Merge 7a3c838657 into c5098961b0

Now also succeeds with player b12cc44b

test/test_jsinterp.py

@@ -425,6 +425,34 @@ class TestJSInterpreter(unittest.TestCase):
             self._test(jsi, [''], args=['', '-'])
             self._test(jsi, [], args=['', ''])
 
+    def test_slice(self):
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice()}', [0, 1, 2, 3, 4, 5, 6, 7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(0)}', [0, 1, 2, 3, 4, 5, 6, 7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(5)}', [5, 6, 7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(99)}', [])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(-2)}', [7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(-99)}', [0, 1, 2, 3, 4, 5, 6, 7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(0, 0)}', [])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(1, 0)}', [])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(0, 1)}', [0])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(3, 6)}', [3, 4, 5])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(1, -1)}', [1, 2, 3, 4, 5, 6, 7])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(-1, 1)}', [])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(-3, -1)}', [6, 7])
+        self._test('function f(){return "012345678".slice()}', '012345678')
+        self._test('function f(){return "012345678".slice(0)}', '012345678')
+        self._test('function f(){return "012345678".slice(5)}', '5678')
+        self._test('function f(){return "012345678".slice(99)}', '')
+        self._test('function f(){return "012345678".slice(-2)}', '78')
+        self._test('function f(){return "012345678".slice(-99)}', '012345678')
+        self._test('function f(){return "012345678".slice(0, 0)}', '')
+        self._test('function f(){return "012345678".slice(1, 0)}', '')
+        self._test('function f(){return "012345678".slice(0, 1)}', '0')
+        self._test('function f(){return "012345678".slice(3, 6)}', '345')
+        self._test('function f(){return "012345678".slice(1, -1)}', '1234567')
+        self._test('function f(){return "012345678".slice(-1, 1)}', '')
+        self._test('function f(){return "012345678".slice(-3, -1)}', '67')
+
 
 if __name__ == '__main__':
     unittest.main()

test/test_youtube_signature.py

@@ -174,6 +174,14 @@ _NSIG_TESTS = [
         'https://www.youtube.com/s/player/5604538d/player_ias.vflset/en_US/base.js',
         '7X-he4jjvMx7BCX', 'sViSydX8IHtdWA',
     ),
+    (
+        'https://www.youtube.com/s/player/20dfca59/player_ias.vflset/en_US/base.js',
+        '-fLCxedkAk4LUTK2', 'O8kfRq1y1eyHGw',
+    ),
+    (
+        'https://www.youtube.com/s/player/b12cc44b/player_ias.vflset/en_US/base.js',
+        'keLa5R2U00sR9SQK', 'N1OGyujjEwMnLw',
+    ),
 ]
 
 

youtube_dl/extractor/bilibili.py

@@ -38,7 +38,21 @@ class BiliBiliIE(InfoExtractor):
                     '''
 
     _TESTS = [{
-        'url': 'http://www.bilibili.tv/video/av1074402/',
+        'url': 'https://www.bilibili.com/video/BV1N7411A7WU/',
+        'info_dict': {
+            'id': '1N7411A7WU',
+            'ext': 'flv',
+            'title': '[经验分享]观看B站视频时如何开启CC字幕？(B站小白必看)',
+            'description': 'md5:b5b03414a659c29c7f5bce58c725b8ee',
+            'duration': 119.267,
+            'timestamp': 1582321654,
+            'upload_date': '20200221',
+            'uploader': 'Lynda教程和字幕',
+            'uploader_id': '389168097',
+            'subtitles': {'zh-CN': [{'url': 'http:\\u002F\\u002Fi0.hdslb.com\\u002Fbfs\\u002Fsubtitle\\u002F3c648ce057eea5b4e619c5dd9f0f23fbcad728f5.json', 'data': 'md5:ece63ef7a3a45fe2f7183f95f96d3181', 'ext': 'srt'}]}
+        },
+    }, {
+        'url': 'http://www.bilibili.com/video/av1074402/',
         'md5': '5f7d29e1a2872f3df0cf76b1f87d3788',
         'info_dict': {
             'id': '1074402',
@@ -112,6 +126,12 @@ class BiliBiliIE(InfoExtractor):
     _APP_KEY = 'iVGUTjsxvpLeuDCf'
     _BILIBILI_KEY = 'aHRmhWMLkdeMuILqORnYZocwMBpMEOdt'
 
+    def convert_timestamp_format(self, cc_timestamp):
+        ss, mm = "{:.3f}".format(float(cc_timestamp)).split(".")
+        m, s = divmod(int(ss), 60)
+        h, m = divmod(m, 60)
+        return "%02d:%02d:%02d,%s" % (h, m, s, mm)
+
     def _report_error(self, result):
         if 'message' in result:
             raise ExtractorError('%s said: %s' % (self.IE_NAME, result['message']), expected=True)
@@ -155,6 +175,21 @@ class BiliBiliIE(InfoExtractor):
                 self._report_error(js)
             cid = js['result']['cid']
 
+        subtitles = {}
+        for sub in re.findall(r'"lan":\s*"([^"]+)"[^}]+"subtitle_url":\s*"([^"]+)"', webpage, re.DOTALL):
+            subtitles[sub[0]] = [{'url': sub[1]}]
+            j = self._download_json(sub[1].replace("\\u002F", "/"), video_id, note='Downloading raw subtitle: %s' % (sub[0]))
+            srt = []
+            for idx, line in enumerate(j["body"]):
+                from_timestamp = self.convert_timestamp_format(line["from"])
+                to_timestamp = self.convert_timestamp_format(line["to"])
+                srt_tmp = (
+                    """%d\n%s --> %s\n%s\n""" % (idx, from_timestamp, to_timestamp, line["content"])
+                )
+                srt.append(srt_tmp)
+            subtitles[sub[0]][0]["data"] = "\n".join(srt)
+            subtitles[sub[0]][0]["ext"] = "srt"
+
         headers = {
             'Accept': 'application/json',
             'Referer': url
@@ -204,6 +239,7 @@ class BiliBiliIE(InfoExtractor):
                     'id': '%s_part%s' % (video_id, idx),
                     'duration': float_or_none(durl.get('length'), 1000),
                     'formats': formats,
+                    'subtitles': subtitles
                 })
             break
 

youtube_dl/extractor/youtube.py

@@ -1659,17 +1659,46 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
     def _extract_n_function_name(self, jscode):
         func_name, idx = self._search_regex(
             # new: (b=String.fromCharCode(110),c=a.get(b))&&c=nfunc[idx](c)
-            # or:  (b="nn"[+a.D],c=a.get(b))&&(c=nfunc[idx](c)s
+            # or:  (b="nn"[+a.D],c=a.get(b))&&(c=nfunc[idx](c)
-            # old: .get("n"))&&(b=nfunc[idx](b)
+            # or:  (PL(a),b=a.j.n||null)&&(b=nfunc[idx](b)
-            # older: .get("n"))&&(b=nfunc(b)
+            # or:  (b="nn"[+a.D],vL(a),c=a.j[b]||null)&&(c=narray[idx](c),a.set(b,c),narray.length||nfunc("")
+            # old: (b=a.get("n"))&&(b=nfunc[idx](b)(?P<c>[a-z])\s*=\s*[a-z]\s*
+            # older: (b=a.get("n"))&&(b=nfunc(b)
             r'''(?x)
-                (?:\(\s*(?P<b>[a-z])\s*=\s*(?:
+                \((?:[\w$()\s]+,)*?\s*      # (
-                    String\s*\.\s*fromCharCode\s*\(\s*110\s*\)|
+                (?P<b>[a-z])\s*=\s*         # b=
-                    "n+"\[\s*\+?s*[\w$.]+\s*]
+                (?:
-                )\s*,(?P<c>[a-z])\s*=\s*[a-z]\s*)?
+                    (?:                     # expect ,c=a.get(b) (etc)
-                \.\s*get\s*\(\s*(?(b)(?P=b)|"n{1,2}")(?:\s*\)){2}\s*&&\s*\(\s*(?(c)(?P=c)|b)\s*=\s*
+                        String\s*\.\s*fromCharCode\s*\(\s*110\s*\)|
-                (?P<nfunc>[a-zA-Z_$][\w$]*)(?:\s*\[(?P<idx>\d+)\])?\s*\(\s*[\w$]+\s*\)
+                        "n+"\[\s*\+?s*[\w$.]+\s*]
-            ''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx'))
+                    )\s*(?:,[\w$()\s]+(?=,))*|
+                       (?P<old>[\w$]+)      # a (old[er])
+                   )\s*
+                   (?(old)
+                                            # b.get("n")
+                       (?:\.\s*[\w$]+\s*|\[\s*[\w$]+\s*]\s*)*?
+                       (?:\.\s*n|\[\s*"n"\s*]|\.\s*get\s*\(\s*"n"\s*\))
+                       |                    # ,c=a.get(b)
+                       ,\s*(?P<c>[a-z])\s*=\s*[a-z]\s*
+                       (?:\.\s*[\w$]+\s*|\[\s*[\w$]+\s*]\s*)*?
+                       (?:\[\s*(?P=b)\s*]|\.\s*get\s*\(\s*(?P=b)\s*\))
+                   )
+                                            # interstitial junk
+                   \s*(?:\|\|\s*null\s*)?(?:\)\s*)?&&\s*(?:\(\s*)?
+               (?(c)(?P=c)|(?P=b))\s*=\s*   # [c|b]=
+                                            # nfunc|nfunc[idx]
+                   (?P<nfunc>[a-zA-Z_$][\w$]*)(?:\s*\[(?P<idx>\d+)\])?\s*\(\s*[\w$]+\s*\)
+            ''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx'),
+            default=(None, None))
+        # thx bashonly: yt-dlp/yt-dlp/pull/10611
+        if not func_name:
+            self.report_warning('Falling back to generic n function search')
+            return self._search_regex(
+                r'''(?xs)
+                    (?:(?<=[^\w$])|^)       # instead of \b, which ignores $
+                    (?P<name>(?!\d)[a-zA-Z\d_$]+)\s*=\s*function\((?!\d)[a-zA-Z\d_$]+\)
+                    \s*\{(?:(?!};).)+?["']enhanced_except_
+                ''', jscode, 'Initial JS player n function name', group='name')
         if not idx:
             return func_name
 

youtube_dl/jsinterp.py

@@ -925,9 +925,16 @@ class JSInterpreter(object):
                     obj.reverse()
                     return obj
                 elif member == 'slice':
-                    assertion(isinstance(obj, list), 'must be applied on a list')
+                    assertion(isinstance(obj, (list, compat_str)), 'must be applied on a list or string')
-                    assertion(len(argvals) == 1, 'takes exactly one argument')
+                    # From [1]:
-                    return obj[argvals[0]:]
+                    # .slice() - like [:]
+                    # .slice(n) - like [n:] (not [slice(n)]
+                    # .slice(m, n) - like [m:n] or [slice(m, n)]
+                    # [1] https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/slice
+                    assertion(len(argvals) <= 2, 'takes between 0 and 2 arguments')
+                    if len(argvals) < 2:
+                        argvals += (None,)
+                    return obj[slice(*argvals)]
                 elif member == 'splice':
                     assertion(isinstance(obj, list), 'must be applied on a list')
                     assertion(argvals, 'takes one or more arguments')