def format_text(self, text, document, formatting): par_len = 150 applicator = Applicator() cleaner = Cleaner() formatter = Formatter() new_text = self.get_string(text) cleaner.create_sentc_list(new_text) sentc_list = cleaner.get_sentc_list() formatter.set_sentlist(sentc_list) if formatting == "list": formatter.frmt_textlist() elif formatting == "block": formatter.frmt_textblock(par_len=par_len) elif formatting == "string": formatter.frmt_textstring() else: print("format not supported") raise SystemExit format_text = formatter.get_text() applicator.apply_text(format_text, document=document)
def clean_text(self, text, document, sentmin, sentmax, dictionary, cycle): print("applicator selected") applicator = Applicator() cleaner = Cleaner() formatter = Formatter() new_text = self.get_string(text) cycle_list = [ "noalpha", "nodeclare", "excaps", "exletters", "firstperson", "secondperson", "dupwords", "duplicates", "trimsentlist", "checkspelling", "help (or h)" ] def return_cycles(): s = ", " s = s.join(cycle_list) return s cleaner.create_sentc_list(new_text) if cycle == "full": cleaner.remv_noalpha() cleaner.remv_nodeclare() cleaner.remv_excaps() cleaner.remv_exletters() cleaner.remv_firstperson() cleaner.remv_secondperson() cleaner.remv_dupwords() cleaner.remv_duplicates() cleaner.trim_sentlist(sentmin, sentmax) cleaner.check_misspelled(dictionary) elif cycle == "noalpha": cleaner.remv_noalpha() elif cycle == "nodeclare": cleaner.remv_nodeclare() elif cycle == "excaps": cleaner.remv_excaps() elif cycle == "exletters": cleaner.remv_exletters() elif cycle == "firstperson": cleaner.remv_firstperson() elif cycle == "secondperson": cleaner.remv_secondperson() elif cycle == "dupwords": cleaner.remv_dupwords() elif cycle == "duplicates": cleaner.remv_duplicates() elif cycle == "trimsentlist": cleaner.trim_sentlist(sentmin, sentmax) elif cycle == "checkspelling": cleaner.check_misspelled(dictionary) else: print("cycle not found") print("available cycles:") print(return_cycles()) raise SystemExit sentc_list = cleaner.get_sentc_list() formatter.set_sentlist(sentc_list) formatter.frmt_textlist() clean_text = formatter.get_text() applicator.apply_text(text=clean_text, document=document)