python 斗鱼爬虫的基础知识（采集快手用户所有公开作品）

• 代码功能如题：根据快手用户的id来爬取用户所有公开作品，包括图集和视频。
• 原理：其实就是利用基于chromium内核的浏览器自带的devtools对所有请求进行排查找出包含作品链接的请求，然后用代码模拟请求去获得数据，再根据url下载作品保存就行了，包括一些网站的自动注册登录、操作都可以模拟。这个其实应该算是写过爬虫的同学们都知道。

• 废话不多说，上核心代码

def __crawl_user(self, uid): if uid.isdigit(): uid = self.__switch_id(uid) payload = {"operationName": "privateFeedsQuery", "variables": {"principalId": uid, "pcursor": "", "count": 999}, "query": "query privateFeedsQuery($principalId: String, $pcursor: String, $count: Int) {\n privateFeeds(principalId: $principalId, pcursor: $pcursor, count: $count) {\n pcursor\n list {\n id\n thumbnailUrl\n poster\n workType\n type\n useVideoPlayer\n imgUrls\n imgSizes\n magicFace\n musicName\n caption\n location\n liked\n onlyFollowerCanComment\n relativeHeight\n timestamp\n width\n height\n counts {\n displayView\n displayLike\n displayComment\n __typename\n }\n user {\n id\n eid\n name\n avatar\n __typename\n }\n expTag\n __typename\n }\n __typename\n }\n}\n"} res = requests.post(self.__data_url, headers=self.__headers, json=payload) works = json.loads(res.content.decode(encoding='utf-8', errors='strict'))['data']['privateFeeds']['list'] if not os.path.exists("../data"): os.makedirs("../data") # 这两行代码将response写入json供分析 # with open("data/" uid ".json", "w") as fp: # fp.write(json.dumps(works, indent=2)) # 防止该用户在直播，第一个作品默认为直播，导致获取信息为NoneType if works[0]['id'] is None: works.pop(0) name = re.sub(r'[\\/:*?"<>|\r\n] ', "", works[0]['user']['name']) dir = "data/" name "(" uid ")/" # print(len(works)) if not os.path.exists(dir): os.makedirs(dir) # if not os.path.exists(dir ".list"): # print("") print("开始爬取用户 " name "，保存在目录 " dir) print(" 共有" str(len(works)) "个作品") for j in range(len(works)): self.__crawl_work(uid, dir, works[j], j 1) time.sleep(1) print("用户 " name "爬取完成!") print() time.sleep(1)

快手分为五种类型的作品，在作品里面表现为workType属性

• 其中两种图集: vertical和multiple，意味着拼接长图和多图，所有图片的链接在imgUrls里
• 一种单张图片: single 图片链接也在imgUrls里
• K歌: ksong 图片链接一样，不考虑爬取音频...
• 视频: video 需要解析html获得视频链接

def __crawl_work(self, uid, dir, work, wdx): w_type = work['workType'] w_caption = re.sub(r"\s ", " ", work['caption']) w_name = re.sub(r'[\/:*?"<>|\r\n] ', "", w_caption)[0:24] w_time = time.strftime('%Y-%m-%d', time.localtime(work['timestamp'] / 1000)) if w_type == 'vertical' or w_type == 'multiple' or w_type == "single" or w_type == 'ksong': w_urls = work['imgUrls'] l = len(w_urls) print(" " str(wdx) ")图集作品：" w_caption "，" "共有" str(l) "张图片") for i in range(l): p_name = w_time "_" w_name "_" str(i 1) ".jpg" pic = dir p_name if not os.path.exists(pic): r = requests.get(w_urls[i]) r.raise_for_status() with open(pic, "wb") as f: f.write(r.content) print(" " str(i 1) "/" str(l) " 图片 " p_name " 下载成功 √") else: print(" " str(i 1) "/" str(l) " 图片 " p_name " 已存在 √") elif w_type == 'video': w_url = self.__work_url work['id'] res = requests.get(w_url, headers=self.__headers_mobile, params={"fid": 1841409882, "cc": "share_copylink", "shareId": "143108986354"}) html = res.text waitreplace = work['id'] '".*?"srcNoMark":"(.*?)"' v_url = re.findall(waitreplace, html) # pattern = re.compile(r"playUrl", re.MULTILINE | re.DOTALL) # script = soup.find("script", text=pattern) # s = pattern.search(script.text).string # v_url = s.split('playUrl":"')[1].split('.mp4')[0].encode('utf-8').decode('unicode-escape') '.mp4' try: print(" " str(wdx) ")视频作品：" w_caption) except: print(" 这里似乎有点小错误，已跳过") v_name = w_time "_" w_name ".mp4" video = dir v_name if v_url: if not os.path.exists(video): r = requests.get(v_url[0]) r.raise_for_status() with open(video, "wb") as f: f.write(r.content) print(" 视频 " v_name " 下载成功 √") else: print(" 视频 " v_name " 已存在 √") else: print("未找到视频") else: print("错误的类型")

注意事项：

• 不考虑提供列表可选的批量下载功能
• 有需要的合理功能可以issue反馈，看到后会考虑是否修改
• 如果需要自定义自己的需求，可以拿走代码自行修改，喜欢的话给个star给个follow
• 本代码仅供学习使用，不可违反法律爬取视频，以及私自盗用搬运视频，后果自负

项目源码地址 https://github.com/oGsLP/kuaishou-crawler