def main(options):
warn("getting pmid list..." % ())
idlist = get_pmidlist(options)
warn("processing %d ids: %s" % (len(idlist), idlist))
if options.dry_run: exit(0)
fuse = options.fuse
for pmid in idlist:
pubmed = Pubmed(pmid)
warn("pmid is %s" % (pmid))
continue
try:
pubmed.store() # does the fetching automatically
except Exception as e:
warn("%d: caught %s" % (pmid, e))
fuse -= 1
if (fuse == 0): break
exit(0)
def main(options):
warn("getting pmid list..." % ())
idlist=get_pmidlist(options)
warn("processing %d ids: %s" % (len(idlist), idlist))
if options.dry_run: exit(0)
fuse=options.fuse
for pmid in idlist:
pubmed=Pubmed(pmid)
warn("pmid is %s" % (pmid))
continue
try:
pubmed.store() # does the fetching automatically
except Exception as e:
warn("%d: caught %s" % (pmid, e))
fuse-=1
if (fuse==0): break
exit(0)
def get_pubmed_words(pubmed_id):
'''
return a dict in the same format as get_field_words: k=field, v=sanitized list of words
'''
words=dict()
pubmed=Pubmed(pubmed_id).populate()
for tag in Pubmed.text_tags:
try:
words[tag]=getattr(pubmed, tag)
# warn("fu %d: %s -> %s" % (pubmed_id, tag, words[tag]))
except AttributeError as ae:
pass
return words
def test_store(self):
pmid = 18297132
pubmed = Pubmed(pmid)
self.assertEqual(pubmed.pubmed_id, pmid)
try:
os.unlink(pubmed.path())
except OSError:
pass
pubmed.fetch()
self.assertTrue(os.access(pubmed.path(), os.R_OK))
self.assertEqual(os.path.getsize(pubmed.path()), 17990)
def test_store(self):
pmid=18297132
pubmed=Pubmed(pmid)
self.assertEqual(pubmed.pubmed_id, pmid)
try: os.unlink(pubmed.path())
except OSError: pass
pubmed.fetch()
self.assertTrue(os.access(pubmed.path(), os.R_OK))
self.assertEqual(os.path.getsize(pubmed.path()), 17990)
def test_store(self):
warn("\n")
pmid = 18297132
pubmed = Pubmed(pmid)
self.assertIsInstance(pubmed, Pubmed)
self.assertEqual(pubmed.pubmed_id, pmid)
pubmed.remove()
pubmed.store()
self.assertTrue(os.access(pubmed.path(), os.R_OK))
cursor = Pubmed.mongo().find({'pubmed_id': pmid})
self.assertEqual(cursor.count(), len(Pubmed.text_tags))
tag2count = {
'MeshHeading': 22,
'AbstractText': 247,
'ArticleTitle': 15
}
for record in cursor:
tag = record['tag']
self.assertEqual(len(record['words']), tag2count[tag])
def test_store(self):
warn("\n")
pmid=18297132
pubmed=Pubmed(pmid)
self.assertIsInstance(pubmed, Pubmed)
self.assertEqual(pubmed.pubmed_id, pmid)
pubmed.remove()
pubmed.store()
self.assertTrue(os.access(pubmed.path(), os.R_OK))
cursor=Pubmed.mongo().find({'pubmed_id':pmid})
self.assertEqual(cursor.count(), len(Pubmed.text_tags))
tag2count={'MeshHeading':22,
'AbstractText':247,
'ArticleTitle':15}
for record in cursor:
tag=record['tag']
self.assertEqual(len(record['words']), tag2count[tag])