def build_detectors(
cls, detector: DocumentFieldMultilineRegexDetector,
log_function: Callable) -> List[CsvRegexpsDetectionRow]:
try:
df = detector.get_as_pandas_df()
except Exception as e:
cls.log_error(
log_function,
f'CSV data is corrupted for field {detector.field_id}', e)
return []
if df.shape[0] == 0:
return []
if df.shape[1] != 2:
cls.log_error(
log_function, 'CSV data is has wrong number ' +
f'of columns ({df.shape[1]}) for field {detector.field_id}')
return []
detectors = [] # type: List[CsvRegexpsDetectionRow]
for i, row in df.iterrows():
try:
detected_value = row[0]
detector_reg_raw = row[1]
if detector.regexps_pre_process_lower:
reg_pattern = re.compile(detector_reg_raw, re.IGNORECASE)
else:
reg_pattern = re.compile(detector_reg_raw)
detectors.append(
CsvRegexpsDetectionRow(reg_pattern, detected_value,
detector))
except Exception as e:
cls.log_error(
log_function, 'CSV data is corrupted for field ' +
f'{detector.field_id} at line #{i}', e)
return detectors
def test_combine_dfs(self):
# first row is the same, second row - same value, another pattern
# third row - same pattern, another value, last row - brand new
another_text = """
,value,pattern
0,"Big Bank & Company (004578) (Knight, Bobby (Charlotte); Bryant, Koby (Charlotte); Williams, Gary (Charlotte); Johnson, Magic (Charlotte); Lobo, Rebecca (Charlotte))","\bbig\s{1,5}bank\s{1,5}.{1,5}\s{1,5}company\s{1,5}(004578)\b"
1,"Family Name Limited (173437) (Tanner, Rebecca (Houston); Saget, Bob (Houston))","\bfamily\s{1,5}guy(173437)\b"
2,"Eye-Eyes Communications (018951)","\ball\s{1,5}eyes\s{1,5}communications\s{1,5}(018951)\b"
3,"John Smith Archives, LLC d/b/a Charlie (085292) (Flay, Bobby (New York))","\bcharlie\s{1,5}(085292)\b"
"""
with StringIO(another_text) as cs_stream:
df = pd.read_csv(cs_stream, usecols=[1, 2])
detector = DocumentFieldMultilineRegexDetector()
detector.csv_content = self.csv_text
detector.update_checksum()
detector.combine_with_dataframe(df)
df_new = detector.get_as_pandas_df()
row_val = [] # type: List[Tuple[str, str]]
for i, row in df_new.iterrows():
row_val.append((
row[0],
row[1],
))
self.assertEqual(8, len(row_val))
self.assertTrue((
'John Smith Archives, LLC d/b/a Charlie (085292) (Flay, Bobby (New York))',
'\bcharlie\s{1,5}(085292)\b',
) in row_val)
self.assertTrue((
'Big Bank & Company (004578) (Knight, Bobby (Charlotte); Bryant, Koby '
+
'(Charlotte); Williams, Gary (Charlotte); Johnson, Magic (Charlotte); '
+ 'Lobo, Rebecca (Charlotte))',
'\bbig\s{1,5}bank\s{1,5}.{1,5}\s{1,5}company\s{1,5}(004578)\b',
) in row_val)
self.assertTrue((
'Family Name Limited (173437) (Tanner, Rebecca (Houston); Saget, Bob (Houston))',
'\bfamily\s{1,5}guy(173437)\b',
) in row_val)
self.assertTrue((
'Family Name Limited (173437) (Tanner, Rebecca (Houston); Saget, Bob (Houston))',
'\bfamily\s{1,5}name(173437)\b',
) in row_val)
self.assertTrue((
'Eye-Eyes Communications (018951)',
'\ball\s{1,5}eyes\s{1,5}communications\s{1,5}(018951)\b',
) in row_val)
# this one is replaced
self.assertFalse((
'All Eyes Communications (018951) (Moore, Michael (New York); Tarantino, Quentin '
+
'(San Francisco); Lee, Spike (New York); Levinson, Barry (Charlotte))',
'\ball\s{1,5}eyes\s{1,5}communications\s{1,5}(018951)\b',
) in row_val)
def test_get_as_pd(self):
detector = DocumentFieldMultilineRegexDetector()
detector.csv_content = self.csv_text
df = detector.get_as_pandas_df()
self.assertIsNotNone(df)
self.assertEqual((
6,
2,
), df.shape)