[Metafacture] Regex, conjunction

[Christoph Böhme]

Hi Pascal

I created a branch [1] on my Metafacture fork which contains three new
Metamorph collectors. These implement quantors to simplify expressing
conditions in Metamorph. They are called <all>, <any>, and <none>.

It would be great if you could give them a try and let me know whether
they are helpful.

The next step is to add an <if> statement which allows something like
this to execute a combine only if a condition is met:

<combine ...>
  <if>
    <all>
      <data ... />
    </all>
  </if>
  <data .. />
</combine>

Best,
Christoph

[1] https://github.com/cboehme/metafacture-core/tree/quantors-for-metamorph


Am 24.11.2013 11:11, schrieb Christoph Böhme:
> Hi Pascal
> 
> A not-operation can be implemented with the following bit of code inside
> a combine-element:
> 
> <choose name="" value="">
> 	<data source="051.">
> 		<!-- This is condition B: -->
> 		<regexp match="(^[^a|])" />
> 		<constant value="B_MATCHES" />
> 	</data>
> 	<data source="_id">
> 		<!-- This can be any field that is available in all
> 		     records: -->
> 		<constant value="B_DOES_NOT_MATCH" />
> 	</data>
> 	<postprocess>
> 		<!-- This bit implements the not-operation. The
> 		     choose-element only fires if it has NOT selected
> 		     B_MATCHES: -->
> 		<equals string="B_DOES_NOT_MATCH" />
> 	</postprocess>
> </choose>
> 
> It is not straight forward but I think it should allow you to write your
> "A andnot B" condition.
> 
> Writing Metamorph scripts which include conditions seems to be one of
> the most unintuitive parts of Metamorph at the moment. I think we should
> consider implementing a more expressive approach to do this. Using
> <combine> for this is not really straight-forward. I created an issue
> for this [1]. The issue outlines two possible ways of implementing it.
> 
> Best,
> Christoph
> 
> [1] https://github.com/culturegraph/metafacture-core/issues/151
> 
> 
> Am 15.11.2013 11:25, schrieb Christoph, Pascal:
>> Hi *,
>>
>> I don't know how to do this with metamorph:
>>
>> "If A andnot B => pass to receiver"
>>
>> Details:
>> Having two entities: entity A has to match a regex, and B has also a regex. The
>> whole should be true if A matches and B does NOT match[1]. Furthermore, if B
>> does not exist at all, the whole should also be true.
>>
>> Thus far I only managed using function "combine" with parameter
>> flushWith="record", like this:
>>
>> <combine name="foo" value="bar" flushWith="record">
>> 	<data source="051.">
>> 		<regexp match="(^[^a|])"/>
>> 	</data>
>> 	<data source="050.">
>> 		<regexp match="^a.*"/>
>> 	</data>
>> </combine>
>>
>> I got my "foo bar" if:
>>
>> - 050. matches AND 051. is missing
>>
>> and that's it, because in Regexp.java nothing is fired (aka "passed to
>> receiver") if the regex does not match. And no, I can't formulate the negative
>> regex as a positive regex.
>>
>> Is there a possibility to do what I want?
>> I enhanced the Regexp class with :
>>
>> if (this.getNamedValueReceiver().getClass().equals(Combine.class) && !match) {
>> getNamedValueReceiver().receive(name, NO_MATCH, source, recordCount, entityCount);
>> }
>>
>> and adopted the Combine class. It's working , but it does not seem apropriate.
>> Anything I missed ?
>>
>> cheers, pascal
>> _______________________________________________
>> Metafacture mailing list
>> Metafacture at lists.dnb.de
>> http://lists.dnb.de/mailman/listinfo/metafacture
>>
> _______________________________________________
> Metafacture mailing list
> Metafacture at lists.dnb.de
> http://lists.dnb.de/mailman/listinfo/metafacture
>