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
* fixed build
This commit is contained in:
Kilian Schuettler 2023-05-09 16:50:52 +02:00
parent c151b4a94b
commit 0443e21858
4 changed files with 190 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
redaction-service-v1/redaction-service-api-v1/pom.xml
View File

@ -40,4 +40,22 @@
</exclusions> </exclusions>
</dependency> </dependency>
</dependencies> </dependencies>
<build>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-compiler-plugin</artifactId>
<configuration>
<annotationProcessorPaths>
<path>
<groupId>org.projectlombok</groupId>
<artifactId>lombok</artifactId>
<version>${lombok.version}</version>
</path>
</annotationProcessorPaths>
</configuration>
</plugin>
</plugins>
</build>
</project> </project>

71
redaction-service-v1/redaction-service-server-v1/pom.xml
View File

@ -132,4 +132,75 @@
<scope>test</scope> <scope>test</scope>
</dependency> </dependency>
</dependencies> </dependencies>
<build>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-compiler-plugin</artifactId>
<configuration>
<annotationProcessorPaths>
<path>
<groupId>org.projectlombok</groupId>
<artifactId>lombok</artifactId>
<version>${lombok.version}</version>
</path>
</annotationProcessorPaths>
</configuration>
</plugin>
<plugin>
<!-- generate git.properties for exposure in /info -->
<groupId>pl.project13.maven</groupId>
<artifactId>git-commit-id-plugin</artifactId>
<executions>
<execution>
<goals>
<goal>revision</goal>
</goals>
<configuration>
<generateGitPropertiesFile>true</generateGitPropertiesFile>
<gitDescribe>
<tags>true</tags>
</gitDescribe>
</configuration>
</execution>
</executions>
</plugin>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-jar-plugin</artifactId>
<executions>
<execution>
<id>original-jar</id>
<goals>
<goal>jar</goal>
</goals>
<configuration>
<classifier>original</classifier>
</configuration>
</execution>
</executions>
</plugin>
<plugin>
<!-- repackages the generated jar into a runnable fat-jar and makes it
executable -->
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-maven-plugin</artifactId>
<executions>
<execution>
<goals>
<goal>repackage</goal>
</goals>
<configuration>
<executable>true</executable>
</configuration>
</execution>
</executions>
</plugin>
</plugins>
</build>
</project> </project>

101
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/graph/nodes/Table.java
View File

@ -40,6 +40,12 @@ public class Table implements SemanticNode {
Set<RedactionEntity> entities = new HashSet<>(); Set<RedactionEntity> entities = new HashSet<>();
/**
* Streams all entities in this table, that appear in a row, which contains any of the provided strings.
*
* @param strings Strings to check whether a row contains them
* @return Stream of all entities in this table, that appear in a row, which contains any of the provided strings
*/
public Stream<RedactionEntity> streamEntitiesWhereRowContainsStringsIgnoreCase(List<String> strings) { public Stream<RedactionEntity> streamEntitiesWhereRowContainsStringsIgnoreCase(List<String> strings) {
return IntStream.range(0, numberOfRows) return IntStream.range(0, numberOfRows)
@ -51,6 +57,13 @@ public class Table implements SemanticNode {
} }
/**
* Checks whether the specified row contains all the provided strings.
*
* @param row the row to check as an Integer, must be smaller than numberOfRows
* @param strings a list of strings to check for
* @return true, if all strings appear in the provided row
*/
public boolean rowContainsStringsIgnoreCase(Integer row, List<String> strings) { public boolean rowContainsStringsIgnoreCase(Integer row, List<String> strings) {
String rowText = streamRow(row).map(TableCell::buildTextBlock).collect(new TextBlockCollector()).getSearchText().toLowerCase(); String rowText = streamRow(row).map(TableCell::buildTextBlock).collect(new TextBlockCollector()).getSearchText().toLowerCase();
@ -58,6 +71,13 @@ public class Table implements SemanticNode {
} }
/**
* Streams all entities which appear in a row where at least one cell has the provided header and the provided value.
*
* @param header the header value to search for
* @param value the string which the table cell should contain
* @return a stream of all entities, which appear in a row where at least one cell has the provided header and the provided value.
*/
public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndValue(String header, String value) { public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndValue(String header, String value) {
List<Integer> vertebrateStudyCols = streamHeaders().filter(headerNode -> headerNode.containsString(header)).map(TableCell::getCol).toList(); List<Integer> vertebrateStudyCols = streamHeaders().filter(headerNode -> headerNode.containsString(header)).map(TableCell::getCol).toList();
@ -66,6 +86,13 @@ public class Table implements SemanticNode {
} }
/**
* Streams all entities which appear in a row where at least one cell has the provided header and any provided value.
*
* @param header the header value to search for
* @param values the strings which the table cell should contain
* @return a stream of all entities, which appear in a row where at least one cell has the provided header and any provided value.
*/
public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndAnyValue(String header, List<String> values) { public Stream<RedactionEntity> streamEntitiesWhereRowHasHeaderAndAnyValue(String header, List<String> values) {
List<Integer> vertebrateStudyCols = streamHeaders().filter(headerNode -> headerNode.containsString(header)).map(TableCell::getCol).toList(); List<Integer> vertebrateStudyCols = streamHeaders().filter(headerNode -> headerNode.containsString(header)).map(TableCell::getCol).toList();
@ -76,6 +103,12 @@ public class Table implements SemanticNode {
} }
/**
* Streams all entities in this table, that appear in a row, which contains at least one entity with any of the provided types.
*
* @param types type strings to check whether a row contains an entity like them
* @return Stream of all entities in this table, that appear in a row, which contains at least one entity with any of the provided types.
*/
public Stream<RedactionEntity> streamEntitiesWhereRowContainsEntitiesOfType(List<String> types) { public Stream<RedactionEntity> streamEntitiesWhereRowContainsEntitiesOfType(List<String> types) {
List<Integer> rowsWithEntityOfType = getEntities().stream() List<Integer> rowsWithEntityOfType = getEntities().stream()
@ -90,6 +123,13 @@ public class Table implements SemanticNode {
} }
/**
* Returns a TableCell at the provided row and column location.
*
* @param row int representing the row, must be smaller than numberOfRows
* @param col int representing the col, must be smaller than numberOfCols
* @return TableCell at the provided location in the table
*/
public TableCell getCell(int row, int col) { public TableCell getCell(int row, int col) {
if (numberOfRows - row < 0 || numberOfCols - col < 0) { if (numberOfRows - row < 0 || numberOfCols - col < 0) {
@ -100,12 +140,22 @@ public class Table implements SemanticNode {
} }
/**
* Streams all TableCells in this Table row-wise.
*
* @return Stream of all TableCells
*/
public Stream<TableCell> streamTableCells() { public Stream<TableCell> streamTableCells() {
return streamChildren().map(node -> (TableCell) node); return streamChildren().map(node -> (TableCell) node);
} }
/**
* Streams all TableCells in this Table which have the provided header row-wise.
*
* @return Stream of all TableCells which have the provided header
*/
public Stream<TableCell> streamTableCellsWithHeader(String header) { public Stream<TableCell> streamTableCellsWithHeader(String header) {
return streamHeaders().filter(tableCellNode -> tableCellNode.buildTextBlock().getSearchText().contains(header)) return streamHeaders().filter(tableCellNode -> tableCellNode.buildTextBlock().getSearchText().contains(header))
@ -115,48 +165,99 @@ public class Table implements SemanticNode {
} }
/**
* Streams all TableCells belonging to the provided column from top down.
*
* @param col int representing the column
* @return Stream of all TableCell in the provided column
*/
public Stream<TableCell> streamCol(int col) { public Stream<TableCell> streamCol(int col) {
return IntStream.range(0, numberOfRows).boxed().map(row -> getCell(row, col)); return IntStream.range(0, numberOfRows).boxed().map(row -> getCell(row, col));
} }
/**
* Streams all TableCells belonging to the provided row from left to right.
*
* @param row int representing the row
* @return Stream of all TableCell in the provided row
*/
public Stream<TableCell> streamRow(int row) { public Stream<TableCell> streamRow(int row) {
return IntStream.range(0, numberOfCols).boxed().map(col -> getCell(row, col)); return IntStream.range(0, numberOfCols).boxed().map(col -> getCell(row, col));
} }
/**
* Streams all TableCells row-wise and filters them with header == true.
*
* @return Stream of all TableCells with header == true
*/
public Stream<TableCell> streamHeaders() { public Stream<TableCell> streamHeaders() {
return streamTableCells().filter(TableCell::isHeader); return streamTableCells().filter(TableCell::isHeader);
} }
/**
* Streams all TableCells of the provided row and column and filters them with header == true.
*
* @param row int representing the row
* @param col int representing the column
* @return Stream of all TableCells with header == true in the provided row or col
*/
public Stream<TableCell> streamHeadersForCell(int row, int col) { public Stream<TableCell> streamHeadersForCell(int row, int col) {
return Stream.concat(streamRow(row), streamCol(col)).filter(TableCell::isHeader); return Stream.concat(streamRow(row), streamCol(col)).filter(TableCell::isHeader);
} }
/**
* Streams all Headers and checks if any equal the provided string.
*
* @param header string to check the headers for
* @return true, if at least one header equals the provided string
*/
public boolean hasHeader(String header) { public boolean hasHeader(String header) {
return streamHeaders().anyMatch(tableCellNode -> tableCellNode.buildTextBlock().getSearchText().strip().equals(header)); return streamHeaders().anyMatch(tableCellNode -> tableCellNode.buildTextBlock().getSearchText().strip().equals(header));
} }
/**
* Checks if this table has a column with the provided header and any of the table cells in that column contain the provided value.
*
* @param header string to find header cells
* @param value string to check cells with provided header
* @return true, if this table has a column with the provided header and any of the table cells in that column contain the provided value
*/
public boolean hasRowWithHeaderAndValue(String header, String value) { public boolean hasRowWithHeaderAndValue(String header, String value) {
return streamTableCellsWithHeader(header).anyMatch(tableCellNode -> tableCellNode.containsString(value)); return streamTableCellsWithHeader(header).anyMatch(tableCellNode -> tableCellNode.containsString(value));
} }
/**
* Checks if this table has a column with the provided header and any of the table cells in that column contains any of the provided values.
*
* @param header string to find header cells
* @param values List of strings to check cells with provided header
* @return true, if this table has a column with the provided header and any of the table cells in that column contains any of the provided values.
*/
public boolean hasRowWithHeaderAndAnyValue(String header, List<String> values) { public boolean hasRowWithHeaderAndAnyValue(String header, List<String> values) {
return streamTableCellsWithHeader(header).anyMatch(tableCellNode -> tableCellNode.containsAnyString(values)); return streamTableCellsWithHeader(header).anyMatch(tableCellNode -> tableCellNode.containsAnyString(values));
} }
/**
* Finds all entities of the provided type, which appear in the same row that the provided entity appears in.
*
* @param type the type of entities to search for
* @param redactionEntity the entity, which appears in the row to search
* @return List of all entities of the provided type, which appear in the same row that the provided entity appears in.
*/
public List<RedactionEntity> getEntitiesOfTypeInSameRow(String type, RedactionEntity redactionEntity) { public List<RedactionEntity> getEntitiesOfTypeInSameRow(String type, RedactionEntity redactionEntity) {
return redactionEntity.getIntersectingNodes() return redactionEntity.getIntersectingNodes()

14
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/layoutparsing/document/graph/nodes/TableCell.java
View File

@ -6,7 +6,6 @@ import java.util.HashSet;
import java.util.List; import java.util.List;
import java.util.Map; import java.util.Map;
import java.util.Set; import java.util.Set;
import java.util.stream.Stream;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.DocumentTree; import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.DocumentTree;
import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.entity.RedactionEntity; import com.iqser.red.service.redaction.v1.server.layoutparsing.document.graph.entity.RedactionEntity;
@ -76,17 +75,4 @@ public class TableCell implements SemanticNode {
return treeId + ": " + NodeType.TABLE_CELL + ": " + buildTextBlock().buildSummary(); return treeId + ": " + NodeType.TABLE_CELL + ": " + buildTextBlock().buildSummary();
} }
public boolean hasHeader(String headerString) {
return getHeaders().anyMatch(header -> header.buildTextBlock().getSearchText().strip().equals(headerString));
}
private Stream<TableCell> getHeaders() {
Table table = (Table) getParent();
return table.streamHeadersForCell(row, col);
}
} }