You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@manifoldcf.apache.org by "Karl Wright (JIRA)" <ji...@apache.org> on 2017/09/26 06:43:03 UTC
[jira] [Commented] (CONNECTORS-1459) Tika service wrong
Content-Type
[ https://issues.apache.org/jira/browse/CONNECTORS-1459?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16180331#comment-16180331 ]
Karl Wright commented on CONNECTORS-1459:
-----------------------------------------
Hi [~julienFL], the only concern I have is this:
{code}
+ document.setMimeType(metadataValue);
{code}
The mime type we send down the chain from the transformer should represent the content in the data stream. Since that's been extracted by Tika already, it should *not* be what was discovered by Tika, but rather "text/plain".
> Tika service wrong Content-Type
> -------------------------------
>
> Key: CONNECTORS-1459
> URL: https://issues.apache.org/jira/browse/CONNECTORS-1459
> Project: ManifoldCF
> Issue Type: Bug
> Components: Tika service connector
> Affects Versions: ManifoldCF 2.8.1
> Reporter: Julien Massiera
> Priority: Minor
> Fix For: ManifoldCF 2.8.1
>
> Attachments: CONNECTORS-1459.patch
>
>
> I noticed that the standard behaviour of the Tika extractor connector is to replace the existing "Content-Type" metadata by the one it founds. This behaviour is not implemented in the Tika service connector which just adds a new metadata entry instead of replacing the existing one. The consequence is that two values are available for the "Content-Type" metadata but only the first one is kept by the connector (which can also be considered as a bug ? this is the case for both the Tika extractor connector and the Tika service connector).
> So depending on the source connector, the resulting "Content-Type" may be wrong if for example the original provided one is "application/octet-stream"
> I will provide a patch for this bug
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)