feat(tiktok): 增加合集下载功能

增加单个合集批量下载功能, 增加主页下所有合集批量下载功能

TikTok.py

@@ -34,7 +34,7 @@ class TikTok(object):
         self.headers = {
         'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/109.0.0.0 Safari/537.36',
         'referer': 'https://www.douyin.com/',
-        'Cookie': '__ac_signature=_02B4Z6wo00f01CEKaogAAIDBqkHxaCCYIyghKm4AAGu9c3; s_v_web_id=verify_ledo1j1t_0NwhDQFJ_nLca_42o5_8tAA_T8CWm5E2M6LF; msToken=%s;odin_tt=324fb4ea4a89c0c05827e18a1ed9cf9bf8a17f7705fcc793fec935b637867e2a5a9b8168c885554d029919117a18ba69;' % self.utils.generate_random_str(107)
+        'Cookie': 'ttwid=1|sGp2L-Krm46cXHcK7BsKghavVeVQIIOYtQInA1LV0-w|1676899557|3e483426230c481bd34f4d6529d6252372c154b75be7d4a2baec8edbfd0a742c; __ac_signature=_02B4Z6wo00f01CEKaogAAIDBqkHxaCCYIyghKm4AAGu9c3; s_v_web_id=verify_ledo1j1t_0NwhDQFJ_nLca_42o5_8tAA_T8CWm5E2M6LF; msToken=%s;odin_tt=324fb4ea4a89c0c05827e18a1ed9cf9bf8a17f7705fcc793fec935b637867e2a5a9b8168c885554d029919117a18ba69;' % self.utils.generate_random_str(107)
         }
 
 
@@ -60,6 +60,8 @@ class TikTok(object):
         # https://www.iesdouyin.com/share/video/7037827546599263488/?region=CN&mid=6939809470193126152&u_code=j8a5173b&did=MS4wLjABAAAA1DICF9-A9M_CiGqAJZdsnig5TInVeIyPdc2QQdGrq58xUgD2w6BqCHovtqdIDs2i&iid=MS4wLjABAAAAomGWi4n2T0H9Ab9x96cUZoJXaILk4qXOJlJMZFiK6b_aJbuHkjN_f0mBzfy91DX1&with_sec_did=1&titleType=title&schema_type=37&from_ssr=1&utm_source=copy&utm_campaign=client_share&utm_medium=android&app=aweme
         # 用户 第一步解析出来的链接是share/user/{sec_uid}
         # https://www.iesdouyin.com/share/user/MS4wLjABAAAA06y3Ctu8QmuefqvUSU7vr0c_ZQnCqB0eaglgkelLTek?did=MS4wLjABAAAA1DICF9-A9M_CiGqAJZdsnig5TInVeIyPdc2QQdGrq58xUgD2w6BqCHovtqdIDs2i&iid=MS4wLjABAAAAomGWi4n2T0H9Ab9x96cUZoJXaILk4qXOJlJMZFiK6b_aJbuHkjN_f0mBzfy91DX1&with_sec_did=1&sec_uid=MS4wLjABAAAA06y3Ctu8QmuefqvUSU7vr0c_ZQnCqB0eaglgkelLTek&from_ssr=1&u_code=j8a5173b&timestamp=1674540164&ecom_share_track_params=%7B%22is_ec_shopping%22%3A%221%22%2C%22secuid%22%3A%22MS4wLjABAAAA-jD2lukp--I21BF8VQsmYUqJDbj3FmU-kGQTHl2y1Cw%22%2C%22enter_from%22%3A%22others_homepage%22%2C%22share_previous_page%22%3A%22others_homepage%22%7D&utm_source=copy&utm_campaign=client_share&utm_medium=android&app=aweme
+        # 合集
+        # https://www.douyin.com/collection/7093490319085307918
         urlstr = str(r.request.path_url)
 
         if "/share/user/" in urlstr:
@@ -75,6 +77,10 @@ class TikTok(object):
             # 获取作品 aweme_id
             key = re.findall('video/(\d+)?', urlstr)[0]
             key_type = "aweme"
+        elif "/collection/" in urlstr:
+            # 获取作品 aweme_id
+            key = re.findall('collection/(\d+)?', urlstr)[0]
+            key_type = "mix"
         elif "live.douyin.com" in r.url:
             key = r.url.replace('https://live.douyin.com/', '')
             key_type = "live"
@@ -82,13 +88,13 @@ class TikTok(object):
         if key is None or key_type is None:
             print('[  错误  ]:输入链接有误！无法获取 id\r')
             return key_type, key
-        print('[  提示  ]:作品或者用户的 id = %s\r' % key)
 
         return key_type, key
 
     # 传入 aweme_id
     # 返回 数据 字典
     def getAwemeInfo(self, aweme_id):
+        print('[  提示  ]:正在请求的作品 id = %s\r\n' % aweme_id)
         if aweme_id is None:
             return None
 
@@ -126,11 +132,12 @@ class TikTok(object):
     # 传入 url 支持 https://www.iesdouyin.com 与 https://v.douyin.com
     # mode : post | like 模式选择 like为用户点赞 post为用户发布
     def getUserInfo(self, sec_uid, mode="post", count=35):
+        print('[  提示  ]:正在请求的用户 id = %s\r\n' % sec_uid)
         if sec_uid is None:
             return None
 
         max_cursor = 0
-        self.awemeList = []
+        awemeList = []
 
         print("[  提示  ]:正在获取所有作品数据请稍后...\r")
         print("[  提示  ]:会进行多次请求，等待时间较长...\r\n")
@@ -145,7 +152,7 @@ class TikTok(object):
                 url = self.urls.USER_FAVORITE_A + self.utils.getXbogus(
                     url=f'sec_user_id={sec_uid}&count={count}&max_cursor={max_cursor}&aid=1128&version_name=23.5.0&device_platform=android&os_version=2333')
             else:
-                print("[  错误  ]:模式选择错误, 仅支持post和like, 请检查后重新运行!\r")
+                print("[  错误  ]:模式选择错误, 仅支持post、like、mix, 请检查后重新运行!\r")
                 return None
 
             while True:
@@ -164,7 +171,7 @@ class TikTok(object):
                 aweme_id = aweme["aweme_id"]
                 # 深拷贝 dict 不然list里面全是同样的数据
                 datanew, dataraw = self.getAwemeInfo(aweme_id)
-                self.awemeList.append(copy.deepcopy(datanew))
+                awemeList.append(copy.deepcopy(datanew))
 
             # 更新 max_cursor
             max_cursor = datadict["max_cursor"]
@@ -174,11 +181,12 @@ class TikTok(object):
                 print("[  提示  ]:所有作品数据获取完成...\r\n")
                 break
             else:
-                print("[  提示  ]:第 " + str(times) + " 次请求成功...\r")
+                print("[  提示  ]:第 " + str(times) + " 次请求成功...\r\n")
 
-        return self.awemeList
+        return awemeList
 
     def getLiveInfo(self, web_rid: str):
+        print('[  提示  ]:正在请求的直播间 id = %s\r\n' % web_rid)
 
         # web_rid = live_url.replace('https://live.douyin.com/', '')
 
@@ -256,6 +264,98 @@ class TikTok(object):
         print('[   📺   ]:复制链接使用下载工具下载')
         return self.result.liveDict
 
+    def getMixInfo(self, mix_id: str, count=35):
+        print('[  提示  ]:正在请求的合集 id = %s\r\n' % mix_id)
+        if mix_id is None:
+            return None
+
+        cursor = 0
+        awemeList = []
+
+        print("[  提示  ]:正在获取合集下的所有作品数据请稍后...\r")
+        print("[  提示  ]:会进行多次请求，等待时间较长...\r\n")
+        times = 0
+        while True:
+            times = times + 1
+            print("[  提示  ]:正在进行第 " + str(times) + " 次请求...\r")
+
+            url = 'https://www.douyin.com/aweme/v1/web/mix/aweme/?' + self.utils.getXbogus(
+                url=f'device_platform=webapp&aid=6383&os_version=10&version_name=17.4.0&mix_id={mix_id}&cursor={cursor}&count={count}')
+
+            while True:
+                # 接口不稳定, 有时服务器不返回数据, 需要重新获取
+                try:
+                    res = requests.get(url=url, headers=self.headers)
+                    datadict = json.loads(res.text)
+                    print('[  提示  ]:本次请求返回 ' + str(len(datadict["aweme_list"])) + ' 条数据')
+                    if datadict is not None:
+                        break
+                except Exception as e:
+                    print("[  警告  ]:接口未返回数据, 正在重新请求!\r")
+
+            for aweme in datadict["aweme_list"]:
+                # 获取 aweme_id
+                aweme_id = aweme["aweme_id"]
+                # 深拷贝 dict 不然list里面全是同样的数据
+                datanew, dataraw = self.getAwemeInfo(aweme_id)
+                awemeList.append(copy.deepcopy(datanew))
+
+            # 更新 max_cursor
+            cursor = datadict["cursor"]
+
+            # 退出条件
+            if datadict["has_more"] == 0 or datadict["has_more"] == False:
+                print("\r\n[  提示  ]:合集下所有作品数据获取完成...\r\n")
+                break
+            else:
+                print("[  提示  ]:第 " + str(times) + " 次请求成功...\r\n")
+
+        return awemeList
+
+    def getUserAllMixInfo(self, sec_uid, count=35):
+        print('[  提示  ]:正在请求的用户 id = %s\r\n' % sec_uid)
+        if sec_uid is None:
+            return None
+
+        cursor = 0
+        mixIdNameDict = {}
+
+        print("[  提示  ]:正在获取所有合集 id 数据请稍后...\r")
+        print("[  提示  ]:会进行多次请求，等待时间较长...\r\n")
+        times = 0
+        while True:
+            times = times + 1
+            print("[  提示  ]:正在进行第 " + str(times) + " 次请求...\r")
+
+            url = self.urls.USER_MIX_LIST + self.utils.getXbogus(
+                url=f'device_platform=webapp&aid=6383&os_version=10&version_name=17.4.0&sec_user_id={sec_uid}&count={count}&cursor={cursor}')
+
+            while True:
+                # 接口不稳定, 有时服务器不返回数据, 需要重新获取
+                try:
+                    res = requests.get(url=url, headers=self.headers)
+                    datadict = json.loads(res.text)
+                    print('[  提示  ]:本次请求返回 ' + str(len(datadict["mix_infos"])) + ' 条数据')
+                    if datadict is not None and datadict["status_code"] == 0:
+                        break
+                except Exception as e:
+                    print("[  警告  ]:接口未返回数据, 正在重新请求!\r")
+
+            for mix in datadict["mix_infos"]:
+                mixIdNameDict[mix["mix_id"]] = mix["mix_name"]
+
+            # 更新 max_cursor
+            cursor = datadict["cursor"]
+
+            # 退出条件
+            if datadict["has_more"] == 0 or datadict["has_more"] == False:
+                print("[  提示  ]:所有合集 id 数据获取完成...\r\n")
+                break
+            else:
+                print("[  提示  ]:第 " + str(times) + " 次请求成功...\r\n")
+
+        return mixIdNameDict
+
     # 来自 https://blog.csdn.net/weixin_43347550/article/details/105248223
     def progressBarDownload(self, url, filepath):
         start = time.time()  # 下载开始时间

TikTokCommand.py

@@ -17,12 +17,13 @@ import argparse
 import os
 import json
 from TikTok import TikTok
+from TikTokUtils import Utils
 
 
 def argument():
     parser = argparse.ArgumentParser(description='抖音批量下载工具 使用帮助')
     parser.add_argument("--link", "-l",
-                        help="1.作品(视频或图集)与个人主页抖音分享链接(删除文案, 保证只有URL, https://v.douyin.com/kcvMpuN/)\r\n"
+                        help="1.作品(视频或图集)、合集、个人主页抖音分享链接(删除文案, 保证只有URL, https://v.douyin.com/kcvMpuN/)\r\n"
                              "2.解析直播网页版网址(https://live.douyin.com/802939216127)",
                         type=str, required=True)
     parser.add_argument("--path", "-p", help="下载保存位置",
@@ -33,7 +34,7 @@ def argument():
                         type=bool, required=False, default=True)
     parser.add_argument("--avatar", "-a", help="是否下载作者的头像(True/False), 默认为True",
                         type=bool, required=False, default=True)
-    parser.add_argument("--mode", "-M", help="link是个人主页时, 设置下载发布的作品(post)或喜欢的作品(like), 默认为post",
+    parser.add_argument("--mode", "-M", help="link是个人主页时, 设置下载发布的作品(post)或喜欢的作品(like)或者用户所有合集(mix), 默认为post",
                         type=str, required=False, default="post")
     args = parser.parse_args()
 
@@ -41,16 +42,33 @@ def argument():
 
 
 def main():
+    utils = Utils()
     args = argument()
     tk = TikTok()
     url = tk.getShareLink(args.link)
     key_type, key = tk.getKey(url)
     if key is None or key_type is None:
         return
-    elif key_type == "user":
+    elif key_type == "user" and args.mode != 'mix':
         datalist = tk.getUserInfo(key, args.mode, 35)
         tk.userDownload(awemeList=datalist, music=args.music, cover=args.cover, avatar=args.avatar,
                         savePath=args.path)
+    elif key_type == "user" and args.mode == 'mix':
+        if not os.path.exists(args.path):
+            os.mkdir(args.path)
+        mixIdNameDict = tk.getUserAllMixInfo(key, 35)
+
+        for mix_id in mixIdNameDict:
+            print(f'\r\n[  提示  ]:正在下载合集 [{mixIdNameDict[mix_id]}] 中的作品\r\n')
+            mix_file_name = utils.replaceStr(mixIdNameDict[mix_id])
+            datalist = tk.getMixInfo(mix_id, 35)
+            tk.userDownload(awemeList=datalist, music=args.music, cover=args.cover, avatar=args.avatar,
+                        savePath=os.path.join(args.path, mix_file_name))
+            print(f'\r\n[  提示  ]:合集 [{mixIdNameDict[mix_id]}] 中的作品下载完成\r\n')
+    elif key_type == "mix":
+        datalist = tk.getMixInfo(key,35)
+        tk.userDownload(awemeList=datalist, music=args.music, cover=args.cover, avatar=args.avatar,
+                        savePath=args.path)
     elif key_type == "aweme":
         datanew, dataraw = tk.getAwemeInfo(key)
         tk.awemeDownload(awemeDict=datanew, music=args.music, cover=args.cover, avatar=args.avatar,

TikTokUrls.py

@@ -39,6 +39,7 @@ class Urls(object):
         self.POST_DETAIL = 'https://www.douyin.com/aweme/v1/web/aweme/detail/?'
 
         # 用户喜欢A
+        # 需要 odin_tt
         self.USER_FAVORITE_A = 'https://www.douyin.com/aweme/v1/web/aweme/favorite/?'
 
         # 用户喜欢B
@@ -59,6 +60,14 @@ class Urls(object):
         # 关注用户作品
         self.FOLLOW_FEED = 'https://www.douyin.com/aweme/v1/web/follow/feed/?'
 
+        # 合集下所有作品
+        # 只需要X-Bogus
+        self.USER_MIX = 'https://www.douyin.com/aweme/v1/web/mix/aweme/?'
+
+        # 用户所有合集列表
+        # 需要 ttwid
+        self.USER_MIX_LIST = 'https://www.douyin.com/aweme/v1/web/mix/list/?'
+
         # X-Bogus Path
         # 60 秒内,请求同一URI累计超过 600 次,封锁IP 300 秒
         self.GET_XB_PATH = 'https://tiktok.199933.xyz/xb'