Compare commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
0497d764ec | ||
|
|
1362e4fbb2 | ||
|
|
665ad40b0b | ||
|
|
e2bcf971c9 | ||
|
|
dacc2f7f43 | ||
|
|
144a9591a2 | ||
|
|
207d9dec97 |
+3
-1
@@ -16,6 +16,8 @@ deploy:
|
||||
reports:
|
||||
dotenv: version.env
|
||||
rules:
|
||||
- if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH
|
||||
- if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH
|
||||
- if: $CI_COMMIT_BRANCH =~ /^release/
|
||||
- if: $CI_COMMIT_TAG
|
||||
pmd:
|
||||
allow_failure: true
|
||||
|
||||
+2
-2
@@ -244,9 +244,9 @@ public class LayoutParsingPipeline {
|
||||
|
||||
private void increaseDocumentStatistics(ClassificationPage classificationPage, ClassificationDocument document) {
|
||||
|
||||
// if (!classificationPage.isLandscape()) {
|
||||
if (!classificationPage.isLandscape()) {
|
||||
document.getFontSizeCounter().addAll(classificationPage.getFontSizeCounter().getCountPerValue());
|
||||
// }
|
||||
}
|
||||
document.getFontCounter().addAll(classificationPage.getFontCounter().getCountPerValue());
|
||||
document.getTextHeightCounter().addAll(classificationPage.getTextHeightCounter().getCountPerValue());
|
||||
document.getFontStyleCounter().addAll(classificationPage.getFontStyleCounter().getCountPerValue());
|
||||
|
||||
+7
@@ -196,6 +196,12 @@ public class TextPositionSequence implements CharSequence {
|
||||
}
|
||||
|
||||
|
||||
public float getTextHeightNoPadding() {
|
||||
|
||||
return textPositions.get(0).getHeightDir();
|
||||
}
|
||||
|
||||
|
||||
@JsonIgnore
|
||||
@JsonAttribute(ignore = true)
|
||||
public float getTextHeight() {
|
||||
@@ -234,6 +240,7 @@ public class TextPositionSequence implements CharSequence {
|
||||
@JsonIgnore
|
||||
@JsonAttribute(ignore = true)
|
||||
public String getFontStyle() {
|
||||
|
||||
if (textPositions.get(0).getFontName() == null) {
|
||||
return "standard";
|
||||
}
|
||||
|
||||
+13
-10
@@ -8,6 +8,7 @@ import java.util.Map;
|
||||
|
||||
import org.springframework.stereotype.Service;
|
||||
|
||||
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.PageInfo;
|
||||
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableCells;
|
||||
import com.knecon.fforesight.service.layoutparser.processor.python_api.model.table.TableServiceResponse;
|
||||
|
||||
@@ -24,24 +25,26 @@ public class CvTableParsingAdapter {
|
||||
Map<Integer, List<TableCells>> tableCells = new HashMap<>();
|
||||
tableServiceResponse.getData()
|
||||
.forEach(tableData -> tableCells.computeIfAbsent(tableData.getPageInfo().getNumber(), tableCell -> new ArrayList<>())
|
||||
.addAll(convertTableCells(tableData.getTableCells())));
|
||||
.addAll(convertTableCells(tableData.getTableCells(), tableData.getPageInfo())));
|
||||
|
||||
return tableCells;
|
||||
}
|
||||
|
||||
|
||||
private Collection<TableCells> convertTableCells(List<TableCells> tableCells) {
|
||||
private Collection<TableCells> convertTableCells(List<TableCells> tableCells, PageInfo pageInfo) {
|
||||
|
||||
List<TableCells> cvParsedTableCells = new ArrayList<>();
|
||||
|
||||
tableCells.forEach(t -> cvParsedTableCells.add(TableCells.builder()
|
||||
.y0(t.getY0())
|
||||
.x1(t.getX1())
|
||||
.y1(t.getY1())
|
||||
.x0(t.getX0())
|
||||
.width(t.getWidth())
|
||||
.height(t.getHeight())
|
||||
.build()));
|
||||
tableCells.stream()
|
||||
.filter(cell -> cell.getWidth() < pageInfo.getWidth() * 0.98 && cell.getHeight() < pageInfo.getHeight() * 0.98)
|
||||
.forEach(t -> cvParsedTableCells.add(TableCells.builder()
|
||||
.y0(t.getY0())
|
||||
.x1(t.getX1())
|
||||
.y1(t.getY1())
|
||||
.x0(t.getX0())
|
||||
.width(t.getWidth())
|
||||
.height(t.getHeight())
|
||||
.build()));
|
||||
|
||||
return cvParsedTableCells;
|
||||
}
|
||||
|
||||
-5
@@ -329,11 +329,6 @@ public class PDFLinesTextStripper extends PDFTextStripper {
|
||||
.getXDirAdj() - (previous.getXDirAdj() + previous.getWidthDirAdj()) < maximumGapSize;
|
||||
}
|
||||
|
||||
// !(sublist.isEmpty() || sublist.size() == 1 && (sublist.get(0).getUnicode().equals(" ") || sublist.get(0).getUnicode().equals("\u00A0") || sublist.get(0)
|
||||
// .getUnicode()
|
||||
// .equals("\t")))
|
||||
|
||||
|
||||
@Override
|
||||
public String getText(PDDocument doc) throws IOException {
|
||||
|
||||
|
||||
+10
-18
@@ -28,15 +28,13 @@ import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPosit
|
||||
*
|
||||
* @author Ben Litchfield
|
||||
*/
|
||||
public class TextPositionSequenceComparator implements Comparator<TextPositionSequence>
|
||||
{
|
||||
public class TextPositionSequenceComparator implements Comparator<TextPositionSequence> {
|
||||
|
||||
@Override
|
||||
public int compare(TextPositionSequence pos1, TextPositionSequence pos2)
|
||||
{
|
||||
public int compare(TextPositionSequence pos1, TextPositionSequence pos2) {
|
||||
// only compare text that is in the same direction
|
||||
int cmp1 = Float.compare(pos1.getDir().getDegrees(), pos2.getDir().getDegrees());
|
||||
if (cmp1 != 0)
|
||||
{
|
||||
if (cmp1 != 0) {
|
||||
return cmp1;
|
||||
}
|
||||
|
||||
@@ -48,25 +46,19 @@ public class TextPositionSequenceComparator implements Comparator<TextPositionSe
|
||||
float pos2YBottom = pos2.getMaxYDirAdj();
|
||||
|
||||
// note that the coordinates have been adjusted so 0,0 is in upper left
|
||||
float pos1YTop = pos1YBottom - pos1.getTextHeight();
|
||||
float pos2YTop = pos2YBottom - pos2.getTextHeight();
|
||||
float pos1YTop = pos1YBottom - pos1.getTextHeightNoPadding();
|
||||
float pos2YTop = pos2YBottom - pos2.getTextHeightNoPadding();
|
||||
|
||||
float yDifference = Math.abs(pos1YBottom - pos2YBottom);
|
||||
|
||||
// we will do a simple tolerance comparison
|
||||
if (yDifference < .1 ||
|
||||
pos2YBottom >= pos1YTop && pos2YBottom <= pos1YBottom ||
|
||||
pos1YBottom >= pos2YTop && pos1YBottom <= pos2YBottom)
|
||||
{
|
||||
if (yDifference < .1 || pos2YBottom >= pos1YTop && pos2YBottom <= pos1YBottom || pos1YBottom >= pos2YTop && pos1YBottom <= pos2YBottom) {
|
||||
return Float.compare(x1, x2);
|
||||
}
|
||||
else if (pos1YBottom < pos2YBottom)
|
||||
{
|
||||
} else if (pos1YBottom < pos2YBottom) {
|
||||
return -1;
|
||||
}
|
||||
else
|
||||
{
|
||||
} else {
|
||||
return 1;
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
+1
-2
@@ -24,8 +24,7 @@ public class MessagingConfiguration {
|
||||
@Bean
|
||||
public Queue layoutparsingResponseQueue() {
|
||||
|
||||
return QueueBuilder.durable(LayoutParsingQueueNames.LAYOUT_PARSING_FINISHED_EVENT_QUEUE)//
|
||||
.withArgument("x-dead-letter-exchange", "").withArgument("x-dead-letter-routing-key", LayoutParsingQueueNames.LAYOUT_PARSING_DLQ).build();
|
||||
return QueueBuilder.durable(LayoutParsingQueueNames.LAYOUT_PARSING_FINISHED_EVENT_QUEUE).build();
|
||||
}
|
||||
|
||||
|
||||
|
||||
+43
-2
@@ -61,12 +61,13 @@ public class PdfSegmentationServiceTest extends AbstractTest {
|
||||
private SectionsBuilderService sectionsBuilderService;
|
||||
|
||||
|
||||
public ClassificationDocument buildClassificationDocument(PDDocument originDocument) {
|
||||
@SneakyThrows
|
||||
public ClassificationDocument buildClassificationDocument(PDDocument originDocument, TableServiceResponse tableServiceResponse) {
|
||||
|
||||
ClassificationDocument classificationDocument = layoutParsingPipeline.parseLayout(LayoutParsingType.REDACT_MANAGER,
|
||||
originDocument,
|
||||
new ImageServiceResponse(),
|
||||
new TableServiceResponse());
|
||||
tableServiceResponse);
|
||||
|
||||
redactManagerClassificationService.classifyDocument(classificationDocument);
|
||||
|
||||
@@ -76,6 +77,13 @@ public class PdfSegmentationServiceTest extends AbstractTest {
|
||||
}
|
||||
|
||||
|
||||
@SneakyThrows
|
||||
public ClassificationDocument buildClassificationDocument(PDDocument originDocument) {
|
||||
|
||||
return buildClassificationDocument(originDocument, new TableServiceResponse());
|
||||
}
|
||||
|
||||
|
||||
@Test
|
||||
public void tablesToHtmlDebugger() throws IOException {
|
||||
|
||||
@@ -88,6 +96,39 @@ public class PdfSegmentationServiceTest extends AbstractTest {
|
||||
}
|
||||
|
||||
|
||||
@Test
|
||||
public void tablesToHtmlDebuggerWithCVResponse() throws IOException {
|
||||
|
||||
ClassPathResource pdfFileResource = new ClassPathResource("files/cv_tables/ScanRotationBorder.pdf");
|
||||
ClassPathResource cvTablesResource = new ClassPathResource("files/cv_tables/ScanRotationBorder.TABLES.json");
|
||||
var tableServiceResponse = objectMapper.readValue(cvTablesResource.getInputStream(), TableServiceResponse.class);
|
||||
|
||||
ClassificationDocument document = buildClassificationDocument(Loader.loadPDF(pdfFileResource.getFile()), tableServiceResponse);
|
||||
|
||||
toHtml(document, "/tmp/ScanRotationBorder.html");
|
||||
}
|
||||
|
||||
|
||||
@Test
|
||||
public void testScanRotationBorderIsIgnored() throws IOException {
|
||||
|
||||
ClassPathResource pdfFileResource = new ClassPathResource("files/cv_tables/ScanRotationBorder.pdf");
|
||||
ClassPathResource cvTablesResource = new ClassPathResource("files/cv_tables/ScanRotationBorder.TABLES.json");
|
||||
var tableServiceResponse = objectMapper.readValue(cvTablesResource.getInputStream(), TableServiceResponse.class);
|
||||
|
||||
ClassificationDocument document = buildClassificationDocument(Loader.loadPDF(pdfFileResource.getFile()), tableServiceResponse);
|
||||
assertThat(document.getSections().stream().flatMap(paragraph -> paragraph.getTables().stream()).collect(Collectors.toList())).isNotEmpty();
|
||||
var tables = document.getSections().stream().flatMap(paragraph -> paragraph.getTables().stream()).toList();
|
||||
|
||||
// Quality of the table parsing is not good, because the file is rotated at scanning.
|
||||
// We only asset that the table border is not the page border.
|
||||
tables.forEach(table -> {
|
||||
assertThat(table.getMinX()).isGreaterThan(75);
|
||||
assertThat(table.getMaxX()).isLessThan(512);
|
||||
});
|
||||
}
|
||||
|
||||
|
||||
@Test
|
||||
@SneakyThrows
|
||||
public void testMapping() {
|
||||
|
||||
+1
File diff suppressed because one or more lines are too long
Reference in New Issue
Block a user