def _process_frame(df_source, config):
"""
Create a result frame
.. seealso::
:`_process_row_series`
:`generate`
Reminder:
- apply() works on a row / column basis of a DataFrame
- applymap() works element-wise on a DataFrame
- map() works element-wise on a Series
"""
df = pd.DataFrame()
# keep the patid from the source
df['PATID'] = df_source[COL_PATID]
for i, rule in enumerate(rulz):
rule_data = rulz.get(rule)
pattern = rule_data['pattern']
required_attr = rule_data['required_attr']
df[rule] = df_source.apply(
lambda x: HashGenerator._process_row_series(
x, rule, pattern, required_attr, config),
axis=1)
# print("Processed frame: \n{}".format(df))
return df
def _process_frame(cls, df_source, config):
"""
Create a result frame
Reminder:
- apply() works on a row / column basis of a DataFrame
- applymap() works element-wise on a DataFrame
- map() works element-wise on a Series
"""
df = pd.DataFrame()
# keep the patid from the source
df['PATID'] = df_source['patid']
for i, rule in enumerate(rulz):
cls.log.debug("Applying rule {}: {}".format(i, rule))
rule_data = rulz.get(rule)
pattern = rule_data['pattern']
required_attr = rule_data['required_attr']
df[rule] = df_source.apply(lambda x: cls._process_row_series(
x, rule, pattern, required_attr, config),
axis=1)
cls.log.debug("Processed frame: \n{}".format(df))
return df
def _validate_config(cls, config):
"""
Helper method for preventing config errors
"""
COLUMN_MAP = config.get('COLUMN_MAP', None)
if COLUMN_MAP is None:
raise ConfigErr(
"Please verify that the config specifies the `COLUMN_MAP` parameter!"
) # noqa
# Verify that every required column is mapped
CONFIGURED_COLS = list(COLUMN_MAP.keys())
diff = list(set(REQUIRED_COLS).difference(CONFIGURED_COLS))
cls.log.info('REQUIRED_COLS: {}'.format(REQUIRED_COLS))
cls.log.info('CONFIGURED_COLS: {}'.format(CONFIGURED_COLS))
if len(diff) > 0:
raise ConfigErr(
'Please verify that the config specifies every column in the `COLUMN_MAP` parameter! Missing: {}'
.format(diff)) # noqa
enabled_rules = config.get('ENABLED_RULES', None)
if not enabled_rules:
raise ConfigErr(
'Please verify that the config specifies the `ENABLED_RULES` parameter!'
) # noqa
for rule_code in enabled_rules:
if rule_code not in rulz:
raise ConfigErr(
'Invalid rule: [{}]! Available codes are: {}'.format(
rule_code, rulz.keys()))
def _validate_config(cls, config):
"""
Helper method for preventing config errors
"""
enabled_rules = config.get('ENABLED_RULES', None)
if not enabled_rules:
raise ConfigErr('Please verify that the config specifies'
' the `ENABLED_RULES` parameter!')
for rule_code in enabled_rules:
if rule_code not in rulz:
raise ConfigErr('Invalid rule code: [{}]!'
' Available codes are: {}'.format(
rule_code, rulz.keys()))