def test_from_data(self):
data = myopen(
os.path.join(os.path.dirname(__file__), '20_ptm_381742.doc'),
'rb').read()
with PlenumProtocolFile.get_from_data(data) as protocol:
if six.PY2:
expected_exception = "'NoneType' object has no attribute 'decode'"
if six.PY3:
expected_exception = "'NoneType' object is not iterable"
expected_data = {
'knesset_num_heb': 'עשרים',
'meeting_num_heb': 'מאתיים-ותשע-עשרה',
"booklet_num_heb": None,
'booklet_meeting_num_heb': 'רי"ט',
'date_string_heb': ('21', 'מרס', '2017'),
'time_string': ('16', '00'),
'datetime': datetime(2017, 3, 21, 16, 0),
"knesset_num": 20,
'booklet_num': expected_exception,
"booklet_meeting_num": 219
}
actual_data = self._get_protocol_data(protocol, expected_data)
self.assertEqual(actual_data, expected_data)
def __init__(self, name, parent_datapackage_path):
self._meeting_schema = PlenumMeetings.get_json_table_schema()
self._protocol_schema = PlenumProtocolFile.get_json_table_schema()
schema = merge_table_schemas(self._meeting_schema,
self._protocol_schema,
{"fields": [{"name": "protocol_original", "type": "string", "description": "original file (without processing), in case of error will be empty"},
{"name": "protocol_antiword_text", "type": "string", "description": "text after antiword processing, in case of error will be empty"},
{"name": "scraper_errors", "type": "string", "description": "comma separated list of errors encountered"}]})
super(PlenumMeetingsResource, self).__init__(name, parent_datapackage_path, schema,
file_fields=["protocol_original", "protocol_antiword_text"])
self.descriptor["plenum_errors"] = []
def test_from_file(self):
with open(os.path.join(os.path.dirname(__file__), '20_ptm_318579.doc')) as f:
with PlenumProtocolFile.get_from_file(f) as protocol:
plenum_protocol_assertions(self, protocol)
def test_from_url(self):
with PlenumProtocolFile.get_from_url('http://fs.knesset.gov.il/20/Plenum/20_ptm_318579.doc') as protocol:
plenum_protocol_assertions(self, protocol)
def test_from_file(self):
with PlenumProtocolFile.get_from_filename(
os.path.join(os.path.dirname(__file__),
'20_ptm_318579.doc')) as protocol:
plenum_protocol_assertions(self, protocol)
def test_from_content(self):
with open(os.path.join(os.path.dirname(__file__),
'20_ptm_318579.doc')) as f:
with PlenumProtocolFile.get_from_data(f.read()) as protocol:
plenum_protocol_assertions(self, protocol)
def test_from_url(self):
with PlenumProtocolFile.get_from_url(
'http://fs.knesset.gov.il/20/Plenum/20_ptm_318579.doc'
) as protocol:
plenum_protocol_assertions(self, protocol)
def __init__(self, url, protocol, date):
self.url = url
self.protocol = PlenumProtocolFile.get_from_data(protocol)
self.date = date
