def disabled_test_factory():
TESTDATA = (
'1-800-RING-RCN (1-800-746-4726)Website:'
'Statement Date:Account Number:Payment Due Date:1'
'Statement Date:Account Number:Payment Due Date:01/31/20172301-')
extractor = ExtractorFactory.get_matching_extractor(TESTDATA)
assert type(extractor) is RcnDateExtractor
def test_factory():
TESTDATA = ('BalanceVested Percent9/30/15 Vested BalanceEMPLOYEE 401(K)'
'$2Performance SummaryYour Personal Rate of '
'Return for the period 7/1/15 through 9/30/15 is -2.61%. '
'Visit us at https://www.planservices.com/')
extractor = ExtractorFactory.get_matching_extractor(TESTDATA)
assert type(extractor) is AscensusDateExtractor
def test_factory():
TESTDATA = (
'Client Services: 800-662-2739December 31, 2017, year-to-date statement'
'DateTransactionAmountSharepriceSharestransactedTotalsharesownedValue'
'Beginning balance on 12/31/2016$12.34537.672'
'3214.354Ending balance on 12/31/2017$23,456.78'
'Client Services: 800-662-2739Page 2 of8'
'Vanguard, P.O. Box 2600Valley Forge, PA 19482-2600')
extractor = ExtractorFactory.get_matching_extractor(TESTDATA)
assert type(extractor) is VanguardDateExtractor
def test_factory():
TESTDATA = (
'Your 360 Savings Activity Account: DJT '
'Annual Percentage Yield Earned: 0.75% '
'ActivityDateAmountBalance'
'Opening Balance02/01/2016$13,213.32'
'My Info section.capitalone360.comInteractive'
'Deposit from VPUTIN02/16/2016$350,000,000.00$350,013,213.32'
'Withdrawal to Electric Checking XXXX259476 / 27 / 2016'
'$(130, 000, 000.00)$220, 013, 213.32'
'Closing Balance02 / 29 / 2016$220, 013, 213.32'
'See below for important information.')
extractor = ExtractorFactory.get_matching_extractor(TESTDATA)
assert type(extractor) is CapitalOne360DateExtractor
def determine_action_for_file(self, filepath):
""" Determine the action required on the specified file:
- Checks for a duplicate using MD5 hash
- Extracts text from the document
- Extracts meaningful data from the document
- Constructs the necessary Action to take on the file and appends
it to the actions list
"""
# TODO: perform this every time? Or only when we find a duplicate target filename?
file_hash = Md5Reader().parse(filepath)
if self.config.hash_only:
print('{} - {}'.format(file_hash, filepath))
return
if file_hash in self.hashes:
self.actions.append(
Action.create_delete_action(
filepath, reason='Duplicate hash: {}'.format(file_hash)))
return
# TODO: can PdfReader accept/process the same streamed data as Md5Reader?
contents = self.reader.parse(filepath)
if self.config.extract_only:
print(contents)
return
extractor = ExtractorFactory.get_matching_extractor(contents)
data = extractor.extract(contents)
data.set_source(filepath)
data.set_hash(file_hash)
new_name = extractor.rename(data)
old_name = filepath.split('/')[-1]
new_path = filepath[0:filepath.rfind('/') + 1] + new_name
self.hashes.append(file_hash)
if old_name == new_name:
action = Action.create_ignore_action(
filepath, reason='Already named correctly')
elif os.path.isfile(new_path):
# TODO: this isn't good enough to avoid an overwrite, as there is no
# coordination between files.
# How about storing the data objects in a hash of target paths?
# Or, key's c
existing_hash = Md5Reader().parse(new_path)
if existing_hash == data.get_hash():
reason = (
'{} matches existing hash ({}) - shared by [{}].'.format(
data.get_source(), existing_hash, new_path))
action = Action.create_delete_action(data.get_source(),
reason=reason)
else:
action = Action.create_ignore_action(
filepath,
reason=
('Target File [{}] already exists. Ignoring to avoid an overwrite.'
.format(new_name)))
else:
action = Action.create_rename_action(filepath, new_path)
self.actions.append(action)
if self.config.verbose:
print('Adding action: {}'.format(action))
def test_factory():
extractor = ExtractorFactory.get_matching_extractor(TESTDATA)
assert type(extractor) is SocialSecurityDateExtractor
def test_factory():
extractor = ExtractorFactory.get_matching_extractor(TESTDATA)
assert isinstance(extractor, EXTRACTOR_UNDER_TEST)
def test_factory():
extractor = ExtractorFactory.get_matching_extractor(TESTDATA)
assert type(extractor) is HanscomDateExtractor
def test_factory():
extractor = ExtractorFactory.get_matching_extractor(TESTDATA)
assert type(extractor) is EXPECTED_EXTRACTOR
def test_factory():
extractor = ExtractorFactory.get_matching_extractor(TESTDATA)
assert type(extractor) is AmazonPrimeDateExtractor