Re: Consistent metadata

[Shay Banon <ki...@gmail.com>]

Hi fellows,

   Just wanted to check in and see if this has progressed since I last
asked? Maybe I should open a jira if there isn't one already? Its really
problematic for me using Tika in order to index documents not having it. For
example, having consistent date format across all parsers, and a constant
set of META data associated with each parser is a must otherwise searching
on the produced documents is problematic.... .

-shay.banon

On Mon, Apr 5, 2010 at 1:10 AM, Shay Banon <ki...@gmail.com> wrote:

> Hi,
>
>   From a brief review of the different parsers Tika provides, it seems like
> the parser basically add metadata as the underlying system provides them. It
> would be great to try and streamline the metadata creation so some will be
> consistent across parsers. For example, author, title, date (with ISO
> format). This means, for example, if you create a Lucene Document out of it,
> then you know what to search on regardless of the type.
>
>   Also, it would be great if each parser would document which metadata it
> is adding, and it which format (number, string, date).
>
>   What do you think?
>
> cheers,
> shay.banon
>

Re: Consistent metadata

[Shay Banon <ki...@gmail.com>]

Great, thanks!. I will get latest and play with it.

On Sat, Nov 6, 2010 at 11:54 PM, Nick Burch <ni...@alfresco.com> wrote:

> On Sat, 6 Nov 2010, Shay Banon wrote:
>
>> Just wanted to check in and see if this has progressed since I last
>> asked?
>>
>
> There has been some, I'd suggest you try a recent SVN checkout, then open a
> JIRA if you spot any more cases where two parsers give different responses
> for the same effective input data
>
>
>  For example, having consistent date format across all parsers
>>
>
> This one has been fixed - as of a month or two back, all the dates should
> be strongly typed and with a converter wherever possible
>
> Nick
>

Re: Consistent metadata

[Nick Burch <ni...@alfresco.com>]

On Sat, 6 Nov 2010, Shay Banon wrote:
> Just wanted to check in and see if this has progressed since I last
> asked?

There has been some, I'd suggest you try a recent SVN checkout, then open 
a JIRA if you spot any more cases where two parsers give different 
responses for the same effective input data

> For example, having consistent date format across all parsers

This one has been fixed - as of a month or two back, all the dates should 
be strongly typed and with a converter wherever possible

Nick