Hi,
my website makes use of URL's like
http://website/dir/index.jsp?content=foo/bar.html
The result of a request like the one above is text/html containing
relative URL's like
<a href="foo/foobar.html">foo</a>
When udmsearch tries to index foo/foobar.html it constructs the
absolute URL of the document badly resulting in a query to
http://website/dir/index.jsp?content=foo/foo/foobar.html
This happens because a '/' is encountered in the query-part of
the original URL - which should *not* be considered the path of
the URL (cf src/parseurl.c). The appended patch gets rid of
that problem by temporarily removing the query-portion of
the file-part while parsing path/file-parts.
The patched version constructs the correct document-url
http://website/dir/foo/foobar.html
Any comment is appreciated.
Jörg Zanger
------
SAP Basis Administration, Abt. IT
Schöck Bauteile GmbH, D-76534 Baden-Baden
Tel.: +49 7223 967-355, Fax: +49 7223 967-352
____________________________________________
--- parseurl.c Wed Aug 16 18:38:48 2000
+++ parseurl.c.orig Thu Jul 6 12:26:59 2000
@@ -9,7 +9,7 @@
#include "udm_utils.h"
int UdmParseURL(UDM_URL *url,char *s){
-char *schema,*anchor,*file,*query;
+char *schema,*anchor,*file;
if(strlen(s)>=UDM_URLSIZE)
return(UDM_PARSEURL_LONG);
@@ -130,16 +130,10 @@
strcpy(url->filename,url->path);
strcpy(url->path,"");
}
-
- /* temporarily truncate path to querystring */
- if((query=strchr(url->path,'?')))
- *(query) = 0;
if((file=strrchr(url->path,'/'))&&(strcmp(file,"/"))){
- if(query) *(query) = '?';
strcpy(url->filename,file+1);
*(file+1)=0;
}
- if(query) *(query) = '?';
return(0);
}
______________
If you want to unsubscribe send "unsubscribe udmsearch"
to [EMAIL PROTECTED]