[PDFBOX-3872] PDFToImage conversion garbled characters on Linux but Windows OK

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Not A Bug
• Affects Version/s: 2.0.5
• Component/s: Text extraction
• Environment: Oracle Linux 3.8.13-118.14.1.el6uek.x86_64 + JRE 8 u131; -versus-; Windows 7 x64 + JRE 8 u131
• Assignee: Unassigned
• Reporter: Ed Sawdon
• Created: 14/Jul/17 13:40
• Updated: 17/Jul/17 09:53
• Resolved: 17/Jul/17 09:53

Description

I have lots of PDF documents with CID/Identity-H encoding which convert to images just fine on a Windows 7 platform, but the conversion on Linux causes garbled text in the image.

I have permission from Blue Chyp Ltd to share a culprit PDF publicly in the PDFBox community - the first attachment.

Running "java -jar pdfbox-app-2.0.5.jar PDFToImage BC600_S1_P3_A708__A.PDF" on Linux gives an image with garbled text - the second attachment.

However, running the same command on Windows 7 gives a faithful jpeg image - the third attachment.

I used the 64 bit JRE 8 update 131 In both the Windows and Linux environments.

I used to suffer from lots of PDF->Image conversions giving garbled text at PDFBox1.8. Upgrading to PDFBox2.0 has fixed most of these issues, except for this odd one with identity-H encoding with the Linux JRE.