Ferret talk - May 2007 - bilingual site: exclude fields set from query

Hi all,

Is there a way to have searches no use some indexed fields, when
processing a query?

context:
I have a model Foo that holds some information in two languages :
  - text1_nl, text2_nl, text3_nl
and
  - text1_en, text2_en, text3_en
Some other fields are common to both languages and indexed as well
   - first_name, last_name

Depending on the visitor language choice I need to exclude the first
three, or last three fields when query processing. Is this doable
relatively simply?
I guess I could use two indexes, but I''d like to keep using
acts_as_ferret if possible.


TIA.


Alain Ravet
--------
http://blog.ravet.com

I would setup your models differently, using Single-Table Inheritance
(STI).  (You can get more info on it by Googling "rails single table
inheritance".)  For your example, you''d have a table with columns
like:

id, type, first_name, last_name, text1, text2, text3

Then, you''d have a model like:

class Text << ActiveRecord::Base
   acts_as_ferret(... declare all fields here )
end

You''d then subclass this with two other classes:

class EnglishText << Text
end

class DutchText << Text
end

Then, to search one or the other, use:

EnglishText.find_by_contents("english query")

DutchText.find_by_contents("dutch query")

That''s a rough idea without exact code, but it should hopefully get you
started.

Thanks,

Doug

On 5/16/07, Alain Ravet <alain.ravet+ferret at gmail.com>
wrote:
>
> Hi all,
>
> Is there a way to have searches no use some indexed fields, when
> processing a query?
>
> context:
> I have a model Foo that holds some information in two languages :
>   - text1_nl, text2_nl, text3_nl
> and
>   - text1_en, text2_en, text3_en
> Some other fields are common to both languages and indexed as well
>    - first_name, last_name
>
> Depending on the visitor language choice I need to exclude the first
> three, or last three fields when query processing. Is this doable
> relatively simply?
> I guess I could use two indexes, but I''d like to keep using
> acts_as_ferret if possible.
>
>
> TIA.
>
>
> Alain Ravet
> --------
> http://blog.ravet.com
> _______________________________________________
> Ferret-talk mailing list
> Ferret-talk at rubyforge.org
> http://rubyforge.org/mailman/listinfo/ferret-talk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://rubyforge.org/pipermail/ferret-talk/attachments/20070516/23c3c226/attachment-0001.html

Doug

 > I would setup your models differently, using Single-Table Inheritance

STI would be impractical (I don''t want to add a useless
"type"
column), but plain inheritance sounds like a very good idea.

 > class Text << ActiveRecord::Base
 >    acts_as_ferret(... declare all fields here )
 > end>
 > You''d then subclass this with two other classes:
 >
 > class EnglishText << Text
 > end
 > class DutchText << Text
 > end

I guess you meant :

 class Text << ActiveRecord::Base
 end
 class EnglishText << Text
    acts_as_ferret :fields => [name, text_en]
 end
 class DutchText << Text
    acts_as_ferret :fields => [name, text_nl]
 end

That would indeed allow to call :
 > EnglishText.find_by_contents("english query")
 > DutchText.find_by_contents ("dutch query")


Clean and simple.
Thanks a lot.

Alain Ravet
----
http://blog.ravet.com

On Wed, May 16, 2007 at 08:55:46PM +0200, Alain Ravet
wrote:
> Hi all,
> 
> Is there a way to have searches no use some indexed fields, when
> processing a query?
> 
> context:
> I have a model Foo that holds some information in two languages :
>   - text1_nl, text2_nl, text3_nl
> and
>   - text1_en, text2_en, text3_en
> Some other fields are common to both languages and indexed as well
>    - first_name, last_name
> 
> Depending on the visitor language choice I need to exclude the first
> three, or last three fields when query processing. Is this doable
> relatively simply?
> I guess I could use two indexes, but I''d like to keep using
> acts_as_ferret if possible.
A query string like

text1_nl|text2_nl|text3_nl|first_name|last_name:query

will only search for query in the named fields.

Jens

-- 
Jens Kr?mer
webit! Gesellschaft f?r neue Medien mbH
Schnorrstra?e 76 | 01069 Dresden
Telefon +49 351 46766-0 | Telefax +49 351 46766-66
kraemer at webit.de | www.webit.de
 
Amtsgericht Dresden | HRB 15422
GF Sven Haubold, Hagen Malessa

Jens,

 > A query string like
 >    text1_nl|text2_nl|text3_nl|first_name|last_name:query
 > will only search for query in the named fields.

Thanks.

Is this the proper syntax for multi-words queries:

  first_name|last_name:(jo* OR va*)
?

Alain

Hey Alain,

finally jumped on board? :-) hope to see you in berlin in a few month ..
> Is this the proper syntax for multi-words queries:
>
>   first_name|last_name:(jo* OR va*)
i would suggest to use the OO-Syntax for more complex queries .. like  
that:

query = BooleanQuery.new
[ :first_name, :last_name ].each do |key|
   query.add_query( PrefixQuery.new( key, ''jo'' ), :should )
end


for a people-livesearch, i do something like [1]:

def build_livesearch_query( fields, q )
   q = filter_special_characters( q ) # custom method
   query = BooleanQuery.new
   fields.uniq.each do |field|
     sq = SpanNearQuery.new(:in_order => true, :slop => 1)
     terms = analyze_query(q, field) # custom method
     terms.each do |term|
       sq << SpanPrefixQuery.new( field, term )
     end
     query << sq
   end
   query
end

i call that like

build_livesearch_query( [:first_name, :last_name, :email, q )     # q  
is an array of search words

Ben


[1] all of the code at http://bugs.omdb.org/browser/trunk/lib/omdb/ 
ferret/local_search.rb

Hi Ben,

  > Hey Alain,
  > finally jumped on board? :-) hope to see you in berlin in a few months

Can''t wait to register. My credit card is on the starting-blocks.


 > > Is this the proper syntax for multi-words queries:
 > >   first_name|last_name:(jo* OR va*)
 >
 > i would suggest to use the OO-Syntax for more complex queries .. like
 > query = BooleanQuery.new


I''m refactoring/upgrading an "old" (1-year) app, where I used
to do
all the ferret index stuff myself (callbacks, queries, ..), to the
mucho simpler interface offered by acts_as_ferret. AFAIK, aaf only
works with string queries, so if there is a string-based solution that
works, I''d rather use this one. If there is none, I''ll bite
the bullet
and follow your advice. "A man''s got to do, what a man''s
got to do"


Alain

-- 
Alain Ravet
--------
http://blog.ravet.com

>
> I''m refactoring/upgrading an "old" (1-year) app, where I
used to do
> all the ferret index stuff myself (callbacks, queries, ..), to the
> mucho simpler interface offered by acts_as_ferret. AFAIK, aaf only
> works with string queries, so if there is a string-based solution that
> works, I''d rather use this one. If there is none, I''ll
bite the bullet
> and follow your advice. "A man''s got to do, what a
man''s got to do"
Alain,

i''m no AAF expert, but afaik most methods of aaf are using
find_id_by_contents.. which is accepting a query parameter..

that query parameter gets forwarded to ferrets own search_each.

search_each accepts strings or ferret query objects.. so i would
assume that you can pass a OO-Query to AAF. And anything else
would have surprised me.. as Jens code is quite sophisticated :-)


Ben