metamorph: use and combination of reset, flushWith and sameEntity attributes

[guenterh]

• since some time I'm struggling with the appropriate use of the reset, flushWith and sameEntity attributes on the various collector artefacts but also on the entity tag of Metamorph
• Mostly after some try and error definitions I got a result which was the one I expected but I had difficulties to explain the clear differences between the examples
• Hopefully I can now provide a distinguished example which might be helpful to discuss the mechanisms behind the implementation
• This directory contains the examples.
• flushWith and reset=true is used in the MorphScript_case_1 whereas sameEntity=true and reset=false delivers the correct result in this Morph script_case_2
• the output files are for case 1 and case 2
• it seems combine uses the sameEntity=true attribute for collecting the various values of an Entity in combination with an If Statement (and no inside collectors) while concat needs the flushWith Attribute in combination with IF and combine with reset=true attribute as the outer collector
• During our get Together meeting in HH I already mentioned my troubles with this and a helpful voice in the audience (I couldn't see as Skype participant... (edit: me,@dr0i)) was optimistic for being able to give more background infomation
• By the way these examples are part of my attempt to transform our XSLT based transformations for Solr SearchDocuments into a Metafacture/Metamorph implementations. My main motivation for this is the question: Where are the limits and differences of Metamorph compared to traditional XSLT procedures. If you are interested how it looks like as XSLT you can find it here Hints are welcome if you know more elegant possibilities to express match ranges matches(@code, '[b-su-z8]') in XSLT compared to the way I have done this

Thanks for any replies!!

Günter

[dr0i]

Hints are welcome if you know more elegant possibilities to express match ranges matches(@code, '[b-su-z8]') in XSLT compared to the way I have

Not exactly sure but if you meant to abbreviate this:

<data source="811??.b|811??.c|811??.d" />

you can use character classes:

  <data source="811??.[bcd]" />

[dr0i]

Re: case1 vs. case2: A bit complex, this. I admit to not totally forsee the result of a complex combination of attributes and their parameters myself! So at that point I just modify&run&test until the result is satisfactory.

I don't understand what the problem is: Since the result of the two cases are also very different (case1: every matched field of all entities are entity-indistinguishable put into a set, while case2 puts one concatenated string per entity into the set) you evidently use two different approaches.

Nevertheless here some remarks: In case1 you use <data source="950??.e"/> in the concat section while flushing with (in principle) every 950-subfield (I think). As 950??.e happens to be the last entry of all 950-entities and it must be concatenated before the concatenated string is flushed ("saturation" - I think this is so, not exactly sure) the "flushing" coincides with what would be better declared as ... flushWith="950??.e" > (because of the clearness of it).

It's sometimes hard to explain and understand complex data transformations, I know. I would find it easier if you would have a question like "I have this data and want this outcome - how to do it?". But I gather you want more of the deeper understanding which I find hard for myself. Maybe you haven't read http://b3e.net/metamorph-book/latest/datamodel.html yet? I haven't yet - seems to be interesting.

[dr0i]

Closing this. Note that some of the problems described here may be solved by newer documentation at https://github.com/metafacture/metafacture-core/wiki/Metamorph-Cookbook.