[jira] [Commented] (NUTCH-2606) MIME detection is wrong for plain-text documents send as Content-Type "application/msword"

["Sebastian Nagel (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/NUTCH-2606?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16648893#comment-16648893 ] 

Sebastian Nagel commented on NUTCH-2606:
----------------------------------------

The debugger brought the stupid reason to the light: a detected type {{text/plain}} is ignored and hence cannot overwrite a wrong type sent by the server, see [MimeUtil, line 203|https://github.com/apache/nutch/blob/2c69694/src/java/org/apache/nutch/util/MimeUtil.java#L203]. This logic has been added with NUTCH-767 because "the type for empty content is auto-detected as "text/plain" and this value overrides the hint from the Content-Type header". That's not the case anymore: Empty documents are not detected as {{text/plain}} by Tika, except they're named with the extension {{.txt}}.

> MIME detection is wrong for plain-text documents send as Content-Type "application/msword"
> ------------------------------------------------------------------------------------------
>
>                 Key: NUTCH-2606
>                 URL: https://issues.apache.org/jira/browse/NUTCH-2606
>             Project: Nutch
>          Issue Type: Bug
>    Affects Versions: 1.14
>            Reporter: Sebastian Nagel
>            Priority: Minor
>             Fix For: 1.16
>
>
> Plain-text documents send as Content-Type "application/msword" are tried to parse as Word documents. The MIME detection should be fixed, so that these are correctly identified as plain-text documents. See NUTCH-2603 and https://www.atnf.csiro.au/computing/software/gipsy/doc/update.doc



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)