From 80d88a1331720436bdfa85806709481d48368870 Mon Sep 17 00:00:00 2001
From: Mort Yao <soi@mort.ninja>
Date: Sun, 26 Jul 2020 15:06:03 +0200
Subject: [PATCH] [tiktok] fix extraction

---
 src/you_get/extractors/tiktok.py | 26 ++++++++++++++++++--------
 tests/test.py                    | 15 +++++++++++----
 2 files changed, 29 insertions(+), 12 deletions(-)

diff --git a/src/you_get/extractors/tiktok.py b/src/you_get/extractors/tiktok.py
index 632e44c2..c2a0eb8d 100644
--- a/src/you_get/extractors/tiktok.py
+++ b/src/you_get/extractors/tiktok.py
@@ -6,15 +6,25 @@ from ..common import *
 
 def tiktok_download(url, output_dir='.', merge=True, info_only=False, **kwargs):
     html = get_html(url, faker=True)
-    title = r1(r'<title.*?>(.*?)</title>', html)
-    video_id = r1(r'/video/(\d+)', url) or r1(r'musical\?id=(\d+)', html)
-    title = '%s [%s]' % (title, video_id)
-    source = r1(r'<video .*?src="([^"]+)"', html) or r1(r'"contentUrl":"([^"]+)"', html)
-    mime, ext, size = url_info(source)
 
-    print_info(site_info, title, mime, size)
-    if not info_only:
-        download_urls([source], title, ext, size, output_dir, merge=merge)
+    data = r1(r'<script id="__NEXT_DATA__".*?>(.*?)</script>', html)
+    info = json.loads(data)
+    videoData = info['props']['pageProps']['videoData']
+    urls = videoData['itemInfos']['video']['urls']
+    videoId = videoData['itemInfos']['id']
+    uniqueId = videoData['authorInfos'].get('uniqueId')
+    nickName = videoData['authorInfos'].get('nickName')
+
+    for i, url in enumerate(urls):
+        title = '%s [%s]' % (nickName or uniqueId, videoId)
+        if len(urls) > 1:
+            title = '%s [%s]' % (title, i)
+
+        mime, ext, size = url_info(url)
+
+        print_info(site_info, title, mime, size)
+        if not info_only:
+            download_urls([url], title, ext, size, output_dir=output_dir, merge=merge)
 
 site_info = "TikTok.com"
 download = tiktok_download
diff --git a/tests/test.py b/tests/test.py
index 38db15dd..00bd4cbb 100644
--- a/tests/test.py
+++ b/tests/test.py
@@ -9,7 +9,8 @@ from you_get.extractors import (
     missevan,
     acfun,
     bilibili,
-    soundcloud
+    soundcloud,
+    tiktok
 )
 
 
@@ -53,9 +54,15 @@ class YouGetTests(unittest.TestCase):
             'https://soundcloud.com/keiny-pham/impure-bird', info_only=True
         )
         ## playlist
-        soundcloud.download(
-            'https://soundcloud.com/anthony-flieger/sets/cytus', info_only=True
-        )
+        #soundcloud.download(
+        #    'https://soundcloud.com/anthony-flieger/sets/cytus', info_only=True
+        #)
+
+    def tests_tiktok(self):
+        tiktok.download('https://www.tiktok.com/@nmb48_official/video/6850796940293164290', info_only=True)
+        tiktok.download('https://t.tiktok.com/i18n/share/video/6850796940293164290/', info_only=True)
+        tiktok.download('https://vt.tiktok.com/UGJR4R/', info_only=True)
+
 
 if __name__ == '__main__':
     unittest.main()