[rules-users] Check if fact is subset of items in the rule
Michael Anstis
michael.anstis at gmail.com
Fri Oct 15 04:58:02 EDT 2010
Thanks for the extra detail Wolfgang.
2010/10/15 Wolfgang Laun <wolfgang.laun at gmail.com>
> All simple field constraints (the expressions you write between commas in a
> pattern) must follow the same
> syntactic pattern, which is
>
> [ <variable> ':' ] <fieldname> <restriction>
>
> This means that you may use a binding (e.g. "$foo :"), but the next thing
> must be a field name from the
> fact class. The <restriction> can have many forms; one of them is a name
> denoting an operator
> followed by the other operand, where you have a little more freedom, e.g.,
> by using a bound variable
> or an access to a field or other forms.
>
> Hence, not being able to reverse the operands of subsetOf is not a compiler
> bug; it is a (perhaps
> regrettable) consequence of the overall structure of constraints.
>
> Thus, if you need an operator supersetOf (which is to subsetOf just like >=
> is to <=) you'll have
> to define it, too. This is simple, since it is practically the same as
> subsetOf, with some operand
> swapping down in the works.
>
> Cheers
> Wolfgang
>
>
>
> 2010/10/14 Kumar Pandey <kumar.pandey at gmail.com>
>
> Thanks for this tip.
>> This works but only if the second rule's $f.numbers is on the RHS and
>> numbers is on the LHS of the subsetOf operator.
>> i.e Element(name="Super set - Rule 1", numbers subsetOf $f.numbers)
>>
>> Which is what I want in my case (to check if values in the fact is subset
>> of values in the rule).
>>
>> If I wanted it the other way round though the compiler seems to
>> disapprove.
>>
>> I get a compiler error like so --
>>
>> java.lang.RuntimeException: Unable to compile rule
>> Unable to determine the used declarations.
>> [Error: unknown class or illegal statement:
>> org.mvel2.ParserContext at 1aed5f9]
>> [Near : {... $f.numbers subsetOf dat ....}]
>>
>> If you try the Wolfgang's example and in the rule swap check like so - s1
>> disjoint numbers
>>
>> e2: Element( this != e1, n2 : name, s1 disjoint numbers )
>>
>> then you get error like so
>>
>> Unable to create Field Extractor for 's1' of '[ClassObjectType
>> class=com.test.context.Element]' in rule 'find disjoint pairs' : [Rule
>> name='find disjoint pairs']
>>
>> Sounds like a compiler bug as the operation should be commutative right?
>>
>>
>> Thanks
>> Kumar
>> 2010/10/13 Michael Anstis <michael.anstis at gmail.com>
>>
>>> OK, unless I'm missing something else you should be able to do something
>>> like this:-
>>>
>>> rule "setup superset - Rule 1"
>>> when
>>> then
>>> Element ss = new Element("Super set - Rule 1", 1, 2, 3);
>>> insert(ss);
>>> end
>>>
>>> rule "rule 1"
>>> when
>>> $f : Fact()
>>> Element(name="Super set - Rule 1", $f.numbers subsetOf numbers)
>>> then
>>> //Do something
>>> end
>>>
>>> This uses Wolfgang's Element Class from his example creating custom
>>> operators and assumes Fact exposes a "numbers" property which returns a
>>> Collection. This remains untested but I have no reason to doubt Wolfgang.
>>>
>>> With kind regards,
>>>
>>> Mike
>>>
>>>
>>>
>>>
>>>
>>> 2010/10/13 Kumar Pandey <kumar.pandey at gmail.com>
>>>
>>>> Well the problem is in forming the DRL :)
>>>> My use case is each rule represents an item with its specific property.
>>>> Fact is a user property and I want to filter the items for this user based
>>>> on the users incoming property as facts.
>>>>
>>>>
>>>> I have been able to get it working like so but eval is not optimum for
>>>> Rete. I was hoping for a better solution that would utilize the Rete
>>>> optimization.
>>>>
>>>> Here's my unit test DRL. Note, I have a isNotSubset method in my fact
>>>> object and I'm getting the list from rule and doing a check.
>>>>
>>>> package com.test
>>>> import com.rule.OfferFact;
>>>> global java.util.List list
>>>>
>>>> rule rule_not_sub_set
>>>> when
>>>> $context : Fact()
>>>> eval($context.isNotSubset("SINGER","DANCER", "DOCTOR"))
>>>> then
>>>> list.add(drools.getRule().getName());
>>>> end
>>>>
>>>>
>>>> 2010/10/13 Michael Anstis <michael.anstis at gmail.com>
>>>>
>>>>> Not as clear as I'd hoped; can you provide an actual rule DRL?
>>>>>
>>>>> 2010/10/13 Kumar Pandey <kumar.pandey at gmail.com>
>>>>>
>>>>> Michael
>>>>>> Here are some examples
>>>>>>
>>>>>> Rule1 has "str1", "Str2", "Str3"
>>>>>>
>>>>>> Rule2 has "str2", "Str3", "Str5", "Str6", "Str7"
>>>>>>
>>>>>> Rule3 has "str1", "Str3", "Str6"
>>>>>>
>>>>>> etc.
>>>>>>
>>>>>> With fact1 that has "str1", "Str3" , since this is subset of values
>>>>>> in Rule1 and Rule3, they should not fire but Rule2 should fire
>>>>>>
>>>>>> With fact2 that has "str2", "Str6", "Str7", Rule 2 should not fire and
>>>>>> Rule1 and Rule3 should fire.
>>>>>>
>>>>>> Hope this is clear.
>>>>>>
>>>>>> Thanks
>>>>>> Kumar
>>>>>>
>>>>>> 2010/10/13 Michael Anstis <michael.anstis at gmail.com>
>>>>>>
>>>>>>> Thanks, please provide a specific example rule too.
>>>>>>>
>>>>>>> Don't forget to keep your posts to the mailing list for the good of
>>>>>>> the community.
>>>>>>>
>>>>>>>
>>>>>>> On 12 October 2010 22:18, Kumar Pandey <kumar.pandey at gmail.com>wrote:
>>>>>>>
>>>>>>>> Michale
>>>>>>>> Thanks for the response.
>>>>>>>> Here's the link for the thread .
>>>>>>>>
>>>>>>>>
>>>>>>>> http://drools-java-rules-engine.46999.n3.nabble.com/Check-if-fact-is-subset-of-items-in-the-rule-td1680013.html#a1680013
>>>>>>>>
>>>>>>>> and the original about matching strings in two arrays.
>>>>>>>>
>>>>>>>> http://drools-java-rules-engine.46999.n3.nabble.com/Matching-strings-in-two-arrays-td1602511.html#a1602511
>>>>>>>>
>>>>>>>>
>>>>>>>> My use case is that I could have hundreds of rule and each rule
>>>>>>>> could have its own set of strings.
>>>>>>>> A fact object is run through these rules to see which ones are
>>>>>>>> fired.
>>>>>>>> One of the condition to check is that the a list in the fact is not
>>>>>>>> a subset of list in the rule.
>>>>>>>> That is fire the rule only if list in fact is not a subset of list
>>>>>>>> in rule.
>>>>>>>>
>>>>>>>> Thanks
>>>>>>>> Kumar
>>>>>>>>
>>>>>>>>
>>>>>>>> On Tue, Oct 12, 2010 at 12:38 PM, Michael Anstis <
>>>>>>>> michael.anstis at gmail.com> wrote:
>>>>>>>>
>>>>>>>>> OK, I admit I don't have the original thread anymore.
>>>>>>>>>
>>>>>>>>> If I am not wrong (which is always a possibility) for Wolfgang's
>>>>>>>>> operator to work you'd need to externalise the superset from the rule into
>>>>>>>>> WorkingMemory. You could have a rule with higher salience construct the
>>>>>>>>> superset WM fact.
>>>>>>>>>
>>>>>>>>> If you don't mind re-posting or providing a link to the complete
>>>>>>>>> thread (on Nabble or somewhere) I'll happily try to help further.
>>>>>>>>>
>>>>>>>>> On 12 October 2010 17:23, <kumar.pandey at gmail.com> wrote:
>>>>>>>>>
>>>>>>>>>> <quote author='Michael Anstis-2'>
>>>>>>>>>>
>>>>>>>>>> Wolfgang gave a great solution.
>>>>>>>>>> </quote>
>>>>>>>>>> Don't know if I'm missing something obvious here. I have a
>>>>>>>>>> superset in the rule itself. Each rule has a superset list. In this case how
>>>>>>>>>> would I use Wolfgang's solution. Its comparing through two arrays in
>>>>>>>>>> runtime. I have not been able to construct an array construct with specific
>>>>>>>>>> values in the rule itself.
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> rules-users mailing list
>>>>>>> rules-users at lists.jboss.org
>>>>>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> rules-users mailing list
>>>>>> rules-users at lists.jboss.org
>>>>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> rules-users mailing list
>>>>> rules-users at lists.jboss.org
>>>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>>>
>>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> rules-users mailing list
>>>> rules-users at lists.jboss.org
>>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>>
>>>>
>>>
>>> _______________________________________________
>>> rules-users mailing list
>>> rules-users at lists.jboss.org
>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>
>>>
>>
>>
>> --
>> Kumar Pandey
>> http://www.linkedin.com/in/kumarpandey
>>
>> _______________________________________________
>> rules-users mailing list
>> rules-users at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/rules-users
>>
>>
>
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/rules-users/attachments/20101015/f73d640d/attachment.html
More information about the rules-users
mailing list