Pull request #33 : Fix entity span in table rows and detection of headers in rotated tables

Merge in RED/redaction-service from bugfix/rowspan-and-header-in-rotated-table-fix to master * commit '4954aafed78e06484531d6264bdf215176ee0ef2': Reduce log level. Remove unused import Adjust test to added rule and fix vertical header propagation for row > 2 Fix entity span in table rows and detection of headers in rotated tables
Reduce log level.
2020-08-25 17:30:57 +02:00 · 2020-08-25 17:21:56 +02:00 · 2020-08-25 15:45:24 +02:00 · 2020-08-25 15:41:44 +02:00 · 2020-08-25 13:50:34 +02:00 · 2020-08-24 16:15:44 +02:00
22 changed files with 788 additions and 387 deletions
@@ -0,0 +1,16 @@
+package com.iqser.red.service.redaction.v1.server.redaction.model;
+
+import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
+
+import lombok.RequiredArgsConstructor;
+import lombok.Value;
+
+@Value
+@RequiredArgsConstructor
+public class CellValue {
+
+    TextBlock textBlock;
+
+    int rowSpanStart;
+
+}
@@ -3,6 +3,8 @@ package com.iqser.red.service.redaction.v1.server.redaction.model;
 import java.util.ArrayList;
 import java.util.List;

+import com.iqser.red.service.redaction.v1.server.parsing.model.TextPositionSequence;
+
 import lombok.Data;
 import lombok.EqualsAndHashCode;

@@ -16,6 +18,7 @@ public class Entity {
    private boolean redaction;
    private String redactionReason;
    private List<EntityPositionSequence> positionSequences = new ArrayList<>();
+    private List<TextPositionSequence> targetSequences;
    private Integer start;
    private Integer end;

@@ -1,6 +1,7 @@
 package com.iqser.red.service.redaction.v1.server.redaction.model;

 import java.util.ArrayList;
+import java.util.Collections;
 import java.util.List;
 import java.util.regex.Pattern;

@@ -25,9 +26,16 @@ public class SearchableText {
    }


-    @SuppressWarnings("checkstyle:ModifiedControlVariable")
    public List<EntityPositionSequence> getSequences(String searchString, boolean caseInsensitive) {

+        return getSequences(searchString, caseInsensitive, null);
+
+    }
+
+    @SuppressWarnings("checkstyle:ModifiedControlVariable")
+    public List<EntityPositionSequence> getSequences(String searchString, boolean caseInsensitive,
+                                                     List<TextPositionSequence> sequencesSubList) {
+
        String normalizedSearchString;
        if (caseInsensitive) {
            normalizedSearchString = searchString.toLowerCase();
@@ -40,37 +48,50 @@ public class SearchableText {

        List<TextPositionSequence> crossSequenceParts = new ArrayList<>();
        List<EntityPositionSequence> finalMatches = new ArrayList<>();
-        for (int i = 0; i < sequences.size(); i++) {
-            TextPositionSequence partMatch = new TextPositionSequence(sequences.get(i).getPage());
-            for (int j = 0; j < sequences.get(i).length(); j++) {

-                if (i > 0 && j == 0 && sequences.get(i).charAt(0, caseInsensitive) == ' ' && sequences.get(i - 1)
-                        .charAt(sequences.get(i - 1).length() - 1, caseInsensitive) == ' ' || j > 0 && sequences.get(i)
-                        .charAt(j, caseInsensitive) == ' ' && sequences.get(i).charAt(j - 1, caseInsensitive) == ' ') {
-                    if (j == sequences.get(i).length() - 1 && counter != 0 && !partMatch.getTextPositions().isEmpty()) {
+        List<TextPositionSequence> searchSpace;
+        if (sequencesSubList != null) {
+            int subListIndex = Collections.indexOfSubList(sequences, sequencesSubList);
+            if (subListIndex != -1) {
+                searchSpace = sequences.subList(subListIndex, subListIndex + sequencesSubList.size());
+            } else {
+                searchSpace = sequences;
+            }
+        } else {
+            searchSpace = sequences;
+        }
+
+        for (int i = 0; i < searchSpace.size(); i++) {
+            TextPositionSequence partMatch = new TextPositionSequence(searchSpace.get(i).getPage());
+            for (int j = 0; j < searchSpace.get(i).length(); j++) {
+
+                if (i > 0 && j == 0 && searchSpace.get(i).charAt(0, caseInsensitive) == ' ' && searchSpace.get(i - 1)
+                        .charAt(searchSpace.get(i - 1).length() - 1, caseInsensitive) == ' ' || j > 0 && searchSpace.get(i)
+                        .charAt(j, caseInsensitive) == ' ' && searchSpace.get(i).charAt(j - 1, caseInsensitive) == ' ') {
+                    if (j == searchSpace.get(i).length() - 1 && counter != 0 && !partMatch.getTextPositions().isEmpty()) {
                        crossSequenceParts.add(partMatch);
                    }
                    continue;
                }

-                if (j == 0 && sequences.get(i).charAt(j, caseInsensitive) != ' ' && i != 0 && sequences.get(i - 1)
-                        .charAt(sequences.get(i - 1)
+                if (j == 0 && searchSpace.get(i).charAt(j, caseInsensitive) != ' ' && i != 0 && searchSpace.get(i - 1)
+                        .charAt(searchSpace.get(i - 1)
                                .length() - 1, caseInsensitive) != ' ' && searchChars[counter] == ' ') {
                    counter++;
                }

-                if (sequences.get(i)
-                        .charAt(j, caseInsensitive) == searchChars[counter] || counter != 0 && sequences.get(i)
+                if (searchSpace.get(i)
+                        .charAt(j, caseInsensitive) == searchChars[counter] || counter != 0 && searchSpace.get(i)
                        .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(sequences.get(i - 1)
-                                    .length() - 1, caseInsensitive)) || j == 0 && i != 0 && sequences.get(i - 1)
-                            .charAt(sequences.get(i - 1).length() - 1, caseInsensitive) != ' ' && sequences.get(i)
+                    if (counter != 0 || i == 0 && j == 0 || j != 0 && isSeparator(searchSpace.get(i)
+                            .charAt(j - 1, caseInsensitive)) || j == 0 && i != 0 && isSeparator(searchSpace.get(i - 1)
+                            .charAt(searchSpace.get(i - 1)
+                                    .length() - 1, caseInsensitive)) || j == 0 && i != 0 && searchSpace.get(i - 1)
+                            .charAt(searchSpace.get(i - 1).length() - 1, caseInsensitive) != ' ' && searchSpace.get(i)
                            .charAt(j, caseInsensitive) != ' ') {
-                        partMatch.add(sequences.get(i).textPositionAt(j));
-                        if (!(j == sequences.get(i).length() - 1 && sequences.get(i)
+                        partMatch.add(searchSpace.get(i).textPositionAt(j));
+                        if (!(j == searchSpace.get(i).length() - 1 && searchSpace.get(i)
                                .charAt(j, caseInsensitive) == '-' && searchChars[counter] != '-')) {
                            counter++;
                        }
@@ -79,19 +100,19 @@ public class SearchableText {
                    if (counter == searchString.length()) {
                        crossSequenceParts.add(partMatch);

-                        if (i == sequences.size() - 1 && j == sequences.get(i).length() - 1 || j != sequences.get(i)
-                                .length() - 1 && isSeparator(sequences.get(i)
-                                .charAt(j + 1, caseInsensitive)) || j == sequences.get(i)
-                                .length() - 1 && isSeparator(sequences.get(i + 1)
-                                .charAt(0, caseInsensitive)) || j == sequences.get(i).length() - 1 && sequences.get(i)
-                                .charAt(j, caseInsensitive) != ' ' && sequences.get(i + 1)
+                        if (i == searchSpace.size() - 1 && j == searchSpace.get(i).length() - 1 || j != searchSpace.get(i)
+                                .length() - 1 && isSeparator(searchSpace.get(i)
+                                .charAt(j + 1, caseInsensitive)) || j == searchSpace.get(i)
+                                .length() - 1 && isSeparator(searchSpace.get(i + 1)
+                                .charAt(0, caseInsensitive)) || j == searchSpace.get(i).length() - 1 && searchSpace.get(i)
+                                .charAt(j, caseInsensitive) != ' ' && searchSpace.get(i + 1)
                                .charAt(0, caseInsensitive) != ' ') {
                            finalMatches.addAll(buildEntityPositionSequence(crossSequenceParts));
                        }

                        counter = 0;
                        crossSequenceParts = new ArrayList<>();
-                        partMatch = new TextPositionSequence(sequences.get(i).getPage());
+                        partMatch = new TextPositionSequence(searchSpace.get(i).getPage());
                    }
                } else {
                    counter = 0;
@@ -99,16 +120,17 @@ public class SearchableText {
                        j--;
                    }
                    crossSequenceParts = new ArrayList<>();
-                    partMatch = new TextPositionSequence(sequences.get(i).getPage());
+                    partMatch = new TextPositionSequence(searchSpace.get(i).getPage());
                }

-                if (j == sequences.get(i).length() - 1 && counter != 0) {
+                if (j == searchSpace.get(i).length() - 1 && counter != 0) {
                    crossSequenceParts.add(partMatch);
                }
            }
        }

        return finalMatches;
+
    }


@@ -7,7 +7,6 @@ import java.util.Map;
 import java.util.Set;
 import java.util.regex.Pattern;

-import org.apache.commons.collections4.CollectionUtils;
 import org.apache.commons.lang3.StringUtils;

 import lombok.Builder;
@@ -31,7 +30,17 @@ public class Section {

    private int sectionNumber;

-    private Map<String, String> tabularData;
+    private Map<String, CellValue> tabularData;
+
+
+    public boolean rowEquals(String headerName, String value){
+        String cleanHeaderName = headerName.replaceAll("\n", "")
+                .replaceAll(" ", "")
+                .replaceAll("-", "");
+
+        return tabularData != null && tabularData.containsKey(cleanHeaderName)
+                && tabularData.get(cleanHeaderName).getTextBlock().getText().equals(value);
+    }


    public boolean contains(String type) {
@@ -161,24 +170,27 @@ public class Section {
    }


-    public void highlightCell(String cellHeader, int ruleNumber) {
+    public void highlightCell(String cellHeader, int ruleNumber, String type) {

-        String value = tabularData.get(cellHeader);
+        String cleanHeaderName = cellHeader.replaceAll("\n", "")
+                .replaceAll(" ", "")
+                .replaceAll("-", "");
+
+        CellValue value = tabularData.get(cleanHeaderName);
        if (value == null) {
            log.warn("Could not find any data for {}.", cellHeader);
        } else {
-            Set<Entity> found = findEntities(value, "must_redact");
-            if (CollectionUtils.isEmpty(found)) {
-                log.warn("Could not identify value {} in row.", value);
-            } else {
-                Entity entity = found.iterator().next();
-                entity.setRedaction(false);
-                entity.setMatchedRule(ruleNumber);
-                entity.setRedactionReason(cellHeader);
-                entities.add(entity);
-            }
+            Entity entity = new Entity(value.getTextBlock()
+                    .getText(), type, value.getRowSpanStart(), value.getRowSpanStart() + value.getTextBlock()
+                    .getText()
+                    .length(), headline, sectionNumber);
+            entity.setRedaction(false);
+            entity.setMatchedRule(ruleNumber);
+            entity.setRedactionReason(cellHeader);
+            entity.setTargetSequences(value.getTextBlock().getSequences()); // Make sure no other cells with same content are highlighted
+            entities.add(entity);
        }

    }

-}
+}
@@ -18,6 +18,7 @@ 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.Paragraph;
 import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
+import com.iqser.red.service.redaction.v1.server.redaction.model.CellValue;
 import com.iqser.red.service.redaction.v1.server.redaction.model.Entity;
 import com.iqser.red.service.redaction.v1.server.redaction.model.EntityPositionSequence;
 import com.iqser.red.service.redaction.v1.server.redaction.model.SearchableText;
@@ -51,24 +52,32 @@ public class EntityRedactionService {
            List<Table> tables = paragraph.getTables();

            for (Table table : tables) {
-                List<String> metadata = table.getHeaders();
                for (List<Cell> row : table.getRows()) {
                    SearchableText searchableRow = new SearchableText();
-                    List<String> cellValues = new ArrayList<>();
+                    Map<String, CellValue> tabularData = new HashMap<>();
+                    int start = 0;
                    for (Cell cell : row) {
-                        if (cell == null || CollectionUtils.isEmpty(cell.getTextBlocks())) {
-                            cellValues.add(null);
+                        if (cell.isHeaderCell() || CollectionUtils.isEmpty(cell.getTextBlocks())) {
                            continue;
                        }
-                        cellValues.add(cell.getTextBlocks().get(0).getText());
                        addSectionToManualRedactions(cell.getTextBlocks(), manualRedactions, table.getHeadline(), sectionNumber);
+                        int cellStart = start;
+                        cell.getHeaderCells().forEach(headerCell -> {
+                            StringBuilder headerBuilder = new StringBuilder();
+                            headerCell.getTextBlocks().forEach(textBlock -> headerBuilder.append(textBlock.getText()));
+                            String headerName = headerBuilder.toString()
+                                    .replaceAll("\n", "")
+                                    .replaceAll(" ", "")
+                                    .replaceAll("-", "");
+                            tabularData.put(headerName, new CellValue(cell.getTextBlocks().get(0), cellStart));
+                        });
+                        start = start + cell.getTextBlocks().get(0).toString().length();
                        for (TextBlock textBlock : cell.getTextBlocks()) {
                            searchableRow.addAll(textBlock.getSequences());
                        }
                    }
                    Set<Entity> rowEntities = findEntities(searchableRow, table.getHeadline(), sectionNumber);

-                    Map<String, String> tabularData = toMap(metadata, cellValues);
                    Section analysedRowSection = droolsExecutionService.executeRules(Section.builder()
                            .entities(rowEntities)
                            .text(searchableRow.getAsStringWithLinebreaks())
@@ -116,35 +125,15 @@ public class EntityRedactionService {
    }


-    private Map<String, String> toMap(List<String> keys, List<String> values) {
-
-        if (keys.size() != values.size()) {
-            log.warn("Cannot merge lists of unequal size, returning empty map.");
-            return new HashMap<>();
-        }
-        Map<String, String> result = new HashMap<>();
-        for (int i = 0; i < keys.size(); i++) {
-            String value = values.get(i);
-            if (value == null) {
-                continue;
-            }
-            result.put(keys.get(i), value);
-        }
-
-        return result;
-
-    }
-
-
    private Set<Entity> clearAndFindPositions(Set<Entity> entities, SearchableText text) {

        removeEntitiesContainedInLarger(entities);

        for (Entity entity : entities) {
            if (dictionaryService.getCaseInsensitiveTypes().contains(entity.getType())) {
-                entity.setPositionSequences(text.getSequences(entity.getWord(), true));
+                entity.setPositionSequences(text.getSequences(entity.getWord(), true, entity.getTargetSequences()));
            } else {
-                entity.setPositionSequences(text.getSequences(entity.getWord(), false));
+                entity.setPositionSequences(text.getSequences(entity.getWord(), false, entity.getTargetSequences()));
            }
        }

@@ -155,7 +144,7 @@ public class EntityRedactionService {
    private Set<Entity> findEntities(SearchableText searchableText, String headline, int sectionNumber) {

        Set<Entity> found = new HashSet<>();
-        if (StringUtils.isEmpty(searchableText.toString()) && StringUtils.isEmpty(headline)) {
+        if (StringUtils.isEmpty(searchableText.toString())) {
            return found;
        }

@@ -1,11 +1,12 @@
 package com.iqser.red.service.redaction.v1.server.segmentation;

 import java.util.ArrayList;
+import java.util.Collections;
 import java.util.Iterator;
 import java.util.List;
+import java.util.stream.Collectors;

 import org.apache.commons.collections4.CollectionUtils;
-import org.apache.commons.lang3.StringUtils;
 import org.springframework.stereotype.Service;

 import com.iqser.red.service.redaction.v1.server.classification.model.Document;
@@ -13,6 +14,7 @@ import com.iqser.red.service.redaction.v1.server.classification.model.Page;
 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.tableextraction.model.AbstractTextContainer;
+import com.iqser.red.service.redaction.v1.server.tableextraction.model.Cell;
 import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table;

@Service
@@ -38,22 +40,27 @@ public class SectionsBuilderService {
                current.setPage(page.getPageNumber());

                if (prev != null && current.getClassification().startsWith("H ") || !document.isHeadlines()) {
-                    Paragraph chunkBlock = buildTextBlock(chunkWords, lastHeadline, previousTable);
+                    Paragraph chunkBlock = buildTextBlock(chunkWords, lastHeadline);
                    chunkBlock.setHeadline(lastHeadline);
                    lastHeadline = current.getText();
-                    if (CollectionUtils.isNotEmpty(chunkBlock.getTables())) {
-                        previousTable = chunkBlock.getTables().get(0);
-                    }
                    chunkBlockList.add(chunkBlock);
                    chunkWords = new ArrayList<>();
+                    if (CollectionUtils.isNotEmpty(chunkBlock.getTables())) {
+                        previousTable = chunkBlock.getTables().get(chunkBlock.getTables().size() - 1);
+                    }
+                }
+                if (current instanceof Table) {
+                    Table table = (Table) current;
+                    // Distribute header information for subsequent tables
+                    mergeTableMetadata(table, previousTable);
+                    previousTable = table;
                }
-
                chunkWords.add(current);
                prev = current;
            }
        }

-        Paragraph chunkBlock = buildTextBlock(chunkWords, lastHeadline, previousTable);
+        Paragraph chunkBlock = buildTextBlock(chunkWords, lastHeadline);
        chunkBlock.setHeadline(lastHeadline);
        chunkBlockList.add(chunkBlock);

@@ -61,7 +68,38 @@ public class SectionsBuilderService {
    }


-    private Paragraph buildTextBlock(List<AbstractTextContainer> wordBlockList, String lastHeadline, Table previousTable) {
+    private void mergeTableMetadata(Table currentTable, Table previousTable) {
+
+        // Distribute header information for subsequent tables
+        if (previousTable != null && hasInvalidHeaderInformation(currentTable) && hasValidHeaderInformation(previousTable)) {
+            List<Cell> previousTableNonHeaderRow = getRowWithNonHeaderCells(previousTable);
+            List<Cell> tableNonHeaderRow = getRowWithNonHeaderCells(currentTable);
+            // Allow merging of tables if header row is separated from first logical non-header row
+            if (previousTableNonHeaderRow.isEmpty() && previousTable.getRowCount() == 1 && previousTable.getRows()
+                    .get(0)
+                    .size() == tableNonHeaderRow.size()) {
+                previousTableNonHeaderRow = previousTable.getRows().get(0).stream().map(cell -> {
+                    Cell fakeCell = new Cell(cell.getPoints()[0], cell.getPoints()[2]);
+                    fakeCell.setHeaderCells(Collections.singletonList(cell));
+                    return fakeCell;
+                }).collect(Collectors.toList());
+            }
+            if (previousTableNonHeaderRow.size() == tableNonHeaderRow.size()) {
+                for (int i = currentTable.getRowCount() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of table
+                    List<Cell> row = currentTable.getRows().get(i);
+                    if (row.size() == tableNonHeaderRow.size() && row.stream()
+                            .allMatch(cell -> cell.getHeaderCells().isEmpty())) {
+                        for (int j = 0; j < row.size(); j++) {
+                            row.get(j).setHeaderCells(previousTableNonHeaderRow.get(j).getHeaderCells());
+                        }
+                    }
+                }
+            }
+        }
+    }
+
+
+    private Paragraph buildTextBlock(List<AbstractTextContainer> wordBlockList, String lastHeadline) {

        Paragraph paragraph = new Paragraph();
        TextBlock textBlock = null;
@@ -84,12 +122,6 @@ public class SectionsBuilderService {
                } else {
                    table.setHeadline("Table in: " + lastHeadline);
                }
-                // Distribute header information for subsequent tables
-                if (previousTable != null && hasInvalidHeaderInformation(table) && hasValidHeaderInformation(previousTable) &&
-                        (previousTable.isVerticalHeader() && previousTable.getRowCount() == table.getRowCount() ||
-                                previousTable.getColCount() == table.getColCount())) {
-                    table.setHeaders(previousTable.getHeaders());
-                }

                if (textBlock != null && !alreadyAdded) {
                    paragraph.getPageBlocks().add(textBlock);
@@ -141,14 +173,32 @@ public class SectionsBuilderService {

    private boolean hasInvalidHeaderInformation(Table table) {

-        if (CollectionUtils.isEmpty(table.getHeaders())) {
-            return true;
-        }
-        if (table.getHeaders().stream().anyMatch(StringUtils::isEmpty)) {
-            return true;
+        return table.getRows().stream()
+                .flatMap(row -> row.stream()
+                        .filter(cell -> CollectionUtils.isNotEmpty(cell.getHeaderCells())))
+                .findAny()
+                .isEmpty();
+
+    }
+
+
+    private List<Cell> getRowWithNonHeaderCells(Table table) {
+
+        for (int i = table.getRowCount() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of table
+            List<Cell> row = table.getRows().get(i);
+            boolean allNonHeader = true;
+            for (Cell cell : row) {
+                if (cell.isHeaderCell()) {
+                    allNonHeader = false;
+                    break;
+                }
+            }
+            if (allNonHeader) {
+                return row;
+            }
        }

-        return false;
+        return Collections.emptyList();

    }

@@ -16,10 +16,14 @@ public class Cell extends Rectangle {

    private List<TextBlock> textBlocks = new ArrayList<>();

+    private List<Cell> headerCells = new ArrayList<>();
+
+    private boolean isHeaderCell;

    public Cell(Point2D topLeft, Point2D bottomRight) {

-        super((float) topLeft.getY(), (float) topLeft.getX(), (float) (bottomRight.getX() - topLeft.getX()), (float) (bottomRight
+        super((float) topLeft.getY(), (float) topLeft.getX(), (float) (bottomRight.getX() - topLeft.getX()),
+                (float) (bottomRight
                .getY() - topLeft.getY()));
    }

@@ -29,4 +33,4 @@ public class Cell extends Rectangle {
        textBlocks.add(textBlock);
    }

-}
+}
@@ -0,0 +1,22 @@
+package com.iqser.red.service.redaction.v1.server.tableextraction.model;
+
+import lombok.RequiredArgsConstructor;
+import lombok.Value;
+
+@Value
+@RequiredArgsConstructor
+public class CellPosition implements Comparable<CellPosition> {
+
+    int row;
+
+    int col;
+
+
+    @Override
+    public int compareTo(CellPosition other) {
+
+        int rowDiff = row - other.row;
+        return rowDiff != 0 ? rowDiff : col - other.col;
+    }
+
+}
@@ -8,12 +8,10 @@ import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
 import java.util.TreeMap;
-import java.util.stream.Collectors;

 import org.apache.commons.collections4.CollectionUtils;

 import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
-import com.iqser.red.service.redaction.v1.server.redaction.utils.TextNormalizationUtilities;
 import com.iqser.red.service.redaction.v1.server.tableextraction.utils.Utils;

 import lombok.Getter;
@@ -31,22 +29,18 @@ public class Table extends AbstractTextContainer {
    @Setter
    private String headline;

-    @Getter
-    private int rowCount;
+    private int unrotatedRowCount;

-    @Getter
-    private int colCount;
+    private int unrotatedColCount;
+
+    private int rowCount = -1;
+
+    private int colCount = -1;

    private final int rotation;

    private List<List<Cell>> rows;

-    @Getter
-    @Setter
-    private List<String> headers;
-
-    @Getter
-    private boolean verticalHeader;

    public Table(List<Cell> cells, Rectangle area, int rotation) {

@@ -65,7 +59,7 @@ public class Table extends AbstractTextContainer {

        if (rows == null) {
            rows = computeRows();
-            headers = computeHeaders();
+            computeHeaders();
        }

        return rows;
@@ -73,72 +67,78 @@ public class Table extends AbstractTextContainer {
    }


+    public int getRowCount() {
+
+        if (rowCount == -1) {
+            rowCount = getRows().size();
+        }
+        return rowCount;
+    }
+
+
+    public int getColCount() {
+
+        if (colCount == -1) {
+            colCount = getRows().stream().mapToInt(List::size).max().orElse(0);
+        }
+        return colCount;
+
+    }
+
+
    /**
     * Detect header cells (either first row or first column):
     * Column is marked as header if cell text is bold and row cell text is not bold.
     * Defaults to row.
     */
-    private List<String> computeHeaders() {
+    private void computeHeaders() {

-        boolean allBold = true;
-        if (rows.isEmpty()) {
-            return Collections.emptyList();
+        if (rows == null) {
+            rows = computeRows();
        }
-        List<Cell> rowCells = rows.get(0);
-        for (Cell cell : rowCells) {
-            if (cell == null || CollectionUtils.isEmpty(cell.getTextBlocks()) ||
-                    !cell.getTextBlocks().get(0).getMostPopularWordStyle().equals("bold")) {
-                allBold = false;
-                break;
-            }
-        }
-        if (!allBold) {
-            allBold = true;
-            List<Cell> firstColCells = new ArrayList<>();
-            for (List<Cell> row : rows) {
-                Cell firstInRow = row.get(0);
-                if (firstInRow == null || CollectionUtils.isEmpty(firstInRow.getTextBlocks()) ||
-                        !firstInRow.getTextBlocks().get(0).getMostPopularWordStyle().equals("bold")) {
-                    allBold = false;
-                    break;
-                }
-                firstColCells.add(firstInRow);
-            }
-            if (allBold) {
-                log.info("Headers are in first column");
-                verticalHeader = true;
-                return firstColCells.stream().map(cell -> {
-                    if (CollectionUtils.isNotEmpty(cell.getTextBlocks())) {
-                        return TextNormalizationUtilities.removeHyphenLineBreaks(cell.getTextBlocks().get(0).getText())
-                                .replaceAll("\n", " ")
-                                .replaceAll(" ", " ");
+        // A bold cell is a header cell as long as every cell to the left/top is bold, too
+        // we move from left to right and top to bottom
+        for (int rowIndex = 0; rowIndex < rows.size(); rowIndex++) {
+            List<Cell> rowCells = rows.get(rowIndex);
+            for (int colIndex = 0; colIndex < rowCells.size(); colIndex++) {
+                Cell cell = rowCells.get(colIndex);
+                List<Cell> cellsToTheLeft = rowCells.subList(0, colIndex);
+                Cell lastHeaderCell = null;
+                for (Cell leftCell : cellsToTheLeft) {
+                    if (leftCell.isHeaderCell()) {
+                        lastHeaderCell = leftCell;
                    } else {
-                        return null;
+                        break;
                    }
-                }).collect(Collectors.toList());
-            } else {
-                log.info("Headers are defaulted in first row.");
-                return rowCells.stream().map(cell -> {
-                    if (cell != null && CollectionUtils.isNotEmpty(cell.getTextBlocks())) {
-                        return TextNormalizationUtilities.removeHyphenLineBreaks(cell.getTextBlocks().get(0).getText())
-                                .replaceAll("\n", " ")
-                                .replaceAll(" ", " ");
-                    } else {
-                        return null;
-                    }
-                }).collect(Collectors.toList());
-            }
-        } else {
-            log.info("Headers are in first row.");
-            return rowCells.stream().map(cell -> {
-                if (CollectionUtils.isNotEmpty(cell.getTextBlocks())) {
-                    return TextNormalizationUtilities.removeHyphenLineBreaks(cell.getTextBlocks().get(0).getText())
-                            .replaceAll("\n", " ")
-                            .replaceAll(" ", " ");
-                } else {
-                    return null;
                }
-            }).collect(Collectors.toList());
+                if (lastHeaderCell != null) {
+                    cell.getHeaderCells().add(lastHeaderCell);
+                }
+                List<Cell> cellsToTheTop = new ArrayList<>();
+                for (int i = 0; i < rowIndex; i++) {
+                    try {
+                        cellsToTheTop.add(rows.get(i).get(colIndex));
+                    } catch (IndexOutOfBoundsException e) {
+                        log.debug("No cell {} in row {}, ignoring.", colIndex, rowIndex);
+                    }
+                }
+                for (Cell topCell : cellsToTheTop) {
+                    if (topCell.isHeaderCell()) {
+                        lastHeaderCell = topCell;
+                    } else {
+                        break;
+                    }
+                }
+                if (lastHeaderCell != null) {
+                    cell.getHeaderCells().add(lastHeaderCell);
+                }
+                if (CollectionUtils.isNotEmpty(cell.getTextBlocks()) && cell.getTextBlocks()
+                        .get(0)
+                        .getMostPopularWordStyle()
+                        .equals("bold")) {
+                    cell.setHeaderCell(true);
+                }
+            }
        }

    }
@@ -148,29 +148,35 @@ public class Table extends AbstractTextContainer {

        List<List<Cell>> rows = new ArrayList<>();
        if (rotation == 90) {
-            for (int i = 0; i < colCount; i++) { // rows
+            for (int i = 0; i < unrotatedColCount; i++) { // rows
                List<Cell> lastRow = new ArrayList<>();
-                for (int j = rowCount - 1; j >= 0; j--) { // cols
+                for (int j = unrotatedRowCount - 1; j >= 0; j--) { // cols
                    Cell cell = cells.get(new CellPosition(j, i));
-                    lastRow.add(cell);
+                    if (cell != null) {
+                        lastRow.add(cell);
+                    }
                }
                rows.add(lastRow);
            }
        } else if (rotation == 270) {
-            for (int i = colCount - 1; i >= 0; i--) { // rows
+            for (int i = unrotatedColCount - 1; i >= 0; i--) { // rows
                List<Cell> lastRow = new ArrayList<>();
-                for (int j = 0; j < rowCount; j++) { // cols
+                for (int j = 0; j < unrotatedRowCount; j++) { // cols
                    Cell cell = cells.get(new CellPosition(i, j));
-                    lastRow.add(cell);
+                    if (cell != null) {
+                        lastRow.add(cell);
+                    }
                }
                rows.add(lastRow);
            }
        } else {
-            for (int i = 0; i < rowCount; i++) {
+            for (int i = 0; i < unrotatedRowCount; i++) {
                List<Cell> lastRow = new ArrayList<>();
-                for (int j = 0; j < colCount; j++) {
+                for (int j = 0; j < unrotatedColCount; j++) {
                    Cell cell = cells.get(new CellPosition(i, j)); // JAVA_8 use getOrDefault()
-                    lastRow.add(cell);
+                    if (cell != null) {
+                        lastRow.add(cell);
+                    }
                }
                rows.add(lastRow);
            }
@@ -183,8 +189,8 @@ public class Table extends AbstractTextContainer {

    private void add(Cell chunk, int row, int col) {

-        rowCount = Math.max(rowCount, row + 1);
-        colCount = Math.max(colCount, col + 1);
+        unrotatedRowCount = Math.max(unrotatedRowCount, row + 1);
+        unrotatedColCount = Math.max(unrotatedColCount, col + 1);

        CellPosition cp = new CellPosition(row, col);
        cells.put(cp, chunk);
@@ -220,20 +226,21 @@ public class Table extends AbstractTextContainer {
            while (rowCells.hasNext()) {
                Cell cell = rowCells.next();
                if (i > 0) {
-                    List<List<Cell>> others = rowsOfCells(si.contains(new Rectangle(cell.getBottom(), si.getBounds()
-                            .getLeft(), cell.getLeft() - si.getBounds().getLeft() + 1, si.getBounds().getBottom() - cell
-                            .getBottom())));
+                    Rectangle rectangle = new Rectangle(cell.getBottom(),
+                            si.getBounds().getLeft(),
+                            cell.getLeft() - si.getBounds().getLeft() + 1,
+                            si.getBounds().getBottom() - cell.getBottom());
+                    List<List<Cell>> others = rowsOfCells(si.contains(rectangle));

                    for (List<Cell> r : others) {
                        jumpToColumn = Math.max(jumpToColumn, r.size());
                    }
-                }

-                while (startColumn != jumpToColumn) {
-                    add(previousNonNullCellForColumnIndex.get(startColumn), i, startColumn);
-                    startColumn++;
+                    while (startColumn != jumpToColumn) {
+                        add(previousNonNullCellForColumnIndex.get(startColumn), i, startColumn);
+                        startColumn++;
+                    }
                }
-
                add(cell, i, startColumn);
                previousNonNullCellForColumnIndex.put(startColumn, cell);
                startColumn++;
@@ -243,26 +250,23 @@ public class Table extends AbstractTextContainer {
    }


-    private static List<List<Cell>> rowsOfCells(List<Cell> cells) {
+    private List<List<Cell>> rowsOfCells(List<Cell> cells) {

-        Cell c;
-        float lastTop;
        List<List<Cell>> rv = new ArrayList<>();
-        List<Cell> lastRow;

        if (cells.isEmpty()) {
            return rv;
        }
-
        cells.sort(Comparator.comparingDouble(Rectangle::getLeft));

-        cells.sort(Collections.reverseOrder((arg0, arg1) -> Float.compare(Utils.round(arg0.getBottom(), 2), Utils.round(arg1
+        cells.sort(Collections.reverseOrder((arg0, arg1) -> Float.compare(Utils.round(arg0.getBottom(), 2),
+                Utils.round(arg1
                .getBottom(), 2))));

        Iterator<Cell> iter = cells.iterator();
-        c = iter.next();
-        lastTop = c.getBottom();
-        lastRow = new ArrayList<>();
+        Cell c = iter.next();
+        float lastTop = c.getBottom();
+        List<Cell> lastRow = new ArrayList<>();
        lastRow.add(c);
        rv.add(lastRow);

@@ -349,51 +353,4 @@ public class Table extends AbstractTextContainer {
        return sb.toString();
    }

-
-    static class CellPosition implements Comparable<CellPosition> {
-
-        CellPosition(int row, int col) {
-
-            this.row = row;
-            this.col = col;
-        }
-
-
-        final int row;
-        final int col;
-
-
-        @Override
-        public int hashCode() {
-
-            return row + 101 * col;
-        }
-
-
-        @Override
-        public boolean equals(Object obj) {
-
-            if (this == obj) {
-                return true;
-            }
-            if (obj == null) {
-                return false;
-            }
-            if (getClass() != obj.getClass()) {
-                return false;
-            }
-            CellPosition other = (CellPosition) obj;
-            return row == other.row && col == other.col;
-        }
-
-
-        @Override
-        public int compareTo(CellPosition other) {
-
-            int rowDiff = row - other.row;
-            return rowDiff != 0 ? rowDiff : col - other.col;
-        }
-
-    }
-
 }
@@ -2,7 +2,6 @@ package com.iqser.red.service.redaction.v1.server.tableextraction.service;

 import java.awt.geom.Point2D;
 import java.util.ArrayList;
-import java.util.Collections;
 import java.util.Comparator;
 import java.util.HashMap;
 import java.util.HashSet;
@@ -25,26 +24,28 @@ import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table;
 import com.iqser.red.service.redaction.v1.server.tableextraction.utils.Utils;

@Service
-@SuppressWarnings("all")
 public class TableExtractionService {

-    public void extractTables(CleanRulings cleanRulings, Page page){
+    public void extractTables(CleanRulings cleanRulings, Page page) {

        List<Cell> cells = findCells(cleanRulings.getHorizontal(), cleanRulings.getVertical());

-        Iterator<AbstractTextContainer> itty = page.getTextBlocks().iterator();
-        while (itty.hasNext()) {
-            TextBlock textBlock = (TextBlock) itty.next();
+        for (AbstractTextContainer abstractTextContainer : page.getTextBlocks()) {
+            TextBlock textBlock = (TextBlock) abstractTextContainer;
            for (Cell cell : cells) {
-                if (cell.intersects(textBlock.getMinX(), textBlock.getMinY(), textBlock.getWidth(), textBlock.getHeight())) {
+                if (cell.intersects(textBlock.getMinX(), textBlock.getMinY(), textBlock.getWidth(),
+                        textBlock.getHeight())) {
                    cell.addTextBlock(textBlock);
                    break;
                }
            }
        }

-        List<Rectangle> spreadsheetAreas = findSpreadsheetsFromCells(cells)
-                .stream()
+        cells = new ArrayList<>(new HashSet<>(cells));
+        Utils.sort(cells, Rectangle.ILL_DEFINED_ORDER);
+
+
+        List<Rectangle> spreadsheetAreas = findSpreadsheetsFromCells(cells).stream()
                .filter(r -> r.getWidth() > 0f && r.getHeight() > 0f)
                .collect(Collectors.toList());

@@ -63,9 +64,9 @@ public class TableExtractionService {
        for (Table table : tables) {
            int position = -1;

-            itty = page.getTextBlocks().iterator();
+            Iterator<AbstractTextContainer> itty = page.getTextBlocks().iterator();
            while (itty.hasNext()) {
-                AbstractTextContainer textBlock = (AbstractTextContainer) itty.next();
+                AbstractTextContainer textBlock = itty.next();
                if (table.contains(textBlock)) {
                    if (position == -1) {
                        position = page.getTextBlocks().indexOf(textBlock);
@@ -79,17 +80,18 @@ public class TableExtractionService {
        }
    }

+
    public List<Cell> findCells(List<Ruling> horizontalRulingLines, List<Ruling> verticalRulingLines) {
+
        List<Cell> cellsFound = new ArrayList<>();
-        Map<Point2D, Ruling[]> intersectionPoints = Ruling.findIntersections(horizontalRulingLines, verticalRulingLines);
+        Map<Point2D, Ruling[]> intersectionPoints = Ruling.findIntersections(horizontalRulingLines,
+                verticalRulingLines);
        List<Point2D> intersectionPointsList = new ArrayList<>(intersectionPoints.keySet());
-        Collections.sort(intersectionPointsList, POINT_COMPARATOR);
-        boolean doBreak;
+        intersectionPointsList.sort(POINT_COMPARATOR);

        for (int i = 0; i < intersectionPointsList.size(); i++) {
            Point2D topLeft = intersectionPointsList.get(i);
            Ruling[] hv = intersectionPoints.get(topLeft);
-            doBreak = false;

            // CrossingPointsDirectlyBelow( topLeft );
            List<Point2D> xPoints = new ArrayList<>();
@@ -106,10 +108,6 @@ public class TableExtractionService {
            }
            outer:
            for (Point2D xPoint : xPoints) {
-                if (doBreak) {
-                    break;
-                }
-
                // is there a vertical edge b/w topLeft and xPoint?
                if (!hv[1].equals(intersectionPoints.get(xPoint)[1])) {
                    continue;
@@ -120,11 +118,9 @@ public class TableExtractionService {
                        continue;
                    }
                    Point2D btmRight = new Point2D.Float((float) yPoint.getX(), (float) xPoint.getY());
-                    if (intersectionPoints.containsKey(btmRight)
-                            && intersectionPoints.get(btmRight)[0].equals(intersectionPoints.get(xPoint)[0])
-                            && intersectionPoints.get(btmRight)[1].equals(intersectionPoints.get(yPoint)[1])) {
+                    if (intersectionPoints.containsKey(btmRight) && intersectionPoints.get(btmRight)[0].equals(intersectionPoints
+                            .get(xPoint)[0]) && intersectionPoints.get(btmRight)[1].equals(intersectionPoints.get(yPoint)[1])) {
                        cellsFound.add(new Cell(topLeft, btmRight));
-                        doBreak = true;
                        break outer;
                    }
                }
@@ -139,7 +135,7 @@ public class TableExtractionService {
    }


-    public List<Rectangle> findSpreadsheetsFromCells(List<? extends Rectangle> cells) {
+    private List<Rectangle> findSpreadsheetsFromCells(List<? extends Rectangle> cells) {
        // via: http://stackoverflow.com/questions/13746284/merging-multiple-adjacent-rectangles-into-one-polygon
        List<Rectangle> rectangles = new ArrayList<>();
        Set<Point2D> pointSet = new HashSet<>();
@@ -147,10 +143,6 @@ public class TableExtractionService {
        Map<Point2D, Point2D> edgesV = new HashMap<>();
        int i = 0;

-        cells = new ArrayList<>(new HashSet<>(cells));
-
-        Utils.sort(cells, Rectangle.ILL_DEFINED_ORDER);
-
        for (Rectangle cell : cells) {
            for (Point2D pt : cell.getPoints()) {
                if (pointSet.contains(pt)) { // shared vertex, remove it
@@ -163,10 +155,10 @@ public class TableExtractionService {

        // X first sort
        List<Point2D> pointsSortX = new ArrayList<>(pointSet);
-        Collections.sort(pointsSortX, X_FIRST_POINT_COMPARATOR);
+        pointsSortX.sort(X_FIRST_POINT_COMPARATOR);
        // Y first sort
        List<Point2D> pointsSortY = new ArrayList<>(pointSet);
-        Collections.sort(pointsSortY, POINT_COMPARATOR);
+        pointsSortY.sort(POINT_COMPARATOR);

        while (i < pointSet.size()) {
            float currY = (float) pointsSortY.get(i).getY();
@@ -203,13 +195,12 @@ public class TableExtractionService {
                    nextVertex = edgesV.get(curr.point);
                    edgesV.remove(curr.point);
                    lastAddedVertex = new PolygonVertex(nextVertex, Direction.VERTICAL);
-                    polygon.add(lastAddedVertex);
                } else {
                    nextVertex = edgesH.get(curr.point);
                    edgesH.remove(curr.point);
                    lastAddedVertex = new PolygonVertex(nextVertex, Direction.HORIZONTAL);
-                    polygon.add(lastAddedVertex);
                }
+                polygon.add(lastAddedVertex);

                if (lastAddedVertex.equals(polygon.get(0))) {
                    // closed polygon
@@ -227,10 +218,10 @@ public class TableExtractionService {

        // calculate grid-aligned minimum area rectangles for each found polygon
        for (List<PolygonVertex> poly : polygons) {
-            float top = java.lang.Float.MAX_VALUE;
-            float left = java.lang.Float.MAX_VALUE;
-            float bottom = java.lang.Float.MIN_VALUE;
-            float right = java.lang.Float.MIN_VALUE;
+            float top = Float.MAX_VALUE;
+            float left = Float.MAX_VALUE;
+            float bottom = Float.MIN_VALUE;
+            float right = Float.MIN_VALUE;
            for (PolygonVertex pt : poly) {
                top = (float) Math.min(top, pt.point.getY());
                left = (float) Math.min(left, pt.point.getX());
@@ -244,69 +235,66 @@ public class TableExtractionService {
    }


-    private static final Comparator<Point2D> X_FIRST_POINT_COMPARATOR = new Comparator<Point2D>() {
-        @Override
-        public int compare(Point2D arg0, Point2D arg1) {
-            int rv = 0;
-            float arg0X = Utils.round(arg0.getX(), 2);
-            float arg0Y = Utils.round(arg0.getY(), 2);
-            float arg1X = Utils.round(arg1.getX(), 2);
-            float arg1Y = Utils.round(arg1.getY(), 2);
+    private static final Comparator<Point2D> X_FIRST_POINT_COMPARATOR = (arg0, arg1) -> {

-            if (arg0X > arg1X) {
-                rv = 1;
-            } else if (arg0X < arg1X) {
-                rv = -1;
-            } else if (arg0Y > arg1Y) {
-                rv = 1;
-            } else if (arg0Y < arg1Y) {
-                rv = -1;
-            }
-            return rv;
+        int rv = 0;
+        float arg0X = Utils.round(arg0.getX(), 2);
+        float arg0Y = Utils.round(arg0.getY(), 2);
+        float arg1X = Utils.round(arg1.getX(), 2);
+        float arg1Y = Utils.round(arg1.getY(), 2);
+
+        if (arg0X > arg1X) {
+            rv = 1;
+        } else if (arg0X < arg1X) {
+            rv = -1;
+        } else if (arg0Y > arg1Y) {
+            rv = 1;
+        } else if (arg0Y < arg1Y) {
+            rv = -1;
        }
+        return rv;
    };

+    private static final Comparator<Point2D> POINT_COMPARATOR = (arg0, arg1) -> {

-    private static final Comparator<Point2D> POINT_COMPARATOR = new Comparator<Point2D>() {
-        @Override
-        public int compare(Point2D arg0, Point2D arg1) {
-            int rv = 0;
-            float arg0X = Utils.round(arg0.getX(), 2);
-            float arg0Y = Utils.round(arg0.getY(), 2);
-            float arg1X = Utils.round(arg1.getX(), 2);
-            float arg1Y = Utils.round(arg1.getY(), 2);
+        int rv = 0;
+        float arg0X = Utils.round(arg0.getX(), 2);
+        float arg0Y = Utils.round(arg0.getY(), 2);
+        float arg1X = Utils.round(arg1.getX(), 2);
+        float arg1Y = Utils.round(arg1.getY(), 2);

-
-            if (arg0Y > arg1Y) {
-                rv = 1;
-            } else if (arg0Y < arg1Y) {
-                rv = -1;
-            } else if (arg0X > arg1X) {
-                rv = 1;
-            } else if (arg0X < arg1X) {
-                rv = -1;
-            }
-            return rv;
+        if (arg0Y > arg1Y) {
+            rv = 1;
+        } else if (arg0Y < arg1Y) {
+            rv = -1;
+        } else if (arg0X > arg1X) {
+            rv = 1;
+        } else if (arg0X < arg1X) {
+            rv = -1;
        }
+        return rv;
    };

-
    private enum Direction {
-        HORIZONTAL,
-        VERTICAL
+        HORIZONTAL, VERTICAL
    }

    static class PolygonVertex {
+
        Point2D point;
        Direction direction;

-        public PolygonVertex(Point2D point, Direction direction) {
+
+        PolygonVertex(Point2D point, Direction direction) {
+
            this.direction = direction;
            this.point = point;
        }

+
        @Override
        public boolean equals(Object other) {
+
            if (this == other) {
                return true;
            }
@@ -316,15 +304,21 @@ public class TableExtractionService {
            return this.point.equals(((PolygonVertex) other).point);
        }

+
        @Override
        public int hashCode() {
+
            return this.point.hashCode();
        }

+
        @Override
        public String toString() {
-            return String.format("%s[point=%s,direction=%s]", this.getClass().getName(), this.point.toString(), this.direction.toString());
+
+            return String.format("%s[point=%s,direction=%s]", this.getClass()
+                    .getName(), this.point.toString(), this.direction.toString());
        }
+
    }

 }
@@ -227,6 +227,7 @@ public class RedactionIntegrationTest {
    @Test
    public void noExceptionShouldBeThrownForAnyFiles() throws IOException {

+        System.out.println("noExceptionShouldBeThrownForAnyFiles");
        ClassLoader loader = getClass().getClassLoader();
        URL url = loader.getResource("files");
        File[] files = new File(url.getPath()).listFiles();
@@ -266,6 +267,7 @@ public class RedactionIntegrationTest {
    @Test
    public void redactionTest() throws IOException {

+        System.out.println("redactionTest");
        long start = System.currentTimeMillis();
        ClassPathResource pdfFileResource = new ClassPathResource("files/Trinexapac/96 Trinexapac-ethyl_RAR_09_Volume_3CA_B-7_2018-02-23.pdf");

@@ -289,8 +291,9 @@ public class RedactionIntegrationTest {
    @Test
    public void testTableRedaction() throws IOException {

+        System.out.println("testTableRedaction");
        long start = System.currentTimeMillis();
-        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Single Table.pdf");
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Metolachlor/S-Metolachlor_RAR_02_Volume_2_2018-09-06.pdf");

        RedactionRequest request = RedactionRequest.builder()
                .document(IOUtils.toByteArray(pdfFileResource.getInputStream()))
@@ -311,6 +314,7 @@ public class RedactionIntegrationTest {
    @Test
    public void testManualRedaction() throws IOException {

+        System.out.println("testManualRedaction");
        long start = System.currentTimeMillis();
        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Single Table.pdf");

@@ -345,6 +349,7 @@ public class RedactionIntegrationTest {
    @Test
    public void classificationTest() throws IOException {

+        System.out.println("classificationTest");
        ClassPathResource pdfFileResource = new ClassPathResource("files/Fludioxonil/51 " +
                "Fludioxonil_RAR_02_Volume_2_2018-02-21.pdf");

@@ -363,6 +368,7 @@ public class RedactionIntegrationTest {
    @Test
    public void sectionsTest() throws IOException {

+        System.out.println("sectionsTest");
        ClassPathResource pdfFileResource = new ClassPathResource("files/Fludioxonil/51 " +
                "Fludioxonil_RAR_02_Volume_2_2018-02-21.pdf");

@@ -381,6 +387,7 @@ public class RedactionIntegrationTest {
    @Test
    public void htmlTablesTest() throws IOException {

+        System.out.println("htmlTablesTest");
        ClassPathResource pdfFileResource = new ClassPathResource("files/Fludioxonil/51 " +
                "Fludioxonil_RAR_02_Volume_2_2018-02-21.pdf");

@@ -399,6 +406,7 @@ public class RedactionIntegrationTest {
    @Test
    public void htmlTableRotationTest() throws IOException {

+        System.out.println("htmlTableRotationTest");
        ClassPathResource pdfFileResource = new ClassPathResource("files/Metolachlor/S" +
                "-Metolachlor_RAR_02_Volume_2_2018-09-06.pdf");

@@ -10,13 +10,16 @@ import java.io.InputStream;
 import java.io.InputStreamReader;
 import java.net.URL;
 import java.nio.charset.StandardCharsets;
+import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.Collections;
 import java.util.HashSet;
 import java.util.Set;
+import java.util.concurrent.atomic.AtomicLong;

 import org.apache.commons.io.IOUtils;
 import org.apache.pdfbox.pdmodel.PDDocument;
+import org.junit.Before;
 import org.junit.Test;
 import org.junit.runner.RunWith;
 import org.kie.api.KieServices;
@@ -45,14 +48,15 @@ import com.iqser.red.service.redaction.v1.server.redaction.model.Entity;
 import com.iqser.red.service.redaction.v1.server.redaction.utils.ResourceLoader;
 import com.iqser.red.service.redaction.v1.server.segmentation.PdfSegmentationService;

-@RunWith(SpringRunner.class)
@SpringBootTest
+@RunWith(SpringRunner.class)
 public class EntityRedactionServiceTest {

    private static final String DEFAULT_RULES = loadFromClassPath("drools/rules.drl");
    private static final String NAME_CODE = "name";
    private static final String ADDRESS_CODE = "address";

+    private static final AtomicLong DICTIONARY_VERSION = new AtomicLong();
    @MockBean
    private DictionaryClient dictionaryClient;

@@ -112,21 +116,157 @@ public class EntityRedactionServiceTest {
                .document(IOUtils.toByteArray(pdfFileResource.getInputStream()))
                .build();

+        DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
+                .entries(Arrays.asList("Casey, H.W.", "O’Loughlin,  C.K.", "Salamon, C.M.", "Smith, S.H."))
+                .build();
+        when(dictionaryClient.getVersion()).thenReturn(DICTIONARY_VERSION.incrementAndGet());
+        when(dictionaryClient.getDictionaryForType(NAME_CODE)).thenReturn(dictionaryResponse);
+        DictionaryResponse addressResponse = DictionaryResponse.builder()
+                .entries(Collections.singletonList("Toxigenics, Inc., Decatur, IL 62526, USA"))
+                .build();
+        when(dictionaryClient.getDictionaryForType(ADDRESS_CODE)).thenReturn(addressResponse);
+
+        try (PDDocument pdDocument = PDDocument.load(new ByteArrayInputStream(redactionRequest.getDocument()))) {
+            Document classifiedDoc = pdfSegmentationService.parseDocument(pdDocument);
+            entityRedactionService.processDocument(classifiedDoc, null);
+            assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
+            assertThat(classifiedDoc.getEntities().get(1)).hasSize(7);// 3 names, 1 address, 1 Y and 2 N entities
+        }
+    }
+
+
+    @Test
+    public void testTrueNegativesInTable() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Cyprodinil/40 Cyprodinil - EU AIR3 - LCA Section 1" +
+                " Supplement - Identity of the active substance - Reference list.pdf");
+        when(dictionaryClient.getVersion()).thenReturn(DICTIONARY_VERSION.incrementAndGet());
+        DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
+                .entries(new ArrayList<>(ResourceLoader.load("dictionaries/names.txt")))
+                .build();
+        when(dictionaryClient.getDictionaryForType(NAME_CODE)).thenReturn(dictionaryResponse);
+        DictionaryResponse addressResponse = DictionaryResponse.builder()
+                .entries(new ArrayList<>(ResourceLoader.load("dictionaries/addresses.txt")))
+                .build();
+        when(dictionaryClient.getDictionaryForType(ADDRESS_CODE)).thenReturn(addressResponse);
+        try (PDDocument pdDocument = PDDocument.load(pdfFileResource.getInputStream())) {
+            Document classifiedDoc = pdfSegmentationService.parseDocument(pdDocument);
+            entityRedactionService.processDocument(classifiedDoc, null);
+            assertThat(classifiedDoc.getEntities()
+                    .entrySet()
+                    .stream()
+                    .noneMatch(entry -> entry.getValue().stream().anyMatch(e -> e.getMatchedRule() == 9))).isTrue();
+        }
+        pdfFileResource = new ClassPathResource("files/Compounds/27 A8637C - EU AIR3 - MCP Section 1 - Identity of " +
+                "the plant protection product.pdf");
+        try (PDDocument pdDocument = PDDocument.load(pdfFileResource.getInputStream())) {
+            Document classifiedDoc = pdfSegmentationService.parseDocument(pdDocument);
+            entityRedactionService.processDocument(classifiedDoc, null);
+            assertThat(classifiedDoc.getEntities()
+                    .entrySet()
+                    .stream()
+                    .noneMatch(entry -> entry.getValue().stream().anyMatch(e -> e.getMatchedRule() == 9))).isTrue();
+        }
+    }
+
+    @Test
+    public void testFalsePositiveInWrongCell() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Row With Ambiguous Redaction.pdf");
+        when(dictionaryClient.getVersion()).thenReturn(DICTIONARY_VERSION.incrementAndGet());
+        DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
+                .entries(new ArrayList<>(ResourceLoader.load("dictionaries/names.txt")))
+                .build();
+        when(dictionaryClient.getDictionaryForType(NAME_CODE)).thenReturn(dictionaryResponse);
+        DictionaryResponse addressResponse = DictionaryResponse.builder()
+                .entries(new ArrayList<>(ResourceLoader.load("dictionaries/addresses.txt")))
+                .build();
+        when(dictionaryClient.getDictionaryForType(ADDRESS_CODE)).thenReturn(addressResponse);
+        try (PDDocument pdDocument = PDDocument.load(pdfFileResource.getInputStream())) {
+            Document classifiedDoc = pdfSegmentationService.parseDocument(pdDocument);
+            entityRedactionService.processDocument(classifiedDoc, null);
+            assertThat(classifiedDoc.getEntities()).hasSize(1); // two pages
+            assertThat(classifiedDoc.getEntities().get(1).stream()
+                    .filter(entity -> entity.getMatchedRule() == 9)
+                    .count()).isEqualTo(10);
+        }
+
+    }
+
+
+    @Test
+    public void headerPropagation() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Header Propagation.pdf");
+
+        DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
+                .entries(Arrays.asList("Bissig R.", "Thanei P."))
+                .build();
+
+        when(dictionaryClient.getVersion()).thenReturn(DICTIONARY_VERSION.incrementAndGet());
+        when(dictionaryClient.getDictionaryForType(NAME_CODE)).thenReturn(dictionaryResponse);
+        DictionaryResponse addressResponse = DictionaryResponse.builder()
+                .entries(Collections.singletonList("Novartis Crop Protection AG, Basel, Switzerland"))
+                .build();
+        when(dictionaryClient.getDictionaryForType(ADDRESS_CODE)).thenReturn(addressResponse);
+
+        try (PDDocument pdDocument = PDDocument.load(pdfFileResource.getInputStream())) {
+            Document classifiedDoc = pdfSegmentationService.parseDocument(pdDocument);
+            entityRedactionService.processDocument(classifiedDoc, null);
+            assertThat(classifiedDoc.getEntities()).hasSize(2); // two pages
+            assertThat(classifiedDoc.getEntities().get(1).stream().filter(entity -> entity.getMatchedRule() == 9).count()).isEqualTo(8);
+            assertThat(classifiedDoc.getEntities().get(2).stream().filter(entity -> entity.getMatchedRule() == 9).count()).isEqualTo(4);
+        }
+    }
+
+
+    @Test
+    public void testNGuideline() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Empty Tabular Data.pdf");
+
+        DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
+                .entries(Collections.singletonList("Aldershof S."))
+                .build();
+
+        when(dictionaryClient.getVersion()).thenReturn(DICTIONARY_VERSION.incrementAndGet());
+        when(dictionaryClient.getDictionaryForType(NAME_CODE)).thenReturn(dictionaryResponse);
+        DictionaryResponse addressResponse = DictionaryResponse.builder()
+                .entries(Collections.singletonList("Novartis Crop Protection AG, Basel, Switzerland"))
+                .build();
+        when(dictionaryClient.getDictionaryForType(ADDRESS_CODE)).thenReturn(addressResponse);
+
+        try (PDDocument pdDocument = PDDocument.load(pdfFileResource.getInputStream())) {
+            Document classifiedDoc = pdfSegmentationService.parseDocument(pdDocument);
+            entityRedactionService.processDocument(classifiedDoc, null);
+            assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
+            assertThat(classifiedDoc.getEntities().get(1).stream().filter(entity -> entity.getMatchedRule() == 8).count()).isEqualTo(6);
+        }
+    }
+
+
+    @Before
+    public void stubRedaction() {
        String tableRules = "package drools\n" +
                "\n" +
                "import com.iqser.red.service.redaction.v1.server.redaction.model.Section\n" +
                "\n" +
                "global Section section\n" +
+                "rule \"8: Not redacted because Vertebrate Study = N\"\n" +
+                "    when\n" +
+                "        Section(rowEquals(\"Vertebrate study Y/N\", \"N\"))\n" +
+                "    then\n" +
+                "        section.redactNot(\"name\", 8, \"Not redacted because row is not a vertebrate study\");\n" +
+                "        section.redactNot(\"address\", 8, \"Not redacted because row is not a vertebrate study\");\n" +
+                "        section.highlightCell(\"Vertebrate study Y/N\", 8, \"hint_only\");\n" +
+                "    end\n" +
                "rule \"9: Redact Authors and Addresses in Reference Table, if it is a Vertebrate study\"\n" +
                "    when\n" +
-                "        Section(tabularData != null && tabularData.size() > 0\n" +
-                "            && tabularData.containsKey(\"Vertebrate study Y/N\")\n" +
-                "            && tabularData.get(\"Vertebrate study Y/N\").equals(\"Y\")\n" +
-                "        )\n" +
+                "        Section(rowEquals(\"Vertebrate study Y/N\", \"Y\"))\n" +
                "    then\n" +
                "        section.redact(\"name\", 9, \"Redacted because row is a vertebrate study\");\n" +
-                "        section.redact(\"address\", 9, \"Redacted because rows is a vertebrate study\");\n" +
-                "        section.highlightCell(\"Vertebrate study Y/N\", 9);\n" +
+                "        section.redact(\"address\", 9, \"Redacted because row is a vertebrate study\");\n" +
+                "        section.highlightCell(\"Vertebrate study Y/N\", 9, \"must_redact\");\n" +
                "    end";
        when(rulesClient.getVersion()).thenReturn(1L);
        when(rulesClient.getRules()).thenReturn(new RulesResponse(tableRules));
@@ -135,22 +275,9 @@ public class EntityRedactionServiceTest {
                        TypeResult.builder().type(NAME_CODE).color(new float[]{1, 1, 0}).build(),
                        TypeResult.builder().type(ADDRESS_CODE).color(new float[]{0, 1, 1}).build()))
                .build();
+        when(dictionaryClient.getVersion()).thenReturn(DICTIONARY_VERSION.incrementAndGet());
        when(dictionaryClient.getAllTypes()).thenReturn(typeResponse);
-        DictionaryResponse dictionaryResponse = DictionaryResponse.builder()
-                .entries(Arrays.asList("Casey, H.W.", "O’Loughlin,  C.K.", "Salamon, C.M.", "Smith, S.H."))
-                .build();
-        when(dictionaryClient.getDictionaryForType(NAME_CODE)).thenReturn(dictionaryResponse);
-        DictionaryResponse addressResponse = DictionaryResponse.builder()
-                .entries(Collections.singletonList("Toxigenics, Inc., Decatur, IL 62526, USA"))
-                .build();
-        when(dictionaryClient.getDictionaryForType(ADDRESS_CODE)).thenReturn(addressResponse);
        when(dictionaryClient.getDefaultColor()).thenReturn(new DefaultColor());
-        try (PDDocument pdDocument = PDDocument.load(new ByteArrayInputStream(redactionRequest.getDocument()))) {
-            Document classifiedDoc = pdfSegmentationService.parseDocument(pdDocument);
-            entityRedactionService.processDocument(classifiedDoc, null);
-            assertThat(classifiedDoc.getEntities()).hasSize(1); // one page
-            assertThat(classifiedDoc.getEntities().get(1)).hasSize(5); // 4 out of 5 entities recognized on page 1
-        }
    }


@@ -0,0 +1,189 @@
+package com.iqser.red.service.redaction.v1.server.segmentation;
+
+import static org.assertj.core.api.Assertions.assertThat;
+
+import java.io.IOException;
+import java.util.Collections;
+import java.util.List;
+import java.util.stream.Collectors;
+
+import org.apache.pdfbox.pdmodel.PDDocument;
+import org.junit.Test;
+import org.junit.runner.RunWith;
+import org.kie.api.runtime.KieContainer;
+import org.springframework.beans.factory.annotation.Autowired;
+import org.springframework.boot.test.context.SpringBootTest;
+import org.springframework.boot.test.mock.mockito.MockBean;
+import org.springframework.core.io.ClassPathResource;
+import org.springframework.test.context.junit4.SpringRunner;
+
+import com.iqser.red.service.redaction.v1.server.classification.model.Document;
+import com.iqser.red.service.redaction.v1.server.classification.service.BlockificationService;
+import com.iqser.red.service.redaction.v1.server.tableextraction.model.Cell;
+import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table;
+import com.iqser.red.service.redaction.v1.server.tableextraction.service.RulingCleaningService;
+import com.iqser.red.service.redaction.v1.server.tableextraction.service.TableExtractionService;
+
+@SpringBootTest
+@RunWith(SpringRunner.class)
+public class PdfSegmentationServiceTest {
+
+    @Autowired
+    private PdfSegmentationService pdfSegmentationService;
+
+    @Autowired
+    private RulingCleaningService rulingCleaningService;
+
+    @Autowired
+    private TableExtractionService tableExtractionService;
+
+    @Autowired
+    private BlockificationService blockificationService;
+
+    @MockBean
+    private KieContainer kieContainer;
+
+
+    @Test
+    public void testPDFSegmentationWithComplexTable() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Spanning Cells.pdf");
+
+        try (PDDocument pdDocument = PDDocument.load(pdfFileResource.getInputStream())) {
+            Document document = pdfSegmentationService.parseDocument(pdDocument);
+            assertThat(document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())).isNotEmpty();
+            Table table = document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())
+                    .get(0);
+            assertThat(table.getColCount()).isEqualTo(6);
+            assertThat(table.getRowCount()).isEqualTo(13);
+            assertThat(table.getRows().stream().mapToInt(List::size).sum()).isEqualTo(6 * 13);
+        }
+    }
+
+
+    @Test
+    public void testTableExtraction() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Merge Table.pdf");
+
+        try (PDDocument pdDocument = PDDocument.load(pdfFileResource.getInputStream())) {
+            Document document = pdfSegmentationService.parseDocument(pdDocument);
+            assertThat(document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())).isNotEmpty();
+            Table firstTable = document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())
+                    .get(0);
+            assertThat(firstTable.getColCount()).isEqualTo(8);
+            assertThat(firstTable.getRowCount()).isEqualTo(1);
+            Table secondTable = document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())
+                    .get(1);
+            assertThat(secondTable.getColCount()).isEqualTo(8);
+            assertThat(secondTable.getRowCount()).isEqualTo(2);
+            List<List<Cell>> firstTableHeaderCells = firstTable.getRows()
+                    .get(0)
+                    .stream()
+                    .map(Collections::singletonList)
+                    .collect(Collectors.toList());
+            assertThat(secondTable.getRows().stream()
+                    .allMatch(row -> row.stream()
+                            .map(Cell::getHeaderCells)
+                            .collect(Collectors.toList())
+                            .equals(firstTableHeaderCells)))
+                    .isTrue();
+        }
+    }
+
+
+    @Test
+    public void testMultiPageMetadataPropagation() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Merge Multi Page Table.pdf");
+
+        try (PDDocument pdDocument = PDDocument.load(pdfFileResource.getInputStream())) {
+            Document document = pdfSegmentationService.parseDocument(pdDocument);
+            assertThat(document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())).isNotEmpty();
+            Table firstTable = document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())
+                    .get(0);
+            assertThat(firstTable.getColCount()).isEqualTo(9);
+            assertThat(firstTable.getRowCount()).isEqualTo(5);
+            Table secondTable = document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())
+                    .get(1);
+            assertThat(secondTable.getColCount()).isEqualTo(9);
+            assertThat(secondTable.getRowCount()).isEqualTo(6);
+            List<List<Cell>> firstTableHeaderCells = firstTable.getRows()
+                    .get(firstTable.getRowCount() - 1)
+                    .stream()
+                    .map(Cell::getHeaderCells)
+                    .collect(Collectors.toList());
+            assertThat(secondTable.getRows().stream()
+                    .allMatch(row -> row.stream()
+                            .map(Cell::getHeaderCells)
+                            .collect(Collectors.toList())
+                            .equals(firstTableHeaderCells)))
+                    .isTrue();
+        }
+    }
+
+
+    @Test
+    public void testHeaderCellsForRotatedTable() throws IOException {
+
+        ClassPathResource pdfFileResource = new ClassPathResource("files/Minimal Examples/Rotated Table Headers.pdf");
+
+        try (PDDocument pdDocument = PDDocument.load(pdfFileResource.getInputStream())) {
+            Document document = pdfSegmentationService.parseDocument(pdDocument);
+            assertThat(document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())).isNotEmpty();
+            Table firstTable = document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())
+                    .get(0);
+            assertThat(firstTable.getColCount()).isEqualTo(8);
+            assertThat(firstTable.getRowCount()).isEqualTo(1);
+            Table secondTable = document.getParagraphs()
+                    .stream()
+                    .flatMap(paragraph -> paragraph.getTables().stream())
+                    .collect(Collectors.toList())
+                    .get(1);
+            assertThat(secondTable.getColCount()).isEqualTo(8);
+            assertThat(secondTable.getRowCount()).isEqualTo(6);
+            List<List<Cell>> firstTableHeaderCells = firstTable.getRows()
+                    .get(0)
+                    .stream()
+                    .map(Collections::singletonList)
+                    .collect(Collectors.toList());
+            assertThat(secondTable.getRows().stream()
+                    .allMatch(row -> row.stream()
+                            .map(Cell::getHeaderCells)
+                            .collect(Collectors.toList())
+                            .equals(firstTableHeaderCells)))
+                    .isTrue();
+        }
+    }
+
+}
@@ -49,64 +49,72 @@ rule "5: Do not redact in guideline sections"
 		section.redactNot("address", 5, "Section is a guideline section.");
 	end

-rule "6: Redact if must redact entry is found"
-	when
-		eval(section.contains("must_redact")==true);
-	then
-		section.redact("name", 6, "must_redact entry was found.");
-		section.redact("address", 6, "must_redact entry was found.");
-	end
-
-
-rule "7: Redact contact information, if applicant is found"
+rule "6: Redact contact information, if applicant is found"
 	when
 		eval(section.headlineContainsWord("applicant") || section.getText().contains("Applicant"));
 	then
-		section.redactLineAfter("Name:", "address", 7, "Applicant information was found");
-		section.redactBetween("Address:", "Contact", "address", 7, "Applicant information was found");
-		section.redactLineAfter("Contact point:",  "address", 7, "Applicant information was found");
-		section.redactLineAfter("Phone:", "address", 7, "Applicant information was found");
-		section.redactLineAfter("Fax:", "address", 7, "Applicant information was found");
-		section.redactLineAfter("Tel.:", "address", 7, "Applicant information was found");
-		section.redactLineAfter("Tel:", "address", 7, "Applicant information was found");
-		section.redactLineAfter("E-mail:", "address", 7, "Applicant information was found");
-		section.redactLineAfter("Email:", "address", 7, "Applicant information was found");
-		section.redactLineAfter("Contact:", "address", 7, "Applicant information was found");
-		section.redactLineAfter("Telephone number:", "address", 7, "Applicant information was found");
-		section.redactLineAfter("Fax number:", "address", 7, "Applicant information was found");
-		section.redactLineAfter("Telephone:", "address", 7, "Applicant information was found");
-		section.redactBetween("No:", "Fax", "address", 7, "Applicant information was found");
-		section.redactBetween("Contact:", "Tel.:", "address", 7, "Applicant information was found");
+		section.redactLineAfter("Name:", "address", 6, "Applicant information was found");
+		section.redactBetween("Address:", "Contact", "address", 6, "Applicant information was found");
+		section.redactLineAfter("Contact point:",  "address", 6, "Applicant information was found");
+		section.redactLineAfter("Phone:", "address", 6, "Applicant information was found");
+		section.redactLineAfter("Fax:", "address", 6, "Applicant information was found");
+		section.redactLineAfter("Tel.:", "address", 6, "Applicant information was found");
+		section.redactLineAfter("Tel:", "address", 6, "Applicant information was found");
+		section.redactLineAfter("E-mail:", "address", 6, "Applicant information was found");
+		section.redactLineAfter("Email:", "address", 6, "Applicant information was found");
+		section.redactLineAfter("Contact:", "address", 6, "Applicant information was found");
+		section.redactLineAfter("Telephone number:", "address", 6, "Applicant information was found");
+		section.redactLineAfter("Fax number:", "address", 6, "Applicant information was found");
+		section.redactLineAfter("Telephone:", "address", 6, "Applicant information was found");
+		section.redactBetween("No:", "Fax", "address", 6, "Applicant information was found");
+		section.redactBetween("Contact:", "Tel.:", "address", 6, "Applicant information was found");
 	end

-rule "8: Redact contact information, if Producer is found"
+rule "7: Redact contact information, if Producer is found"
 	when
 		eval(section.getText().toLowerCase().contains("producer of the plant protection") || section.getText().toLowerCase().contains("producer of the active substance") || section.getText().contains("Manufacturer of the active substance") || section.getText().contains("Manufacturer:") || section.getText().contains("Producer or producers of the active substance"));
 	then
-		section.redactLineAfter("Name:", "address", 8, "Producer was found");
-		section.redactBetween("Address:", "Contact", "address", 8, "Producer was found");
-		section.redactBetween("Contact:", "Phone", "address", 8, "Producer was found");
-		section.redactBetween("Contact:", "Telephone number:", "address", 8, "Producer was found");
-		section.redactBetween("Address:", "Manufacturing", "address", 8, "Producer was found");
-		section.redactLineAfter("Telephone:", "address", 8, "Producer was found");
-		section.redactLineAfter("Phone:", "address", 8, "Producer was found");
-		section.redactLineAfter("Fax:", "address", 8, "Producer was found");
-		section.redactLineAfter("E-mail:", "address", 8, "Producer was found");
-		section.redactLineAfter("Contact:", "address", 8, "Producer was found");
-		section.redactLineAfter("Fax number:", "address", 8, "Producer was found");
-		section.redactLineAfter("Telephone number:", "address", 8, "Producer was found");
-		section.redactLineAfter("Tel:", "address", 8, "Producer was found");
-		section.redactBetween("No:", "Fax", "address", 8, "Producer was found");
+		section.redactLineAfter("Name:", "address", 7, "Producer was found");
+		section.redactBetween("Address:", "Contact", "address", 7, "Producer was found");
+		section.redactBetween("Contact:", "Phone", "address", 7, "Producer was found");
+		section.redactBetween("Contact:", "Telephone number:", "address", 7, "Producer was found");
+		section.redactBetween("Address:", "Manufacturing", "address", 7, "Producer was found");
+		section.redactLineAfter("Telephone:", "address", 7, "Producer was found");
+		section.redactLineAfter("Phone:", "address", 7, "Producer was found");
+		section.redactLineAfter("Fax:", "address", 7, "Producer was found");
+		section.redactLineAfter("E-mail:", "address", 7, "Producer was found");
+		section.redactLineAfter("Contact:", "address", 7, "Producer was found");
+		section.redactLineAfter("Fax number:", "address", 7, "Producer was found");
+		section.redactLineAfter("Telephone number:", "address", 7, "Producer was found");
+		section.redactLineAfter("Tel:", "address", 7, "Producer was found");
+		section.redactBetween("No:", "Fax", "address", 7, "Producer was found");
 	end

-rule "9: Redact Authors and Addresses in Reference Table, if it is a Vertebrate study"
+
+rule "8: Not redacted because Vertebrate Study = N"
    when
-        Section(tabularData != null
-            && tabularData.containsKey("Vertebrate study Y/N")
-            && tabularData.get("Vertebrate study Y/N").equals("Y")
-        )
+        Section(rowEquals("Vertebrate study Y/N", "N"))
    then
-        section.redact("name", 9, "Redacted because row is a vertebrate study");
-        section.redact("address", 9, "Redacted because rows is a vertebrate study");
-        section.highlightCell("Vertebrate study Y/N", 9);
-    end
+        section.redactNot("name", 8, "Not redacted because row is not a vertebrate study");
+        section.redactNot("address", 8, "Not redacted because row is not a vertebrate study");
+        section.highlightCell("Vertebrate study Y/N", 8, "hint_only");
+    end
+
+
+rule "9: Redact if must redact entry is found"
+	when
+		eval(section.contains("must_redact")==true);
+	then
+		section.redact("name", 9, "must_redact entry was found.");
+		section.redact("address", 9, "must_redact entry was found.");
+	end
+
+
+rule "10: Redact Authors and Addresses in Reference Table, if it is a Vertebrate study"
+    when
+        Section(rowEquals("Vertebrate study Y/N", "Y"))
+    then
+        section.redact("name", 10, "Redacted because row is a vertebrate study");
+        section.redact("address", 10, "Redacted because row is a vertebrate study");
+        section.highlightCell("Vertebrate study Y/N", 10, "must_redact");
+    end
Author	SHA1	Message	Date
Thierry Goeckel	81ea2e91ef	Pull request #33 : Fix entity span in table rows and detection of headers in rotated tables Merge in RED/redaction-service from bugfix/rowspan-and-header-in-rotated-table-fix to master * commit '4954aafed78e06484531d6264bdf215176ee0ef2': Reduce log level. Remove unused import Adjust test to added rule and fix vertical header propagation for row > 2 Fix entity span in table rows and detection of headers in rotated tables	2020-08-25 17:30:57 +02:00
Thierry Göckel	4954aafed7	Reduce log level.	2020-08-25 17:21:56 +02:00
Thierry Göckel	0baf48698d	Remove unused import	2020-08-25 15:45:24 +02:00
Thierry Göckel	3c590dcf1d	Adjust test to added rule and fix vertical header propagation for row > 2	2020-08-25 15:41:44 +02:00
Thierry Göckel	848c506c3f	Fix entity span in table rows and detection of headers in rotated tables	2020-08-25 13:50:34 +02:00
Thierry Goeckel	6483e637c6	Pull request #32 : Remove '-' in headernames Merge in RED/redaction-service from tablesHeader3 to master * commit '272c7cc2287b146664b8bb277c3731262ac78a19': Remove '-' in headernames	2020-08-24 16:15:44 +02:00
deiflaender	272c7cc228	Remove '-' in headernames	2020-08-24 15:39:50 +02:00
Thierry Goeckel	5586531c33	Pull request #31 : Made rules for table more stabil and flexible Merge in RED/redaction-service from tablesRules2 to master * commit '559c42154276ea861c09cf29b71d9bb9c17e6e87': Made rules for table more stabil and flexible	2020-08-24 14:52:29 +02:00
deiflaender	559c421542	Made rules for table more stabil and flexible	2020-08-24 14:45:30 +02:00
Dominique Eiflaender	d026f4e1db	Pull request #30 : Fix test and logic of same page table merge for one row tables Merge in RED/redaction-service from bugfix/fix-test-table-merge to master * commit '38ddb1d9c8dbf0e83d40ec916f1ebeb6d0f86f42': Fix test and logic of same page table merge for one row tables	2020-08-24 14:22:41 +02:00
Thierry Göckel	38ddb1d9c8	Fix test and logic of same page table merge for one row tables	2020-08-24 14:06:34 +02:00
Thierry Goeckel	92f0af9d21	Pull request #28 : Highlight N in vertebrate study tables and fixed rule order Merge in RED/redaction-service from tableRules to master * commit 'afa26895c8e78f261ec3700c515f412d8cf605ce': Fixed test problem Fixed pmd Highlight N in vertebrate study tables and fixed rule order	2020-08-24 13:15:11 +02:00
Dominique Eiflaender	b24922aa60	Pull request #29 : Fix multi-page table merge Merge in RED/redaction-service from bugfix/multi-page-table-merge to master * commit 'baa703928f50e53a66f925fcf72082df188482a1': Fix multi-page table merge	2020-08-24 13:09:43 +02:00
deiflaender	afa26895c8	Fixed test problem	2020-08-24 13:05:23 +02:00
Thierry Göckel	baa703928f	Fix multi-page table merge	2020-08-24 13:03:03 +02:00
deiflaender	85667c91e4	Fixed pmd	2020-08-24 12:36:49 +02:00
deiflaender	8a471d0771	Highlight N in vertebrate study tables and fixed rule order	2020-08-24 12:24:48 +02:00
Dominique Eiflaender	bb1112d0d7	Pull request #27 : Fix merging of tables with single header row table Merge in RED/redaction-service from bugfix/merge-single-row-tables to master * commit '00a960ee2308e7ed7da23ebd3e85f5ef9db14610': Fix merging of tables with single header row table	2020-08-24 12:10:53 +02:00
Thierry Göckel	00a960ee23	Fix merging of tables with single header row table	2020-08-24 11:43:19 +02:00
Dominique Eiflaender	8c08bb3664	Pull request #25 : RED-101 & RED-102: Make table structure detection more robust Merge in RED/redaction-service from bugfix/RED-101 to master * commit '76369f13f8d41154f3dd3690af81b9567ca1133e': Remove redundant comment RED-101: Add features as requested in PR Add unit test for table structure requirements Make table structure detection more robust	2020-08-24 11:09:17 +02:00
Cheng Zhu	14b4f4ab8a	Pull request #26 : Feature/RED-116: replace gin4-commons with red-commons Merge in RED/redaction-service from feature/RED-116 to master * commit '4206e708a20a862083f03d27a992e8661c44abc2': RED-116: replace gin4-commons with red-commons RED-116: replace gin4-commons with red-commons	2020-08-20 17:22:36 +02:00
Thierry Göckel	76369f13f8	Remove redundant comment	2020-08-18 20:53:41 +02:00
Thierry Göckel	1fff4f7eb0	RED-101: Add features as requested in PR - Add rule for non-vertebrate studies - Don't highlight ambiguous cell values - Rewrite rules in more human-readable fashion	2020-08-18 20:35:40 +02:00
Thierry Göckel	c7f5b4a280	Add unit test for table structure requirements	2020-08-18 20:35:17 +02:00
Thierry Göckel	4236fa05cc	Make table structure detection more robust	2020-08-13 17:48:37 +02:00