Found another pattern "]], [[", they still just a few.
--
Nuno Tavares
DRI, Consultoria Informática
Telef: +351 936 184 086
Nuno Tavares escreveu:
Dear Vicenç [I believe all others should ignore this
message],
Yes, I've just noticed, e.g., "[[Ogassa]] i [[Ribes de Freser]]". My
(professional) opinion is that this is a huge mistake, since that field
represents 2 values, which is a clear violation of standard database
rules. Having this in mind, I'd suggest splitting the object in two.
I found a pattern for "multiple municipalities in the same object",
which seems to be "]] i [[", and there seems to be only 6 of them. You
can just wipe out one of them from the field, or manually duplicate the
record them for splitting - the latter being more complex, think of a
good ID disambiguation though. Regarding link disambiguations, you can
just add a column for that - in a "working table", as they are of no use
for a listing like this (alas, that kind of data is only relevant if you
plan to regenerate the lists).
We had similar problems - we even had descriptions names in wiki syntax
already, our lists date back to 2006. As I wanted to preserve the links
and original "names" (designations), I've done some transformations on
the data before it gets on the list - this kind of approach was actually
the reason to have our tools/plist [1] - notice multiple links near the
end of each description, and notice that, if you export to wiki format,
you'll find municipalities disambiguated as well.
[1]
http://www.wikilovesmonuments.org.pt/tools/plist/?region=Braga
That's why you currently can "regenerate" the lists with their original
descriptions and links.
This is, however, a more complex approach which I won't have time to
assist you, but I may point you directions if you know ETL.