I updated the patch to also handle "AND". Yes it solves the issue.

Colm.

On Fri, Sep 2, 2016 at 1:38 PM, Francesco Chicchiriccò <ilgro...@apache.org>
wrote:

> ----- Il 2-set-16, alle 14:34, Colm O hEigeartaigh <cohei...@apache.org>
> ha scritto:
>
> Yep. I've attached a proposed patch to SYNCOPE-929. It simply makes sure
> that OR expressions are encapsulated in parentheses. It feels like a bit of
> a hack though :-)
>
>
> I've taken a look at your patch (guess that AND deserves the same
> treatment...): have you already checked if it actually solves this issue?
>
> On Fri, Sep 2, 2016 at 12:59 PM, Francesco Chicchiriccò <
> ilgro...@apache.org> wrote:
>
>> ----- Il 2-set-16, alle 13:33, Colm O hEigeartaigh <cohei...@apache.org>
>> ha scritto:
>>
>> Hi Francesco,
>>
>> I've dug into this a bit deeper. The logic in SearchCondVisitor is fine I
>> think. The problem is in JPAAnySearchDao when the SearchCond is converted
>> into a String. Using the example I gave, 
>> "(city==dublin,state==abc);country==ireland",
>> yields the following String:
>>
>> SELECT DISTINCT any_id FROM user_search_attr WHERE schema_id='city' AND
>> stringvalue=?1
>>     OR any_id IN ( SELECT DISTINCT any_id FROM user_search_attr WHERE
>> schema_id='state' AND stringvalue=?2)
>>     AND any_id IN ( SELECT DISTINCT any_id FROM user_search_attr WHERE
>> schema_id='country' AND stringvalue=?3)
>>
>> So we have WHERE X OR Y AND Z, instead of WHERE (X OR Y) AND Z, and so it
>> incorrectly evaluates Y AND Z first. The code in JPAAnySearchDao is quite
>> complex, I'll keep digging to see if I can fix it, unless you have an easy
>> fix?
>>
>> I believe the key point is around
>>
>> https://github.com/apache/syncope/blob/master/core/
>> persistence-jpa/src/main/java/org/apache/syncope/core/
>> persistence/jpa/dao/JPAAnySearchDAO.java#L475-L487
>>
>> Regards.
>>
>>
>> On Thu, Sep 1, 2016 at 5:20 PM, Francesco Chicchiriccò <
>> ilgro...@apache.org> wrote:
>>
>>> Hi,
>>> I have just tried the same sample as above and obtained the same results.
>>>
>>> The FIQL query is parsed into a SearchCond instance [1] by
>>> SearchCondVisitor [2]: at a first glance, I don't see any reason why
>>> parentheses should be ignored, but maybe it's only because I am quite tired
>>> at the moment :-)
>>>
>>> If you're opening an issue, please set "affects-version" to 2.0.0.M5 and
>>> "fix-for-version" to 2.0.0, thanks.
>>>
>>> Regards.
>>>
>>> [1] https://github.com/apache/syncope/blob/master/core/
>>> persistence-api/src/main/java/org/apache/syncope/core/
>>> persistence/api/dao/search/SearchCond.java
>>> [2] https://github.com/apache/syncope/blob/master/core/
>>> persistence-api/src/main/java/org/apache/syncope/core/
>>> persistence/api/search/SearchCondVisitor.java
>>>
>>> On 01/09/2016 18:18, Colm O hEigeartaigh wrote:
>>>
>>> Hi Iurii,
>>>
>>> Yep there is an issue here. Could you open a JIRA for it?
>>>
>>> I can reproduce with two users with attributes (city/country/state):
>>>
>>> alice: dublin/ireland/abc
>>> bob: dublin/canada/xyz
>>>
>>> Using a FIQL of "city==dublin;country==ireland" using
>>> AnyQuery.Builder() correctly yields one user ("alice"). However using
>>> "(city==dublin,state==abc);country==ireland" yields both users
>>> incorrectly. It appears that it might be ignoring the country part and just
>>> returning the result of the brackets.
>>>
>>> Colm.
>>>
>>> On Thu, Sep 1, 2016 at 4:21 PM, Iurii Smyrnov <ismyr...@talend.com>
>>> wrote:
>>>
>>>>
>>>> Hi Syncope Devs,
>>>>   Could you please check Does Syncope support braces " ( ) " in a
>>>> FIQL request ?
>>>>
>>>> I faced with an issue: it seems Syncope ignores braces in the request,
>>>> for example (get groups items):
>>>> - request is : (name==testA,name==testB);name==testC
>>>> - the same request (http encoded) is: http://192.
>>>> 168.99.100:9080/syncope/rest/groups?fiql=
>>>> <http://192.168.99.100:9080/syncope/rest/groups?fiql=name%3D%3DtestA%2Cname%3D%3DtestB3Bname%3D%3DtestC>
>>>> %28name%3D%3DtestA%2Cname%3D%3DtestB%29%3Bname%3D%3DtestC
>>>> <http://192.168.99.100:9080/syncope/rest/groups?fiql=name%3D%3DtestA%2Cname%3D%3DtestB3Bname%3D%3DtestC>
>>>>
>>>> In this case response should return zero items , but it returns Group
>>>> item with name=testA  - the same result if there would not been braces like
>>>> below:
>>>> - request is : name==testA,name==testB;name==testC or FIQL encode
>>>> name%3D%3DtestA%2Cname%3D%3DtestB%3Bname%3D%3DtestC
>>>> <http://192.168.99.100:9080/syncope/rest/groups?fiql=name%3D%3DtestA%2Cname%3D%3DtestB3Bname%3D%3DtestC>
>>>>
>>>>
>>>> It seems Syncope ignores braces "( )".
>>>>
>>>> --
>>> Francesco Chicchiriccò
>>>
>>> Tirasa - Open Source Excellencehttp://www.tirasa.net/
>>>
>>> Involved at The Apache Software Foundation:
>>> member, Syncope PMC chair, Cocoon PMC, Olingo PMC,
>>> CXF Committer, OpenJPA Committer, PonyMail 
>>> PPMChttp://home.apache.org/~ilgrosso/
>>>
>>>
>>
>>
>> --
>> Colm O hEigeartaigh
>>
>> Talend Community Coder
>> http://coders.talend.com
>>
>>
>>
>> --
>> Francesco Chicchiriccò
>>
>> Tirasa - Open Source Excellence
>> http://www.tirasa.net/
>>
>> Involved at The Apache Software Foundation:
>> member, Syncope PMC chair, Cocoon PMC, Olingo PMC,
>> CXF Committer, OpenJPA Committer, PonyMail PPMC
>> http://home.apache.org/~ilgrosso/
>>
>
>
>
> --
> Colm O hEigeartaigh
>
> Talend Community Coder
> http://coders.talend.com
>
>
>
> --
> Francesco Chicchiriccò
>
> Tirasa - Open Source Excellence
> http://www.tirasa.net/
>
> Involved at The Apache Software Foundation:
> member, Syncope PMC chair, Cocoon PMC, Olingo PMC,
> CXF Committer, OpenJPA Committer, PonyMail PPMC
> http://home.apache.org/~ilgrosso/
>



-- 
Colm O hEigeartaigh

Talend Community Coder
http://coders.talend.com

Reply via email to