class ParserTests (TestCase):
"""A fixture for setting up common tangle of objects used during parsing.
Hopefully this will change (or go away) as parsing is refactored.
"""
context = Context()
def __init__ (self, name, source_name):
TestCase.__init__(self, name)
source_catalog = self.context.get_source_catalog()
source = source_catalog.get_source_by_name(source_name)
self.mapped_terms = source_catalog.get_mapped_terms(source.uid)
self.parser = Parser(source)
def parse (self, file_name):
return self.parser.parse(file_name, self.mapped_terms, False)
def parse_email(self, file_name):
return self.parser.parse(file_name, self.mapped_terms, True)
def get_mapped_metadata(self, record):
return record.get_mapped_metadata(self.mapped_terms)
class ParserTests(TestCase):
"""A fixture for setting up common tangle of objects used during parsing.
Hopefully this will change (or go away) as parsing is refactored.
"""
context = Context()
def __init__(self, name, source_name):
TestCase.__init__(self, name)
source_catalog = self.context.get_source_catalog()
source = source_catalog.get_source_by_name(source_name)
self.mapped_terms = source_catalog.get_mapped_terms(source.uid)
self.parser = Parser(source)
def parse(self, file_name):
return self.parser.parse(file_name, self.mapped_terms, False)
def parse_email(self, file_name):
return self.parser.parse(file_name, self.mapped_terms, True)
def get_mapped_metadata(self, record):
return record.get_mapped_metadata(self.mapped_terms)
def __init__ (self, name, source_name):
TestCase.__init__(self, name)
source_catalog = self.context.get_source_catalog()
source = source_catalog.get_source_by_name(source_name)
self.mapped_terms = source_catalog.get_mapped_terms(source.uid)
self.parser = Parser(source)
def __init__(self, name, source_name):
TestCase.__init__(self, name)
source_catalog = self.context.get_source_catalog()
source = source_catalog.get_source_by_name(source_name)
self.mapped_terms = source_catalog.get_mapped_terms(source.uid)
self.parser = Parser(source)
if options.canary_id:
rows = [
[
options.canary_id,
],
]
else:
# get all active queuedrecord ids
cursor.execute("""
SELECT uid
FROM queued_records
""")
rows = cursor.fetchall()
parser = Parser(pubmed_source)
for row in rows:
qr = QueuedRecord(context, row[0])
print 'Fetching pubmed data for ui %s' % qr.unique_identifier
pm_data = pubmed_search.fetch(qr.unique_identifier)
fetched_records = parser.parse(mapped_terms=complete_mapping,
is_email=False,
data=pm_data)
if len(fetched_records) != 1:
print 'Fetch for %s (%s) found %s records, ignoring' % (
ui, qr.uid, len(fetched_records))
else:
print 'Orig metadata:', qr.metadata
fetched_rec = fetched_records[0]
print 'Fetched metadata:', fetched_rec.metadata
