[PDFBOX-2837] PDFBox creates files with EBCDIC code on z/OS

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.8.9, 1.8.10, 2.0.0
• Fix Version/s: 1.8.10, 2.0.0
• Component/s: Writing
• Environment: OS: Unix Service running on Z/OS.; JRE: JRE 1.6.0 IBM J9 2.4 z/OS s390-31
• Assignee: Tilman Hausherr
• Reporter: Tilman Hausherr
• Created: 20/Jun/15 10:30
• Updated: 23/Jul/15 06:35
• Resolved: 22/Jun/15 06:36

Description

A PDF file created on z/OS (an IBM mainframe os) has this

xref
ð ù
0000000000 65535 f

the two chars are EBCDIC code for 0 9. The cause is this code that outputs in the system default encoding:

    private void writeXrefRange(long x, long y) throws IOException
    {
        getStandardOutput().write(String.valueOf(x).getBytes());
        getStandardOutput().write(SPACE);
        getStandardOutput().write(String.valueOf(y).getBytes());
        getStandardOutput().writeEOL();
    } 

getBytes should use the specific charset, e.g. getBytes(Charsets.ISO_8859_1).