On Fri, May 23, 2003 at 12:09:36AM -0700, Timo Haberkern wrote:
> #IndexContents XML* .ml .sxw .sxc .sxg
> StoreDescription XML* <text:p> 320
> StoreDescription HTML <body> 320
> StoreDescription TXT 320
> My problem is that no description is shown in the searchresult even if i
> use the -p swishdescription argument.
This is a common source of confusion (and probably doesn't need to be).
You have to say:
IndexContents HTML* .html .htm
then StoreDescription HTML* <body> 1000
(Remember HTML* == HTML2 if libxml2 is installed, otherwise HTML* == HTML.)
The reason (which is not a good reason) is that StoreDescription is set for
a document type (parser, really). But you have to use DefaultContents or
IndexContents to assign a type to a file.
In other words, without DefaultContents or IndexContents there is no type
assigned to the file. Swish still indexes the file (and defaults to using
the HTML* parser), but since there is no type assigned to the file none of
the StoreDescription types match.
Life would be easier if swish-e used content-types instead of HTML*. Use
mime.types to map file extensions to mime types. Then base filter
selection, parsers, etc. on the mime types.
Received on Fri May 23 07:56:22 2003