main.py

# -*- coding: utf-8 -*-
# filename: main.py
import re

import jieba
import numpy as np
from flask import Flask, jsonify, request
from keras.layers import Embedding, Dense, Bidirectional, Conv1D, GRU, BatchNormalization, Activation, Dropout,MaxPool1D
from keras.models import Sequential
from keras.preprocessing.sequence import pad_sequences
from sklearn.externals import joblib

from Attention import Attention
import tensorflow as tf

global graph,models
graph = tf.get_default_graph()

class Region(object):
    def __init__(self):
        self.tokenizer = joblib.load('tokenizer_final.model')
        self.model = self.cnn_rnn_attention()
        self.restr = r'[0-9\s+\.\!\/_,$%^*();?:\-<>《》【】+\"\']+|[+——！，；。？：、~@#￥%……&*（）]+'

    def cnn_rnn_attention(self):
        model = Sequential([
            Embedding(20000 + 1, 64, input_shape=(80,)),
            Conv1D(64, 3, padding='same'),
            BatchNormalization(),
            Activation('relu'),
            Bidirectional(GRU(128, return_sequences=True, reset_after=True), merge_mode='sum'),
            Attention(64),
            Dropout(0.5),
            Dense(2, activation='softmax')
        ])
        model.load_weights('model.h5')
        return model

    def prdected(self, text):
        resu = text.replace('|', '').replace('&nbsp;', '').replace('ldquo', '').replace('rdquo',
                                                                                        '').replace(
            'lsquo', '').replace('rsquo', '').replace('“', '').replace('”', '').replace('〔', '').replace('〕', '')
        resu = re.split(r'\s+', resu)
        dr = re.compile(r'<[^>]+>', re.S)
        dd = dr.sub('', ''.join(resu))
        line = re.sub(self.restr, '', dd)
        seg_list = jieba.lcut(line)
        sequences = self.tokenizer.texts_to_sequences([seg_list])
        start = 0
        pred = []
        for i in range(int(len(sequences[0]) / 80) + 1):
            data = [sequences[0][start:start + 80]]
            data = pad_sequences(data, maxlen=80)
            pred.append(self.model.predict(data).tolist()[0][1])
            start += 80
        return 1 - np.mean(pred)


models = Region()
server = Flask(__name__)


@server.route('/pos_neg', methods=['post'])
def reg():
    content = request.values.get('content')
    if (content is not None) and (content != ""):
        with graph.as_default():
            result = jsonify({"result": models.prdected(content),"status":"1"})
    else:
        result = {"result": "", "status": "0"}
    return result


server.run(host='0.0.0.0', port=5000, debug=True, use_reloader=False)