[
https://issues.apache.org/jira/browse/TIKA-4012?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17711432#comment-17711432
]
Tim Allison edited comment on TIKA-4012 at 4/12/23 4:14 PM:
------------------------------------------------------------
*Categories for "missing" attachments*
A) Duplicates in the name tree (see above)
B) Duplication across name tree and annotations/actions
JEAYAWVYMGG74UFHIJEWDHHGKCMVLFHR
C) In the name tree for EmbeddedFiles, a direct file (/Type=/F) is attached.
There is no /Filespec
PQHSN7BOJAB57BVQ6P5OT3PVEKPOMGLT
249447
699266
D) The /Type=Filespec entry is a direct object in the names tree
PDFBOX-2408-3.pdf
*Cases where we get extra*
A) bad caps in names dictionary: eMBEDDEDfILES
COKYA5ETJWTAWHZUWIMZJCDH6HHAOHHK
B) Actual /AF attached to a page with an /AFRelationship!!!
ADBE_Contact_Private
REDHAT/1692355-1720581/REDHAT-1697878-0.pdf
evince-1165-1.pdf
was (Author: [email protected]):
*Categories for "missing" attachments*
A) Duplicates in the name tree (see above)
B) Duplication across name tree and annotations/actions
JEAYAWVYMGG74UFHIJEWDHHGKCMVLFHR
C) In the name tree for EmbeddedFiles, a direct file (/Type=/F) is attached.
There is no /Filespec
PQHSN7BOJAB57BVQ6P5OT3PVEKPOMGLT
249447
699266
*Cases where we get extra*
A) bad caps in names dictionary: eMBEDDEDfILES
COKYA5ETJWTAWHZUWIMZJCDH6HHAOHHK
B) Actual /AF attached to a page with an /AFRelationship!!!
ADBE_Contact_Private
REDHAT/1692355-1720581/REDHAT-1697878-0.pdf
evince-1165-1.pdf
> Improve extraction of embedded documents in PDFs
> ------------------------------------------------
>
> Key: TIKA-4012
> URL: https://issues.apache.org/jira/browse/TIKA-4012
> Project: Tika
> Issue Type: New Feature
> Reporter: Tim Allison
> Priority: Major
> Attachments: pdfbox-new-attachments-reports.tgz
>
>
> We're currently processing the EmbeddedFiles entry in the name tree and
> annotations to look for file spec dictionaries. Unfortunately, PDFs may embed
> files in lots of other places. The newly free 2.0 spec makes this abundantly
> and painfully clear.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
