def build_dictionary(self, dataset_id):
texts = DbTextIterator(self.queryset)
tokenized_texts = self.tokenizer(texts, *self.filters)
dataset = Dataset.objects.get(pk=dataset_id)
return Dictionary._create_from_texts(tokenized_texts=tokenized_texts,
name=self.name,
minimum_frequency=self.minimum_frequency,
dataset=dataset,
settings=self.get_dict_settings())
def build_dictionary(self, dataset_id):
texts = DbTextIterator(self.queryset)
tokenized_texts = self.tokenizer(texts, self.lemmatizer, *self.filters)
dataset = Dataset.objects.get(pk=dataset_id)
return Dictionary._create_from_texts(
tokenized_texts=tokenized_texts,
name=self.name,
minimum_frequency=self.minimum_frequency,
dataset=dataset,
settings=self.get_dict_settings())
def build_dictionary(self):
texts = DbTextIterator(self.queryset, textfield=self.textfield)
tokenized_texts = self.tokenizer(texts, stoplist=self.stoplist)
return Dictionary._create_from_texts(
tokenized_texts=tokenized_texts,
name=self.name,
minimum_frequency=self.minimum_frequency,
dataset=self.queryset.model.__name__,
settings=self.get_dict_settings())