def clean(self, dictionaries):
"""Cleaning process. Categorizes words as pass/fail/indeterminate.
Returns:
list: 3 list containg allowed, not alowed and indeterminate words
"""
#print self.raw_data, "before we remove"
temp = self.raw_data.replace(",", " ") # remove commas temporariarly
dates, non_dates = ptchk.check_for_dates(temp)
allowed, not_allowed, indeterminate = \
ptchk.check_for_words(non_dates,
dictionaries)
not_allowed = not_allowed + dates
return allowed, not_allowed, indeterminate
def test_pattern(self):
#test to see if pattern check finds obvious names
dictionary = ptchk.Dictionary()
allowed, not_allowed, indeterminate = \
ptchk.check_for_words("Aristotle Lewis Rajah Seth",
dictionary.export_dicts())
#apparently no one names their children aristole or rajah in america
expected_allowed = ['Aristotle', 'Rajah']
expected_not_allowed = ['Seth']
expected_indeterminate = ['Lewis']
#True only if output matches expected output
pattern_pass = allowed == expected_allowed and \
not_allowed == expected_not_allowed and \
indeterminate == expected_indeterminate
if not pattern_pass:
print "expected allowed:\n %s\n got\n %s" \
% (expected_allowed, allowed)
print "expected not allowed:\n %s\n got\n %s" \
% (expected_not_allowed, not_allowed)
print "expeted indeterminate:\n %s\n got\n %s" \
% (expected_indeterminate, indeterminate)
self.failUnless(pattern_pass, "Unexpected pattern match")
