94 lines
3.3 KiB
Python
94 lines
3.3 KiB
Python
from src.config import config, redis, tokenizer, trigram_repository
|
|
from src.utils import strings_has_equal_letters, capitalize
|
|
|
|
|
|
class ReplyGenerator:
|
|
"""
|
|
Handles generation of responses for user message
|
|
"""
|
|
def __init__(self):
|
|
self.redis = redis
|
|
self.tokenizer = tokenizer
|
|
self.trigram_repository = trigram_repository
|
|
|
|
self.max_wrds = config.getint('grammar', 'max_wrds')
|
|
self.max_msgs = config.getint('grammar', 'max_msgs')
|
|
|
|
self.stop_word = config['grammar']['stop_word']
|
|
self.separator = config['grammar']['separator']
|
|
self.endsen = config['grammar']['endsen']
|
|
|
|
def generate(self, message):
|
|
"""
|
|
Generates response based on given message
|
|
|
|
:param message: Message
|
|
:return:
|
|
- response (a message)
|
|
- empty string (if response == message)
|
|
"""
|
|
|
|
words = self.tokenizer.extract_words(message)
|
|
|
|
# TODO explain this
|
|
pairs = [trigram[:-1] for trigram in self.tokenizer.split_to_trigrams(words)]
|
|
|
|
# TODO explain why it returns what it returns
|
|
messages = [self.__generate_best_message(chat_id=message.chat_id, pair=pair) for pair in pairs]
|
|
longest_message = max(messages, key=len) if len(messages) else None
|
|
|
|
if longest_message and strings_has_equal_letters(longest_message, ''.join(words)):
|
|
return None
|
|
|
|
return longest_message
|
|
|
|
def __generate_best_message(self, chat_id, pair):
|
|
# TODO explain this method
|
|
best_message = ''
|
|
for _ in range(self.max_msgs):
|
|
generated = self.__generate_sentence(chat_id=chat_id, pair=pair)
|
|
if len(generated) > len(best_message):
|
|
best_message = generated
|
|
# TODO explain the concept of the BEST message
|
|
return best_message
|
|
|
|
def __generate_sentence(self, chat_id, pair):
|
|
# TODO explain this method
|
|
gen_words = []
|
|
key = self.separator.join(pair)
|
|
|
|
# TODO explain this loop
|
|
for _ in range(self.max_wrds):
|
|
words = key.split(self.separator)
|
|
|
|
gen_words.append(words[1] if len(gen_words) == 0 else words[1])
|
|
|
|
next_word = self.trigram_repository.get_random_reply(chat_id, key)
|
|
if next_word is None:
|
|
break
|
|
|
|
key = self.separator.join(words[1:] + [next_word])
|
|
|
|
# TODO explain what's last word
|
|
# Append last word unless it is in the list already
|
|
last_word = key.split(self.separator)[-1]
|
|
if last_word not in gen_words:
|
|
gen_words.append(last_word)
|
|
|
|
# If all words are equal (if set(words) == words[0]), leave just 1 word
|
|
if len(set(gen_words)) == 1:
|
|
gen_words = list(set(gen_words))
|
|
|
|
gen_words = [w for w in gen_words if w != self.stop_word]
|
|
# TODO maybe move wordlist preparations to some function?
|
|
|
|
# TODO maybe move generating the sentence to a function?
|
|
sentence = ' '.join(gen_words).strip()
|
|
if sentence[-1:] not in self.endsen:
|
|
# TODO explain this pls:
|
|
sentence += self.tokenizer.random_end_sentence_token()
|
|
# sentence = capitalize(sentence)
|
|
# TODO my intuition tells me we shouldn't return fun(obj), but IDK really
|
|
|
|
return sentence
|