爬虫代码

2 months ago · 181f0e3a3f
parent 6d3ff14738
commit 181f0e3a3f
1 changed files with 101 additions and 0 deletions
--- a/final.py
+++ b/final.py
@ -0,0 +1,101 @@
+import re
+import random
+import requests
+import datetime
+import sys
+import io
+import json
+import openpyxl
+from bs4 import BeautifulSoup
+import time
+import pandas
+import wordcloud
+import numpy
+import jieba
+import pylint
+information_list = []#用来保存爬取的弹幕
+#解决一下怎么爬出BV号，每个网站的BV号不同，根据不同的BV号我们可以对不同的视频进行操作
+bvnum_list = []#存BV号，有了BV号再循环导出弹幕
+cid_list = []#存cid号
+def findbv():
+    temp_list =[]
+    for i in range(1,10):
+        url = f'https://search.bilibili.com/video?keyword=2024%E5%B7%B4%E9%BB%8E%E5%A5%A5%E8%BF%90%E4%BC%9A&from_source=webtop_search&spm_id_from=333.1007&search_source=3&page={i}'
+        headers = {
+            "User-Agent" : "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36 Edg/128.0.0.0",
+            "Cookie" : "i-wanna-go-back=-1; rpdid=|(RYkm|Ykul0J'uY~YmYlYl~; DedeUserID=474817892; DedeUserID__ckMd5=2ad2b4cb2126ef63; buvid4=9AE18FFB-B9C4-8389-DF6A-72C0E6BCC34D43412-023022115-vdGT0WAv%2FuJ7%2FRzFm7Eg7A%3D%3D; buvid_fp_plain=undefined; LIVE_BUVID=AUTO4716769855903121; CURRENT_BLACKGAP=0; enable_web_push=DISABLE; buvid3=913910D1-D049-0F77-02E0-21B1093ACF0601725infoc; b_nut=1708668801; b_ut=5; _uuid=65739108E-2B75-9642-2E59-826F71099D7E203422infoc; header_theme_version=CLOSE; hit-dyn-v2=1; FEED_LIVE_VERSION=V_WATCHLATER_PIP_WINDOW3; SESSDATA=709ab107%2C1729957738%2C3372c%2A42CjB0lzRiXKvwOi-Op4_kAeXHcFSeDW2GtWNhiW33Y4bLFvKzOUP5fUvm7CyuyD9Y1OISVkVDUkozNXlxUGIwLVZiNTNrTE9RMHhyeW1ycEJBZWZoa2VJYjd0WnN6UmNKTFVFWDZYWUI4NmpuYXNIT0k0T1ozNjFQVzBRbEdCaDliWGtFRjdFTkd3IIEC; bili_jct=717efb0472a4b30b96a2e6a470bdec42; sid=6a6r71lf; is-2022-channel=1; CURRENT_FNVAL=4048; fingerprint=254d2c4cf86e9c4fe76c3c00ffd0de23; CURRENT_QUALITY=80; buvid_fp=e440b4a9c5e0b9af9ad8950245b201f7; bili_ticket=eyJhbGciOiJIUzI1NiIsImtpZCI6InMwMyIsInR5cCI6IkpXVCJ9.eyJleHAiOjE3MjY1Nzc5NjIsImlhdCI6MTcyNjMxODcwMiwicGx0IjotMX0.ruBBQK0YAJ7TyuLBgdfVijbpzBhfwGTglNFMMAmhv-c; bili_ticket_expires=1726577902; PVID=6; home_feed_column=5; browser_resolution=1707-837; b_lsid=610579158_191F9ED6DE3; bp_t_offset_474817892=977706466930589696"
+        } 
+        result = requests.get(url=url,headers= headers)
+        result.encoding = "utf-8" 
+        tempresult = result.text
+        tempnum = re.findall(r'bvid:"(.*?)"', tempresult)
+        temp_list.extend(tempnum)
+    return temp_list
+#接下来抓取cid
+def getcid(list):
+    temp_cid_list = []
+    try:
+        for i in range(300):
+            temp_url = f'https://api.bilibili.com/x/player/pagelist?bvid={list[i]}&jsonp=jsonp'
+            temp_headers = {
+                "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36 Edg/128.0.0.0",
+                'Cookie': "i-wanna-go-back=-1; rpdid=|(RYkm|Ykul0J'uY~YmYlYl~; DedeUserID=474817892; DedeUserID__ckMd5=2ad2b4cb2126ef63; buvid4=9AE18FFB-B9C4-8389-DF6A-72C0E6BCC34D43412-023022115-vdGT0WAv%2FuJ7%2FRzFm7Eg7A%3D%3D; buvid_fp_plain=undefined; LIVE_BUVID=AUTO4716769855903121; CURRENT_BLACKGAP=0; enable_web_push=DISABLE; buvid3=913910D1-D049-0F77-02E0-21B1093ACF0601725infoc; b_nut=1708668801; b_ut=5; _uuid=65739108E-2B75-9642-2E59-826F71099D7E203422infoc; header_theme_version=CLOSE; hit-dyn-v2=1; FEED_LIVE_VERSION=V_WATCHLATER_PIP_WINDOW3; SESSDATA=709ab107%2C1729957738%2C3372c%2A42CjB0lzRiXKvwOi-Op4_kAeXHcFSeDW2GtWNhiW33Y4bLFvKzOUP5fUvm7CyuyD9Y1OISVkVDUkozNXlxUGIwLVZiNTNrTE9RMHhyeW1ycEJBZWZoa2VJYjd0WnN6UmNKTFVFWDZYWUI4NmpuYXNIT0k0T1ozNjFQVzBRbEdCaDliWGtFRjdFTkd3IIEC; bili_jct=717efb0472a4b30b96a2e6a470bdec42; sid=6a6r71lf; is-2022-channel=1; CURRENT_FNVAL=4048; fingerprint=254d2c4cf86e9c4fe76c3c00ffd0de23; CURRENT_QUALITY=80; buvid_fp=e440b4a9c5e0b9af9ad8950245b201f7; bili_ticket=eyJhbGciOiJIUzI1NiIsImtpZCI6InMwMyIsInR5cCI6IkpXVCJ9.eyJleHAiOjE3MjY1Nzc5NjIsImlhdCI6MTcyNjMxODcwMiwicGx0IjotMX0.ruBBQK0YAJ7TyuLBgdfVijbpzBhfwGTglNFMMAmhv-c; bili_ticket_expires=1726577902; PVID=6; home_feed_column=5; browser_resolution=1707-837; b_lsid=610579158_191F9ED6DE3; bp_t_offset_474817892=977706466930589696"
+            }
+            respone = requests.get(url = temp_url,headers = temp_headers)
+            respone.encoding = 'utf-8'
+            if respone.status_code == 200:
+                temp_cid_list.extend(re.findall(r'"cid":(.*?),',respone.text))
+    except requests.RequestException as e:
+        print(f"An error occurred: {e}")
+        
+    return temp_cid_list
+
+def get_information(list):
+    sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf8')
+    temp_information_list = []
+    for i in range(300):
+        temp_url = f'https://api.bilibili.com/x/v1/dm/list.so?oid={list[i]}'
+        temp_headers = {
+            "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36 Edg/128.0.0.0",
+            'Cookie': "i-wanna-go-back=-1; rpdid=|(RYkm|Ykul0J'uY~YmYlYl~; DedeUserID=474817892; DedeUserID__ckMd5=2ad2b4cb2126ef63; buvid4=9AE18FFB-B9C4-8389-DF6A-72C0E6BCC34D43412-023022115-vdGT0WAv%2FuJ7%2FRzFm7Eg7A%3D%3D; buvid_fp_plain=undefined; LIVE_BUVID=AUTO4716769855903121; CURRENT_BLACKGAP=0; enable_web_push=DISABLE; buvid3=913910D1-D049-0F77-02E0-21B1093ACF0601725infoc; b_nut=1708668801; b_ut=5; _uuid=65739108E-2B75-9642-2E59-826F71099D7E203422infoc; header_theme_version=CLOSE; hit-dyn-v2=1; FEED_LIVE_VERSION=V_WATCHLATER_PIP_WINDOW3; SESSDATA=709ab107%2C1729957738%2C3372c%2A42CjB0lzRiXKvwOi-Op4_kAeXHcFSeDW2GtWNhiW33Y4bLFvKzOUP5fUvm7CyuyD9Y1OISVkVDUkozNXlxUGIwLVZiNTNrTE9RMHhyeW1ycEJBZWZoa2VJYjd0WnN6UmNKTFVFWDZYWUI4NmpuYXNIT0k0T1ozNjFQVzBRbEdCaDliWGtFRjdFTkd3IIEC; bili_jct=717efb0472a4b30b96a2e6a470bdec42; sid=6a6r71lf; is-2022-channel=1; CURRENT_FNVAL=4048; fingerprint=254d2c4cf86e9c4fe76c3c00ffd0de23; CURRENT_QUALITY=80; buvid_fp=e440b4a9c5e0b9af9ad8950245b201f7; bili_ticket=eyJhbGciOiJIUzI1NiIsImtpZCI6InMwMyIsInR5cCI6IkpXVCJ9.eyJleHAiOjE3MjY1Nzc5NjIsImlhdCI6MTcyNjMxODcwMiwicGx0IjotMX0.ruBBQK0YAJ7TyuLBgdfVijbpzBhfwGTglNFMMAmhv-c; bili_ticket_expires=1726577902; PVID=6; home_feed_column=5; browser_resolution=1707-837; b_lsid=610579158_191F9ED6DE3; bp_t_offset_474817892=977706466930589696"
+        }
+        try:
+            response = requests.get(url=temp_url, headers=temp_headers)
+            response.encoding = 'utf-8'
+            temp_list = re.findall('<d p=".*?">(.*?)</d>',response.text)
+            temp_information_list.extend(temp_list)
+        except requests.RequestException as e:
+            print(f"An error occurred: {e}")
+        time.sleep(0.5) 
+    return temp_information_list
+
+
+
+bvnum_list = findbv()
+cid_list = getcid(bvnum_list)
+information_list = get_information(cid_list)
+for i in range(len(information_list)):
+    with open("E:\python1\\requirement1.txt","a",encoding= "utf-8") as f: #/r要做出区分
+        f.write(information_list[i])
+        f.write("\n")
+f = open("E:\python1\\requirement1.txt",encoding= "utf-8")
+txt = f.read()
+string = ' '.join(jieba.lcut(txt))
+wc = wordcloud.WordCloud(
+    width = 500,
+    height = 500,
+    background_color= 'white',
+    scale = 10,
+    font_path = 'msyh.ttc',
+)
+wc.generate(string)
+wc.to_file("wordcloud.png")
+df = pandas.DataFrame(information_list)
+df.to_excel('spider1.xlsx',index = False)
+
+
+
+
+
+
+