Hello,
Here's a query that get's me almost there - I'm having trouble on next steps to
improve:
SELECT id,
regexp_replace(
marc,
E'.+?<datafield tag="856".+?><subfield code="u">([^<]+).*$',
E'\\1'
) as URL
FROM biblio.record_entry
WHERE marc LIKE '%<datafield tag="856"%' and deleted = 'False' and id > '0'
limit 200
Problem:
It appears that for records that have more than one URL, the URL column returns
the full marcxml data value instead of URLs. I need one record id, URL row for
each URL in the db (and of course not the normalized-for-indexing version
obtainable from the relevant metabib table).
Eg.
8214622 http://ir.lib.sfu.ca/handle/1892/10598
8214443 <record xmlns:xsi="http://www.w3.org/2001/X..... [this record contains
multiple 856s]
8213549 http://apps1.gdr.nrcan.gc.ca/mirage/mirage_list_e.php?id=247370
Any thoughts on how to improve this query to support the proper extraction of
id, raw URLs that includes proper extraction from multi-856 records too?
Thanks,
George
George Duimovich
NRCan Library / Bibliothèque de RNCan
