def test_null_matches(self):
matches = matchers.best_match('', US_STATES, top_n=2)
self.assertEqual(len(matches), 2)
matches = matchers.best_match('nothing', [], top_n=5)
self.assertEqual([], matches)
matches = matchers.best_match(None, [], top_n=5)
self.assertEqual([], matches)
def test_null_matches(self):
matches = matchers.best_match('', US_STATES, top_n=2)
self.assertEqual(len(matches), 2)
matches = matchers.best_match('nothing', [], top_n=5)
self.assertEqual([], matches)
matches = matchers.best_match(None, [], top_n=5)
self.assertEqual([], matches)
def test_case_insensitivity(self):
"""Make sure we disregard case when doing comparisons."""
fake_comp = 'TeST'
fake_categories = ['test', 'thing', 'face']
match, percent = matchers.best_match(
fake_comp, fake_categories, top_n=1
)[0]
self.assertEqual(match, 'test')
self.assertEqual(percent, 100)
def test_multiple_matches(self):
"""tests that multiple matches come back"""
state = 'Ilinois'
matches = matchers.best_match(state, US_STATES, top_n=6)
self.assertEqual(len(matches), 6)
first_match = matches[0]
second_match = matches[1]
self.assertEqual(first_match[0], 'illinois')
self.assertGreater(first_match[1], 90)
self.assertLess(second_match[1], 90)
def test_multiple_matches(self):
"""tests that multiple matches come back"""
state = 'Ilinois'
matches = matchers.best_match(state, US_STATES, top_n=6)
self.assertEqual(len(matches), 6)
first_match = matches[0]
second_match = matches[1]
self.assertEqual(first_match[1], 'illinois')
self.assertGreater(first_match[2], 90)
self.assertLess(second_match[2], 90)
def test_case_insensitivity(self):
"""Make sure we disregard case when doing comparisons."""
fake_comp = 'TeST'
fake_categories = ['test', 'thing', 'face']
match, percent = matchers.best_match(fake_comp,
fake_categories,
top_n=1)[0]
self.assertEqual(match, 'test')
self.assertEqual(percent, 100)
def __init__(self, raw_columns, dest_columns, previous_mapping=None, map_args=None,
threshold=0):
"""
:param raw_columns: list of str. The column names we're trying to map.
:param dest_columns: list of str. The columns we're mapping to.
:param previous_mapping: Method that contains previous mapped columns
:param map_args: .. todo: document as I have no idea what this is doing.
:param thresh: int, Minimum value of the matching confidence to allow for matching.
"""
self.data = {}
for raw in raw_columns:
attempt_best_match = False
# We want previous mappings to be at the top of the list.
if previous_mapping and callable(previous_mapping):
args = map_args or []
# Mapping will look something like this -- [u'table', u'field', 100]
mapping = previous_mapping(raw, *args)
if mapping:
self.add_mappings(raw, [mapping], True)
else:
attempt_best_match = True
else:
attempt_best_match = True
# Only enter this flow if we haven't already selected a result. Ignore blank columns
# with conf of 100 since a conf of 100 signifies the user has saved that mapping.
if attempt_best_match:
# convert raw fields spaces into underscores because that is what is in the database
raw_test = raw.replace(' ', '_')
# try some alternatives to the raw column in specific cases
# (e.g. zip => postal code). Hack for now, but should make this some global
# config or organization specific config
if raw_test.lower() == 'zip' or raw_test.lower() == 'zip_code':
raw_test = 'postal_code'
if raw_test.lower() == 'gba':
raw_test = 'gross_floor_area'
if raw_test.lower() == 'building_address':
raw_test = 'address_line_1'
matches = matchers.best_match(raw_test, dest_columns, top_n=5)
# go get the top 5 matches. format will be [('PropertyState', 'building_count', 62), ...]
self.add_mappings(raw, matches)
# convert this to an exception and catch it some day...
index = 0
while self.duplicates and index < 10:
index += 1
_log.debug("Index: {} with duplicates: {}".format(index, self.duplicates))
for k, v in self.duplicates.iteritems():
self.resolve_duplicate(k, v)
if threshold > 0:
self.apply_threshold(threshold)
def __init__(self, raw_columns, dest_columns, previous_mapping=None, map_args=None,
threshold=0):
"""
:param raw_columns: list of str. The column names we're trying to map.
:param dest_columns: list of str. The columns we're mapping to.
:param previous_mapping: Method that contains previous mapped columns
:param map_args: .. todo: document as I have no idea what this is doing.
:param thresh: int, Minimum value of the matching confidence to allow for matching.
"""
self.data = {}
for raw in raw_columns:
attempt_best_match = False
# We want previous mappings to be at the top of the list.
if previous_mapping and callable(previous_mapping):
args = map_args or []
# Mapping will look something like this -- [u'table', u'field', 100]
mapping = previous_mapping(raw, *args)
if mapping:
self.add_mappings(raw, [mapping], True)
else:
attempt_best_match = True
else:
attempt_best_match = True
# Only enter this flow if we haven't already selected a result. Ignore blank columns
# with conf of 100 since a conf of 100 signifies the user has saved that mapping.
if attempt_best_match:
# convert raw fields spaces into underscores because that is what is in the database
raw_test = raw.replace(' ', '_')
# try some alternatives to the raw column in specific cases
# (e.g. zip => postal code). Hack for now, but should make this some global
# config or organization specific config
if raw_test.lower() == 'zip' or raw_test.lower() == 'zip_code':
raw_test = 'postal_code'
if raw_test.lower() == 'gba':
raw_test = 'gross_floor_area'
if raw_test.lower() == 'building_address':
raw_test = 'address_line_1'
matches = matchers.best_match(raw_test, dest_columns, top_n=5)
# go get the top 5 matches. format will be [('PropertyState', 'building_count', 62), ...]
self.add_mappings(raw, matches)
# convert this to an exception and catch it some day...
index = 0
while self.duplicates and index < 10:
index += 1
_log.debug("Index: {} with duplicates: {}".format(index, self.duplicates))
for k, v in self.duplicates.iteritems():
self.resolve_duplicate(k, v)
if threshold > 0:
self.apply_threshold(threshold)
def __init__(self, raw_columns, dest_columns, previous_mapping=None, map_args=None,
default_mappings=None,
threshold=0):
"""
:param raw_columns: list of str. The column names we're trying to map.
:param dest_columns: list of str. The columns we're mapping to.
:param previous_mapping: Method that contains previous mapped columns
.. code:
The expectation is that our callable always gets passed a raw key. If
it finds a match, it returns the raw_column and score.
previous_mapping('example field', *map_args) ->
('field_1', 0.93)
:param map_args: Arguments to pass into the previous_mapping method (e.g. Organization ID)
:param default_mappings: dict of mappings. Use these mappings if the column is not found in the previous mapping call
:param threshold: int, Minimum value of the matching confidence to allow for matching.
:return dict: {'raw_column': ('dest_column', score), 'raw_column_2': ('dest_column_2',...)}
"""
self.data = {}
for raw in raw_columns:
attempt_best_match = False
# We want previous mappings to be at the top of the list.
if previous_mapping and callable(previous_mapping):
args = map_args or []
# Mapping will look something like this -- ['table', 'field', 100]
mapping = previous_mapping(raw, *args)
if mapping:
self.add_mappings(raw, [mapping], True)
elif default_mappings and raw in default_mappings:
self.add_mappings(raw, [default_mappings[raw]], True)
else:
attempt_best_match = True
else:
attempt_best_match = True
# Only enter this flow if we haven't already selected a result. Ignore blank columns
# with conf of 100 since a conf of 100 signifies the user has saved that mapping.
if attempt_best_match:
# convert raw fields spaces into underscores because that is what is in the database
raw_test = raw.replace(' ', '_')
# try some alternatives to the raw column in specific cases
# (e.g. zip => postal code). Hack for now, but should make this some global
# config or organization specific config
if raw_test.lower() == 'zip' or raw_test.lower() == 'zip_code':
raw_test = 'postal_code'
if raw_test.lower() == 'gba':
raw_test = 'gross_floor_area'
if raw_test.lower() == 'building_address':
raw_test = 'address_line_1'
if raw_test.lower() == 'ubi':
raw_test = 'jurisdiction_tax_lot_id'
matches = matchers.best_match(raw_test, dest_columns, top_n=5)
# go get the top 5 matches. format will be [('PropertyState', 'building_count', 62), ...]
self.add_mappings(raw, matches)
# convert this to an exception and catch it some day.
index = 0
while self.duplicates and index < 10:
index += 1
_log.debug("Index: %s with duplicates: %s" % (index, self.duplicates))
for k, v in self.duplicates.items():
self.resolve_duplicate(k, v)
if threshold > 0:
self.apply_threshold(threshold)
