Compare commits

..
Author SHA1 Message Date
Dominique Eifländer 21255e9854 Test old pattern 2021-10-14 15:09:53 +02:00
7 changed files with 12 additions and 25 deletions
@@ -21,7 +21,6 @@ public class RedTextPosition {
private float YDirAdj;
private float width;
private float heightDir;
private float widthDirAdj;
// not used in reanalysis
@JsonIgnore
@@ -1,7 +1,5 @@
package com.iqser.red.service.redaction.v1.server.parsing.model;
import static java.util.stream.Collectors.toSet;
import com.fasterxml.jackson.annotation.JsonIgnore;
import com.fasterxml.jackson.annotation.JsonIgnoreProperties;
import com.iqser.red.service.redaction.v1.model.Point;
@@ -246,15 +244,7 @@ public class TextPositionSequence implements CharSequence {
posXInit = textPositions.get(0).getPageWidth() - textPositions.get(0).getYDirAdj() - 2;
posXEnd = textPositions.get(0).getPageWidth() - textPositions.get(textPositions.size() - 1)
.getYDirAdj() + height;
} else if(textPositions.get(0).getRotation() == 0 && textPositions.stream().map(t -> t.getY()).collect(toSet()).size() > 1) {
posYInit = textPositions.get(0).getPageHeight() - getX1();
posYEnd = textPositions.get(0).getPageHeight() - getX2() - textPositions.get(0)
.getWidthDirAdj() - textPositions.get(textPositions.size() - 1).getWidthDirAdj() - 3;
posXInit = textPositions.get(0).getPageWidth() - textPositions.get(0).getYDirAdj() - 2;
posXEnd = textPositions.get(0).getPageWidth() - textPositions.get(textPositions.size() - 1)
.getYDirAdj() + height;
}
else {
} else {
posXEnd = textPositions.get(textPositions.size() - 1)
.getXDirAdj() + textPositions.get(textPositions.size() - 1).getWidth() + 1;
posYInit = textPositions.get(0).getPageHeight() - textPositions.get(0).getYDirAdj() - 2;
@@ -65,13 +65,13 @@ public class Dictionary {
public boolean containsValue(String type, String value) {
return localAccessMap.containsKey(type) && localAccessMap.get(type)
.getValues(false)
.getEntries().stream().map(DictionaryEntry::getValue).collect(toSet())
.contains(value) || localAccessMap.containsKey(type) && localAccessMap.get(type)
.getValues(true)
.getLocalEntries()
.contains(value) || localAccessMap.containsKey(RECOMMENDATION_PREFIX + type) && localAccessMap.get(RECOMMENDATION_PREFIX + type)
.getValues(false)
.getEntries().stream().map(DictionaryEntry::getValue).collect(toSet())
.contains(value) || localAccessMap.containsKey(RECOMMENDATION_PREFIX + type) && localAccessMap.get(RECOMMENDATION_PREFIX + type)
.getValues(true)
.getLocalEntries()
.contains(value);
}
@@ -209,10 +209,10 @@ public class EntityRedactionService {
for (DictionaryModel model : dictionary.getDictionaryModels()) {
if (model.isCaseInsensitive()) {
EntitySearchUtils.addOrAddEngine(found, EntitySearchUtils.find(lowercaseInputString, model.getValues(local), model
.getType(), headline, sectionNumber, !local, model.isDossierDictionary(), local ? Engine.RULE : Engine.DICTIONARY));
.getType(), headline, sectionNumber, !local, model.isDossierDictionary(), Engine.DICTIONARY));
} else {
EntitySearchUtils.addOrAddEngine(found, EntitySearchUtils.find(searchableString, model.getValues(local), model
.getType(), headline, sectionNumber, !local, model.isDossierDictionary(), local ? Engine.RULE : Engine.DICTIONARY));
.getType(), headline, sectionNumber, !local, model.isDossierDictionary(), Engine.DICTIONARY));
}
}
@@ -146,14 +146,11 @@ public class EntitySearchUtils {
}
var existing = existingOptional.get();
log.info("existing: {}", existing);
log.info("found: {} ", found);
if(existing.getType().equals(found.getType())){
existing.getEngines().addAll(found.getEngines());
} else if (dictionary.getDictionaryRank(existing.getType()) <= dictionary.getDictionaryRank(found.getType())){
found.getEngines().addAll(existing.getEngines());
if (dictionary.getDictionaryRank(existing.getType()) <= dictionary.getDictionaryRank(found.getType())) {
entities.remove(found);
entities.add(found);
} else {
existing.getEngines().addAll(found.getEngines());
}
} else {
entities.add(found);
@@ -11,7 +11,8 @@ public class Patterns {
public static Map<String, Pattern> patternCache = new HashMap<>();
public static Pattern AUTHOR_TABLE_SPITTER = Pattern.compile("(((((di)|(van)) )|[A-Z]’)?[A-ZÄÖÜ][\\wäöüéèê]{2,500}( ?[A-ZÄÖÜ]{1,2}\\.){1,3})|(((((di)|(van)) )|[A-Z]’)?[A-ZÄÖÜ][\\wäöüéèê]{2,500}( ?[A-ZÄÖÜ]{1,2} ){1,3})");
public static Pattern AUTHOR_TABLE_SPITTER = Pattern.compile("((((di)|(van)) )|[A-Z]’)?[A-ZÄÖÜ][\\wäöüéèê]{2,}( ?[A-ZÄÖÜ]{1,2}\\.)+|((((di)|(van)) )|[A-Z]’)?[A-ZÄÖÜ][\\wäöüéèê]{2,}( ?[A-ZÄÖÜ]{1,2} )+");
public Pattern getCompiledPattern(String pattern, boolean caseInsensitive) {