Merge dc7a5390ce into c5098961b0

[Youtube] Rework n function extraction pattern
Now also succeeds with player b12cc44b
2025-01-08 12:30:10 +09:00 · 2024-08-21 22:22:45 -04:00 · 2024-08-06 20:59:09 +01:00 · 2024-08-06 20:51:38 +01:00 · 2024-08-01 19:18:34 +01:00 · 2021-10-11 22:19:21 +03:00
6 changed files with 125 additions and 13 deletions
--- a/test/test_jsinterp.py
+++ b/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()
--- a/test/test_youtube_signature.py
+++ b/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',
+    ),
 ]


--- a/youtube_dl/extractor/cspan.py
+++ b/youtube_dl/extractor/cspan.py
@ -1,8 +1,13 @@
 from __future__ import unicode_literals

+import os
 import re

 from .common import InfoExtractor
+from ..compat import (
+    compat_str,
+    compat_urllib_parse_unquote
+)
 from ..utils import (
    determine_ext,
    ExtractorError,
@ -23,6 +28,7 @@ from .ustream import UstreamIE


 class CSpanIE(InfoExtractor):
+    _BASIC_URL = r'https?://(?:www\.)?c-span\.org/'
    _VALID_URL = r'https?://(?:www\.)?c-span\.org/video/\?(?P<id>[0-9a-f]+)'
    IE_DESC = 'C-SPAN'
    _TESTS = [{
@ -81,6 +87,29 @@ class CSpanIE(InfoExtractor):
    }]
    BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/%s_%s/index.html?videoId=%s'

+    @classmethod
+    def is_basic_url(cls, url):
+        if '_BASIC_URL_RE' not in cls.__dict__:
+            cls._BASIC_URL_RE = re.compile(cls._BASIC_URL)
+
+        return cls._BASIC_URL_RE.match(url) is not None
+
+    @classmethod
+    def get_basic_url(cls, url):
+        return "/".join(url.rstrip('/').split('/')[:3])
+
+    @classmethod
+    def _get_id(cls, url):
+        if '_ID_URL_RE' not in cls.__dict__:
+            cls._ID_URL_RE = re.compile(cls._ID_URL)
+
+        if cls._ID_URL_RE.match(url) is not None:
+            m = cls._ID_URL_RE.match(url)
+            assert m
+            return compat_str(m.group('id'))
+        else:
+            return compat_urllib_parse_unquote(os.path.splitext(url.rstrip('/').split('/')[-1])[0])
+
    def _real_extract(self, url):
        video_id = self._match_id(url)
        video_type = None
--- a/youtube_dl/extractor/generic.py
+++ b/youtube_dl/extractor/generic.py
@ -50,6 +50,7 @@ from .nexx import (
    NexxIE,
    NexxEmbedIE,
 )
+from .cspan import CSpanIE
 from .nbc import NBCSportsVPlayerIE
 from .ooyala import OoyalaIE
 from .rutv import RUTVIE
@ -3766,6 +3767,16 @@ class GenericIE(InfoExtractor):
            if entry_info_dict.get('formats'):
                self._sort_formats(entry_info_dict['formats'])

+            if CSpanIE.is_basic_url(url):
+                basic_url = CSpanIE.get_basic_url(url)
+                for f in entry_info_dict['formats']:
+                    f.setdefault('http_headers', {})['referer'] = basic_url + "/"
+                    f.setdefault('http_headers', {})['origin'] = basic_url
+                    f.setdefault('http_headers', {})['accept'] = "*/*"
+                    f.setdefault('http_headers', {})['sec-fetch-dest'] = "empty"
+                    f.setdefault('http_headers', {})['sec-fetch-mode'] = "cors"
+                    f.setdefault('http_headers', {})['sec-fetch-site'] = "cross-site"
+
            entries.append(entry_info_dict)

        if len(entries) == 1:
--- a/youtube_dl/extractor/youtube.py
+++ b/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
-            # old: .get("n"))&&(b=nfunc[idx](b)
-            # older: .get("n"))&&(b=nfunc(b)
+            # or:  (b="nn"[+a.D],c=a.get(b))&&(c=nfunc[idx](c)
+            # or:  (PL(a),b=a.j.n||null)&&(b=nfunc[idx](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*(?:
-                    String\s*\.\s*fromCharCode\s*\(\s*110\s*\)|
-                    "n+"\[\s*\+?s*[\w$.]+\s*]
-                )\s*,(?P<c>[a-z])\s*=\s*[a-z]\s*)?
-                \.\s*get\s*\(\s*(?(b)(?P=b)|"n{1,2}")(?:\s*\)){2}\s*&&\s*\(\s*(?(c)(?P=c)|b)\s*=\s*
-                (?P<nfunc>[a-zA-Z_$][\w$]*)(?:\s*\[(?P<idx>\d+)\])?\s*\(\s*[\w$]+\s*\)
-            ''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx'))
+                \((?:[\w$()\s]+,)*?\s*      # (
+                (?P<b>[a-z])\s*=\s*         # b=
+                (?:
+                    (?:                     # expect ,c=a.get(b) (etc)
+                        String\s*\.\s*fromCharCode\s*\(\s*110\s*\)|
+                        "n+"\[\s*\+?s*[\w$.]+\s*]
+                    )\s*(?:,[\w$()\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

--- a/youtube_dl/jsinterp.py
+++ b/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(len(argvals) == 1, 'takes exactly one argument')
-                    return obj[argvals[0]:]
+                    assertion(isinstance(obj, (list, compat_str)), 'must be applied on a list or string')
+                    # From [1]:
+                    # .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')
Author	SHA1	Message	Date
Grabien	075a535c6b	Merge `dc7a5390ce` into `c5098961b0`	2024-08-21 22:22:45 -04:00
dirkf	c5098961b0	[Youtube] Rework n function extraction pattern Now also succeeds with player b12cc44b	2024-08-06 20:59:09 +01:00
dirkf	dbc08fba83	[jsinterp] Improve slice implementation for player b12cc44b Partly taken from yt-dlp/yt-dlp#10664, thx seproDev Fixes #32896	2024-08-06 20:51:38 +01:00
Aiur Adept	71223bff39	[Youtube] Fix nsig extraction for player 20dfca59 (#32891 ) * dirkf's patch for nsig extraction * add generic search per yt-dlp/yt-dlp/pull/10611 - thx bashonly --------- Co-authored-by: dirkf <fieldhouse@gmx.net>	2024-08-01 19:18:34 +01:00
Grabien	dc7a5390ce	C-SPAN plugin imports fixes	2021-10-11 22:19:21 +03:00
Grabien	f8f3b6f2bf	C-Span links support improvements	2021-09-27 22:22:39 +03:00