def createSentenceStructureDict(self):
def isSentStruct(item):
(k, v) = item
# is_sent_struct = (df.SENT_STRUCT_START_SYMB in k)
is_sent_struct = ('s)/MX1' in k)
# is_sent_struct = (re.search(r'it is as if', k) is not None)
# is_sent_struct = (df.ENDING_WITH.search(k) is not None)
return is_sent_struct
temp_dict={}
temp_set = list(filter(isSentStruct, self.items()))
# temp_set = [(x, y) for (x, y) in self.items() if df.SENT_STRUCT_START_SYMB in x]
# temp_set = [(x, y) for (x, y) in self.items() if '${`' in x]
temp_list = sorted(temp_set, key=lambda x: len(x[0]), reverse=True)
for key, value in temp_set:
value = self.replaceTranRef(value)
key_pattern = cm.creatSentRecogniserPattern(key)
dict_sl_mm, dict_sl_word_list = cm.createSentRecogniserRecord(key)
dict_tl_mm, dict_tl_word_list = cm.createSentRecogniserRecord(value)
entry = {key_pattern: (key, dict_sl_word_list, dict_sl_mm, value, dict_tl_word_list, dict_tl_mm)}
temp_dict.update(entry)
temp_dict = OrderedDict(sorted(list(temp_dict.items()), key=lambda x: len(x[0]), reverse=True))
# print_list = [(x, y[0]) for (x, y) in temp_dict.items()]
# pp(print_list, width=200)
self.sentence_struct_dict = NoCaseDict(temp_dict)
def setupRecords(self):
fname = INP.currentframe().f_code.co_name
dict_tl_list = None
try:
if not self.dict_sl_rec:
self.dict_sl_rec, self.dict_sl_wordlist = cm.createSentRecogniserRecord(
self.dict_sl_txt)
if not self.recog_pattern:
self.recog_pattern = re.compile(cm.formPattern(
self.dict_sl_wordlist),
flags=re.I)
if not self.dict_tl_rec:
self.dict_tl_rec, dict_tl_list = cm.createSentRecogniserRecord(
self.dict_tl_txt)
if not self.dict_tl_wordlist:
self.dict_tl_wordlist = self.dict_tl_rec.getSubEntriesAsList()
sent_tl_list = CP.deepcopy(self.dict_tl_wordlist)
self.sent_tl_rec = CP.copy(self.dict_tl_rec)
self.sent_tl_rec.clear()
self.sent_tl_rec.update(sent_tl_list)
except Exception as e:
# df.LOG(f'{e}', error=True)
self.is_sent_struct = False
self.is_sent_struct = bool(self.recog_pattern)
self.setupSentSLRecord()
