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

RED-6009 - Document Tree Structure

Browse Source 
* added more javadoc
This commit is contained in:
Kilian Schuettler 2023-05-09 14:01:19 +02:00
parent 088a5cde1a
commit e31a7b1a2f
8 changed files with 58 additions and 64 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

77
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/graph/nodes/SemanticNode.java
View File

@ -21,10 +21,10 @@ public interface SemanticNode {
/**
* Searches all Nodes located underneath this Node in the DocumentTree and concatenates their AtomicTextBlocks into a single TextBlockEntity.
* So, for a ClassificationSection all TextBlocks of Subsections, Paragraphs, and Tables are concatenated into a single TextBlockEntity
* So, for a Section all TextBlocks of Subsections, Paragraphs, and Tables are concatenated into a single TextBlockEntity
* If the Node is Terminal, the TerminalTextBlock will be returned instead.
*
* @return ClassificationTextBlock containing all AtomicTextBlocks that are located under this Node.
* @return TextBlock containing all AtomicTextBlocks that are located under this Node.
*/
TextBlock buildTextBlock();
@ -39,7 +39,7 @@ public interface SemanticNode {
/**
* Each AtomicTextBlock is assigned a page, so to get the pages this node appears on, it collects the PageNodes from each AtomicTextBlock belonging to this node's ClassificationTextBlock.
* Each AtomicTextBlock is assigned a page, so to get the pages this node appears on, it collects the PageNodes from each AtomicTextBlock belonging to this node's TextBlock.
*
* @return Set of PageNodes this node appears on.
*/
@ -50,7 +50,7 @@ public interface SemanticNode {
/**
* Each AtomicTextBlock is assigned a page, so to get the pages for this boundary, it collects the PageNodes from each AtomicTextBlock belonging to this node's ClassificationTextBlock.
* Each AtomicTextBlock is assigned a page, so to get the pages for this boundary, it collects the PageNodes from each AtomicTextBlock belonging to this node's TextBlock.
*
* @return Set of PageNodes this node appears on.
*/
@ -64,7 +64,9 @@ public interface SemanticNode {
/**
* @return the DocumentTree of the ClassificationDocument this node belongs to
* Returns the DocumentTree Object.
*
* @return the DocumentTree of the Document this node belongs to
*/
DocumentTree getDocumentTree();
@ -129,11 +131,11 @@ public interface SemanticNode {
/**
* Terminal means a SemanticNode has direct access to a ClassificationTextBlock, by default this is false and must be overridden.
* Terminal means a SemanticNode has direct access to a TextBlock, by default this is false and must be overridden.
* Currently only Sections, Images, and Tables are not terminal.
* A TableCell might be Terminal depending on its area compared to the page.
*
* @return boolean, indicating if a Node has direct access to a ClassificationTextBlock
* @return boolean, indicating if a Node has direct access to a TextBlock
*/
default boolean isTerminal() {
@ -142,7 +144,7 @@ public interface SemanticNode {
/**
* Terminal means a SemanticNode has direct access to a ClassificationTextBlock, by default this is false and must be overridden.
* Terminal means a SemanticNode has direct access to a TextBlock, by default this is false and must be overridden.
* Currently only Sections and Tables are not terminal.
*
* @return AtomicTextBlock
@ -153,24 +155,47 @@ public interface SemanticNode {
}
/**
* Should only be used during construction of the Graph. Sets the TerminalTextBlock of this SemanticNode.
*
* @param textBlock the TextBlock to set as the terminalTextBlock of this SemanticNode
*/
default void setTerminalTextBlock(TextBlock textBlock) {
throw new UnsupportedOperationException();
}
/**
* Checks whether this SemanticNode has any Entity of the provided type.
*
* @param type string representing the type of entity to check for
* @return true, if this SemanticNode has at least one Entity of the provided type
*/
default boolean hasEntitiesOfType(String type) {
return getEntities().stream().anyMatch(redactionEntity -> redactionEntity.getType().equals(type));
}
/**
* Returns a List of Entities in this SemanticNode which are of the provided type such as "CBI_author".
*
* @param type string representing the type of entities to return
* @return List of RedactionEntities of any the type
*/
default List<RedactionEntity> getEntitiesOfType(String type) {
return getEntities().stream().filter(redactionEntity -> redactionEntity.getType().equals(type)).toList();
}
/**
* Returns a List of Entities in this SemanticNode which have any of the provided types such as "CBI_author".
*
* @param types A list of strings representing the types of entities to return
* @return List of RedactionEntities of any provided type
*/
default List<RedactionEntity> getEntitiesOfType(List<String> types) {
return getEntities().stream().filter(redactionEntity -> redactionEntity.isAnyType(types)).toList();
@ -195,7 +220,9 @@ public interface SemanticNode {
/**
* @return true, if this node's ClassificationTextBlock is not empty
* Checks if the SemanticNode contains any text.
*
* @return true, if this node's TextBlock is not empty
*/
default boolean hasText() {
@ -204,6 +231,8 @@ public interface SemanticNode {
/**
* Checks whether this SemanticNode contains the provided String.
*
* @param string A String which the TextBlock might contain
* @return true, if this node's TextBlock contains the string
*/
@ -214,6 +243,8 @@ public interface SemanticNode {
/**
* Checks whether this SemanticNode contains all the provided Strings.
*
* @param strings A List of Strings which the TextBlock might contain
* @return true, if this node's TextBlock contains all strings
*/
@ -224,8 +255,10 @@ public interface SemanticNode {
/**
* @param string A String which the ClassificationTextBlock might contain
* @return true, if this node's ClassificationTextBlock contains the string
* Checks whether this SemanticNode contains all the provided Strings ignoring case.
*
* @param string A String which the TextBlock might contain
* @return true, if this node's TextBlock contains the string ignoring case
*/
default boolean containsStringIgnoreCase(String string) {
@ -234,8 +267,10 @@ public interface SemanticNode {
/**
* @param strings A List of Strings which the ClassificationTextBlock might contain
* @return true, if this node's ClassificationTextBlock contains any of the strings
* Checks whether this SemanticNode contains any of the provided Strings.
*
* @param strings A List of Strings which the TextBlock might contain
* @return true, if this node's TextBlock contains any of the strings
*/
default boolean containsAnyString(List<String> strings) {
@ -244,8 +279,10 @@ public interface SemanticNode {
/**
* @param strings A List of Strings which the ClassificationTextBlock might contain
* @return true, if this node's ClassificationTextBlock contains any of the strings
* Checks whether this SemanticNode contains any of the provided Strings ignoring case.
*
* @param strings A List of Strings which the TextBlock might contain
* @return true, if this node's TextBlock contains any of the strings
*/
default boolean containsAnyStringIgnoreCase(List<String> strings) {
@ -286,7 +323,7 @@ public interface SemanticNode {
/**
* recursively streams all SemanticNodes located underneath this node in the DocumentTree in order.
* Recursively streams all SemanticNodes located underneath this node in the DocumentTree in order.
*
* @return Stream of all SubNodes
*/
@ -297,7 +334,9 @@ public interface SemanticNode {
/**
* @return Boundary of this Node's ClassificationTextBlock
* The Boundary is the start and end string offsets in the reading order of the document.
*
* @return Boundary of this Node's TextBlock
*/
default Boundary getBoundary() {
@ -307,7 +346,7 @@ public interface SemanticNode {
/**
* If this Node is Terminal it will calculate the boundingBox of its TerminalTextBlock, otherwise it will calculate the Union of the BoundingBoxes of all its Children.
* If called on the ClassificationDocument, it will return the cropbox of each page
* If called on the Document, it will return the cropbox of each page
*
* @return Rectangle2D fully encapsulating this Node for each page.
*/
@ -356,7 +395,7 @@ public interface SemanticNode {
/**
* @param bBoxPerPage initial empty BoundingBox
* @return The union of all BoundingBoxes of the ClassificationTextBlock of this node
* @return The union of all BoundingBoxes of the TextBlock of this node
*/
private Map<Page, Rectangle2D> getBBoxFromTerminalTextBlock(Map<Page, Rectangle2D> bBoxPerPage) {

23
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/conditions/SemanticNodeRuleConditions.java
View File

@ -1,23 +0,0 @@
package com.iqser.red.service.redaction.v1.server.redaction.conditions;
import java.util.List;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.nodes.SemanticNode;
import lombok.experimental.UtilityClass;
@UtilityClass
public class SemanticNodeRuleConditions {
public boolean hasEntitiesOfType(SemanticNode semanticNode, String type) {
return semanticNode.getEntities().stream().anyMatch(redactionEntity -> redactionEntity.getType().equals(type));
}
public boolean hasEntitiesOfType(SemanticNode semanticNode, List<String> types) {
return types.stream().allMatch(type -> hasEntitiesOfType(semanticNode, type));
}
}

17
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/utils/Liszt.java
View File

@ -1,17 +0,0 @@
package com.iqser.red.service.redaction.v1.server.redaction.utils;
import java.util.List;
/**
* Drools fucked up and introduced a bug with newer versions of java, use this class instead of java.util.List.of(e1, e2) or else.
* java.lang.IncompatibleClassChangeError: Method 'java.util.List java.util.List.of(java.lang.Object, java.lang.Object)' must be InterfaceMethodref constant
* Will be thrown on LHS of rules, but only sometimes, I was not able to recreate this bug in a different project.
*/
public class Liszt {
public static <E> List<E> of(E e1, E e2) {
return List.of(e1, e2);
}
}

1
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/manual_redaction_rules.drl
View File

@ -6,7 +6,6 @@ import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.u
import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.data.mapper.PropertiesMapper.parseImageType;
import java.util.List;
import com.iqser.red.service.redaction.v1.server.redaction.utils.Liszt;
import java.util.LinkedList;
import java.util.HashSet;

1
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/prod_syngenta_new.drl
View File

@ -6,7 +6,6 @@ import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.u
import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.data.mapper.PropertiesMapper.parseImageType;
import java.util.List;
import com.iqser.red.service.redaction.v1.server.redaction.utils.Liszt;
import java.util.LinkedList;
import java.util.HashSet;

1
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/rules.drl
View File

@ -6,7 +6,6 @@ import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.u
import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.data.mapper.PropertiesMapper.parseImageType;
import java.util.List;
import com.iqser.red.service.redaction.v1.server.redaction.utils.Liszt;
import java.util.LinkedList;
import java.util.HashSet;

1
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/rules_v2.drl
View File

@ -6,7 +6,6 @@ import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.u
import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.data.mapper.PropertiesMapper.parseImageType;
import java.util.List;
import com.iqser.red.service.redaction.v1.server.redaction.utils.Liszt;
import java.util.LinkedList;
import java.util.HashSet;

1
redaction-service-v1/redaction-service-server-v1/src/test/resources/performance/dictionaries/EFSA_sanitisation_GFL_v1/rules.drl
View File

@ -6,7 +6,6 @@ import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.u
import static com.iqser.red.service.redaction.v1.server.layoutparsing.document.data.mapper.PropertiesMapper.parseImageType;
import java.util.List;
import com.iqser.red.service.redaction.v1.server.redaction.utils.Liszt;
import java.util.LinkedList;
import java.util.HashSet;