[ 
https://issues.apache.org/jira/browse/TIKA-1074?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13585082#comment-13585082
 ] 

Michael McCandless commented on TIKA-1074:
------------------------------------------

bq. My app needs to extract text even from corrupt documents.

That's exactly the intent here as well.

bq. Currently I am setting ParseContext with a custom AutoDetectParser that, 
when an exception is hit, e.g. visiting an embedded, catches the exception, 
logs it AND extracts raw/binary strings from the problematic doc (or embedded)

Wait, the exceptions that this change now catches & logs is in the
decoding an OLE10 embedded entry (into its byte[] data), not in
actually parsing of the resulting byte[] data.  If the exception is
hit later when we recurse into parseEmbedded, the exception is still
thrown as before, so your custom AutoDetectParser will still
see/handle the exception.

But I think this is separately a good idea (an AutoDetectParser
logging & continuing by default): is this something you could possibly
contribute...?

Do you have an example corrupted document?  We could test before/after
this change and see.

                
> Extraction should continue if an exception is hit visiting an embedded 
> document
> -------------------------------------------------------------------------------
>
>                 Key: TIKA-1074
>                 URL: https://issues.apache.org/jira/browse/TIKA-1074
>             Project: Tika
>          Issue Type: Improvement
>          Components: parser
>            Reporter: Michael McCandless
>            Assignee: Michael McCandless
>             Fix For: 1.4
>
>         Attachments: TIKA-1074.patch, TIKA-1074.patch
>
>
> Spinoff from TIKA-1072.
> In that issue, a problematic document (still not sure if document is corrupt, 
> or possible POI bug) caused an exception when visiting the embedded documents.
> If I change Tika to suppress that exception, the rest of the document 
> extracts fine.
> So somehow I think we should be more robust here, and maybe log the 
> exception, or save/record the exception(s) somewhere so after parsing the app 
> could decide what to do about them ...

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to