I have been looking at the modules code and what I noticed is that the search is always done within the body, title etc fields and not the language specific fields (body_LANGUAGE, title_LANGUAGE etc) which results into problematic searches with non-english languages.
For example, with Greek language, body_el is using the text_el field type which I've manually set to use the GreekStemFilterFactory in my schema.xml. Unfortunately though, the change I did plays no role at all because the field being searched is the body field which uses the text field type (which has the English Stemmer) and as a result, instead of being able to have a non accented-sensitive search with my Greek documents, i must use accents because the Greek Stemmer was never used.
I am not sure if am i am missing something here but if what i say is true, then what is the point of language specific fields if they are not being used in search? I am looking at apachesolr_multilingual_apachesolr_modify_query and I don't see anything being done in there to use the language specific fields for searching, only for highlighting comparisons and spell checking.
Comments
Comment #1
mkalkbrennerPlease don't open so many issues. You're basic issue is handled at #1288144: CCK Multilanguage values which still waits for your answer.
The language specific fields get queried by the apachesolr module itself according to your configuration at /admin/settings/apachesolr/query-fields
So we have to solve the issue why the language specific fields don't occur at this settings page, nothing else ...
Comment #2
arekanderu commentedMy apologies first of all, i didn't mean to over do it but i am not sure if the #1288144: CCK Multilanguage values is related to what I am saying because If i look at the query log, i see the body field being queried instead of body_el. Shouldn't this field be queried even If i am having the CCK fields issue or do you think they are related?
Example query i see:
If you think they are related then just close this issue :)
Comment #3
mkalkbrennerThat's not true. In your log I see
But I also see
Didn't you "omit" body ("Body text - the full, rendered content:") at "/admin/settings/apachesolr/query-fields"?
In your settings body is 40 times more important than body_el.
Comment #4
arekanderu commentedIt seems that i forgot to set to "Omit" the "Body text inside H4, H5, or H6 tags". Everything now seems to work properly.
Thanks and my apologies again.
Comment #5
mkalkbrennerYour welcome.
I know that using this module is not that easy because it has to deal with the design decisions of the apachesolr module itself ;-)
I close this issue. But it would be nice if you can reply to #1288144: CCK Multilanguage values.