[
https://issues.apache.org/jira/browse/TIKA-1928?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15224073#comment-15224073
]
Nick Burch commented on TIKA-1928:
----------------------------------
I've just tried this, and I can't reproduce it using the Tika CLI App and a
freshly renamed test PDF:
{code}
$ java -jar tika-app-1.13-SNAPSHOT.jar --detect Testing##.pdf
application/pdf
{code}
What operating system are you seeing this on? Can you reproduce this using the
Tika CLI App? Can you turn your problem into a junit test case?
> Filename detection misses when a # is in a filename
> ---------------------------------------------------
>
> Key: TIKA-1928
> URL: https://issues.apache.org/jira/browse/TIKA-1928
> Project: Tika
> Issue Type: Bug
> Components: detector
> Affects Versions: 1.12
> Environment: java 8
> Reporter: Jean Coudon
> Priority: Minor
>
> If there is a pound character in a filename it will be detected as
> application/octet-stream instead of the proper type that is detected without
> the filename containing the pound.
> {code:java}
> Metadata metadata = new Metadata();
> Tika tika = new Tika();
> metadata.add(Metadata.RESOURCE_NAME_KEY, "test#.pdf");
> // tika uses NameDetector if first parameter == null
> System.out.println(tika.detect(null, metadata));
> // prints application/octet-stream instead of application/pdf
> {code}
> Tested for application/pdf and application/xml.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
