[PDFBOX-2102] Characters swallowed on COSString.getString()

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.8.5, 1.8.6, 2.0.0
• Fix Version/s: 1.8.6, 2.0.0
• Component/s: Parsing
• Assignee: Jeremias Maerki
• Reporter: Jeremias Maerki
• Created: 29/May/14 10:12
• Updated: 22/Jun/14 14:34
• Resolved: 01/Jun/14 17:53

Description

PDFBOX-1437 seems to have introduced a regression that causes characters like \n to be swallowed when COSString.getString() is called. PDFDocEncoding doesn't handle all valid characters.

testStr = "Line1\nLine2\nLine3\n";
COSString lineFeedString = new COSString(testStr);
assertEquals(testStr, lineFeedString.getString());

//Same as previous but this time as a dictionary value
lineFeedString = new COSString(true);
for (int i = 0; i < testStr.length(); i++) {
    lineFeedString.append(testStr.charAt(i));
}
assertEquals(testStr, lineFeedString.getString()); //currently fails

Direct link to the change causing the regression:
http://svn.apache.org/viewvc?view=revision&revision=1406628

Issue Links

• is broken by: PDFBOX-1437 Title invalidly read in DocumentInformation