--unaccent option does not work

Sorry, I forgot to said that the versión of PostgreSQL is 9.1.

Amit Parik, why you maked this error as duplicated?

They are both related to unaccent, but they are not the same AFAIAC...

Hello Damian,

Sorry for inconvenience.

Thanks!

I think that the problem is when searching in Products, because it uses a subquery here. The subqueries are not parsed correctly to use unaccent(). I made a few changes in the 'expression.py' files and now it is working. I include the new 'expression.py' file that makes things work correctly.

In the file 'openerp/osv/expression.py' In line 639, the lines:

else:
    subselect += ' AND value ' + sql_operator + instr + \
         ') UNION (' \
         ' SELECT id' \
         ' FROM "' + working_table._table + '"' \
         ' WHERE "' + left + '" ' + sql_operator + instr + ")"

should be replaced by the lines:

else:
    if sql_operator in ['like','ilike'] and self.has_unaccent:
        subselect += ' AND value ' + sql_operator + instr + \
             ') UNION (' \
             ' SELECT id' \
             ' FROM "' + working_table._table + '"' \
             ' WHERE unaccent("' + left + '") ' + sql_operator + ' unaccent(' + instr + "))"
    else:
        subselect += ' AND value ' + sql_operator + instr + \
             ') UNION (' \
             ' SELECT id' \
             ' FROM "' + working_table._table + '"' \
             ' WHERE "' + left + '" ' + sql_operator + instr + ")"

With this modifications, if searching in products, the unaccent() function is used.

Also I think (but not quite sure about it) that in line 625 the lines:

subselect = '( SELECT res_id' \
         ' FROM ir_translation' \
         ' WHERE name = %s' \
         ' AND lang = %s' \
         ' AND type = %s'

should be changed with something like this:

if self.has_unaccent:
    subselect = '( SELECT res_id' \
             ' FROM ir_translation' \
             ' WHERE unaccent(name) = unaccent(%s)' \
             ' AND lang = %s' \
             ' AND type = %s'
else:
    subselect = '( SELECT res_id' \
             ' FROM ir_translation' \
             ' WHERE name = %s' \
             ' AND lang = %s' \
             ' AND type = %s'

I include Vo Minh Thu and Olivier Dony since they committed the unaccent() support in launchpad. They may better understand the implication of this changes in the code.

There is any news regarding this bug??

I don't know why it is still undecided, the problem seems really clear and easy to test. Also the solution should highlight the problem here.

Is there any way I can contribute to fix this in the server?

This bug has been reported like 40days ago, it seems to me a very important one since countries that use accents are unable to search some stuffs. Also I already point out a possible solution, that will take 5 mins to implement.

Is there any way I can accelerate this process? Is it feasible I propose a merge for the core of OpenERP? (I already propose meges in addons but when unnoticed)

Sorry Olivier Dony I made a merge proposal but I had an incorrect indentation. I delete the previous merge proposal, correct the indentation and redid the proposal.

I am not sure how to finish the merge and what I have to wait for, so the bug is fixed correctly in the nightly build, can you give me some hints?

Hi Damián,

Thanks for reporting the bug with great details and taking the time to submit a patch!

> Sorry Olivier Dony I made a merge proposal but I had an incorrect
> indentation. I delete the previous merge proposal, correct the indentation
> and redid the proposal.

No problem, though in general there is no need to delete and re-create a merge proposal, as they are automatically updated whenever you modify the proposed branch.
I also manually linked your branch with the bug because the link was not created automatically (you probably forgot to tag your commit with `--fixes`. See also [1] )

> I am not sure how to finish the merge and what I have to wait for, so the
> bug is fixed correctly in the nightly build, can you give me some hints?

You don't need to do anything else. Thu did the original implementation, so he will review your patch soon, and he will give you feedback on the merge proposal. He will also assess the risk/benefit ratio of the patch, to see whether it can be applied on the 6.1 stable branch safely [2].

Thanks for your contributions!

[1] For all the gory details regarding merge proposals and how to make them, see the documentation (there's a complete example including --fixes): http://doc.openerp.com/v6.1/contribute/15_guidelines/contribution_guidelines.html

[2] See the OpenERP Bug management policy: http://doc.openerp.com/v6.1/contribute/11_bug_tracker.html#bug-management-policy

Thanks for the bug report and the merge prop. I've added a new merge prop. correcting a small mistake you made, and it should make its way to a new 6.1 build.

Grate, thanks Thu!

It will be released in an upcoming nightly build?? (I still don't understand perfectly which is the code that generates the nightly build)

Thu, the merge propose that solve this problem has not been committed yet right?

You know if this will take much longer?

BR

Damián

Fixed in 7.0 via revid 5268 <email address hidden>