def testDelete(self):
self.addThree(date.today())
Citation.delete([1, 2])
count = 0
for m in self.sess.query(Citation):
self.assertEqual(3, m.pmid)
count += 1
self.assertEqual(1, count)
def testDelete(self):
self.addThree(date.today())
Citation.delete([1, 2])
count = 0
for m in self.sess.query(Citation):
self.assertEqual(3, m.pmid)
count += 1
self.assertEqual(1, count)
def testSelect(self):
self.addThree(date.today())
count = 0
for row in Citation.select([1, 2], ['journal']):
self.assertEqual('Journal {}'.format(row['pmid']), row['journal'])
count += 1
self.assertEqual(2, count)
def testSelect(self):
self.addThree(date.today())
count = 0
for row in Citation.select([1, 2], ['journal']):
self.assertEqual('Journal {}'.format(row['pmid']), row['journal'])
count += 1
self.assertEqual(2, count)
def DefaultCitation(pmid=1,
status='MEDLINE',
title='title',
journal='journal',
pub_date='1990 published',
created=date.today()):
return Citation(pmid, status, title, journal, pub_date, created)
def testSelectAll(self):
d = date.today()
self.addThree(d)
count = 0
for row in Citation.selectAll([1, 2]):
self.assertEqual('Journal {}'.format(row['pmid']), row['journal'])
self.assertEqual(d, row['created'])
self.assertEqual('MEDLINE', row[1])
count += 1
self.assertEqual(2, count)
def testSelectAll(self):
d = date.today()
self.addThree(d)
count = 0
for row in Citation.selectAll([1, 2]):
self.assertEqual('Journal {}'.format(row['pmid']), row['journal'])
self.assertEqual(d, row['created'])
self.assertEqual('MEDLINE', row[1])
count += 1
self.assertEqual(2, count)
def testInsertMultiple(self):
d = date.today()
data = {Citation.__tablename__: [
dict(pmid=1, status='MEDLINE', title='Title 1',
journal='Journal 1', pub_date='PubDate', created=d),
dict(pmid=2, status='MEDLINE', title='Title 2',
journal='Journal 2', pub_date='PubDate', created=d),
dict(pmid=3, status='MEDLINE', title='Title 3',
journal='Journal 3', pub_date='PubDate', created=d)
]}
Citation.insert(data)
count = 0
for m in self.sess.query(Citation):
if m.pmid not in (1, 2, 3):
self.fail(m)
else:
count += 1
self.assertEqual(3, count)
def testModifiedBefore(self):
today = date.today()
yesterday = today - timedelta(days=1)
m1 = DefaultCitation()
m2 = DefaultCitation(2)
m1.modified = today
m2.modified = yesterday
self.sess.add(m1)
self.sess.add(m2)
self.sess.commit()
self.assertListEqual([2], list(Citation.modifiedBefore([1, 2], date.today())))
def testInsert(self):
data = {
Citation.__tablename__: [
dict(pmid=1, status='MEDLINE', title='Title', journal='Journal',
pub_date='PubDate', created=date.today())
],
Abstract.__tablename__: [
dict(pmid=1, source='NLM')
],
Section.__tablename__: [
dict(pmid=1, source='NLM', seq=1, name='Title', content='The title.')
],
Author.__tablename__: [
dict(pmid=1, pos=1, name='Author')
],
Descriptor.__tablename__: [
dict(pmid=1, num=1, major=True, name='descriptor')
],
Qualifier.__tablename__: [
dict(pmid=1, num=1, sub=1, major=True, name='descriptor')
],
Identifier.__tablename__: [
dict(pmid=1, namespace='ns', value='id')
],
Database.__tablename__: [
dict(pmid=1, name='name', accession='accession')
],
Chemical.__tablename__: [
dict(pmid=1, idx=1, name='name')
],
Keyword.__tablename__: [
dict(pmid=1, owner='NLM', cnt=1, major=True, value='name')
],
}
Citation.insert(data)
for m in self.sess.query(Citation):
self.assertEqual(1, m.pmid)
for t in Citation.CHILDREN:
self.assertEqual(1, list(self.sess.query(t))[0].pmid)
def testModifiedBefore(self):
today = date.today()
yesterday = today - timedelta(days=1)
m1 = DefaultCitation()
m2 = DefaultCitation(2)
m1.modified = today
m2.modified = yesterday
self.sess.add(m1)
self.sess.add(m2)
self.sess.commit()
self.assertListEqual([2],
list(Citation.modifiedBefore([1, 2],
date.today())))
def testInsert(self):
data = {
Citation.__tablename__: [
dict(pmid=1,
status='MEDLINE',
year=1990,
title='Title',
journal='Journal',
pub_date='1990 PubDate',
created=date.today())
],
Abstract.__tablename__: [dict(pmid=1, source='NLM')],
Section.__tablename__: [
dict(pmid=1,
source='NLM',
seq=1,
name='Title',
content='The title.')
],
Author.__tablename__: [dict(pmid=1, pos=1, name='Author')],
Descriptor.__tablename__:
[dict(pmid=1, num=1, major=True, name='descriptor')],
Qualifier.__tablename__:
[dict(pmid=1, num=1, sub=1, major=True, name='descriptor')],
Identifier.__tablename__:
[dict(pmid=1, namespace='ns', value='id')],
Database.__tablename__:
[dict(pmid=1, name='name', accession='accession')],
Chemical.__tablename__: [dict(pmid=1, idx=1, name='name')],
Keyword.__tablename__:
[dict(pmid=1, owner='NLM', cnt=1, major=True, value='name')],
}
Citation.insert(data)
for m in self.sess.query(Citation):
self.assertEqual(1, m.pmid)
for t in Citation.CHILDREN:
self.assertEqual(1, list(self.sess.query(t))[0].pmid)
def testInsertMultiple(self):
d = date.today()
data = {
Citation.__tablename__: [
dict(pmid=1,
status='MEDLINE',
year=1990,
title='Title 1',
journal='Journal 1',
pub_date='1990 PubDate',
created=d),
dict(pmid=2,
status='MEDLINE',
year=1990,
title='Title 2',
journal='Journal 2',
pub_date='1990 PubDate',
created=d),
dict(pmid=3,
status='MEDLINE',
year=1990,
title='Title 3',
journal='Journal 3',
pub_date='1990 PubDate',
created=d)
]
}
Citation.insert(data)
count = 0
for m in self.sess.query(Citation):
if m.pmid not in (1, 2, 3):
self.fail(m)
else:
count += 1
self.assertEqual(3, count)
def MedlineCitation(self, element):
options = {}
for name, key in (('DateCompleted', 'completed'),
('DateCreated', 'created'), ('DateRevised',
'revised')):
e = element.find(name)
if e is not None:
options[key] = ParseDate(e)
created = options['created']
del options['created']
status = element.get('Status')
journal = element.find('MedlineJournalInfo').find(
'MedlineTA').text.strip()
article = element.find('Article')
title = article.find('ArticleTitle').text
# ArticleTitle element tags should never be empty when in fact they are sometimes,
# in violation of the MEDLINE DTD; For example, while PMID 22536004 has a
# VernacularTitle, its ArticleTitle element is empty.
if title is None and article.find('VernacularTitle') is not None:
logging.info(
'PMID %d had no ArticleTitle; using VernacularTitle instead' %
self.pmid)
title = article.find('VernacularTitle').text
if title is None:
logging.warning('could not find any title for PMID %d' % self.pmid)
title = 'UNKNOWN' # NB: a DB constraint ensures titles are never empty!
title = title.strip()
pub_date = self.parsePubDate(
article.find('Journal/JournalIssue/PubDate'))
issue = self.parseIssue(article.find('Journal/JournalIssue'))
pagination = self.parsePagination(
article.find('Pagination/MedlinePgn'))
if issue:
options['issue'] = issue
if pagination:
options['pagination'] = pagination
return Citation(self.pmid, status, title, journal, pub_date, created,
**options)
DATA = [
Abstract(1, 'NLM'),
Section(1, 'NLM', 1, 'Abstract', 'The Abstract 1'),
Section(1, 'NLM', 2, 'Abstract', 'The Abstract 2'),
Descriptor(1, 1, 'd_name', True),
Descriptor(1, 2, 'd_name'),
Qualifier(1, 1, 1, 'q_name', True),
Author(1, 1, 'first'),
Author(1, 2, 'last'),
Identifier(1, 'ns', 'id'),
Database(1, 'name', 'accession'),
PublicationType(1, 'some'),
PublicationType(1, 'another'),
Chemical(1, 1, 'name', 'uid'),
Keyword(1, 'NOTNLM', 1, 'name', True),
Citation(1, 'MEDLINE', 'title', 'journal', '1990 pub_date', date.today()),
]
class ParserMock:
def __init__(self, instances):
self.instances = instances
def parse(self, _):
for i in self.instances:
yield i
class TestDump(unittest.TestCase):
def setUp(self):
self.out = {
def testExisting(self):
self.addThree(date.today())
self.assertListEqual([1, 3], list(Citation.existing([1, 3, 5])))
def testExisting(self):
self.addThree(date.today())
self.assertListEqual([1, 3], list(Citation.existing([1, 3, 5])))
def testRequireModifiedDateOrNone(self):
m = Citation(*self.defaults)
m.modified = ''
self.sess.add(m)
self.assertRaises(StatementError, self.sess.commit)
def testRequireModifiedDateOrNone(self):
m = Citation(*self.defaults)
m.modified = ''
self.sess.add(m)
self.assertRaises(StatementError, self.sess.commit)
def testToString(self):
d = date.today()
r = Citation(1, 'MEDLINE', 'title', 'journal\\.', '1990 PubDate', d)
line = "1\tMEDLINE\t1990\ttitle\tjournal\\\\.\t1990 PubDate\t\\N\t\\N\t{}\t\\N\t\\N\t{}\n".format(
d.isoformat(), d.isoformat())
self.assertEqual(line, str(r))