[
https://issues.apache.org/jira/browse/PDFBOX-5965?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17947268#comment-17947268
]
Tilman Hausherr edited comment on PDFBOX-5965 at 4/25/25 9:24 AM:
------------------------------------------------------------------
Position = size of stream = 16389 = 0x4005, bytes to rewind: 7
in rewind(): CURRENT bytes -1, LAST bytes 5, NEXT bytes -1 this means we've
already lost the "real" LAST (=previous) buffer at some time before.
was (Author: tilman):
Position = size of stream = 16389 = 0x4005, bytes to rewind: 7
in rewind(): CURRENT bytes -1, LAST bytes 5, NEXT bytes -1 this means we've
already lost the "real" previous buffer at some time before.
> Rewind in NonSeekableRandomAccessReadInputStream malfunction near end of file
> -----------------------------------------------------------------------------
>
> Key: PDFBOX-5965
> URL: https://issues.apache.org/jira/browse/PDFBOX-5965
> Project: PDFBox
> Issue Type: Bug
> Components: IO
> Affects Versions: 3.0.4 PDFBox
> Reporter: Tilman Hausherr
> Assignee: Tilman Hausherr
> Priority: Major
> Labels: regression
> Fix For: 3.0.5 PDFBox, 4.0.0
>
> Attachments: POPPLER-208-p2.pdf
>
>
> A user has contacted me with a problem that involves a token count difference
> of parsed content stream tokens between PDFBox 2.0.33 and 3.0.4. It happens
> when an inline image is close to the end of the file, and is followed by two
> "Q" operators. The two "Q" operators are ignored despite the rewind() in
> PDFStreamParser. The problem goes away by replacing {{new
> NonSeekableRandomAccessReadInputStream(decoderStream)}} with {{new
> RandomAccessReadBuffer(decoderStream)}} in {{PDPage}}. It does not happen in
> PDFBox 3.0.2.
> I have a file that reproduces the problem but I doubt I can share it, I will
> try to find or create something.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
