python爬虫,爬取豆瓣电影《芳华》电影短评,分词生成云图。

782次阅读
没有评论

项目github地址:https://github.com/kocor01/spider_cloub/

Python版本为3.6

最近突然想玩玩云图,动手写了个简单的爬虫,搭建了简单的爬虫架构

爬虫爬取最近比较火的电影《芳华》分词后生成云图

使用了 jieba分词,云图用wordcloud生成python爬虫,爬取豆瓣电影《芳华》电影短评,分词生成云图。

用了朋友的2B姿势的自拍照片简单的P了下(为了不暴露,P成全黑的),作为生成云图的底图模板

云图底图模板:

python爬虫,爬取豆瓣电影《芳华》电影短评,分词生成云图。

生成的云图效果:

python爬虫,爬取豆瓣电影《芳华》电影短评,分词生成云图。

爬虫基础框架

python爬虫,爬取豆瓣电影《芳华》电影短评,分词生成云图。

spider_main.py  爬虫入口
url_manager.py  URL管理器
html_downloader.py  网页下载器
html_parser.py  数据提取器
html_outputer.py  数据处理器
word_cloud.py  云图生成器

extra_dict文件夹文件如下:

python爬虫,爬取豆瓣电影《芳华》电影短评,分词生成云图。

li.png   云图底图模板

simhei.ttf   生成云图的字体文件

str.txt   爬取的电影短评

stop_words.txt  分词排除的词

cut_str.txt   jieba分词后文件

yun.png  最后生成的云图

代码如下:

spider_main.py  爬虫入口

#coding:utf-8

import url_manager,html_parser,html_outputer,html_downloader,word_cloud

class SpiderMain(object):

def __init__(self): # URL管理器 self.urls = url_manager.UrlManager() # 网页下载器 self.downloader = html_downloader.HtmlDownloader() # 数据提取器 self.parser = html_parser.HtmlParser() # 数据处理器 self.outputer = html_outputer.HtmlOutputer() # 云图生成器 self.cloud = word_cloud.Wordcloud()

def craw(self, root_url): count =1 # 爬虫入口URL self.urls.add_new_url(root_url) # 待爬取URL wait_url = self.urls.has_new_url()

if wait_url is not None: while wait_url: try: # 获取一个待爬取URL new_url = self.urls.get_new_url() print("carw %d : %s" % (count, new_url)) # 爬取页面 html_cont = self.downloader.download(new_url) # 数据提取 new_url, new_datas = self.parser.parser(new_url, html_cont) # 添加新待爬取URL self.urls.add_new_url(new_url) # 数据加工处理 self.outputer.collect_data(new_datas) # 爬虫循环控制 if count == 10: break

count = count + 1 except: print("craw failed")

# 数据加工输出 self.outputer.process_data() #print("finish")

# 分词 self.outputer.cut_str()

# 生成云图 self.cloud.make() print("finish")

if __name__ == "__main__": # 爬虫入口URL root_url = "https://movie.douban.com/subject/26862829/comments?status=P" obj_spider = SpiderMain() # 启动爬虫 obj_spider.craw(root_url)

url_manager.py   URL管理器

#coding:utf-8

class UrlManager(object):

def __init__(self): self.new_urls = set() self.old_urls = set()

def add_new_url(self, url): if url is None: return if url not in self.new_urls and url not in self.old_urls: self.new_urls.add(url)

def add_new_urls(self, urls): if urls is None or len(urls) == 0: return for url in urls: self.add_new_url(url)

def has_new_url(self): return len(self.new_urls) != 0

def get_new_url(self): new_url = self.new_urls.pop() self.old_urls.add(new_url) return new_url

html_downloader.py   网页下载器

#coding:utf-8

import urllib.request

class HtmlDownloader(object):

def download(self, url): if url is None: return None

request = urllib.request.Request(url) request.add_header("user-agent", "Mozilla/5.0") response = urllib.request.urlopen(url)

if response.getcode() != 200: return None

return response.read()

html_parser.py   数据提取器

#coding:utf-8

import http.cookiejar from bs4 import BeautifulSoup import re import urllib.parse

class HtmlParser(object):

def parser(self, page_url, content):

if page_url is None or content is None: return

soup = BeautifulSoup(content, "html.parser", from_encoding='utf-8') new_url = self._get_new_url(page_url, soup) new_datas = self._get_new_datas(page_url, soup) return new_url, new_datas

def _get_new_url(self, page_url, soup):

new_url = soup.find('div', id="paginator").find('a', class_="next").get('href') new_full_url = urllib.parse.urljoin(page_url, new_url) return new_full_url

def _get_new_datas(self, page_url, soup): res_datas = set() contents = soup.find_all('div', class_="comment-item") for content in contents: res_datas.add(content.find('div', class_="comment").find('p').get_text())

return res_datas

html_outputer.py    数据处理器

#coding:utf-8

import pymysql import jieba.analyse

class HtmlOutputer(object):

def __init__(self): self.datas = []

def collect_data(self, data): res_datas = set()

if data is None: return for d in data: self.datas.append(d)

def process_data(self): #print(len(self.datas)) file_object = open('./extra_dict/str.txt', 'w',encoding='utf-8',errors='ignore') data_str = '' for data in self.datas: #data_str += data file_object.write(data)

#print(data_str) file_object.close()

def cut_str(self): content = open('./extra_dict/str.txt',encoding='utf-8',errors='ignore').read() jieba.analyse.set_stop_words("./extra_dict/stop_words.txt") tags = jieba.analyse.extract_tags(content, topK=1000,withWeight=True) file_object = open('./extra_dict/cut_str.txt', 'w') for v, n in tags: #权重是小数,为了凑整,乘了一万 #print(v + '\t' + str(int(n * 10000))) data_str = v + '\t' + str(int(n * 10000)) + '\n' file_object.write(data_str) file_object.close()

word_cloud.py    云图生成器

from os import path from PIL import Image import numpy as np import matplotlib.pyplot as plt

from wordcloud import WordCloud, STOPWORDS, ImageColorGenerator

class Wordcloud(object):

def make(self): d = path.dirname(__file__)

# Read the whole text. text = open(path.join(d, './extra_dict/cut_str.txt')).read()

# read the mask / color image taken from alice_coloring = np.array(Image.open(path.join(d, "./extra_dict/li.png"))) stopwords = set(STOPWORDS) stopwords.add("said")

wc = WordCloud(font_path="./extra_dict/simhei.ttf",background_color="white", max_words=2000, mask=alice_coloring, stopwords=stopwords, max_font_size=40, random_state=42) # generate word cloud wc.generate(text)

# create coloring from image image_colors = ImageColorGenerator(alice_coloring)

# show plt.imshow(wc, interpolation="bilinear") plt.axis("off") plt.figure() # recolor wordcloud and show # we could also give color_func=image_colors directly in the constructor plt.imshow(wc.recolor(color_func=image_colors), interpolation="bilinear") plt.axis("off") plt.figure() plt.imshow(alice_coloring, cmap=plt.cm.gray, interpolation="bilinear") plt.axis("off") wc.to_file(path.join(d, "./extra_dict/yun.png")) plt.show()

神龙|纯净稳定代理IP免费测试>>>>>>>>天启|企业级代理IP免费测试>>>>>>>>IPIPGO|全球住宅代理IP免费测试

相关文章:

版权声明:Python教程2022-10-25发表,共计5268字。
新手QQ群:570568346,欢迎进群讨论 Python51学习