Make sure you also use the PropertyNames directive in your config file.
I had a similar question when I first started with SWISHE:
----------------
I recant; I had to re-index (I added all the requisite WordCharacter
parameters
to my config file). Now it works as expected. Thanks, David!
pek
Peter Karman wrote:
>> Thanks. That didn't help, unfortunately. I get the same results: nada.
>>
>> pek
>>
>> David L Norris wrote:
>>
>>
>
>>>>On Thu, 2003-07-10 at 09:58, Peter Karman wrote:
>>>>
>>>>
>>
>>>>>>I'm indexing with the -S prog option. It seems to be working fine
from the
>>>>>>indexing point of view, since the pubnumber is successfully
captured from the
>>>>>>HTML files. However, I seem to be missing the trick when it comes
to searching
>>>>>>on pubnumber content. I've read the documentation (several times).
>>
>>>>
>>>>
>>>>You might simply need to add _-. to WordCharacters.
>>>>See WORDCHARS definition in swish-e/src/config.h if the below
>>>>WordCharacters line is scrambled in my email (and ignore the line
>>>>wrapping).
>>>>
>>>>
>>>>Try this config and see what happens:
>>>>
>>>>% cat swish-default.conf
>>>># include the PUBNUMBER meta value for each file
>>>>MetaNames PUBNUMBER
>>>>
>>>># save the PUBNUMBER in the index so its search-able
>>>>PropertyNames PUBNUMBER
>>>>
>>>># minimum word length
>>>>MinWordLimit 3
>>>>
>>>># Valid characters found in "words"
>>>>WordCharacters
>>>>_-.0123456789abcdefghijklmnopqrstuvwxyzB*B:CCCCC CC
>>>>CCCCCCCCC C"C$C&C(C*C,C.C0C2C4C6C9C;C=C?
>>>>
>>>>
>>>>
>
>>
>>
Julie Wetherill waxed lyrical on 1/15/04 5:10 PM:
> Apologies for what must be a newbie question, but I didn't find anything in
> the SWISH-e archives.
>
> I'm testing the indexing/retrieval of metadata in my documents (html and
> pdf). I set up a swish-e conf file that includes this directive:
>
> MetaNames description keywords
>
> and have a small number of html files containing metadata tags, for example:
>
> <meta name="description" content="acquisitions">
> <meta name="keywords" content="printing">
>
> When I run a search from the command line that includes a metaname, e.g.:
>
> swish-e -w description=acquisitions -f metadata.index
>
> I get the error:
>
> # SWISH format: 2.2.3
> # Search words: description=acquisitions
> err: Unknown metaname: 'description'
>
> I know the indexing of these documents is working. I can search for the
> metadata values as plain text and get results. But searching by any
> metaname that I define in the conf file always fails with the above error.
>
> I have tried enclosing description=acquisitions in single/double quotes,
> same outcome. I also have pdfs that by default contain some Dublin Core
> metadata tags (e.g., "dc:description"). I'm hoping these also can be
> handled via the MetaNames directive, but I can't even get the simple html
> metadata tags to work. What is wrong here? And assuming I can fix the
> simple case, does anyone know if the DC metadata will work using MetaNames?
>
> Thanks for any help you can provide. --julie
>
>
>
>
> ===============================================================
> Julie Wetherill
> Office for Information System
> Harvard University Library
> 1280 Massachusetts Ave., Suite 404
> Cambridge, MA 02138
>
--
Peter Karman - Software Publications Engineer - Cray Inc
phone: 651-605-9009 - mailto:karman@cray.com
Received on Thu Jan 15 23:24:01 2004