def fetch_opensky_pids (self):
"""
implement a solrQuery Feed to generate pids
(doi:* AND date:[2014-01-01 TO *])
- should fetch 5184 (VERIFIED)
"""
query = 'mods_identifier_doi_mt:*'
query += ' AND keyDate:[2014-01-01T00:00:00Z TO *]'
# query += ' AND keyDate:[{} TO *]'.format(get_utc_time('2014-01-01'))
# print 'QUERY: {}'.format(query)
args = {
'params' :
{
# 'q': affiliation_clause + ' AND ' + date_clause,
# 'q': 'mods_identifier_doi_mt:*'
'q': query
},
'baseUrl': CONFIG.get("fedora", "SERVER") + CONFIG.get("fedora", "SOLR_PATH"),
}
feed = feeds.SolrSearchFeed(**args)
print '{} in feed'.format(len(feed.pids))
return feed.pids
def getFeed():
args = {
'params' :
{
'q': 'mods_note_funding_s:*'
},
'baseUrl': CONFIG.get("fedora", "SERVER") + CONFIG.get("fedora", "SOLR_PATH"),
}
feed = feeds.SolrSearchFeed(**args)
feed.numFound = len(feed.pids)
print 'feed: %d (%d)' % (len(feed.pids), feed.size())
feed.pids.sort()
# for i, pid in enumerate(feed.pids):
# print '- %d - %s' % (i, pid)
return feed
def update_old_funder_records():
"""
The Kuali backfill updated the records that also had Kuali IDs, but there are
452 that still have old-school funder info (mods_name_corporate_funder_s:*).
<name type="corporate">
<namePart>National Science Foundation (NSF)</namePart>
<role>
<roleTerm type="text" authority="marcrelator">funder</roleTerm>
</role>
</name>
<note type="funding">National Science Foundation (NSF): 1852977</note>
Now we want to remove the corporate funder info, and try to normalize the
award_ids with Kuali-verfied versions. If ids can't be normlized thus, then
add a 'displayLabel' = 'Legacy funding data' attribute
"""
dowrites = 0
args = {
'params': {
'q': 'mods_name_corporate_funder_s:*'
},
'baseUrl':
CONFIG.get("fedora", "SERVER") + CONFIG.get("fedora", "SOLR_PATH"),
}
feed = feeds.SolrSearchFeed(**args)
print 'feed has {}'.format(len(feed.pids))
NotesMODS.dowrites = dowrites
for pid in feed.pids:
print pid
mods_xml = get_datastream(pid, 'MODS')
mods = NotesMODS(mods_xml, pid)
# print 'BEFORE Backfill'
# mods.show_notes()
mods.do_back_fill([])