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

DM-165: Renamed isOnPage to onPage to prevent not working drools optimizations

Browse Source
This commit is contained in:
deiflaender 2023-08-07 13:19:47 +02:00
parent a9c02540c2
commit ea36f31df4
4 changed files with 9 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
redaction-service-v1/redaction-service-server-v1/build.gradle.kts
View File

@ -12,7 +12,7 @@ plugins {
description = "redaction-service-server-v1" description = "redaction-service-server-v1"
val layoutParserVersion = "0.18.0" val layoutParserVersion = "0.23.0"
val jacksonVersion = "2.15.2" val jacksonVersion = "2.15.2"
val droolsVersion = "8.42.0.Final" val droolsVersion = "8.42.0.Final"
val pdfBoxVersion = "3.0.0-alpha2" val pdfBoxVersion = "3.0.0-alpha2"

2
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/document/graph/nodes/SemanticNode.java
View File

@ -85,7 +85,7 @@ public interface SemanticNode {
} }
default boolean isOnPage(int pageNumber) { default boolean onPage(int pageNumber) {
return getPages().stream().anyMatch(page -> page.getNumber() == pageNumber); return getPages().stream().anyMatch(page -> page.getNumber() == pageNumber);
} }

4
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/adapter/CustomEntityCreationAdapter.java
View File

@ -143,10 +143,10 @@ public class CustomEntityCreationAdapter {
private Map<String, List<RedactionEntity>> findAllPossibleEntitiesAndGroupByValue(SemanticNode node, Set<Integer> pageNumbers, Set<String> entryValues) { private Map<String, List<RedactionEntity>> findAllPossibleEntitiesAndGroupByValue(SemanticNode node, Set<Integer> pageNumbers, Set<String> entryValues) {
if (!pageNumbers.stream().allMatch(node::isOnPage)) { if (!pageNumbers.stream().allMatch(node::onPage)) {
throw new IllegalArgumentException(format("SemanticNode \"%s\" does not contain these pages %s, it has pages: %s", throw new IllegalArgumentException(format("SemanticNode \"%s\" does not contain these pages %s, it has pages: %s",
node, node,
pageNumbers.stream().filter(pageNumber -> !node.isOnPage(pageNumber)).toList(), pageNumbers.stream().filter(pageNumber -> !node.onPage(pageNumber)).toList(),
node.getPages())); node.getPages()));
} }
SearchImplementation searchImplementation = new SearchImplementation(entryValues, true); SearchImplementation searchImplementation = new SearchImplementation(entryValues, true);

10
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/documine_flora.drl
View File

@ -63,7 +63,7 @@ query "getFileAttributes"
rule "H.0.0 retract table of contents page" rule "H.0.0 retract table of contents page"
when when
$page: Page(getMainBodyTextBlock().getSearchText().contains("........") || (getMainBodyTextBlock().getSearchText().contains("APPENDICES") && getMainBodyTextBlock().getSearchText().contains("TABLES"))) $page: Page(getMainBodyTextBlock().getSearchText().contains("........") || (getMainBodyTextBlock().getSearchText().contains("APPENDICES") && getMainBodyTextBlock().getSearchText().contains("TABLES")))
$node: SemanticNode(isOnPage($page.getNumber()), !isOnPage($page.getNumber() -1)) $node: SemanticNode(onPage($page.getNumber()), !onPage($page.getNumber() -1))
then then
retract($node); retract($node);
end end
@ -347,7 +347,7 @@ rule "DOC.5.0: Strain"
rule "DOC.7.0: study title by document structure" rule "DOC.7.0: study title by document structure"
when when
$table: Table(isOnPage(1), $table: Table(onPage(1),
(containsString("Final Report") || containsString("SPL")), (containsString("Final Report") || containsString("SPL")),
numberOfRows == 1, numberOfRows == 1,
numberOfCols == 1) numberOfCols == 1)
@ -361,7 +361,7 @@ rule "DOC.7.0: study title by document structure"
rule "DOC.7.1: study title" rule "DOC.7.1: study title"
when when
$table: Table(isOnPage(1), (containsString("Final Report") || containsString("SPL"))) $table: Table(onPage(1), (containsString("Final Report") || containsString("SPL")))
then then
entityCreationService.byRegexWithLineBreaksIgnoreCase("(?<=\\n)[\\w\\W]{1,300}(?=\\nFinal Report)", "title", EntityType.ENTITY, $table).findFirst().ifPresent(entity -> { entityCreationService.byRegexWithLineBreaksIgnoreCase("(?<=\\n)[\\w\\W]{1,300}(?=\\nFinal Report)", "title", EntityType.ENTITY, $table).findFirst().ifPresent(entity -> {
entity.apply("DOC.7.1", "Title found", "n-a"); entity.apply("DOC.7.1", "Title found", "n-a");
@ -371,8 +371,8 @@ rule "DOC.7.1: study title"
rule "DOC.7.2: study title" rule "DOC.7.2: study title"
when when
not Table(isOnPage(1), (containsString("Final Report") || containsString("SPL"))) not Table(onPage(1), (containsString("Final Report") || containsString("SPL")))
$section: Section(isOnPage(1), (containsString("Final Report") || containsString("SPL"))) $section: Section(onPage(1), (containsString("Final Report") || containsString("SPL")))
then then
entityCreationService.byRegexWithLineBreaksIgnoreCase("(?<=\\n)[\\w\\W]{1,300}(?=\\nFinal Report)", "title", EntityType.ENTITY, $section).findFirst().ifPresent(entity -> { entityCreationService.byRegexWithLineBreaksIgnoreCase("(?<=\\n)[\\w\\W]{1,300}(?=\\nFinal Report)", "title", EntityType.ENTITY, $section).findFirst().ifPresent(entity -> {
entity.apply("DOC.7.2", "Title found", "n-a"); entity.apply("DOC.7.2", "Title found", "n-a");