Search multiple indexes

pirwin · Post by **pirwin** » Tue Jun 07, 2005 5:01 pm

I work with balegaspi and we do have a thunderstone search appliance. Here's our situation.

We have a folder that contains subfolders grouped by classification. fake example:

library/
library/files/
library/books/
library/documents/
library/sheets/

what we are doing now is we have an index on library and all of its subfolders. however, users are wanting three things. (1) to be able to combine all indexes together in one search, (2) to be able to combine certain indexes together, but not all (i.e. files and documents but not books and sheets, depending on what they are looking for), and (3) they want to be able to select them individually for just searching sheets, for example. the types of files are varied in the different subfolders, and we can not easily change this directory structure, and each subfolder has hundreds of subfolders in it and thousands of files dating back to the 80's.

how do we (a) combine certain indexes together, and (b) still allow individual searching? how would you do this with metasearches?

Post by **mark** » Thu Jun 09, 2005 2:07 pm

You can setup each index separately. Then setup meta searches for different combinations of those profiles.
There's currently no way for a user to select which sub-profiles to search together so you'd have to create meta profiles for each desired combination.

If you only ever wanted to search one area at a time you could use categories in a single profile.

watterson · Post by **watterson** » Fri Jun 10, 2005 12:55 pm

I have been watching this thread as I want to be able to do the same thing that pirwin wants to do. We have not switched to the appliance on our production site yet, but we have been doing something similar with the 2.51 version of webinator we have now. I was looking at something with the appliance and would wonder if this would work.

We have an profile that crawls our entire site, named www. We have "portals" on our site, /hst, /jwst, /resources, etc. With 2.51 we use "uq" to restrict our search results to a specific url base and it worked very well.

In the walk settings, if you set a category with the name jwst with the corresponding url pattern to http://www.stsci.edu/jwst/, then go to the search live and click "only jwst", then look at the source, the form that is created looks like this:

<FORM METHOD=get ACTION="/texis/search">
<INPUT TYPE="HIDDEN" NAME="dropXSL" VALUE="">
<INPUT TYPE=hidden NAME=pr VALUE="www-external">
<INPUT TYPE=hidden NAME=prox VALUE="page">
<INPUT TYPE=hidden NAME=rorder VALUE="500">
<INPUT TYPE=hidden NAME=rprox VALUE="500">
<INPUT TYPE=hidden NAME=rdfreq VALUE="250">
<INPUT TYPE=hidden NAME=rwfreq VALUE="750">
<INPUT TYPE=hidden NAME=rlead VALUE="500">
<INPUT TYPE=hidden NAME=sufs VALUE="1">
<INPUT TYPE=hidden NAME=order VALUE="r">
<TR><TH ALIGN=RIGHT NOWRAP><FONT FACE="Arial,Helvetica" SIZE="-1">Search for this:</FONT></TH><TD NOWRAP><FONT FACE="Arial,Helvetica"><INPUT SIZE=30 name=query value=""><select name=cq>
<OPTION VALUE="" SELECTED>Everything
<OPTION VALUE=1 >Only jwst

</select><INPUT TYPE=submit name=submit VALUE="Submit"> </FONT></TD></TR>
</FORM>
</TABLE>

Can this html be manipulated to search the /jwst portal only without any input from the user?

Mike

Post by **John** » Fri Jun 10, 2005 1:19 pm

Either move the SELECTED from the Everything option to the jwst option, or if you don't want to supply the user any choice replace the whole <select name=cq> ... </select> block with <input type="hidden" name="cq" value="1">

You can also use the uq variable with the appliance.

watterson · Post by **watterson** » Fri Jun 10, 2005 2:33 pm

John,

That is good to know, thanks. That will help us a great deal as we roll the appliance to production. Is there any documentation as to what the list of variables are available for the appliance?

Also, I tried to use the uq variable and I have not been able to get it to restrict its results and on top of that it seems to only return .doc and .pdf files, no html.

This is the code I am using to test:

<html><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><title>Search: </title></head><body bgcolor=white><FONT FACE="Arial,Helvetica" SIZE="-1"><p>
</FONT><p>
<TABLE WIDTH="600" BORDER=0 CELLSPACING=0 CELLPADDING=0>
<FORM METHOD=get ACTION="http://search.stsci.edu/texis/search">
<INPUT TYPE="HIDDEN" NAME="pr" VALUE="www-external">
<INPUT TYPE="HIDDEN" NAME="dropXSL" VALUE="">
<INPUT TYPE="HIDDEN" NAME="uq" VALUE="http://www.stsci.edu/jwst/%">
<TR><TH ALIGN=RIGHT NOWRAP><FONT FACE="Arial,Helvetica" SIZE="-1">Search for this:</FONT></TH><TD NOWRAP><FONT FACE="Arial,Helvetica"><INPUT SIZE=30 name=query value=""><INPUT TYPE=submit name=submit VALUE="Submit"> </FONT></TD></TR>
</FORM>
</TABLE>
</body></html>

Is that any apparent problem with this?

Mike

Post by **John** » Fri Jun 10, 2005 2:46 pm

Trying a search for "modem" seems to return html. Are most of the documents PDF or Word? It looks as if the list is not currently documented, however if you contact support directly they may be able to help, or at least provide guidance for specific situations.