python爬虫及结巴分词《攀登者》影评分析

《攀登者》影评爬取及分析

0、项目结构

其中simkai.ttf为字体文件，Windows查看系统自带的字体

C:\Windows\Fonts

一、爬取豆瓣影评数据

# -*- coding: utf-8 -*-
"""爬取豆瓣影评"""
import requests
from lxml import etree
import time

url = "https://movie.douban.com/subject/30413052/comments?start=%d&limit=20&sort=new_score&status=P"

#请求头
headers = {'Host': 'movie.douban.com',
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:70.0) Gecko/20100101 Firefox/70.0',
'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
'Accept-Language': 'zh-CN,zh;q=0.8,zh-TW;q=0.7,zh-HK;q=0.5,en-US;q=0.3,en;q=0.2',
#'Accept-Encoding': 'gzip, deflate, br',
'Connection': 'keep-alive',
'Cookie': 'bid=TXwfIvNFTRE; douban-fav-remind=1; __gads=ID=e042951d078c30b3:T=1570518321:S=ALNI_Mbp-ZmoryuBFEnTQy24mwdf0B89ig; __utma=30149280.1448315194.1570518324.1570518324.1572927825.2; __utmz=30149280.1570518324.1.1.utmcsr=baidu|utmccn=(organic)|utmcmd=organic; _pk_id.100001.4cf6=589509e524ead00f.1572927824.1.1572927824.1572927824.; _pk_ses.100001.4cf6=*; __utmb=30149280.0.10.1572927825; __utmc=30149280; __utma=223695111.1094105223.1572927825.1572927825.1572927825.1; __utmb=223695111.0.10.1572927825; __utmc=223695111; __utmz=223695111.1572927825.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none); ap_v=0,6.0',
'Upgrade-Insecure-Requests': '1',
'Cache-Control': 'max-age=0'}

if __name__ == '__main__':

    f = open("./climb.csv", mode="w", encoding='utf-8')
    f.write("author\tcomment\tvotes\n")

    #start:0,20,40,...,200
    for i in range(11):#range左闭右开
        #1拼接url，只能获取前11页数据
        if i == 10:#最后一页
            url_climb = url%(200)
        else:
            url_climb = url%(i*20)

        #2发起请求，设置编码，获取文本内容
        response = requests.get(url_climb, headers = headers)
        response.encoding = "utf-8"
        text = response.text

        #存储
        #with open("./climb.html", mode="w", encoding="utf-8") as f:
        #    f.write(text)

        #使用etree解析
        html = etree.HTML(text)
        comments = html.xpath('//div[@id="comments"]/div[@class="comment-item"]')
        for comment in comments:
            #获取评论人
            author = comment.xpath('./div[@class="avatar"]/a/@title')[0].strip()
            #获取评论内容
            p = comment.xpath('.//span[@class="short"]/text()')[0].strip()

            #获取这条评论对应的点赞数
            vote = comment.xpath('.//span[@class="votes"]/text()')[0].strip()

            #print(author, p, vote)
            f.write("%s\t%s\t%s\n" % (author,p,vote))

        #打印提示信息，并休眠一秒，反爬虫
        print("第%d页的数据保存成功" % (i+1))
        time.sleep(1)

    f.close()

二、对评论信息进行情感分析

# -*- coding: utf-8 -*-
"""
pandas:python data analysis lib,返回值为DataFrame（行，列），行是样本，列为属性
"""
import pandas as pd
from snownlp import SnowNLP

# 显示所有列
pd.set_option('display.max_columns', None)

def convert(comment):
    """将传入的评论进行情感分析"""
    snow = SnowNLP(str(comment))
    sentiments = snow.sentiments#0(消极评论)-1（积极评论）
    return sentiments

if __name__ =='__main__':

    data = pd.read_csv('./climb.csv', '\t')
    #print(data.head(), "\n", data.shape)

    #获取评论数据，进行情感分析，DataFrame就会新增加一列名为‘情感评分’的数据
    data['情感评分'] = data.comment.apply(convert)
    data.sort_values(by='情感评分', ascending=False, inplace=True)

    #保存数据
    data.to_csv('./climb_snownlp.csv', sep='\t', index=False, encoding='utf-8')

    print(data[:5])
    print(data[-5:])

三、对评论数据进行jieba分词，生成关键词条形图和词云

# -*- coding: utf-8 -*-

import pandas as pd
import jieba
from jieba import analyse
import matplotlib.pyplot as plt
import numpy as np
import wordcloud
from PIL import Image

if __name__ == '__main__':
    data = pd.read_csv('./climb.csv', sep='\t')

    #列表生成式，获取所有评论信息
    comments = ';'.join([str(c) for c in data['comment'].tolist()])
    #print(comments)

    #使用jieba库对文本进行分词，返回的是生成器
    gen_ret = jieba.cut(comments)
    seg_words = '/'.join(gen_ret)
    #print(seg_words)

    #对分好的词进行分析,topK返回的关键词个数，withWeight带着权重
    tags_ret = analyse.extract_tags(seg_words, topK=500, withWeight=True)
    #print(tags_ret)
    #将数据转换成DataFrame
    df_ret = pd.DataFrame(tags_ret, columns=['词语', '重要性'])
    df_ret.sort_values(by='重要性', ascending=False, inplace=True)#根据重要性降序排列
    #print(df_ret)

    #可视化，500个词语，选取前20个分析
    plt.barh(y=np.arange(0,20), width=df_ret[:20]['重要性'][::-1])
    plt.ylabel('Importance')
    plt.yticks(np.arange(0,20), labels=df_ret[:20]['词语'][::-1], fontproperties='KaiTi')
    #保存条形图!!!保存代码一定要写在show之前,dpi表示屏幕像素密度
    plt.savefig('./条形图_20个keyword.jpg', dpi=200)
    plt.show()

    #词云操作
    bg = np.array(Image.open('./bg.jpg'))#词云的图片
    words = dict(tags_ret)#将标签转为词典
    cloud = wordcloud.WordCloud(width=1200, height=968,
                        font_path='./simkai.ttf',#字体路径
                        background_color='white', mask=bg,
                        max_words=500, max_font_size=150)
    #生成词云图片
    word_cloud = cloud.generate_from_frequencies(words)
    plt.figure(figsize=(12,12))
    plt.imshow(word_cloud)
    #词云保存
    plt.savefig('./攀登者词云.jpg', dpi=200)
    plt.show()