第五关电影票房数据爬取

新建文本文档 (3).txt

@@ -0,0 +1,24 @@
+import requests
+
+from lxml import etree
+headers={
+    'Cookie': 'Hm_lvt_e71d0b417f75981e161a94970becbb1b=1623642627; DIDA642a4585eb3d6e32fdaa37b44468fb6c=cbpkv0aghe6jjfjcgnt6kta7g6; remember=0; time=MTEzNTI2LjIxNjM0Mi4xMDI4MTYuMTA3MTAwLjExMTM4NC4yMDc3NzQuMTE5OTUyLjExMTM4NC4xMDQ5NTguMTE1NjY4LjEwNzEwMC4xMDkyNDIuMTE1NjY4LjExMTM4NC4xMDkyNDIuMTAyODE2LjExNzgxMC4xMTM1MjYuMA%3D%3D; Hm_lpvt_e71d0b417f75981e161a94970becbb1b=1623643075',
+    'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.101 Safari/537.36 Edg/91.0.864.48'
+}
+for i in range(0,11):
+    url = 'http://58921.com/alltime?page={}'.format(i)
+    wood=requests.get(url=url,headers=headers)
+    wood.encoding=wood.apparent_encoding
+    # print(wood)
+
+    word=etree.HTML(wood.text)
+    all_list=word.xpath('//div[@class="table-responsive"]/table[@class="center_table movie_box_office_stats_table table table-bordered table-condensed"]//tr')
+    for li in all_list:
+        wa=li.xpath('./td[2]/text()')
+        title=li.xpath('./td[3]/a/text()')
+        year=li.xpath('./td[last()-1]/text()')
+        if wa !=[]:
+            text=wa[0]+'|'+title[0]+'|'+year[0]
+            print(text)
+            with open('wood.txt','a+',encoding='utf-8') as f:
+                f.write(text+'\n')