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

Fix style

Browse Source 
Fix style.

Fix style.

Fix style and naming

Fix style, naming and field modifier

Fix style and remove warning suppression
This commit is contained in:
Thierry Göckel 2020-08-10 12:04:36 +02:00
parent 81048dcc9f
commit 695564d162
7 changed files with 75 additions and 59 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Paragraph.java
View File

@ -10,7 +10,6 @@ import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table;
import lombok.Data;
import lombok.NoArgsConstructor;
@Data
@NoArgsConstructor
public class Paragraph {
@ -18,10 +17,12 @@ public class Paragraph {
private List<AbstractTextContainer> pageBlocks = new ArrayList<>();
private String headline;
public SearchableText getSearchableText(){
public SearchableText getSearchableText() {
SearchableText searchableText = new SearchableText();
pageBlocks.forEach(block -> {
if(block instanceof TextBlock){
if (block instanceof TextBlock) {
searchableText.addAll(((TextBlock) block).getSequences());
}
});
@ -29,14 +30,15 @@ public class Paragraph {
}
public List<Table> getTables(){
public List<Table> getTables() {
List<Table> tables = new ArrayList<>();
pageBlocks.forEach(block -> {
if(block instanceof Table){
if (block instanceof Table) {
tables.add((Table) block);
}
});
return tables;
}
}
}

34
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/StringFrequencyCounter.java
View File

@ -5,43 +5,45 @@ import java.util.Map;
import lombok.Getter;
/**
*
*/
public class StringFrequencyCounter {
@Getter
Map<String, Integer> countPerValue = new HashMap<>();
private final Map<String, Integer> countPerValue = new HashMap<>();
public void add(String value){
if(!countPerValue.containsKey(value)){
public void add(String value) {
if (!countPerValue.containsKey(value)) {
countPerValue.put(value, 1);
} else {
countPerValue.put(value, countPerValue.get(value) + 1);
}
}
public void addAll(Map<String, Integer> otherCounter){
for(Map.Entry<String, Integer> entry: otherCounter.entrySet()){
if(countPerValue.containsKey(entry.getKey())){
countPerValue.put(entry.getKey(), countPerValue.get(entry.getKey())+ entry.getValue());
public void addAll(Map<String, Integer> otherCounter) {
for (Map.Entry<String, Integer> entry : otherCounter.entrySet()) {
if (countPerValue.containsKey(entry.getKey())) {
countPerValue.put(entry.getKey(), countPerValue.get(entry.getKey()) + entry.getValue());
} else {
countPerValue.put(entry.getKey(), entry.getValue());
}
}
}
public String getMostPopular(){
public String getMostPopular() {
Map.Entry<String, Integer> mostPopular = null;
for(Map.Entry<String, Integer> entry: countPerValue.entrySet()){
if(mostPopular == null){
for (Map.Entry<String, Integer> entry : countPerValue.entrySet()) {
if (mostPopular == null) {
mostPopular = entry;
} else if(entry.getValue() > mostPopular.getValue()){
} else if (entry.getValue() > mostPopular.getValue()) {
mostPopular = entry;
}
}
return mostPopular != null ? mostPopular.getKey() : null;
}
}
}

32
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/BlockificationService.java
View File

@ -29,20 +29,16 @@ public class BlockificationService {
float minX = 1000, maxX = 0, minY = 1000, maxY = 0;
TextPositionSequence prev = null;
for (TextPositionSequence word : textPositions) {
boolean lineSeparation = minY - word.getY2() > word.getHeight() * 1.25;
boolean startFromTop = word.getY1() > maxY + word.getHeight();
if (prev != null &&
(lineSeparation
|| startFromTop
|| word.getRotation() == 0 && isSplittedByRuling(maxX, minY, word.getX1(), word.getY1(), verticalRulingLines)
|| word.getRotation() == 0 && isSplittedByRuling(minX, minY, word.getX1(), word.getY2(), horizontalRulingLines)
|| word.getRotation() == 90 && isSplittedByRuling(maxX, minY, word.getX1(), word.getY1(), horizontalRulingLines)
|| word.getRotation() == 90 && isSplittedByRuling(minX, minY, word.getX1(), word.getY2(), verticalRulingLines)
)) {
if (prev != null && (lineSeparation || startFromTop || word.getRotation() == 0 && isSplittedByRuling(maxX, minY, word
.getX1(), word.getY1(), verticalRulingLines) || word.getRotation() == 0 && isSplittedByRuling(minX, minY, word
.getX1(), word.getY2(), horizontalRulingLines) || word.getRotation() == 90 && isSplittedByRuling(maxX, minY, word
.getX1(), word.getY1(), horizontalRulingLines) || word.getRotation() == 90 && isSplittedByRuling(minX, minY, word
.getX1(), word.getY2(), verticalRulingLines))) {
TextBlock cb1 = buildTextBlock(chunkWords);
chunkBlockList1.add(cb1);
@ -100,11 +96,12 @@ public class BlockificationService {
styleFrequencyCounter.add(wordBlock.getFontStyle());
if (textBlock == null) {
textBlock = new TextBlock(wordBlock.getX1(), wordBlock.getX2(), wordBlock.getY1(), wordBlock.getY2(), wordBlockList, wordBlock.getRotation());
textBlock = new TextBlock(wordBlock.getX1(), wordBlock.getX2(), wordBlock.getY1(), wordBlock.getY2(), wordBlockList, wordBlock
.getRotation());
} else {
TextBlock spatialEntity = textBlock.union(wordBlock);
textBlock.resize(spatialEntity.getMinX(), spatialEntity.getMinY(),
spatialEntity.getWidth(), spatialEntity.getHeight());
textBlock.resize(spatialEntity.getMinX(), spatialEntity.getMinY(), spatialEntity.getWidth(), spatialEntity
.getHeight());
}
}
@ -122,6 +119,7 @@ public class BlockificationService {
private boolean isSplittedByRuling(float previousX2, float previousY1, float currentX1, float currentY1, List<Ruling> rulingLines) {
for (Ruling ruling : rulingLines) {
if (ruling.intersectsLine(previousX2, previousY1, currentX1, currentY1)) {
return true;
@ -133,7 +131,6 @@ public class BlockificationService {
public Rectangle calculateBodyTextFrame(List<Page> pages, FloatFrequencyCounter documentFontSizeCounter, boolean landscape) {
float minX = 10000;
float maxX = -100;
float minY = 10000;
@ -147,7 +144,6 @@ public class BlockificationService {
for (AbstractTextContainer container : page.getTextBlocks()) {
if (container instanceof TextBlock) {
TextBlock textBlock = (TextBlock) container;
if (textBlock.getMostPopularWordFont() == null || textBlock.getMostPopularWordStyle() == null) {
@ -179,16 +175,15 @@ public class BlockificationService {
}
}
if (container instanceof Table) {
Table table = (Table) container;
for (List<Cell> row : table.getRows()) {
for (Cell column : row) {
for (Cell cell : row) {
if (column == null || column.getTextBlocks() == null) {
if (cell == null || cell.getTextBlocks() == null) {
continue;
}
for (TextBlock textBlock : column.getTextBlocks()) {
for (TextBlock textBlock : cell.getTextBlocks()) {
if (textBlock.getMinX() < minX) {
minX = textBlock.getMinX();
}
@ -211,5 +206,4 @@ public class BlockificationService {
return new Rectangle(minY, minX, maxX - minX, maxY - minY);
}
}

18
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/SearchableText.java
View File

@ -4,14 +4,14 @@ import java.util.ArrayList;
import java.util.List;
import java.util.UUID;
import java.util.regex.Pattern;
import java.util.stream.Collectors;
import com.iqser.red.service.redaction.v1.server.parsing.model.TextPositionSequence;
import com.iqser.red.service.redaction.v1.server.redaction.utils.TextNormalizationUtilities;
@SuppressWarnings("all")
public class SearchableText {
private List<TextPositionSequence> sequences = new ArrayList<>();
private final List<TextPositionSequence> sequences = new ArrayList<>();
public void add(TextPositionSequence textPositionSequence) {
@ -64,9 +64,9 @@ public class SearchableText {
.charAt(j, caseInsensitive) == '-') {
if (counter != 0 || i == 0 && j == 0 || j != 0 && isSeparator(sequences.get(i)
.charAt(j - 1, caseInsensitive)) || j == 0 && i != 0 && isSeparator(sequences.get(i - 1)
.charAt(j - 1, caseInsensitive)) || j == 0 && isSeparator(sequences.get(i - 1)
.charAt(sequences.get(i - 1)
.length() - 1, caseInsensitive)) || j == 0 && i != 0 && sequences.get(i - 1)
.length() - 1, caseInsensitive)) || j == 0 && sequences.get(i - 1)
.charAt(sequences.get(i - 1).length() - 1, caseInsensitive) != ' ' && sequences.get(i)
.charAt(j, caseInsensitive) != ' ') {
partMatch.add(sequences.get(i).textPositionAt(j));
@ -163,7 +163,7 @@ public class SearchableText {
return TextNormalizationUtilities.removeHyphenLineBreaks(sb.toString())
.replaceAll("\n", " ")
.replaceAll(" ", " ");
.replaceAll(" {2}", " ");
}
@ -187,4 +187,10 @@ public class SearchableText {
return sb.append("\n").toString();
}
}
public List<String> getAsTabularData() {
return sequences.stream().map(TextPositionSequence::toString).collect(Collectors.toList());
}
}

1
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/PdfSegmentationService.java
View File

@ -29,7 +29,6 @@ import lombok.extern.slf4j.Slf4j;
@Slf4j
@Service
@RequiredArgsConstructor
@SuppressWarnings("PMD")
public class PdfSegmentationService {
private final RulingCleaningService rulingCleaningService;

8
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/tableextraction/model/Cell.java
View File

@ -16,11 +16,17 @@ public class Cell extends Rectangle {
private List<TextBlock> textBlocks = new ArrayList<>();
public Cell(Point2D topLeft, Point2D bottomRight) {
super((float) topLeft.getY(), (float) topLeft.getX(), (float) (bottomRight.getX() - topLeft.getX()), (float) (bottomRight.getY() - topLeft.getY()));
super((float) topLeft.getY(), (float) topLeft.getX(), (float) (bottomRight.getX() - topLeft.getX()), (float) (bottomRight
.getY() - topLeft.getY()));
}
public void addTextBlock(TextBlock textBlock) {
textBlocks.add(textBlock);
}
}

27
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/tableextraction/model/RectangleSpatialIndex.java
View File

@ -8,25 +8,28 @@ import org.locationtech.jts.index.strtree.STRtree;
import com.iqser.red.service.redaction.v1.server.tableextraction.utils.Utils;
@SuppressWarnings("all")
public class RectangleSpatialIndex<T extends Rectangle> {
private final STRtree si = new STRtree();
private final List<T> rectangles = new ArrayList<>();
public void add(T te) {
rectangles.add(te);
si.insert(new Envelope(te.getLeft(), te.getRight(), te.getBottom(), te.getTop()), te);
}
public List<T> contains(Rectangle r) {
List<T> intersection = si.query(new Envelope(r.getLeft(), r.getRight(), r.getTop(), r.getBottom()));
public List<T> contains(Rectangle rectangle) {
List<T> intersection = si.query(new Envelope(rectangle.getLeft(), rectangle.getRight(), rectangle.getTop(), rectangle
.getBottom()));
List<T> rv = new ArrayList<T>();
for (T ir: intersection) {
if (r.contains(ir)) {
for (T ir : intersection) {
if (rectangle.contains(ir)) {
rv.add(ir);
}
}
@ -34,18 +37,22 @@ public class RectangleSpatialIndex<T extends Rectangle> {
Utils.sort(rv, Rectangle.ILL_DEFINED_ORDER);
return rv;
}
public List<T> intersects(Rectangle r) {
List rv = si.query(new Envelope(r.getLeft(), r.getRight(), r.getTop(), r.getBottom()));
return rv;
}
/**
* Minimum bounding box of all the Rectangles contained on this RectangleSpatialIndex
*
*
* @return a Rectangle
*/
public Rectangle getBounds() {
return Rectangle.boundingBoxOf(rectangles);
}