[
https://issues.apache.org/jira/browse/TIKA-1374?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tim Allison updated TIKA-1374:
------------------------------
Description:
Embedded files in PDFs can be found by the general all purpose key we
currently use via PDFBox: "F". However, embedded documents can also be stored
under OS specific keys: "DOS", "Mac" and "Unix".
[~lehmi] confirmed on the PDFBox users
[list|http://mail-archives.apache.org/mod_mbox/pdfbox-users/201407.mbox/%[email protected]%3e]
that we might be missing embedded documents if we're not trying the OS
specific keys as well. As Andreas points out, according to the spec the OS
specific keys shouldn't be used any more, but I think we should support
extraction for them.
My proposal is to pull all documents that are available by any of the four keys
(well, via getEmbeddedFile<OS>() in PDFBox). This has the downside of
potentially extracting basically duplicate documents, but I'd prefer to err on
the side of extracting everything.
The code fix is trivial, and I'll try to commit it today. However, it will
take me a bit of time to generate a test file that stores files under the OS
specific keys. So, if anyone has an ASF-friendly file available or wants to
take the task of generating one, please do.
was:
Embedded files in PDFs can be found by the general all purpose key we
currently use via PDFBox: "EF/F". However, embedded documents can also be
stored under OS specific keys: "EF/DOS", "EF/Mac" and "EF/Unix".
[~lehmi] confirmed on the PDFBox users
[list|http://mail-archives.apache.org/mod_mbox/pdfbox-users/201407.mbox/%[email protected]%3e]
that we might be missing embedded documents if we're not trying the OS
specific keys as well. As Andreas points out, according to the spec the OS
specific keys shouldn't be used any more, but I think we should support
extraction for them.
My proposal is to pull all documents that are available by any of the four keys
(well, via getEmbeddedFile<OS>() in PDFBox). The code fix is trivial, and I'll
try to commit it today. However, it will take me a bit of time to generate a
test file that stores files under the OS specific keys. So, if anyone has an
ASF-friendly file available or wants to take the task of generating one, please
do.
> Need to add code to look for OS-specific keys for embedded files within PDFs
> ----------------------------------------------------------------------------
>
> Key: TIKA-1374
> URL: https://issues.apache.org/jira/browse/TIKA-1374
> Project: Tika
> Issue Type: Improvement
> Components: parser
> Reporter: Tim Allison
> Assignee: Tim Allison
> Priority: Minor
> Fix For: 1.6
>
>
> Embedded files in PDFs can be found by the general all purpose key we
> currently use via PDFBox: "F". However, embedded documents can also be
> stored under OS specific keys: "DOS", "Mac" and "Unix".
> [~lehmi] confirmed on the PDFBox users
> [list|http://mail-archives.apache.org/mod_mbox/pdfbox-users/201407.mbox/%[email protected]%3e]
> that we might be missing embedded documents if we're not trying the OS
> specific keys as well. As Andreas points out, according to the spec the OS
> specific keys shouldn't be used any more, but I think we should support
> extraction for them.
> My proposal is to pull all documents that are available by any of the four
> keys (well, via getEmbeddedFile<OS>() in PDFBox). This has the downside of
> potentially extracting basically duplicate documents, but I'd prefer to err
> on the side of extracting everything.
> The code fix is trivial, and I'll try to commit it today. However, it will
> take me a bit of time to generate a test file that stores files under the OS
> specific keys. So, if anyone has an ASF-friendly file available or wants to
> take the task of generating one, please do.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
