Fix PdfSweep failure on documents with ligatures

DEVSIX-1940

Author: Alan Goo

kernel/src/main/java/com/itextpdf/kernel/pdf/canvas/parser/listener/CharacterRenderInfo.java

@@ -72,19 +72,16 @@ static StringConversionInfo mapString(List<CharacterRenderInfo> cris) {
         for (int i = 0; i < cris.size(); i++) {
             CharacterRenderInfo chunk = cris.get(i);
             if (lastChunk == null) {
-                indexMap.put(sb.length(), i);
-                sb.append(chunk.getText());
+                putCharsWithIndex(chunk.getText(), i, indexMap, sb);
             } else {
                 if (chunk.sameLine(lastChunk)) {
                     // we only insert a blank space if the trailing character of the previous string wasn't a space, and the leading character of the current string isn't a space
                     if (chunk.getLocation().isAtWordBoundary(lastChunk.getLocation()) && !chunk.getText().startsWith(" ") && !chunk.getText().endsWith(" ")) {
                         sb.append(' ');
                     }
-                    indexMap.put(sb.length(), i);
-                    sb.append(chunk.getText());
+                    putCharsWithIndex(chunk.getText(), i, indexMap, sb);
                 } else {
-                    indexMap.put(sb.length(), i);
-                    sb.append(chunk.getText());
+                    putCharsWithIndex(chunk.getText(), i, indexMap, sb);
                 }
             }
             lastChunk = chunk;
@@ -95,12 +92,18 @@ static StringConversionInfo mapString(List<CharacterRenderInfo> cris) {
         return ret;
     }
 
+    private static void putCharsWithIndex(final CharSequence seq, int index, final Map<Integer, Integer> indexMap, StringBuilder sb) {
+        int charCount = seq.length();
+        for (int i = 0; i < charCount; i++) {
+            indexMap.put(sb.length(), index);
+            sb.append(seq.charAt(i));
+        }
+    }
+
     public CharacterRenderInfo(TextRenderInfo tri) {
         super(tri == null ? "" : tri.getText(), tri == null ? null : getLocation(tri));
         if (tri == null)
             throw new IllegalArgumentException("TextRenderInfo argument is not nullable.");
-        if (tri.getText().length() != 1)
-            throw new IllegalArgumentException("CharacterRenderInfo objects represent a single character. They should not be made from TextRenderInfo objects containing more than a single character of text.");
 
         // determine bounding box
         float x0 = tri.getDescentLine().getStartPoint().get(0);

kernel/src/main/java/com/itextpdf/kernel/pdf/canvas/parser/listener/RegexBasedLocationExtractionStrategy.java

@@ -110,9 +110,25 @@ public int compare(IPdfTextLocation l1, IPdfTextLocation l2) {
             }
         });
 
+        // ligatures can produces same rectangle
+        removeDuplicates(retval);
+
         return retval;
     }
 
+    private void removeDuplicates(List<IPdfTextLocation> sortedList) {
+        IPdfTextLocation lastItem = null;
+        int orgSize = sortedList.size();
+        for (int i = orgSize - 1; i >= 0; i--) {
+            IPdfTextLocation currItem = sortedList.get(i);
+            Rectangle currRect = currItem.getRectangle();
+            if (lastItem != null && currRect.equalsWithEpsilon(lastItem.getRectangle())) {
+                sortedList.remove(currItem);
+            }
+            lastItem = currItem;
+        }
+    }
+
     @Override
     public void eventOccurred(IEventData data, EventType type) {
         if (data instanceof TextRenderInfo) {

kernel/src/test/java/com/itextpdf/kernel/pdf/canvas/parser/listener/RegexBasedLocationExtractionStrategyTest.java

@@ -42,6 +42,7 @@ This file is part of the iText (R) project.
  */
 package com.itextpdf.kernel.pdf.canvas.parser.listener;
 
+import com.itextpdf.kernel.geom.Rectangle;
 import com.itextpdf.kernel.pdf.PdfDocument;
 import com.itextpdf.kernel.pdf.PdfReader;
 import com.itextpdf.kernel.pdf.canvas.parser.PdfCanvasProcessor;
@@ -84,11 +85,11 @@ public void test01() throws IOException {
         }
 
         // compare
-        Assert.assertEquals(locationList.size(), 1);
+        Assert.assertEquals(1, locationList.size());
 
         IPdfTextLocation loc = locationList.get(0);
 
-        Assert.assertEquals(loc.getText(), "{{Signature}}");
+        Assert.assertEquals("{{Signature}}", loc.getText());
         Assert.assertEquals(23, (int) loc.getRectangle().getX());
         Assert.assertEquals(375, (int) loc.getRectangle().getY());
         Assert.assertEquals(55, (int) loc.getRectangle().getWidth());
@@ -97,4 +98,113 @@ public void test01() throws IOException {
         // close
         pdfDocument.close();
     }
+
+
+    // https://jira.itextsupport.com/browse/DEVSIX-1940
+    // text is 'calligraphy' and 'll' is composing a ligature
+
+    @Test
+    public void testLigatureBeforeLigature() throws IOException {
+        System.out.println(new File(sourceFolder).getAbsolutePath());
+
+        PdfDocument pdfDocument = new PdfDocument(new PdfReader(sourceFolder + "ligature.pdf"));
+
+        // build strategy
+        RegexBasedLocationExtractionStrategy extractionStrategy = new RegexBasedLocationExtractionStrategy("ca");
+
+        // get locations
+        List<IPdfTextLocation> locationList = new ArrayList<>();
+        for (int x = 1; x <= pdfDocument.getNumberOfPages(); x++) {
+            new PdfCanvasProcessor(extractionStrategy).processPageContent(pdfDocument.getPage(x));
+            for(IPdfTextLocation location : extractionStrategy.getResultantLocations()) {
+                if(location != null) {
+                    locationList.add(location);
+                }
+            }
+        }
+
+        // compare
+        Assert.assertEquals(1, locationList.size());
+
+        IPdfTextLocation loc = locationList.get(0);
+
+        Assert.assertEquals("ca", loc.getText());
+        Rectangle rect = loc.getRectangle();
+        Assert.assertEquals(36, rect.getX(), 0.0001);
+        Assert.assertEquals(655.4600, rect.getY(), 0.0001);
+        Assert.assertEquals(25.1000, rect.getWidth(), 0.0001);
+        Assert.assertEquals(20, rect.getHeight(), 0.0001);
+
+        pdfDocument.close();
+    }
+
+    @Test
+    public void testLigatureCrossLigature() throws IOException {
+        System.out.println(new File(sourceFolder).getAbsolutePath());
+
+        PdfDocument pdfDocument = new PdfDocument(new PdfReader(sourceFolder + "ligature.pdf"));
+
+        // build strategy
+        RegexBasedLocationExtractionStrategy extractionStrategy = new RegexBasedLocationExtractionStrategy("al");
+
+        // get locations
+        List<IPdfTextLocation> locationList = new ArrayList<>();
+        for (int x = 1; x <= pdfDocument.getNumberOfPages(); x++) {
+            new PdfCanvasProcessor(extractionStrategy).processPageContent(pdfDocument.getPage(x));
+            for(IPdfTextLocation location : extractionStrategy.getResultantLocations()) {
+                if(location != null) {
+                    locationList.add(location);
+                }
+            }
+        }
+
+        // compare
+        Assert.assertEquals(1, locationList.size());
+
+        IPdfTextLocation loc = locationList.get(0);
+
+        Assert.assertEquals("al", loc.getText());
+        Rectangle rect = loc.getRectangle();
+        Assert.assertEquals(48.7600, rect.getX(), 0.0001);
+        Assert.assertEquals(655.4600, rect.getY(), 0.0001);
+        Assert.assertEquals(25.9799, rect.getWidth(), 0.0001);
+        Assert.assertEquals(20, rect.getHeight(), 0.0001);
+
+        pdfDocument.close();
+    }
+
+    @Test
+    public void testLigatureInLigature() throws IOException {
+        System.out.println(new File(sourceFolder).getAbsolutePath());
+
+        PdfDocument pdfDocument = new PdfDocument(new PdfReader(sourceFolder + "ligature.pdf"));
+
+        // build strategy
+        RegexBasedLocationExtractionStrategy extractionStrategy = new RegexBasedLocationExtractionStrategy("l");
+
+        // get locations
+        List<IPdfTextLocation> locationList = new ArrayList<>();
+        for (int x = 1; x <= pdfDocument.getNumberOfPages(); x++) {
+            new PdfCanvasProcessor(extractionStrategy).processPageContent(pdfDocument.getPage(x));
+            for(IPdfTextLocation location : extractionStrategy.getResultantLocations()) {
+                if(location != null) {
+                    locationList.add(location);
+                }
+            }
+        }
+
+        // compare
+        Assert.assertEquals(1, locationList.size());
+
+        IPdfTextLocation loc = locationList.get(0);
+
+        Assert.assertEquals("l", loc.getText());
+        Rectangle rect = loc.getRectangle();
+        Assert.assertEquals(61.0999, rect.getX(), 0.0001);
+        Assert.assertEquals(655.4600, rect.getY(), 0.0001);
+        Assert.assertEquals(13.6399, rect.getWidth(), 0.0001);
+        Assert.assertEquals(20, rect.getHeight(), 0.0001);
+
+        pdfDocument.close();
+    }
 }