[
https://issues.apache.org/jira/browse/TIKA-3496?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17386428#comment-17386428
]
Nick Burch commented on TIKA-3496:
----------------------------------
If there is no timezone stored in the original file, I don't think we should be
adding one, we should be honest to users that it's a timezone-naive datetime.
That means returning it as just date+time, no timezone
If a given system we're passing data to refuses to accept timezone-naive dates,
our least-worst option is to slap the local machine's local zone on
If the file format has timezone info we should get our parsers to extract it,
but if the format lacks a timezone I don't think there's really anything more
we can/should do at the metadata properties level
> Dates should have a timezone?
> -----------------------------
>
> Key: TIKA-3496
> URL: https://issues.apache.org/jira/browse/TIKA-3496
> Project: Tika
> Issue Type: Bug
> Reporter: Tim Allison
> Priority: Major
>
> In working on the Solr pipe emitter, I noticed that some dates as stored in
> our Metadata do not have a timezone, which causes a problem for Solr.
> I noticed this issue in a JPEG with date: "2011-06-11T09:30:54".
> In a comment in our JPEG parser, I see:
> {noformat}
> // Unless we have GPS time we don't know the time zone so date must be set
> // as ISO 8601 datetime without timezone suffix (no Z or +/-)
> {noformat}
> So, the question is should we try to add a timezone (arbitrarily assign 'Z')
> in the Solr (and OpenSearch) emitter or should we store the date as if it
> were Z in the JPEG parser?
> Or do something else?
> The challenge with doing anything on the emitter side, is that we aren't
> currently storing the property type in the metadata. So, at emit time, we
> only have string keys and string values. We can't easily guess which fields
> should be a date in order to reformat for the sake of Solr. We could make a
> request to Solr/OpenSearch to figure out what the field types are, but that
> seems really awful...
> Ideas?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
