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
18 changed files with 368 additions and 192 deletions
@@ -5,7 +5,7 @@
    <parent>
        <artifactId>platform-dependency</artifactId>
        <groupId>com.iqser.red</groupId>
-        <version>1.0.1</version>
+        <version>1.0.2</version>
    </parent>
    <modelVersion>4.0.0</modelVersion>

@@ -32,21 +32,21 @@
            <dependency>
                <groupId>com.iqser.red</groupId>
                <artifactId>platform-commons-dependency</artifactId>
-                <version>1.0.0</version>
+                <version>1.1.0</version>
                <scope>import</scope>
                <type>pom</type>
            </dependency>

-                <dependency>
-                    <groupId>org.apache.pdfbox</groupId>
-                    <artifactId>pdfbox</artifactId>
-                    <version>${pdfbox.version}</version>
-                </dependency>
-                <dependency>
-                    <groupId>org.apache.pdfbox</groupId>
-                    <artifactId>pdfbox-tools</artifactId>
-                    <version>${pdfbox.version}</version>
-                </dependency>
+            <dependency>
+                <groupId>org.apache.pdfbox</groupId>
+                <artifactId>pdfbox</artifactId>
+                <version>${pdfbox.version}</version>
+            </dependency>
+            <dependency>
+                <groupId>org.apache.pdfbox</groupId>
+                <artifactId>pdfbox-tools</artifactId>
+                <version>${pdfbox.version}</version>
+            </dependency>

        </dependencies>

@@ -3,11 +3,15 @@ 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 RedactionLogEntry {

    private String id;
@@ -63,15 +63,15 @@
        </dependency>
        <!-- commons -->
        <dependency>
-            <groupId>com.iqser.gin4.commons</groupId>
+            <groupId>com.iqser.red.commons</groupId>
            <artifactId>spring-commons</artifactId>
        </dependency>
        <dependency>
-            <groupId>com.iqser.gin4.commons</groupId>
+            <groupId>com.iqser.red.commons</groupId>
            <artifactId>logging-commons</artifactId>
        </dependency>
        <dependency>
-            <groupId>com.iqser.gin4.commons</groupId>
+            <groupId>com.iqser.red.commons</groupId>
            <artifactId>metric-commons</artifactId>
        </dependency>
        <!-- other external -->
@@ -104,7 +104,7 @@
            <scope>test</scope>
        </dependency>
        <dependency>
-            <groupId>com.iqser.gin4.commons</groupId>
+            <groupId>com.iqser.red.commons</groupId>
            <artifactId>test-commons</artifactId>
            <scope>test</scope>
        </dependency>
@@ -20,7 +20,7 @@ import org.springframework.cloud.openfeign.EnableFeignClients;
 import org.springframework.context.annotation.Bean;
 import org.springframework.context.annotation.Import;

-import com.iqser.gin4.commons.spring.DefaultWebMvcConfiguration;
+import com.iqser.red.commons.spring.DefaultWebMvcConfiguration;
 import com.iqser.red.service.configuration.v1.api.model.RulesResponse;
 import com.iqser.red.service.redaction.v1.server.client.RulesClient;
 import com.iqser.red.service.redaction.v1.server.exception.RulesValidationException;
@@ -2,13 +2,13 @@ package com.iqser.red.service.redaction.v1.server.controller;

 import java.time.OffsetDateTime;

+import com.iqser.red.commons.spring.ErrorMessage;
 import org.springframework.http.HttpStatus;
 import org.springframework.web.bind.annotation.ExceptionHandler;
 import org.springframework.web.bind.annotation.ResponseBody;
 import org.springframework.web.bind.annotation.ResponseStatus;
 import org.springframework.web.bind.annotation.RestControllerAdvice;

-import com.iqser.gin4.commons.api.errorhandling.ErrorMessage;
 import com.iqser.red.service.redaction.v1.server.exception.RulesValidationException;

 import lombok.extern.slf4j.Slf4j;
@@ -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;
+
+}
@@ -9,8 +9,6 @@ import java.util.regex.Pattern;

 import org.apache.commons.lang3.StringUtils;

-import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
-
 import lombok.Builder;
 import lombok.Data;
 import lombok.extern.slf4j.Slf4j;
@@ -32,20 +30,16 @@ public class Section {

    private int sectionNumber;

-    private Map<String, TextBlock> tabularData;
+    private Map<String, CellValue> tabularData;


-    public boolean isVertebrateStudy() {
-        return tabularData != null
-                && tabularData.containsKey("Vertebrate study Y/N")
-                && tabularData.get("Vertebrate study Y/N").getText().equals("Y");
-    }
+    public boolean rowEquals(String headerName, String value){
+        String cleanHeaderName = headerName.replaceAll("\n", "")
+                .replaceAll(" ", "")
+                .replaceAll("-", "");

-
-    public boolean isNotVertebrateStudy() {
-        return tabularData != null
-                && tabularData.containsKey("Vertebrate study Y/N")
-                && tabularData.get("Vertebrate study Y/N").getText().equals("N");
+        return tabularData != null && tabularData.containsKey(cleanHeaderName)
+                && tabularData.get(cleanHeaderName).getTextBlock().getText().equals(value);
    }


@@ -176,20 +170,27 @@ public class Section {
    }


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

-        TextBlock 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 {
-            Entity entity = new Entity(value.getText(), "must_redact", 0, value.getText().length(), headline, sectionNumber);
+            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.getSequences()); // Make sure no other cells with same content are highlighted
+            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;
@@ -53,22 +54,27 @@ public class EntityRedactionService {
            for (Table table : tables) {
                for (List<Cell> row : table.getRows()) {
                    SearchableText searchableRow = new SearchableText();
-                    Map<String, TextBlock> tabularData = new HashMap<>();
+                    Map<String, CellValue> tabularData = new HashMap<>();
+                    int start = 0;
                    for (Cell cell : row) {
                        if (cell.isHeaderCell() || CollectionUtils.isEmpty(cell.getTextBlocks())) {
                            continue;
                        }
                        addSectionToManualRedactions(cell.getTextBlocks(), manualRedactions, table.getHeadline(), sectionNumber);
+                        int cellStart = start;
                        cell.getHeaderCells().forEach(headerCell -> {
-                            String headerName = headerCell.getTextBlocks().get(0).getText()
-                                    .replaceAll("\n", " ")
-                                    .replaceAll(" ", " ");
-                            tabularData.put(headerName, cell.getTextBlocks().get(0));
+                            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);

@@ -138,7 +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;
        }

@@ -4,6 +4,7 @@ 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.springframework.stereotype.Service;
@@ -39,22 +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);

@@ -62,7 +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;
@@ -85,22 +122,6 @@ public class SectionsBuilderService {
                } else {
                    table.setHeadline("Table in: " + lastHeadline);
                }
-                // Distribute header information for subsequent tables
-                if (previousTable != null && hasInvalidHeaderInformation(table) && hasValidHeaderInformation(previousTable)) {
-                    List<Cell> previousTableNonHeaderRow = getRowWithNonHeaderCells(previousTable);
-                    List<Cell> tableNonHeaderRow = getRowWithNonHeaderCells(table);
-                    if (previousTableNonHeaderRow.size() == tableNonHeaderRow.size()) {
-                        for (int i = table.getRows().size() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of table
-                            List<Cell> row = table.getRows().get(i);
-                            if (row.size() == previousTableNonHeaderRow.size()
-                                    && row.stream().allMatch(cell -> cell.getHeaderCells().isEmpty())) {
-                                for (int j = 0; j < row.size(); j++) {
-                                    row.get(j).setHeaderCells(previousTableNonHeaderRow.get(j).getHeaderCells());
-                                }
-                            }
-                        }
-                    }
-                }

                if (textBlock != null && !alreadyAdded) {
                    paragraph.getPageBlocks().add(textBlock);
@@ -163,7 +184,7 @@ public class SectionsBuilderService {

    private List<Cell> getRowWithNonHeaderCells(Table table) {

-        for (int i = table.getRows().size() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of 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) {
@@ -29,11 +29,13 @@ 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;

@@ -65,6 +67,25 @@ 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.
@@ -72,100 +93,54 @@ public class Table extends AbstractTextContainer {
     */
    private void computeHeaders() {

+        if (rows == null) {
+            rows = computeRows();
+        }
        // A bold cell is a header cell as long as every cell to the left/top is bold, too
-        cells.forEach((position, cell) -> {
-            List<Cell> cellsToTheLeft = getCellsToTheLeft(position);
-            Cell lastHeaderCell = null;
-            for (Cell leftCell : cellsToTheLeft) {
-                if (CollectionUtils.isNotEmpty(leftCell.getTextBlocks()) && leftCell.getTextBlocks()
+        // 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 {
+                        break;
+                    }
+                }
+                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")) {
-                    lastHeaderCell = leftCell;
-                } else {
-                    break;
+                    cell.setHeaderCell(true);
                }
            }
-            if (lastHeaderCell != null) {
-                cell.getHeaderCells().add(lastHeaderCell);
-            }
-            lastHeaderCell = null;
-            List<Cell> cellsToTheTop = getCellToTheTop(position);
-            for (Cell topCell : cellsToTheTop) {
-                if (CollectionUtils.isNotEmpty(topCell.getTextBlocks()) && topCell.getTextBlocks()
-                        .get(0)
-                        .getMostPopularWordStyle()
-                        .equals("bold")) {
-                    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);
-            }
-        });
-
-    }
-
-
-    private List<Cell> getCellsToTheLeft(CellPosition cellPosition) {
-
-        List<Cell> result = new ArrayList<>();
-        if (cellPosition.getCol() == 0) {
-            return result;
-        }
-        int row = cellPosition.getRow();
-        for (int i = cellPosition.getCol() - 1; i >= 0; i--) {
-            if (cells.get(new CellPosition(row, i)) != null) {
-                result.add(cells.get(new CellPosition(row, i)));
-            } else {
-                Cell spanningCell = null;
-                while (spanningCell == null && row >= 0) {
-                    row--;
-                    spanningCell = cells.get(new CellPosition(row, i));
-                }
-                if (spanningCell != null) {
-                    result.add(spanningCell);
-                }
-                row = cellPosition.getRow();
-            }
        }
-        Collections.reverse(result);
-        return result;
-    }

-
-    private List<Cell> getCellToTheTop(CellPosition cellPosition) {
-
-        List<Cell> result = new ArrayList<>();
-        if (cellPosition.getRow() == 0) {
-            return result;
-        }
-        int col = cellPosition.getCol();
-        for (int i = cellPosition.getRow() - 1; i >= 0; i--) {
-            if (cells.get(new CellPosition(i, col)) != null) {
-                result.add(cells.get(new CellPosition(i, col)));
-            } else {
-                Cell spanningCell = null;
-                while (spanningCell == null && col >= 0) {
-                    col--;
-                    spanningCell = cells.get(new CellPosition(i, col));
-                }
-                if (spanningCell != null) {
-                    result.add(spanningCell);
-                }
-                col = cellPosition.getCol();
-            }
-        }
-        Collections.reverse(result);
-        return result;
    }


@@ -173,9 +148,9 @@ 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));
                    if (cell != null) {
                        lastRow.add(cell);
@@ -184,9 +159,9 @@ public class Table extends AbstractTextContainer {
                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));
                    if (cell != null) {
                        lastRow.add(cell);
@@ -195,9 +170,9 @@ public class Table extends AbstractTextContainer {
                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()
                    if (cell != null) {
                        lastRow.add(cell);
@@ -214,8 +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);
@@ -8,12 +8,6 @@ spring:
  profiles:
    active: kubernetes

-platform.multi-tenancy:
-  enabled: ${multitenancy.enabled:false}
-  tenantFilter:
-    urlPatterns: /redact
-    urlPatternsToIgnore:
-
 management:
  endpoint:
    metrics.enabled: ${monitoring.enabled:false}
@@ -130,7 +130,7 @@ public class EntityRedactionServiceTest {
            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
+            assertThat(classifiedDoc.getEntities().get(1)).hasSize(7);// 3 names, 1 address, 1 Y and 2 N entities
        }
    }

@@ -193,6 +193,7 @@ public class EntityRedactionServiceTest {

    }

+
    @Test
    public void headerPropagation() throws IOException {

@@ -214,7 +215,32 @@ public class EntityRedactionServiceTest {
            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); // FIXME including one false positive "Y"
+            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);
        }
    }

@@ -226,13 +252,21 @@ public class EntityRedactionServiceTest {
                "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(isVertebrateStudy())\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));
@@ -3,6 +3,7 @@ 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;

@@ -18,6 +19,7 @@ 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;
@@ -64,4 +66,124 @@ public class PdfSegmentationServiceTest {
        }
    }

+
+    @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();
+        }
+    }
+
 }
@@ -90,28 +90,31 @@ rule "7: Redact contact information, if Producer is found"
 		section.redactBetween("No:", "Fax", "address", 7, "Producer was found");
 	end

-rule "8: Redact Authors and Addresses in Reference Table, if it is a Vertebrate study"
-    when
-        Section(isVertebrateStudy())
-    then
-        section.redact("name", 8, "Redacted because row is a vertebrate study");
-        section.redact("address", 8, "Redacted because row is a vertebrate study");
-        section.highlightCell("Vertebrate study Y/N", 9);
-    end

-rule "9: Not redacted because Vertebrate Study = N"
+rule "8: Not redacted because Vertebrate Study = N"
    when
-        Section(isNotVertebrateStudy())
+        Section(rowEquals("Vertebrate study Y/N", "N"))
    then
-        section.redactNot("name", 9, "Not redacted because row is not a vertebrate study");
-        section.redactNot("address", 9, "Not redacted because row is not a vertebrate study");
+        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 "10: Redact if must redact entry is found"
+rule "9: Redact if must redact entry is found"
 	when
 		eval(section.contains("must_redact")==true);
 	then
-		section.redact("name", 10, "must_redact entry was found.");
-		section.redact("address", 10, "must_redact entry was found.");
-	end
+		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
cheng	4206e708a2	RED-116: replace gin4-commons with red-commons	2020-08-20 16:52:56 +02:00
cheng	c4164b57fb	RED-116: replace gin4-commons with red-commons	2020-08-20 16:51:02 +02:00
Thierry Göckel	76369f13f8	Remove redundant comment	2020-08-18 20:53:41 +02:00
deiflaender	89ef8234d2	Added missing NoArgsConstructor to RedactionLogEntry, needed for Json Serialization	2020-08-14 11:21:11 +02:00