1 # 使用了线程库
2 import threading
3 # 队列
4 from queue import Queue
5 # 解析库
6 from lxml import etree
7 # 请求处理
8 import requests
9 # json处理
10 import time
11
12
13 class ThreadCrawl(threading.Thread):
14 def __init__(self, thread_name, page_queue, data_queue):
15 #threading.Thread.__init__(self)
16 # 调用父类初始化方法
17 super(ThreadCrawl, self).__init__()
18 # 线程名
19 self.thread_name = thread_name
20 # 页码队列
21 self.page_queue = page_queue
22 # 数据队列
23 self.data_queue = data_queue
24 # 请求报头
25 self.headers = {"User-Agent" : "Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Trident/5.0;"}
26
27 def run(self):
28 print("启动 " + self.thread_name)
29 while not CRAWL_EXIT:
30 try:
31 # 取出一个数字，先进先出
32 # 可选参数block，默认值为True
33 #1. 如果对列为空，block为True的话，不会结束，会进入阻塞状态，直到队列有新的数据
34 #2. 如果队列为空，block为False的话，就弹出一个Queue.empty()异常，
35 page = self.page_queue.get(False)
36 url = "http://www.qiushibaike.com/8hr/page/" + str(page) +"/"
37 #print url
38 content = requests.get(url, headers = self.headers).text
39 time.sleep(1)
40 self.data_queue.put(content)
41 #print len(content)
42 except:
43 pass
44 print("结束 " + self.thread_name)
45
46 class ThreadParse(threading.Thread):
47 def __init__(self, thread_name, data_queue):
48 super(ThreadParse, self).__init__()
49 # 线程名
50 self.thread_name = thread_name
51 # 数据队列
52 self.data_queue = data_queue
53
54 def run(self):
55 print("启动" + self.thread_name)
56 while not PARSE_EXIT:
57 try:
58 html = self.data_queue.get(False)
59 self.parse(html)
60 except:
61 pass
62 print("退出" + self.thread_name)
63
64 def parse(self, html):
65 selector = etree.HTML(html)
66
67 # 返回所有段子的节点位置,contant()模糊查询方法,第一个参数是要匹配的标签,第二个参数是这个标签的部分内容
68 # 每个节点包括一条完整的段子(用户名,段子内容,点赞,评论等)
69 node_list = selector.xpath('//div[contains(@id,"qiushi_tag_")]')
70
71 for node in node_list:
72 # 爬取所有用户名信息
73 # 取出标签里的内容,使用.text方法
74 user_name = node.xpath('./div[@class="author clearfix"]//h2')[0].text
75
76 # 爬取段子内容,匹配规则必须加点 不然还是会从整个页面开始匹配
77 # 注意:如果span标签中有br 在插件中没问题,在代码中会把br也弄进来
78 duanzi_info = node.xpath('.//div[@class="content"]/span')[0].text.strip()
79
80 # 爬取段子的点赞数
81 vote_num = node.xpath('.//span[@class="stats-vote"]/i')[0].text
82
83 # 爬取评论数
84 comment_num = node.xpath('.//span[@class="stats-comments"]//i')[0].text
85
86 # 爬取图片链接
87 # 属性src的值,所以不需要.text
88 img_url = node.xpath('.//div[@class="thumb"]//@src')
89 if len(img_url) > 0:
90 img_url = img_url[0]
91 else:
92 img_url = "无图片"
93
94 self.save_info(user_name, duanzi_info, vote_num, comment_num, img_url)
95
96 def save_info(self, user_name, duanzi_info, vote_num, comment_num, img_url):
97 """把每条段子的相关信息写进字典"""
98 item = {
99 "username": user_name,
100 "content": duanzi_info,
101 "zan": vote_num,
102 "comment": comment_num,
103 "image_url": img_url
104 }
105
106 print(item)
107
108
109
110 CRAWL_EXIT = False
111 PARSE_EXIT = False
112
113
114 def main():
115 # 页码的队列，表示20个页面
116 pageQueue = Queue(20)
117 # 放入1~10的数字，先进先出
118 for i in range(1, 21):
119 pageQueue.put(i)
120
121 # 采集结果(每页的HTML源码)的数据队列，参数为空表示不限制
122 dataQueue = Queue()
123
124 filename = open("duanzi.json", "a")
125 # 创建锁
126 lock = threading.Lock()
127
128 # 三个采集线程的名字
129 crawlList = ["采集线程1号", "采集线程2号", "采集线程3号"]
130 # 存储三个采集线程的列表集合
131 thread_crawl = []
132 for threadName in crawlList:
133 thread = ThreadCrawl(threadName, pageQueue, dataQueue)
134 thread.start()
135 thread_crawl.append(thread)
136
137
138 # 三个解析线程的名字
139 parseList = ["解析线程1号","解析线程2号","解析线程3号"]
140 # 存储三个解析线程
141 thread_parse = []
142 for threadName in parseList:
143 thread = ThreadParse(threadName, dataQueue)
144 thread.start()
145 thread_parse.append(thread)
146
147 # 等待pageQueue队列为空，也就是等待之前的操作执行完毕
148 while not pageQueue.empty():
149 pass
150
151 # 如果pageQueue为空，采集线程退出循环
152 global CRAWL_EXIT
153 CRAWL_EXIT = True
154
155 print("pageQueue为空")
156
157 for thread in thread_crawl:
158 thread.join()
159
160 while not dataQueue.empty():
161 pass
162
163 global PARSE_EXIT
164 PARSE_EXIT = True
165
166 for thread in thread_parse:
167 thread.join()
168
169
170
171 if __name__ == "__main__":
172 main()
爬取效果如下:
如果你和我有共同爱好,我们可以加个好友一起交流哈!
