[
https://issues.apache.org/jira/browse/TIKA-4375?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17922738#comment-17922738
]
Tim Allison edited comment on TIKA-4375 at 1/31/25 3:22 PM:
------------------------------------------------------------
A few observations:
1) fewer exceptions. The small handful of new exceptions are not surprising
2) attachment diffs look good
3) Files now identified as {{model/x.stl.ascii}} have no text extracted -- I
think this is ok? This is the behavior in 3.x. Should we have this format
extend text/plain so that the text is still parsed?
4) A number of files are now being identified as GB18030 instead of UTF-8. From
the common words scores, this looks like a regression on some but an
improvement on others. I'm not sure there's much we can do.
was (Author: talli...@mitre.org):
A few observations:
1) fewer exceptions. The small handful of new exceptions are not surprising
2) attachment diffs look good
3) Files now identified as {{model/x.stl.ascii}} have no text extracted -- I
think this is ok? Should we have this format extend text/plain so that the text
is still parsed?
4) A number of files are now being identified as GB18030 instead of UTF-8. From
the common words scores, this looks like a regression, but I'm not sure there's
much we can do. There are also a number of files now identified as GB18030 that
appear to have improved extraction.
> Regression tests for 2.9.3 release
> ----------------------------------
>
> Key: TIKA-4375
> URL: https://issues.apache.org/jira/browse/TIKA-4375
> Project: Tika
> Issue Type: Task
> Reporter: Tim Allison
> Priority: Major
> Attachments: tika-2.9.2-v-tika-2.9.3-reports.tgz
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
