Pull request #125: RED-1090: Return text with areas and sectionNumber

Merge in RED/redaction-service from RED-1090 to master

* commit '0dd47111c5294d9a9a0a10e174c2832436983205':
  RED-1090: Return text with areas and sectionNumber

redaction-service-v1/redaction-service-api-v1/src/main/java/com/iqser/red/service/redaction/v1/model/AnalyzeResult.java

@@ -14,5 +14,6 @@ public class AnalyzeResult {
     private int numberOfPages;
     private RedactionLog redactionLog;
     private SectionGrid sectionGrid;
+    private Text text;
 
 }

redaction-service-v1/redaction-service-api-v1/src/main/java/com/iqser/red/service/redaction/v1/model/RedactionLogEntry.java

@@ -40,4 +40,7 @@ public class RedactionLogEntry {
     @Builder.Default
     private List<Comment> comments = new ArrayList<>();
 
+    private int startOffset;
+    private int endOffset;
+
 }

redaction-service-v1/redaction-service-api-v1/src/main/java/com/iqser/red/service/redaction/v1/model/SectionArea.java

@@ -0,0 +1,29 @@
+package com.iqser.red.service.redaction.v1.model;
+
+import lombok.AllArgsConstructor;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+import lombok.NonNull;
+import lombok.RequiredArgsConstructor;
+
+@Data
+@RequiredArgsConstructor
+@NoArgsConstructor
+@AllArgsConstructor
+public class SectionArea {
+
+    @NonNull
+    private Point topLeft;
+
+    @NonNull
+    private float width;
+
+    @NonNull
+    private float height;
+
+    @NonNull
+    private int page;
+
+    private String header;
+
+}

redaction-service-v1/redaction-service-api-v1/src/main/java/com/iqser/red/service/redaction/v1/model/SectionText.java

@@ -0,0 +1,25 @@
+package com.iqser.red.service.redaction.v1.model;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+@Data
+@Builder
+@NoArgsConstructor
+@AllArgsConstructor
+public class SectionText {
+
+    private int sectionNumber;
+    private String text;
+
+    private boolean isTable;
+    private String headline;
+
+    private List<SectionArea> sectionAreas = new ArrayList<>();
+
+}

redaction-service-v1/redaction-service-api-v1/src/main/java/com/iqser/red/service/redaction/v1/model/Text.java

@@ -0,0 +1,17 @@
+package com.iqser.red.service.redaction.v1.model;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import lombok.AllArgsConstructor;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+@Data
+@NoArgsConstructor
+@AllArgsConstructor
+public class Text {
+
+    private List<SectionText> sectionTexts = new ArrayList<>();
+
+}

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Document.java

@@ -8,6 +8,7 @@ import java.util.Map;
 import com.iqser.red.service.redaction.v1.model.RedactionLogEntry;
 import com.iqser.red.service.redaction.v1.model.SectionGrid;
 import com.iqser.red.service.redaction.v1.server.redaction.model.Entity;
+import com.iqser.red.service.redaction.v1.model.SectionText;
 
 import lombok.Data;
 import lombok.NoArgsConstructor;
@@ -32,4 +33,6 @@ public class Document {
     private SectionGrid sectionGrid = new SectionGrid();
     private long dictionaryVersion;
     private long rulesVersion;
+
+    private List<SectionText> sectionText = new ArrayList<>();
 }

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/controller/RedactionController.java

@@ -9,6 +9,7 @@ import com.iqser.red.service.redaction.v1.model.RedactionLogEntry;
 import com.iqser.red.service.redaction.v1.model.RedactionRequest;
 import com.iqser.red.service.redaction.v1.model.RedactionResult;
 import com.iqser.red.service.redaction.v1.model.SectionGrid;
+import com.iqser.red.service.redaction.v1.model.Text;
 import com.iqser.red.service.redaction.v1.resources.RedactionResource;
 import com.iqser.red.service.redaction.v1.server.classification.model.Document;
 import com.iqser.red.service.redaction.v1.server.classification.model.Page;
@@ -69,6 +70,7 @@ public class RedactionController implements RedactionResource {
                     .redactionLog(new RedactionLog(classifiedDoc.getRedactionLogEntities(), classifiedDoc.getDictionaryVersion(), classifiedDoc
                     .getRulesVersion(), analyzeRequest.getRuleSetId()))
                     .numberOfPages(classifiedDoc.getPages().size())
+                    .text(new Text(classifiedDoc.getSectionText()))
                     .build();
 
         } catch (Exception e) {

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/model/Entity.java

@@ -39,7 +39,7 @@ public class Entity {
     private String textAfter;
 
 
-    public Entity(String word, String type, boolean redaction, String redactionReason, List<EntityPositionSequence> positionSequences, String headline, int matchedRule, int sectionNumber, String legalBasis, boolean isDictionaryEntry, String textBefore, String textAfter) {
+    public Entity(String word, String type, boolean redaction, String redactionReason, List<EntityPositionSequence> positionSequences, String headline, int matchedRule, int sectionNumber, String legalBasis, boolean isDictionaryEntry, String textBefore, String textAfter, Integer start, Integer end) {
 
         this.word = word;
         this.type = type;
@@ -53,6 +53,8 @@ public class Entity {
         this.isDictionaryEntry = isDictionaryEntry;
         this.textBefore = textBefore;
         this.textAfter = textAfter;
+        this.start = start;
+        this.end = end;
     }
 
 

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/EntityRedactionService.java

@@ -17,6 +17,7 @@ import org.springframework.stereotype.Service;
 
 import com.iqser.red.service.redaction.v1.model.ManualRedactionEntry;
 import com.iqser.red.service.redaction.v1.model.ManualRedactions;
+import com.iqser.red.service.redaction.v1.model.Point;
 import com.iqser.red.service.redaction.v1.model.Rectangle;
 import com.iqser.red.service.redaction.v1.server.classification.model.Document;
 import com.iqser.red.service.redaction.v1.server.classification.model.Footer;
@@ -24,6 +25,7 @@ import com.iqser.red.service.redaction.v1.server.classification.model.Header;
 import com.iqser.red.service.redaction.v1.server.classification.model.Paragraph;
 import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
 import com.iqser.red.service.redaction.v1.server.classification.model.UnclassifiedText;
+import com.iqser.red.service.redaction.v1.model.SectionArea;
 import com.iqser.red.service.redaction.v1.server.redaction.model.CellValue;
 import com.iqser.red.service.redaction.v1.server.redaction.model.Dictionary;
 import com.iqser.red.service.redaction.v1.server.redaction.model.DictionaryModel;
@@ -32,6 +34,7 @@ import com.iqser.red.service.redaction.v1.server.redaction.model.EntityPositionS
 import com.iqser.red.service.redaction.v1.server.redaction.model.SearchableText;
 import com.iqser.red.service.redaction.v1.server.redaction.model.Section;
 import com.iqser.red.service.redaction.v1.server.redaction.model.SectionSearchableTextPair;
+import com.iqser.red.service.redaction.v1.model.SectionText;
 import com.iqser.red.service.redaction.v1.server.redaction.utils.EntitySearchUtils;
 import com.iqser.red.service.redaction.v1.server.tableextraction.model.Cell;
 import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table;
@@ -86,7 +89,7 @@ public class EntityRedactionService {
                         .computeIfAbsent(entry.getKey(), (x) -> new ArrayList<>())
                         .add(new Entity(entity.getWord(), entity.getType(), entity.isRedaction(), entity.getRedactionReason(), entry
                                 .getValue(), entity.getHeadline(), entity.getMatchedRule(), entity.getSectionNumber(), entity
-                                .getLegalBasis(), entity.isDictionaryEntry(), entity.getTextBefore(), entity.getTextAfter()));
+                                .getLegalBasis(), entity.isDictionaryEntry(), entity.getTextBefore(), entity.getTextAfter(), entity.getStart(), entity.getEnd()));
             }
         }
 
@@ -110,29 +113,29 @@ public class EntityRedactionService {
             List<Table> tables = paragraph.getTables();
             for (Table table : tables) {
                 if (table.getColCount() == 2) {
-                    sectionSearchableTextPairs.addAll(processTableAsOneText(table, manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
+                    sectionSearchableTextPairs.addAll(processTableAsOneText(classifiedDoc, table, manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
                 } else {
-                    sectionSearchableTextPairs.addAll(processTablePerRow(table, manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
+                    sectionSearchableTextPairs.addAll(processTablePerRow(classifiedDoc, table, manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
                 }
                 sectionNumber.incrementAndGet();
             }
-            sectionSearchableTextPairs.add(processText(paragraph.getSearchableText(), paragraph.getTextBlocks(), paragraph
+            sectionSearchableTextPairs.add(processText(classifiedDoc, paragraph.getSearchableText(), paragraph.getTextBlocks(), paragraph
                     .getHeadline(), manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
             sectionNumber.incrementAndGet();
         }
 
         for (Header header : classifiedDoc.getHeaders()) {
-            sectionSearchableTextPairs.add(processText(header.getSearchableText(), header.getTextBlocks(), "Header", manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
+            sectionSearchableTextPairs.add(processText(classifiedDoc, header.getSearchableText(), header.getTextBlocks(), "Header", manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
             sectionNumber.incrementAndGet();
         }
 
         for (Footer footer : classifiedDoc.getFooters()) {
-            sectionSearchableTextPairs.add(processText(footer.getSearchableText(), footer.getTextBlocks(), "Footer", manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
+            sectionSearchableTextPairs.add(processText(classifiedDoc, footer.getSearchableText(), footer.getTextBlocks(), "Footer", manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
             sectionNumber.incrementAndGet();
         }
 
         for (UnclassifiedText unclassifiedText : classifiedDoc.getUnclassifiedTexts()) {
-            sectionSearchableTextPairs.add(processText(unclassifiedText.getSearchableText(), unclassifiedText.getTextBlocks(), "", manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
+            sectionSearchableTextPairs.add(processText(classifiedDoc, unclassifiedText.getSearchableText(), unclassifiedText.getTextBlocks(), "", manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
             sectionNumber.incrementAndGet();
         }
 
@@ -169,7 +172,7 @@ public class EntityRedactionService {
     }
 
 
-    private List<SectionSearchableTextPair> processTablePerRow(Table table, ManualRedactions manualRedactions,
+    private List<SectionSearchableTextPair> processTablePerRow(Document classifiedDoc, Table table, ManualRedactions manualRedactions,
                                                                AtomicInteger sectionNumber, Dictionary dictionary,
                                                                boolean local,
                                                                Map<Integer, Set<Entity>> hintsPerSectionNumber) {
@@ -181,10 +184,17 @@ public class EntityRedactionService {
             Map<String, CellValue> tabularData = new HashMap<>();
             int start = 0;
             List<Integer> cellStarts = new ArrayList<>();
+            SectionText sectionText = new SectionText();
             for (Cell cell : row) {
+
                 if (CollectionUtils.isEmpty(cell.getTextBlocks())) {
                     continue;
                 }
+
+                SectionArea sectionArea = new SectionArea(new Point((float) cell.getX(), (float) cell.getY()), (float) cell
+                        .getWidth(), (float) cell.getHeight(), cell.getTextBlocks().get(0).getSequences().get(0).getPage());
+                sectionText.getSectionAreas().add(sectionArea);
+
                 addSectionToManualRedactions(cell.getTextBlocks(), manualRedactions, table.getHeadline(), sectionNumber.intValue());
                 int cellStart = start;
 
@@ -196,6 +206,7 @@ public class EntityRedactionService {
                                 .replaceAll("\n", "")
                                 .replaceAll(" ", "")
                                 .replaceAll("-", "");
+                        sectionArea.setHeader(headerName);
                         tabularData.put(headerName, new CellValue(cell.getTextBlocks(), cellStart));
                     });
                 }
@@ -206,6 +217,7 @@ public class EntityRedactionService {
                 }
                 cellStarts.add(cellStart);
                 start = start + cell.toString().trim().length() + 1;
+
             }
             Set<Entity> rowEntities = findEntities(searchableRow, table.getHeadline(), sectionNumber.intValue(), dictionary, local);
             surroundingWordsService.addSurroundingText(rowEntities, searchableRow, dictionary, cellStarts);
@@ -225,6 +237,14 @@ public class EntityRedactionService {
                     .dictionary(dictionary)
                     .build(), searchableRow));
 
+            if(!local) {
+                sectionText.setText(searchableRow.toString());
+                sectionText.setHeadline(table.getHeadline());
+                sectionText.setSectionNumber(sectionNumber.intValue());
+                sectionText.setTable(true);
+                classifiedDoc.getSectionText().add(sectionText);
+            }
+
             sectionNumber.incrementAndGet();
         }
 
@@ -232,18 +252,26 @@ public class EntityRedactionService {
     }
 
 
-    private List<SectionSearchableTextPair> processTableAsOneText(Table table, ManualRedactions manualRedactions,
+    private List<SectionSearchableTextPair> processTableAsOneText(Document classifiedDoc, Table table, ManualRedactions manualRedactions,
                                                                   AtomicInteger sectionNumber, Dictionary dictionary,
                                                                   boolean local,
                                                                   Map<Integer, Set<Entity>> hintsPerSectionNumber) {
 
         List<SectionSearchableTextPair> sectionSearchableTextPairs = new ArrayList<>();
         SearchableText entireTableText = new SearchableText();
+        SectionText sectionText = new SectionText();
         for (List<Cell> row : table.getRows()) {
             for (Cell cell : row) {
                 if (CollectionUtils.isEmpty(cell.getTextBlocks())) {
                     continue;
                 }
+
+                if(!local) {
+                    SectionArea sectionArea = new SectionArea(new Point((float) cell.getX(), (float) cell.getY()), (float) cell
+                            .getWidth(), (float) cell.getHeight(), cell.getTextBlocks().get(0).getSequences().get(0).getPage());
+                    sectionText.getSectionAreas().add(sectionArea);
+                }
+
                 for (TextBlock textBlock : cell.getTextBlocks()) {
                     entireTableText.addAll(textBlock.getSequences());
                 }
@@ -251,6 +279,7 @@ public class EntityRedactionService {
             }
         }
 
+
         Set<Entity> rowEntities = findEntities(entireTableText, table.getHeadline(), sectionNumber.intValue(), dictionary, local);
         surroundingWordsService.addSurroundingText(rowEntities, entireTableText, dictionary);
 
@@ -268,15 +297,38 @@ public class EntityRedactionService {
                 .dictionary(dictionary)
                 .build(), entireTableText));
 
+        if(!local) {
+            sectionText.setText(entireTableText.toString());
+            sectionText.setHeadline(table.getHeadline());
+            sectionText.setSectionNumber(sectionNumber.intValue());
+            sectionText.setTable(true);
+            classifiedDoc.getSectionText().add(sectionText);
+        }
+
         return sectionSearchableTextPairs;
     }
 
 
-    private SectionSearchableTextPair processText(SearchableText searchableText, List<TextBlock> paragraphTextBlocks,
+    private SectionSearchableTextPair processText(Document classifiedDoc, SearchableText searchableText, List<TextBlock> paragraphTextBlocks,
                                                   String headline, ManualRedactions manualRedactions,
                                                   AtomicInteger sectionNumber, Dictionary dictionary, boolean local,
                                                   Map<Integer, Set<Entity>> hintsPerSectionNumber) {
 
+        if(!local) {
+            SectionText sectionText = new SectionText();
+            for (TextBlock paragraphTextBlock : paragraphTextBlocks) {
+                SectionArea sectionArea = new SectionArea(new Point(paragraphTextBlock.getMinX(), paragraphTextBlock.getMinY()), paragraphTextBlock
+                        .getWidth(), paragraphTextBlock.getHeight(), paragraphTextBlock.getPage());
+                sectionText.getSectionAreas().add(sectionArea);
+            }
+
+            sectionText.setText(searchableText.toString());
+            sectionText.setHeadline(headline);
+            sectionText.setSectionNumber(sectionNumber.intValue());
+            sectionText.setTable(false);
+            classifiedDoc.getSectionText().add(sectionText);
+        }
+
         addSectionToManualRedactions(paragraphTextBlocks, manualRedactions, headline, sectionNumber.intValue());
         Set<Entity> entities = findEntities(searchableText, headline, sectionNumber.intValue(), dictionary, local);
         surroundingWordsService.addSurroundingText(entities, searchableText, dictionary);

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/RedactionLogCreatorService.java

@@ -307,6 +307,8 @@ public class RedactionLogCreatorService {
                 .isDictionaryEntry(entity.isDictionaryEntry())
                 .textAfter(entity.getTextAfter())
                 .textBefore(entity.getTextBefore())
+                .startOffset(entity.getStart())
+                .endOffset(entity.getEnd())
                 .build();
     }
 

redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/SectionsBuilderService.java

@@ -84,9 +84,16 @@ public class SectionsBuilderService {
                 chunkWords.add(current);
                 prev = current;
             }
-            headers.add(new Header(header));
+
-            footers.add(new Footer(footer));
+            if(!header.isEmpty()) {
-            unclassifiedTexts.add(new UnclassifiedText(unclassifiedText));
+                headers.add(new Header(header));
+            }
+            if(!footer.isEmpty()) {
+                footers.add(new Footer(footer));
+            }
+            if(!unclassifiedText.isEmpty()) {
+                unclassifiedTexts.add(new UnclassifiedText(unclassifiedText));
+            }
         }
 
         Paragraph chunkBlock = buildTextBlock(chunkWords, lastHeadline);

redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java

@@ -40,6 +40,7 @@ import org.springframework.context.annotation.Bean;
 import org.springframework.core.io.ClassPathResource;
 import org.springframework.test.context.junit4.SpringRunner;
 
+import com.fasterxml.jackson.databind.ObjectMapper;
 import com.iqser.red.service.configuration.v1.api.model.Colors;
 import com.iqser.red.service.configuration.v1.api.model.DictionaryResponse;
 import com.iqser.red.service.configuration.v1.api.model.RulesResponse;
@@ -59,6 +60,7 @@ import com.iqser.red.service.redaction.v1.model.Rectangle;
 import com.iqser.red.service.redaction.v1.model.RedactionLogEntry;
 import com.iqser.red.service.redaction.v1.model.RedactionRequest;
 import com.iqser.red.service.redaction.v1.model.RedactionResult;
+import com.iqser.red.service.redaction.v1.model.SectionText;
 import com.iqser.red.service.redaction.v1.model.Status;
 import com.iqser.red.service.redaction.v1.server.client.DictionaryClient;
 import com.iqser.red.service.redaction.v1.server.client.RulesClient;
@@ -94,6 +96,9 @@ public class RedactionIntegrationTest {
     @Autowired
     private RedactionController redactionController;
 
+    @Autowired
+    private ObjectMapper objectMapper;
+
     @MockBean
     private RulesClient rulesClient;
 
@@ -408,6 +413,7 @@ public class RedactionIntegrationTest {
             duplicates.entrySet().forEach(entry -> {
                 assertThat(entry.getValue().size()).isEqualTo(1);
             });
+
         }
 
         long end = System.currentTimeMillis();
@@ -438,15 +444,9 @@ public class RedactionIntegrationTest {
     @Test
     public void redactionTest() throws IOException {
 
-//        49 Cyprodinil - EU AIR3 - MCA Section 8 Supplement - Ecotoxicological studies on the active substance.pdf
-//        182 Fludioxonil - EU AIR3 - MCA Section 8 Supplement - Ecotoxicological studies on the active substance.pdf
-//        38 A14325E - EU AIR3 - MCP Section 10 - Ecotoxicological studies on the plant protection product.pdf
-//        91 Trinexapac-ethyl_RAR_01_Volume_1_2018-02-23.pdf
-//        95 Trinexapac-ethyl_RAR_08_Volume_3CA_B-6_2018-01-10.pdf
-
         System.out.println("redactionTest");
         long start = System.currentTimeMillis();
-        ClassPathResource pdfFileResource = new ClassPathResource("files/Cyprodinil/49 Cyprodinil - EU AIR3 - MCA Section 8 Supplement - Ecotoxicological studies on the active substance.pdf");
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Metolachlor/S-Metolachlor_RAR_01_Volume_1_2018-09-06.pdf");
 
         AnalyzeRequest request = AnalyzeRequest.builder()
                 .ruleSetId(TEST_RULESET_ID)
@@ -465,6 +465,33 @@ public class RedactionIntegrationTest {
             fileOutputStream.write(annotateResponse.getDocument());
         }
 
+        try (FileOutputStream fileOutputStream = new FileOutputStream("/tmp/Test.json")) {
+            fileOutputStream.write(objectMapper.writeValueAsBytes(result.getText()));
+        }
+
+        int correctFound = 0;
+        loop:
+        for (RedactionLogEntry redactionLogEntry : result.getRedactionLog().getRedactionLogEntry()) {
+            for (SectionText sectionText : result.getText().getSectionTexts()) {
+                if (redactionLogEntry.getType().equals("image")) {
+                    correctFound++;
+                    continue loop;
+                }
+                if (redactionLogEntry.getSectionNumber() == sectionText.getSectionNumber()) {
+                    String value = sectionText.getText()
+                            .substring(redactionLogEntry.getStartOffset(), redactionLogEntry.getEndOffset());
+                    if (redactionLogEntry.getValue().equalsIgnoreCase(value)) {
+                        correctFound++;
+                    } else {
+                        throw new RuntimeException("WTF");
+                    }
+                }
+            }
+        }
+        assertThat(correctFound).isEqualTo(result.getRedactionLog().getRedactionLogEntry().size());
+
+        System.out.println("correctFound " + correctFound);
+
         long end = System.currentTimeMillis();
 
         System.out.println("duration: " + (end - start));