 +95,8 @@ public class RedactionStorageService {
public static class StorageIdUtils {
public static String getStorageId(String projectId, String fileId, FileType fileType) {
return projectId + "/" + fileId + "." + fileType.name() + fileType.getExtension();
public static String getStorageId(String dossierId, String fileId, FileType fileType) {
return dossierId + "/" + fileId + "." + fileType.name() + fileType.getExtension();
}
}
@@ -2,6 +2,8 @@ package com.iqser.red.service.redaction.v1.server.tableextraction.model;
import com.fasterxml.jackson.annotation.JsonIgnore;
import com.iqser.red.service.redaction.v1.model.Rectangle;
import com.iqser.red.service.redaction.v1.server.classification.model.Orientation;
import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;
@@ -18,6 +20,8 @@ public abstract class AbstractTextContainer {
protected String classification;
protected int page;
private Orientation orientation = Orientation.NONE;
public abstract String getText();
public boolean contains(AbstractTextContainer other) {
@@ -246,8 +246,12 @@ public class Ruling extends Line2D.Float {
public Ruling expand(float amount) {
Ruling r = (Ruling) this.clone();
r.setStart(this.getStart() - amount);
r.setEnd(this.getEnd() + amount);
try {
r.setStart(this.getStart() - amount);
r.setEnd(this.getEnd() + amount);
} catch (UnsupportedOperationException e){
log.warn("Could not expand ruling!");
}
return r;
}
@@ -102,7 +102,7 @@ public class PdfVisualisationService {
contentStream.newLineAtOffset(textBlock.getMinX(), textBlock.getMaxY());
contentStream.showText(textBlock.getClassification());
contentStream.showText(textBlock.getClassification() + textBlock.getOrientation());
contentStream.endText();
}
@@ -9,6 +9,7 @@ import com.iqser.red.service.redaction.v1.model.*;
import com.iqser.red.service.redaction.v1.server.classification.model.SectionText;
import com.iqser.red.service.redaction.v1.server.client.DictionaryClient;
import com.iqser.red.service.redaction.v1.server.client.ImageClassificationClient;
import com.iqser.red.service.redaction.v1.server.client.LegalBasisClient;
import com.iqser.red.service.redaction.v1.server.client.RulesClient;
import com.iqser.red.service.redaction.v1.server.controller.RedactionController;
import com.iqser.red.service.redaction.v1.server.memory.MemoryStats;
@@ -114,6 +115,9 @@ public class RedactionIntegrationTest {
@MockBean
private RabbitTemplate rabbitTemplate;
@MockBean
private LegalBasisClient legalBasisClient;
private final Map<String, List<String>> dictionary = new HashMap<>();
private final Map<String, List<String>> dossierDictionary = new HashMap<>();
private final Map<String, String> typeColorMap = new HashMap<>();
@@ -124,8 +128,8 @@ public class RedactionIntegrationTest {
private final Colors colors = new Colors();
private final Map<String, Long> reanlysisVersions = new HashMap<>();
private final static String TEST_RULESET_ID = "123";
private final static String TEST_PROJECT_ID = "123";
private final static String TEST_DOSSIER_TEMPLATE_ID = "123";
private final static String TEST_DOSSIER_ID = "123";
private final static String TEST_FILE_ID = "123";
@Configuration
@@ -169,21 +173,21 @@ public class RedactionIntegrationTest {
@Before
public void stubClients() {
when(rulesClient.getVersion(TEST_RULESET_ID)).thenReturn(0L);
when(rulesClient.getRules(TEST_RULESET_ID)).thenReturn(new RulesResponse(RULES));
when(rulesClient.getVersion(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(0L);
when(rulesClient.getRules(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(new RulesResponse(RULES));
loadDictionaryForTest();
loadTypeForTest();
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(0L);
when(dictionaryClient.getAllTypes(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(TypeResponse.builder()
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(0L);
when(dictionaryClient.getAllTypes(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(TypeResponse.builder()
.types(getTypeResponse())
.build());
when(dictionaryClient.getVersion(TEST_RULESET_ID, TEST_PROJECT_ID)).thenReturn(0L);
when(dictionaryClient.getAllTypes(TEST_RULESET_ID, TEST_PROJECT_ID)).thenReturn(TypeResponse.builder()
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, TEST_DOSSIER_ID)).thenReturn(0L);
when(dictionaryClient.getAllTypes(TEST_DOSSIER_TEMPLATE_ID, TEST_DOSSIER_ID)).thenReturn(TypeResponse.builder()
.types(List.of(TypeResult.builder()
.type(DOSSIER_REDACTIONS)
.ruleSetId(TEST_RULESET_ID)
.dossierTemplateId(TEST_DOSSIER_TEMPLATE_ID)
.hexColor( "#ffe187")
.isHint(hintTypeMap.get(DOSSIER_REDACTIONS))
.isCaseInsensitive(caseInSensitiveMap.get(DOSSIER_REDACTIONS))
@@ -192,28 +196,28 @@ public class RedactionIntegrationTest {
.build()))
.build());
when(dictionaryClient.getDictionaryForType(VERTEBRATE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(VERTEBRATE, false));
when(dictionaryClient.getDictionaryForType(ADDRESS, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(ADDRESS, false));
when(dictionaryClient.getDictionaryForType(AUTHOR, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(AUTHOR, false));
when(dictionaryClient.getDictionaryForType(SPONSOR, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(SPONSOR, false));
when(dictionaryClient.getDictionaryForType(NO_REDACTION_INDICATOR, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(NO_REDACTION_INDICATOR, false));
when(dictionaryClient.getDictionaryForType(REDACTION_INDICATOR, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(REDACTION_INDICATOR, false));
when(dictionaryClient.getDictionaryForType(HINT_ONLY, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(HINT_ONLY, false));
when(dictionaryClient.getDictionaryForType(MUST_REDACT, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(MUST_REDACT, false));
when(dictionaryClient.getDictionaryForType(PUBLISHED_INFORMATION, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(PUBLISHED_INFORMATION, false));
when(dictionaryClient.getDictionaryForType(TEST_METHOD, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(TEST_METHOD, false));
when(dictionaryClient.getDictionaryForType(PII, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(PII, false));
when(dictionaryClient.getDictionaryForType(RECOMMENDATION_AUTHOR, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(RECOMMENDATION_AUTHOR, false));
when(dictionaryClient.getDictionaryForType(RECOMMENDATION_ADDRESS, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(RECOMMENDATION_ADDRESS, false));
when(dictionaryClient.getDictionaryForType(FALSE_POSITIVE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(FALSE_POSITIVE, false));
when(dictionaryClient.getDictionaryForType(PURITY, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(PURITY, false));
when(dictionaryClient.getDictionaryForType(IMAGE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(IMAGE, false));
when(dictionaryClient.getDictionaryForType(OCR, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(OCR, false));
when(dictionaryClient.getDictionaryForType(LOGO, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(LOGO, false));
when(dictionaryClient.getDictionaryForType(SIGNATURE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(SIGNATURE, false));
when(dictionaryClient.getDictionaryForType(FORMULA, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(FORMULA, false));
when(dictionaryClient.getDictionaryForType(DOSSIER_REDACTIONS, TEST_RULESET_ID, TEST_PROJECT_ID)).thenReturn(getDictionaryResponse(DOSSIER_REDACTIONS, true));
when(dictionaryClient.getColors(TEST_RULESET_ID)).thenReturn(colors);
when(dictionaryClient.getDictionaryForType(VERTEBRATE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(VERTEBRATE, false));
when(dictionaryClient.getDictionaryForType(ADDRESS, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(ADDRESS, false));
when(dictionaryClient.getDictionaryForType(AUTHOR, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(AUTHOR, false));
when(dictionaryClient.getDictionaryForType(SPONSOR, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(SPONSOR, false));
when(dictionaryClient.getDictionaryForType(NO_REDACTION_INDICATOR, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(NO_REDACTION_INDICATOR, false));
when(dictionaryClient.getDictionaryForType(REDACTION_INDICATOR, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(REDACTION_INDICATOR, false));
when(dictionaryClient.getDictionaryForType(HINT_ONLY, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(HINT_ONLY, false));
when(dictionaryClient.getDictionaryForType(MUST_REDACT, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(MUST_REDACT, false));
when(dictionaryClient.getDictionaryForType(PUBLISHED_INFORMATION, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(PUBLISHED_INFORMATION, false));
when(dictionaryClient.getDictionaryForType(TEST_METHOD, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(TEST_METHOD, false));
when(dictionaryClient.getDictionaryForType(PII, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(PII, false));
when(dictionaryClient.getDictionaryForType(RECOMMENDATION_AUTHOR, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(RECOMMENDATION_AUTHOR, false));
when(dictionaryClient.getDictionaryForType(RECOMMENDATION_ADDRESS, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(RECOMMENDATION_ADDRESS, false));
when(dictionaryClient.getDictionaryForType(FALSE_POSITIVE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(FALSE_POSITIVE, false));
when(dictionaryClient.getDictionaryForType(PURITY, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(PURITY, false));
when(dictionaryClient.getDictionaryForType(IMAGE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(IMAGE, false));
when(dictionaryClient.getDictionaryForType(OCR, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(OCR, false));
when(dictionaryClient.getDictionaryForType(LOGO, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(LOGO, false));
when(dictionaryClient.getDictionaryForType(SIGNATURE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(SIGNATURE, false));
when(dictionaryClient.getDictionaryForType(FORMULA, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(FORMULA, false));
when(dictionaryClient.getDictionaryForType(DOSSIER_REDACTIONS, TEST_DOSSIER_TEMPLATE_ID, TEST_DOSSIER_ID)).thenReturn(getDictionaryResponse(DOSSIER_REDACTIONS, true));
when(dictionaryClient.getColors(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(colors);
}
@@ -457,7 +461,7 @@ public class RedactionIntegrationTest {
.stream()
.map(typeColor -> TypeResult.builder()
.type(typeColor.getKey())
.ruleSetId(TEST_RULESET_ID)
.dossierTemplateId(TEST_DOSSIER_TEMPLATE_ID)
.hexColor(typeColor.getValue())
.isHint(hintTypeMap.get(typeColor.getKey()))
.isCaseInsensitive(caseInSensitiveMap.get(typeColor.getKey()))
@@ -521,7 +525,7 @@ public class RedactionIntegrationTest {
Map<String, List<RedactionLogEntry>> duplicates = new HashMap<>();
var redactionLog = redactionStorageService.getRedactionLog(TEST_PROJECT_ID, TEST_FILE_ID);
var redactionLog = redactionStorageService.getRedactionLog(TEST_DOSSIER_ID, TEST_FILE_ID);
redactionLog.getRedactionLogEntry().forEach(entry -> {
duplicates.computeIfAbsent(entry.getId(), v -> new ArrayList<>()).add(entry);
@@ -532,10 +536,10 @@ public class RedactionIntegrationTest {
});
dictionary.get(AUTHOR).add("Drinking water");
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(1L);
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(1L);
AnnotateResponse annotateResponse = redactionController.annotate(AnnotateRequest.builder()
.projectId(TEST_PROJECT_ID)
.dossierId(TEST_DOSSIER_ID)
.fileId(TEST_FILE_ID)
.build());
@@ -577,7 +581,7 @@ public class RedactionIntegrationTest {
Map<String, List<RedactionLogEntry>> duplicates = new HashMap<>();
var redactionLog = redactionStorageService.getRedactionLog(TEST_PROJECT_ID, TEST_FILE_ID);
var redactionLog = redactionStorageService.getRedactionLog(TEST_DOSSIER_ID, TEST_FILE_ID);
redactionLog.getRedactionLogEntry().forEach(entry -> {
duplicates.computeIfAbsent(entry.getId(), v -> new ArrayList<>()).add(entry);
@@ -588,7 +592,7 @@ public class RedactionIntegrationTest {
});
dictionary.get(AUTHOR).add("Drinking water");
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(1L);
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(1L);
long rstart = System.currentTimeMillis();
reanalyzeService.reanalyze(request);
@@ -632,8 +636,8 @@ public class RedactionIntegrationTest {
AnalyzeResult result = reanalyzeService.analyze(request);
var redactionLog = redactionStorageService.getRedactionLog(TEST_PROJECT_ID, TEST_FILE_ID);
var text = redactionStorageService.getText(TEST_PROJECT_ID, TEST_FILE_ID);
var redactionLog = redactionStorageService.getRedactionLog(TEST_DOSSIER_ID, TEST_FILE_ID);
var text = redactionStorageService.getText(TEST_DOSSIER_ID, TEST_FILE_ID);
redactionLog.getRedactionLogEntry().forEach(entry -> {
if (entry.isImage()) {
@@ -646,7 +650,7 @@ public class RedactionIntegrationTest {
System.out.println("first analysis duration: " + (end - start));
try (FileOutputStream fileOutputStream = new FileOutputStream("/tmp/Test.json")) {
fileOutputStream.write(objectMapper.writeValueAsBytes(redactionStorageService.getText(TEST_PROJECT_ID, TEST_FILE_ID)));
fileOutputStream.write(objectMapper.writeValueAsBytes(redactionStorageService.getText(TEST_DOSSIER_ID, TEST_FILE_ID)));
}
int correctFound = 0;
@@ -676,12 +680,12 @@ public class RedactionIntegrationTest {
dictionary.get(AUTHOR).add("physical");
reanlysisVersions.put("physical", 2L);
dictionary.get(VERTEBRATE).add("s-metolachlor");
reanlysisVersions.put("s-metolachlor", 3L);
// dictionary.get(VERTEBRATE).add("s-metolachlor");
// reanlysisVersions.put("s-metolachlor", 3L);
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(3L);
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(3L);
when(dictionaryClient.getDictionaryForType(VERTEBRATE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(VERTEBRATE, false));
when(dictionaryClient.getDictionaryForType(VERTEBRATE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(getDictionaryResponse(VERTEBRATE, false));
start = System.currentTimeMillis();
AnalyzeResult reanalyzeResult = reanalyzeService.reanalyze(request);
@@ -690,7 +694,7 @@ public class RedactionIntegrationTest {
System.out.println("reanalysis analysis duration: " + (end - start));
AnnotateResponse annotateResponse = redactionController.annotate(AnnotateRequest.builder()
.projectId(TEST_PROJECT_ID)
.dossierId(TEST_DOSSIER_ID)
.fileId(TEST_FILE_ID)
.build());
@@ -711,7 +715,7 @@ public class RedactionIntegrationTest {
AnalyzeResult result = reanalyzeService.analyze(request);
AnnotateResponse annotateResponse = redactionController.annotate(AnnotateRequest.builder()
.projectId(TEST_PROJECT_ID)
.dossierId(TEST_DOSSIER_ID)
.fileId(TEST_FILE_ID)
.build());
@@ -782,7 +786,7 @@ public class RedactionIntegrationTest {
AnnotateResponse annotateResponse = redactionController.annotate(AnnotateRequest.builder()
.projectId(TEST_PROJECT_ID)
.dossierId(TEST_DOSSIER_ID)
.fileId(TEST_FILE_ID)
.build());
@@ -801,15 +805,15 @@ public class RedactionIntegrationTest {
public void classificationTest() throws IOException {
System.out.println("classificationTest");
ClassPathResource pdfFileResource = new ClassPathResource("files/Trinexapac/93 Trinexapac-ethyl_RAR_03_Volume_3CA_B-1_2017-03-31.pdf");
ClassPathResource pdfFileResource = new ClassPathResource("files/new/Single Study - Oral (Gavage) Mouse.pdf");
AnalyzeRequest request = prepareStorage(pdfFileResource.getInputStream());
RedactionRequest redactionRequest = RedactionRequest.builder()
.projectId(request.getProjectId())
.dossierId(request.getDossierId())
.fileId(request.getFileId())
.ruleSetId(request.getRuleSetId())
.dossierTemplateId(request.getDossierTemplateId())
.build();
RedactionResult result = redactionController.classify(redactionRequest);
@@ -829,9 +833,9 @@ public class RedactionIntegrationTest {
AnalyzeRequest request = prepareStorage(pdfFileResource.getInputStream());
RedactionRequest redactionRequest = RedactionRequest.builder()
.projectId(request.getProjectId())
.dossierId(request.getDossierId())
.fileId(request.getFileId())
.ruleSetId(request.getRuleSetId())
.dossierTemplateId(request.getDossierTemplateId())
.build();
RedactionResult result = redactionController.sections(redactionRequest);
@@ -851,9 +855,9 @@ public class RedactionIntegrationTest {
AnalyzeRequest request = prepareStorage(pdfFileResource.getInputStream());
RedactionRequest redactionRequest = RedactionRequest.builder()
.projectId(request.getProjectId())
.dossierId(request.getDossierId())
.fileId(request.getFileId())
.ruleSetId(request.getRuleSetId())
.dossierTemplateId(request.getDossierTemplateId())
.build();
RedactionResult result = redactionController.htmlTables(redactionRequest);
@@ -873,9 +877,9 @@ public class RedactionIntegrationTest {
AnalyzeRequest request = prepareStorage(pdfFileResource.getInputStream());
RedactionRequest redactionRequest = RedactionRequest.builder()
.projectId(request.getProjectId())
.dossierId(request.getDossierId())
.fileId(request.getFileId())
.ruleSetId(request.getRuleSetId())
.dossierTemplateId(request.getDossierTemplateId())
.build();
RedactionResult result = redactionController.htmlTables(redactionRequest);
@@ -895,7 +899,7 @@ public class RedactionIntegrationTest {
AnalyzeResult result = reanalyzeService.analyze(request);
var redactionLog = redactionStorageService.getRedactionLog(TEST_PROJECT_ID, TEST_FILE_ID);
var redactionLog = redactionStorageService.getRedactionLog(TEST_DOSSIER_ID, TEST_FILE_ID);
redactionLog.getRedactionLogEntry().forEach(entry -> {
if (!entry.isHint()) {
@@ -916,15 +920,15 @@ public class RedactionIntegrationTest {
private AnalyzeRequest prepareStorage(InputStream stream) {
AnalyzeRequest request = AnalyzeRequest.builder()
.ruleSetId(TEST_RULESET_ID)
.projectId(TEST_PROJECT_ID)
.dossierTemplateId(TEST_DOSSIER_TEMPLATE_ID)
.dossierId(TEST_DOSSIER_ID)
.fileId(TEST_FILE_ID)
.lastProcessed(OffsetDateTime.now())
.build();
var bytes = IOUtils.toByteArray(stream);
storageService.storeObject(RedactionStorageService.StorageIdUtils.getStorageId(TEST_PROJECT_ID, TEST_FILE_ID, FileType.ORIGIN), bytes);
storageService.storeObject(RedactionStorageService.StorageIdUtils.getStorageId(TEST_DOSSIER_ID, TEST_FILE_ID, FileType.ORIGIN), bytes);
return request;
@@ -943,7 +947,7 @@ public class RedactionIntegrationTest {
AnalyzeResult result = reanalyzeService.analyze(request);
AnnotateResponse annotateResponse = redactionController.annotate(AnnotateRequest.builder()
.projectId(TEST_PROJECT_ID)
.dossierId(TEST_DOSSIER_ID)
.fileId(TEST_FILE_ID)
.build());
@@ -7,6 +7,7 @@ import com.iqser.red.service.redaction.v1.server.Application;
import com.iqser.red.service.redaction.v1.server.FileSystemBackedStorageService;
import com.iqser.red.service.redaction.v1.server.classification.model.Document;
import com.iqser.red.service.redaction.v1.server.client.DictionaryClient;
import com.iqser.red.service.redaction.v1.server.client.LegalBasisClient;
import com.iqser.red.service.redaction.v1.server.client.RulesClient;
import com.iqser.red.service.redaction.v1.server.redaction.model.Entity;
import com.iqser.red.service.redaction.v1.server.redaction.utils.EntitySearchUtils;
@@ -78,7 +79,10 @@ public class EntityRedactionServiceTest {
@MockBean
private AmazonS3 amazonS3;
private final static String TEST_RULESET_ID = "123";
@MockBean
private LegalBasisClient legalBasisClient;
private final static String TEST_DOSSIER_TEMPLATE_ID = "123";
@Configuration
@EnableAutoConfiguration(exclude = {RabbitAutoConfiguration.class})
@@ -134,20 +138,20 @@ public class EntityRedactionServiceTest {
DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(Arrays.asList("Casey, H.W.", "O’Loughlin, C.K.", "Salamon, C.M.", "Smith, S.H.")))
.build();
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
DictionaryResponse addressResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(Collections.singletonList("Toxigenics, Inc., Decatur, IL 62526, USA")))
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
DictionaryResponse sponsorResponse = DictionaryResponse.builder()
.entries(Collections.emptyList())
.build();
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
Document classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
assertThat(classifiedDoc.getEntities().get(1)).hasSize(7);// 3 author cells, 1 address, 1 Y and 2 N entities
}
@@ -161,19 +165,19 @@ public class EntityRedactionServiceTest {
DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(Arrays.asList("Casey, H.W.", "O’Loughlin, C.K.", "Salamon, C.M.", "Smith, S.H.")))
.build();
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
DictionaryResponse addressResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(Collections.singletonList("Toxigenics, Inc., Decatur, IL 62526, USA")))
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
DictionaryResponse sponsorResponse = DictionaryResponse.builder()
.entries(Collections.emptyList())
.build();
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
Document classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
assertThat(classifiedDoc.getEntities().get(1)).hasSize(7);// 3 author cells, 1 address, 1 Y and 2 N entities
}
@@ -184,21 +188,21 @@ public class EntityRedactionServiceTest {
ClassPathResource pdfFileResource = new ClassPathResource("files/Cyprodinil/40 Cyprodinil - EU AIR3 - LCA Section 1" +
" Supplement - Identity of the active substance - Reference list.pdf");
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(new ArrayList<>(ResourceLoader.load("dictionaries/CBI_author.txt"))))
.build();
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
DictionaryResponse addressResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(new ArrayList<>(ResourceLoader.load("dictionaries/CBI_address.txt"))))
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
DictionaryResponse sponsorResponse = DictionaryResponse.builder()
.entries(Collections.emptyList())
.build();
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
Document classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()
.entrySet()
.stream()
@@ -206,7 +210,7 @@ public class EntityRedactionServiceTest {
pdfFileResource = new ClassPathResource("files/Compounds/27 A8637C - EU AIR3 - MCP Section 1 - Identity of " +
"the plant protection product.pdf");
classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()
.entrySet()
.stream()
@@ -217,21 +221,21 @@ public class EntityRedactionServiceTest {
public void testFalsePositiveInWrongCell() throws IOException {
ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Row With Ambiguous Redaction.pdf");
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(new ArrayList<>(ResourceLoader.load("dictionaries/CBI_author.txt"))))
.build();
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
DictionaryResponse addressResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(new ArrayList<>(ResourceLoader.load("dictionaries/CBI_address.txt"))))
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
DictionaryResponse sponsorResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(new ArrayList<>(ResourceLoader.load("dictionaries/CBI_sponsor.txt"))))
.build();
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
Document classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
assertThat(classifiedDoc.getEntities().get(1).stream()
.filter(entity -> entity.getMatchedRule() == 9)
@@ -279,26 +283,26 @@ public class EntityRedactionServiceTest {
"\"Reg (EC) No 1107/2009 Art. 63 (2g)\");\n" +
" section.redactBetween(\"Contact:\", \"Tel.:\", \"address\", 6,true, \"Applicant information was found\", \"Reg (EC) No 1107/2009 Art. 63 (2g)\");\n" +
" end";
when(rulesClient.getVersion(TEST_RULESET_ID)).thenReturn(RULES_VERSION.incrementAndGet());
when(rulesClient.getRules(TEST_RULESET_ID)).thenReturn(new RulesResponse(tableRules));
droolsExecutionService.updateRules(TEST_RULESET_ID);
when(rulesClient.getVersion(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(RULES_VERSION.incrementAndGet());
when(rulesClient.getRules(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(new RulesResponse(tableRules));
droolsExecutionService.updateRules(TEST_DOSSIER_TEMPLATE_ID);
ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Applicant Producer Table.pdf");
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(new ArrayList<>(ResourceLoader.load("dictionaries/CBI_author.txt"))))
.build();
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
DictionaryResponse addressResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(new ArrayList<>(ResourceLoader.load("dictionaries/CBI_address.txt"))))
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
DictionaryResponse sponsorResponse = DictionaryResponse.builder()
.entries(Collections.emptyList())
.build();
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
Document classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
assertThat(classifiedDoc.getEntities().get(1).stream()
.filter(entity -> entity.getMatchedRule() == 6)
@@ -318,26 +322,26 @@ public class EntityRedactionServiceTest {
"Section(searchText.toLowerCase().contains(\"batches produced at\"))\n" + " then\n" + " section" +
".redactIfPrecededBy(\"batches produced at\", \"sponsor\", 11, \"Redacted because it represents a " +
"sponsor company\", \"Reg (EC) No 1107/2009 Art. 63 (2g)\");\n" + " end";
when(rulesClient.getVersion(TEST_RULESET_ID)).thenReturn(RULES_VERSION.incrementAndGet());
when(rulesClient.getRules(TEST_RULESET_ID)).thenReturn(new RulesResponse(tableRules));
droolsExecutionService.updateRules(TEST_RULESET_ID);
when(rulesClient.getVersion(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(RULES_VERSION.incrementAndGet());
when(rulesClient.getRules(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(new RulesResponse(tableRules));
droolsExecutionService.updateRules(TEST_DOSSIER_TEMPLATE_ID);
ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/batches_new_line.pdf");
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
DictionaryResponse addressResponse = DictionaryResponse.builder()
.entries(Collections.emptyList())
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
DictionaryResponse authorResponse = DictionaryResponse.builder()
.entries(Collections.emptyList())
.build();
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(authorResponse);
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(authorResponse);
DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(new ArrayList<>(ResourceLoader.load("dictionaries/CBI_sponsor.txt"))))
.build();
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
Document classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
assertThat(classifiedDoc.getEntities().get(1).stream()
.filter(entity -> entity.getMatchedRule() == 11)
@@ -355,19 +359,19 @@ public class EntityRedactionServiceTest {
.entries(toDictionaryEntry(Arrays.asList("Bissig R.", "Thanei P.")))
.build();
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
DictionaryResponse addressResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(Collections.singletonList("Novartis Crop Protection AG, Basel, Switzerland")))
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
DictionaryResponse sponsorResponse = DictionaryResponse.builder()
.entries(Collections.emptyList())
.build();
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
Document classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()).hasSize(2); // two pages
assertThat(classifiedDoc.getEntities().get(1).stream().filter(entity -> entity.getMatchedRule() == 9).count()).isEqualTo(8);
assertThat(classifiedDoc.getEntities().get(2).stream().filter(entity -> entity.getMatchedRule() == 9).count()).isEqualTo(5); // 2 names, 1 address, 2 Y
@@ -378,15 +382,15 @@ public class EntityRedactionServiceTest {
.entries(toDictionaryEntry(Arrays.asList("Tribolet, R.", "Muir, G.", "Kühne-Thu, H.", "Close, C.")))
.build();
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
addressResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(Collections.singletonList("Novartis Crop Protection AG, Basel, Switzerland")))
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
assertThat(classifiedDoc.getEntities().get(1).stream().filter(entity -> entity.getMatchedRule() == 9).count()).isEqualTo(3);
assertThat(classifiedDoc.getEntities().get(1).stream().filter(entity -> entity.getMatchedRule() == 8).count()).isEqualTo(9);
@@ -403,19 +407,19 @@ public class EntityRedactionServiceTest {
.entries(toDictionaryEntry(Collections.singletonList("Aldershof S.")))
.build();
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
DictionaryResponse addressResponse = DictionaryResponse.builder()
.entries(toDictionaryEntry(Collections.singletonList("Novartis Crop Protection AG, Basel, Switzerland")))
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
DictionaryResponse sponsorResponse = DictionaryResponse.builder()
.entries(Collections.emptyList())
.build();
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
Document classifiedDoc = pdfSegmentationService.parseDocument(pdfFileResource.getInputStream());
entityRedactionService.processDocument(classifiedDoc, TEST_RULESET_ID, null, "dossierId");
entityRedactionService.processDocument(classifiedDoc, TEST_DOSSIER_TEMPLATE_ID, null, "dossierId");
assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
assertThat(classifiedDoc.getEntities().get(1).stream().filter(entity -> entity.getMatchedRule() == 8).count()).isEqualTo(6);
}
@@ -442,31 +446,31 @@ public class EntityRedactionServiceTest {
"\"Yes\"))\n" +
" then\n" +
" section.redactCell(\"Author(s)\", 9, \"name\", false, \"Redacted because row is a vertebrate study\", \"Reg (EC) No 1107/2009 Art. 63 (2g)\");\n" +
" section.redact(\"address\", 9, \"Redacted because row is a vertebrate study\", \"Reg (EC) No" +
" section.redact(\"address\", 9, \"Redacted because row is a vertebrate sgitudy\", \"Reg (EC) No" +
" 1107/2009 Art. 63 (2g)\");\n" +
" section.highlightCell(\"Vertebrate study Y/N\", 9, \"must_redact\");\n" +
" end";
when(rulesClient.getVersion(TEST_RULESET_ID)).thenReturn(RULES_VERSION.incrementAndGet());
when(rulesClient.getRules(TEST_RULESET_ID)).thenReturn(new RulesResponse(tableRules));
when(rulesClient.getVersion(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(RULES_VERSION.incrementAndGet());
when(rulesClient.getRules(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(new RulesResponse(tableRules));
TypeResponse typeResponse = TypeResponse.builder()
.types(Arrays.asList(
TypeResult.builder().ruleSetId(TEST_RULESET_ID).type(AUTHOR_CODE).hexColor("#ffff00").build(),
TypeResult.builder().ruleSetId(TEST_RULESET_ID).type(ADDRESS_CODE).hexColor("#ff00ff").build(),
TypeResult.builder().ruleSetId(TEST_RULESET_ID).type(SPONSOR_CODE).hexColor("#00ffff").build()))
TypeResult.builder().dossierTemplateId(TEST_DOSSIER_TEMPLATE_ID).type(AUTHOR_CODE).hexColor("#ffff00").build(),
TypeResult.builder().dossierTemplateId(TEST_DOSSIER_TEMPLATE_ID).type(ADDRESS_CODE).hexColor("#ff00ff").build(),
TypeResult.builder().dossierTemplateId(TEST_DOSSIER_TEMPLATE_ID).type(SPONSOR_CODE).hexColor("#00ffff").build()))
.build();
when(dictionaryClient.getVersion(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getAllTypes(TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(typeResponse);
when(dictionaryClient.getVersion(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(DICTIONARY_VERSION.incrementAndGet());
when(dictionaryClient.getAllTypes(TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(typeResponse);
// Default empty return to prevent NPEs
DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
.build();
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
when(dictionaryClient.getDictionaryForType(AUTHOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(dictionaryResponse);
DictionaryResponse addressResponse = DictionaryResponse.builder()
.build();
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
when(dictionaryClient.getDictionaryForType(ADDRESS_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(addressResponse);
DictionaryResponse sponsorResponse = DictionaryResponse.builder()
.build();
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_RULESET_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
when(dictionaryClient.getDictionaryForType(SPONSOR_CODE, TEST_DOSSIER_TEMPLATE_ID, DictionaryResource.GLOBAL_DOSSIER)).thenReturn(sponsorResponse);
Colors colors = new Colors();
colors.setDefaultColor("#acfc00");
@@ -474,7 +478,7 @@ public class EntityRedactionServiceTest {
colors.setRequestAdd("#04b093");
colors.setRequestRemove("#04b093");
when(dictionaryClient.getColors(TEST_RULESET_ID)).thenReturn(colors);
when(dictionaryClient.getColors(TEST_DOSSIER_TEMPLATE_ID)).thenReturn(colors);
}
@@ -5,6 +5,7 @@ import com.iqser.red.service.redaction.v1.server.Application;
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.service.BlockificationService;
import com.iqser.red.service.redaction.v1.server.client.LegalBasisClient;
import com.iqser.red.service.redaction.v1.server.redaction.model.PdfImage;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Cell;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table;
@@ -62,6 +63,9 @@ public class PdfSegmentationServiceTest {
@MockBean
private RabbitTemplate rabbitTemplate;
@MockBean
private LegalBasisClient legalBasisClient;
@Configuration
@EnableAutoConfiguration(exclude = { RabbitAutoConfiguration.class})
public static class TestConfiguration {
@@ -1652,3 +1652,5 @@ Zoecon Corp.
Zoecon Corp., Palo Alto, USA
Zyma SA
Zyma SA, Nyon, Switzerland
Mambo-Tox Ltd. Biomedical Sciences Building Bassett Crescent East Southampton SO16 7PX UK
Syngenta Environmental Sciences Jealott’s Hill International Research Centre Bracknell, Berkshire RG42 6EY UK