Marmelator/redaction-service
1
0
Fork 0
Code Issues Pull Requests 14 Actions Packages Projects Releases 622 Wiki Activity

Add hint annotation by reg-ex and expand to hint annotation by reg-ex

Browse Source
This commit is contained in:
lmaldacker 2021-02-10 14:17:26 +01:00 committed by Kresnadi Budisantoso
parent 6d9ed080ce
commit 0f263c69b8
4 changed files with 47 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
redaction-service-v1/pom.xml
View File

@ -5,7 +5,7 @@
<parent>
<artifactId>platform-dependency</artifactId>
<groupId>com.iqser.red</groupId>
<version>1.0.2</version>
<version>1.0.3</version>
</parent>
<modelVersion>4.0.0</modelVersion>

42
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/Section.java
View File

@ -140,6 +140,48 @@ public class Section {
}
public void expandToHintAnnotationByRegEx(String type, String pattern, boolean patternCaseInsensitive, int group, String asType) {
Pattern compiledPattern = Patterns.getCompiledPattern(pattern, patternCaseInsensitive);
Set<Entity> expanded = new HashSet<>();
for (Entity entity : entities) {
if (!entity.getType().equals(type) || entity.getTextAfter() == null) {
continue;
}
Matcher matcher = compiledPattern.matcher(entity.getTextAfter());
while (matcher.find()) {
String match = matcher.group(group);
if (StringUtils.isNotBlank(match)) {
expanded.addAll(findEntities(entity.getWord() + match, asType, false, false, 0, null, null));
}
}
}
EntitySearchUtils.addEntitiesWithHigherRank(entities, expanded, dictionary);
EntitySearchUtils.removeEntitiesContainedInLarger(entities);
}
public void addHintAnnotationByRegEx(String pattern, boolean patternCaseInsensitive, int group, String asType) {
Pattern compiledPattern = Patterns.getCompiledPattern(pattern, patternCaseInsensitive);
Matcher matcher = compiledPattern.matcher(searchText);
while (matcher.find()) {
String match = matcher.group(group);
if (StringUtils.isNotBlank(match)) {
Set<Entity> found = findEntities(match.trim(), asType, false, false, 0, null, null);
EntitySearchUtils.addEntitiesWithHigherRank(entities, found, dictionary);
}
}
}
public void redactIfPrecededBy(String prefix, String type, int ruleNumber, String reason, String legalBasis) {
entities.forEach(entity -> {

6
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/allAuthors.drl
View File

@ -316,9 +316,9 @@ rule "26: Redact On behalf of Syngenta Ltd.:"
// --------------------------------------- other rules -------------------------------------------------------------------
rule "27: Redact Purity"
rule "27: Purity Hint"
when
Section(searchText.contains("purity"))
Section(searchText.toLowerCase().contains("purity"))
then
section.redactByRegEx("purity ?:? (([\\d\\.]+)( .{0,4}\\.)? ?%)", true, 1, "purity", 27, "Purity found", "Reg (EC) No 1107/2009 Art. 63 (2a)");
section.addHintAnnotationByRegEx("(purity( of)?( ?:)?) [\\d\\.]+( .{0,4}\\.)? ?%", true, 1, "hint_only");
end

2
redaction-service-v1/redaction-service-server-v1/src/test/resources/log4j2-test.xml
View File

@ -10,7 +10,7 @@
<Root level="warn">
<AppenderRef ref="CONSOLE" />
</Root>
<Logger name="com.iqser.gin4" level="info" />
<Logger name="com.iqser" level="info" />
</Loggers>
</Configuration>