intent/models/dirty.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""
@Time    : 2024/9/13 14:29
@Author  : cao
@File    : dirty.py
@Desc    : 
"""
import logging
import utils
import jieba

logger = logging.getLogger(__name__)


def load_dirty(file):
    with open(file) as f:
        words = [line.strip().split(" ")[0] for line in f]
        return words

dirty = load_dirty(utils.DIRTY_DATA)

def dirty_detect(sentence):
    if sentence in dirty:
        return True, {"status":1, "word":sentence, "query": sentence, "意图类别":"脏话", "intent": "转人工","isFaq":False, "query": sentence}
    words = jieba.lcut(sentence.lower(), cut_all=True)
    for word in words:
        if word in dirty:
            return True, {"status":1,"word": word,"query": sentence, "意图类别":"脏话", "intent": "转人工","isFaq":False, "query": sentence}
    return False, ''




if __name__ == "__main__":
    print(dirty_detect("SB吧，怎么老停水"))