[
https://issues.apache.org/jira/browse/TIKA-4424?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17956997#comment-17956997
]
Tim Allison commented on TIKA-4424:
-----------------------------------
As a side note, the KMZDetector when operating on a file required a single file
with name ending in ".kml" at the root level of the kmz file (following
[https://developers.google.com/kml/documentation/kmzarchives|https://developers.google.com/kml/documentation/kmzarchives).]
). If there were other files at the root level (as in the test file attached
here), this was detected as application/zip.
When streaming however, the only requirement was one kml file at the root
level, but there could be other files at the root level. So, {_}I think{_},
this file would have been detected as {{application/zip}} if file detection
were used, but {{application/vnd.google-earth.kmz}} if streaming.
> Regression in zip-based detection with an InputStream in 3.2.0
> --------------------------------------------------------------
>
> Key: TIKA-4424
> URL: https://issues.apache.org/jira/browse/TIKA-4424
> Project: Tika
> Issue Type: Task
> Components: detector
> Affects Versions: 3.2.0
> Reporter: Tim Allison
> Priority: Major
> Labels: regression
> Fix For: 4.0.0, 3.2.1
>
> Attachments: tika-4424.zip
>
>
> On the user list, Craig Muchinsky and Pontus Amberg noted new problems with
> detection of zip based files.
> Craig noted that this affects InputStream detection, and Pontus noted that
> even if he switched to a TikaInputStream, his kmz file was getting detected
> as a zip.
> This is Pontus' code:
> {noformat}
> Tike.detect(InputStream stream, String name)
> {noformat}
> {noformat}
> pp//org.apache.tika.io.BoundedInputStream.reset(BoundedInputStream.java:115)
> app//org.apache.tika.detect.zip.DefaultZipContainerDetector.detectStreaming(DefaultZipContainerDetector.java:279)
> app//org.apache.tika.detect.zip.DefaultZipContainerDetector.detect(DefaultZipContainerDetector.java:192)
> app//org.apache.tika.detect.CompositeDetector.detect(CompositeDetector.java:84)
> {noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
