svn commit: r1647158 - /pdfbox/trunk/pdfbox/src/main/java/org/apache/pdfbox/util/PDFTextStripper.java

[le...@apache.org]

Author: lehmi
Date: Sun Dec 21 15:37:10 2014
New Revision: 1647158

URL: http://svn.apache.org/r1647158
Log:
PDFBOX-1874: adjust precision to avoid false results when comparing floats as proposed by Yuri Burrows

Modified:
    pdfbox/trunk/pdfbox/src/main/java/org/apache/pdfbox/util/PDFTextStripper.java

Modified: pdfbox/trunk/pdfbox/src/main/java/org/apache/pdfbox/util/PDFTextStripper.java
URL: http://svn.apache.org/viewvc/pdfbox/trunk/pdfbox/src/main/java/org/apache/pdfbox/util/PDFTextStripper.java?rev=1647158&r1=1647157&r2=1647158&view=diff
==============================================================================
--- pdfbox/trunk/pdfbox/src/main/java/org/apache/pdfbox/util/PDFTextStripper.java (original)
+++ pdfbox/trunk/pdfbox/src/main/java/org/apache/pdfbox/util/PDFTextStripper.java Sun Dec 21 15:37:10 2014
@@ -1487,14 +1487,18 @@ public class PDFTextStripper extends PDF
         {
             float yGap = Math.abs(position.getTextPosition().getYDirAdj() -
                     lastPosition.getTextPosition().getYDirAdj());
+            float newYVal = multiplyFloat(getDropThreshold(), maxHeightForLine);
             // do we need to flip this for rtl?
             float xGap = position.getTextPosition().getXDirAdj() -
                     lastLineStartPosition.getTextPosition().getXDirAdj();
-            if (yGap > getDropThreshold()*maxHeightForLine)
+            float newXVal = multiplyFloat(getIndentThreshold(), position.getTextPosition().getWidthOfSpace());
+            float positionWidth = multiplyFloat(0.25f, position.getTextPosition().getWidth());
+
+            if (yGap > newYVal)
             {
                 result = true;
             }
-            else if (xGap > getIndentThreshold()*position.getTextPosition().getWidthOfSpace())
+            else if (xGap > newXVal)
             {
                 // text is indented, but try to screen for hanging indent
                 if (!lastLineStartPosition.isParagraphStart())
@@ -1514,7 +1518,7 @@ public class PDFTextStripper extends PDF
                     result = true;
                 }
             }
-            else if (Math.abs(xGap) < 0.25 * position.getTextPosition().getWidth())
+            else if (Math.abs(xGap) < positionWidth)
             {
                 // current horizontal position is within 1/4 a char of the last
                 // linestart. We'll treat them as lined up.
@@ -1544,6 +1548,12 @@ public class PDFTextStripper extends PDF
         }
     }
 
+    private float multiplyFloat(float value1, float value2)
+    {
+        // multiply 2 floats and truncate the resulting value to 3 decimal places
+        // to avoid wrong results when comparing with another float
+        return Math.round(value1 * value2 * 1000) / 1000f;
+    }
     /**
      * writes the paragraph separator string to the output.
      * @throws IOException if something went wrong