From 5b217541977d67a0b823cd224326339af88adbbe Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Bartosz=20Bia=C5=82as?= Date: Wed, 8 Jun 2022 19:11:24 +0200 Subject: [PATCH 01/21] [SwitchTube] Add new extractor --- docs/supportedsites.md | 3 + youtube_dl/extractor/extractors.py | 5 + youtube_dl/extractor/switchtube.py | 219 +++++++++++++++++++++++++++++ 3 files changed, 227 insertions(+) create mode 100644 youtube_dl/extractor/switchtube.py diff --git a/docs/supportedsites.md b/docs/supportedsites.md index ae2a6b8b0..6a3bab242 100644 --- a/docs/supportedsites.md +++ b/docs/supportedsites.md @@ -910,6 +910,9 @@ - **SVTPage** - **SVTPlay**: SVT Play and Öppet arkiv - **SVTSeries** + - **switchtube** + - **switchtube:channel** + - **switchtube:profile** - **SWRMediathek** - **Syfy** - **SztvHu** diff --git a/youtube_dl/extractor/extractors.py b/youtube_dl/extractor/extractors.py index 751fc38b6..78ea0d9a4 100644 --- a/youtube_dl/extractor/extractors.py +++ b/youtube_dl/extractor/extractors.py @@ -1209,6 +1209,11 @@ from .svt import ( SVTPlayIE, SVTSeriesIE, ) +from .switchtube import ( + SwitchTubeIE, + SwitchTubeProfileIE, + SwitchTubeChannelIE, +) from .swrmediathek import SWRMediathekIE from .syfy import SyfyIE from .sztvhu import SztvHuIE diff --git a/youtube_dl/extractor/switchtube.py b/youtube_dl/extractor/switchtube.py new file mode 100644 index 000000000..3601572b6 --- /dev/null +++ b/youtube_dl/extractor/switchtube.py @@ -0,0 +1,219 @@ +# coding: utf-8 +from __future__ import unicode_literals + +import itertools +import re + +from .common import InfoExtractor +from ..utils import ( + clean_html, + extract_attributes, + float_or_none, + get_element_by_attribute, + int_or_none, + parse_iso8601, + strip_or_none, + urljoin, +) + + +class SwitchTubeIE(InfoExtractor): + _VALID_URL = r'https?://tube\.switch\.ch/videos/(?P[\da-zA-Z]+)' + IE_NAME = 'switchtube' + _TESTS = [{ + 'url': 'https://tube.switch.ch/videos/0T1XfaIFSX', + 'info_dict': { + 'id': '0T1XfaIFSX', + 'title': '2016_ASE_sqC03-Entretien', + 'channel': 'ASE Assistant-e socio-éducatif-ve CFC', + 'channel_url': 'https://tube.switch.ch/channels/bsaer76yoL', + 'channel_id': 'bsaer76yoL', + 'ext': 'mp4', + 'description': None, + 'thumbnail': r're:^https?://tube.switch.ch/image/representations/[\w-]+$', + 'license': 'All rights reserved', + 'creator': 'Jean-Marc Pouly from Eidgenössische Hochschule für Berufsbildung', + 'uploader': 'Jean-Marc Pouly from Eidgenössische Hochschule für Berufsbildung', + 'uploader_url': 'https://tube.switch.ch/profiles/42481', + 'uploader_id': '42481', + 'upload_date': '20220309', + 'timestamp': 1646839068, + } + }, { + 'url': 'https://tube.switch.ch/videos/0cf3886d', + 'info_dict': { + 'id': '0cf3886d', + 'ext': 'mp4', + 'title': 'Introduction: Mini-Batches in On- and Off-Policy Deep Reinforcement Learning', + 'license': 'All rights reserved', + 'description': 'One of the challenges in Deep Reinforcement Learning is to decorrelate the data. How this is possible with replay buffers is explained here.', + 'thumbnail': r're:^https?://tube.switch.ch/image/representations/[\w-]+$', + 'channel': 'CS-456 Artificial Neural Networks', + 'channel_url': 'https://tube.switch.ch/channels/1deb03e0', + 'channel_id': '1deb03e0', + 'timestamp': 1590733406, + 'upload_date': '20200529', + 'creator': 'Wulfram Gerstner from École polytechnique fédérale de Lausanne (EPFL)', + 'uploader': 'Wulfram Gerstner from École polytechnique fédérale de Lausanne (EPFL)', + 'uploader_url': 'https://tube.switch.ch/profiles/94248', + 'uploader_id': '94248', + } + }] + + def _real_extract(self, url): + video_id = self._match_id(url) + webpage = self._download_webpage(url, video_id) + title = self._html_search_regex(r'(?s)]*>(.+?)', webpage, 'title') + + info = { + 'id': video_id, + 'title': title, + 'is_live': False, + } + + info['view_count'] = int_or_none(self._html_search_regex(r'(?s)]*>\s*(\d+)\s+views?\s*

', webpage, + 'view_count', default=None)) + info['license'] = self._html_search_regex(r'''(?s)]*?\bproperty\s*=\s*["'](?:\b[cd]c:license\b\s*){2,}[^>]+>(.+)''', + webpage, 'license', default=None) + info['description'] = strip_or_none(clean_html( + get_element_by_attribute('property', 'dc:description', webpage))) + info['duration'] = float_or_none( + self._search_regex(r'''\bdata-duration\s*=\s*["']([\d.]+)''', + webpage, 'duration', default=None)) + + channel_groups = self._search_regex(r'''(?s)]+?\bhref\s*=\s*["']/channels/(.{4,}?)''', + webpage, 'channel groups', default='') + channel_groups = re.split(r'''(?s)["'][^>]*>\s*''', channel_groups, 1) + if len(channel_groups) == 2: + for i, ch in enumerate(('channel_id', 'channel')): + info[ch] = strip_or_none(channel_groups[i]) + if info['channel_id']: + info['channel_url'] = 'https://tube.switch.ch/channels/' + info['channel_id'] + + def outer_elements_by_attribute(attr, value, html, tag=None, escape_value=True, include_inner=False): + """generate matching HTML element strings + if include_inner, tuples of (element, content)""" + pattern = r'''(?s)(?P<(%s)\b[^>]+?\b%s\s*=\s*("|')%s\b[^>]+>)(?P.*?)''' % \ + (re.escape(tag) if tag is not None else r'\w+', attr, re.escape(value) if escape_value else value) + matches = re.finditer(pattern, html) + for m in matches: + yield m.group('element', 'inner') if include_inner else m.group('element') + + for dt in outer_elements_by_attribute('property', 'dc:date', webpage, tag='span'): + dt = extract_attributes(dt) + if dt.get('class') == 'dt-published': + info['timestamp'] = parse_iso8601(dt.get('content')) + break + + creator_groups = self._search_regex( + r'''(?s)]+?\bclass\s*=\s*("|')(?:(?!\1).)*\bp-author\b(?:(?!\1).)*\1\s*property\s*=\s*["']dc:creator\b[^>]+>\s*(.*?]+?\bclass\s*=\s*["']p-name\b.*).*?''', + webpage, 'creator groups', default='', group=2) + creator_groups = re.match(r'''(?s)]+?\bhref\s*=\s*("|')/profiles/(?P.+?)\1[^>]*>\s*]+>\s*(?P.+?)\s*[,\s]*]+\bclass\s*=\s*["']p-organization-name\b[^>]+>\s*(?P.+?)\s*''', creator_groups) + if creator_groups: + creator_groups = creator_groups.groupdict() + info['uploader'] = info['creator'] = ' from '.join((creator_groups['creator_name'], creator_groups['organization_name'])) + info['uploader_id'] = creator_groups['profile_id'] + info['uploader_url'] = 'https://tube.switch.ch/profiles/' + info['uploader_id'] + + parsed_media_entries = self._parse_html5_media_entries(url, webpage, video_id)[0] + info['thumbnail'] = parsed_media_entries['thumbnail'] + info['formats'] = parsed_media_entries['formats'] + self._sort_formats(info['formats']) + + return info + + +class SwitchTubeProfileIE(InfoExtractor): + _VALID_URL = r'https?://tube\.switch\.ch/profiles/(?P[\da-zA-Z]+)' + IE_NAME = 'switchtube:profile' + _TESTS = [{ + 'url': 'https://tube.switch.ch/profiles/94248', + 'info_dict': { + 'id': '94248', + 'title': 'Wulfram Gerstner', + 'description': None, + }, + 'playlist_mincount': 94, + }] + + @classmethod + def suitable(cls, url): + return False if SwitchTubeIE.suitable(url) else super(SwitchTubeProfileIE, cls).suitable(url) + + def _real_extract(self, url): + channel_id = self._match_id(url) + webpage = self._download_webpage(url, channel_id) + channel_title = self._html_search_regex(r'(?s)]*>(.+?)', webpage, 'title', default=None) + + description = self._html_search_regex(r'''(?s)]+class\s*=\s*("|')p-summary formatted\1[^>]+property\s*=\s*("|')dc:description\2[^>]*>\s*(.+?)\s*''', + webpage, 'description', default=None, group=3) + + entries = [] + next_page = None + for current_page_number in itertools.count(1): + if next_page: + webpage = self._download_webpage(next_page, channel_id, note='Downloading page %d' % (current_page_number, )) + + for _, video_path, video_id, title in re.findall( + r'''(?s)]+\bhref\s*=\s*("|')(/videos/((?:(?!\1).)+?))\1[^>]*>\s*]+\bclass\s*=\s*["']title\b[^>]+>(.+?)''', + webpage): + video_url = urljoin(url, video_path) + if video_url: + entries.append(self.url_result(video_url, ie=SwitchTubeIE.ie_key(), video_id=video_id)) + + next_page = self._search_regex( + r''']+?\bhref\s*=\s*("|')(?P/profiles/%s\?(?:(?!\1).)+)\1[^>]*>\s*Next\s*''' % (channel_id,), + webpage, 'next page', group='path', default=None) + if next_page: + next_page = urljoin(url, next_page) + if not next_page: + break + + return self.playlist_result(entries, channel_id, channel_title, + description) + + +class SwitchTubeChannelIE(InfoExtractor): + _VALID_URL = r'https?://tube\.switch\.ch/channels/(?P[\da-zA-Z]+)' + IE_NAME = 'switchtube:channel' + _TESTS = [{ + 'url': 'https://tube.switch.ch/channels/1deb03e0', + 'info_dict': { + 'id': '1deb03e0', + 'title': 'CS-456 Artificial Neural Networks', + 'description': 'Class on Artificial Neural Networks and Reinforcement Learning designed for EPFL master students in CS and related disciplines.' + }, + 'playlist_mincount': 94, + }] + + @classmethod + def suitable(cls, url): + return False if SwitchTubeIE.suitable(url) else super(SwitchTubeChannelIE, cls).suitable(url) + + def _real_extract(self, url): + channel_id = self._match_id(url) + + entries = [] + for current_page_number in itertools.count(0): + page_url = urljoin(url, '/channels/%s?order=episodes&page=%d' % (channel_id, current_page_number)) + webpage = self._download_webpage(page_url, channel_id, note='Downloading page %d' % (current_page_number + 1, )) + + if current_page_number == 0: + channel_title = self._html_search_regex(r'(?s)]*>(.+?)', webpage, 'title', default=None) + description = self._html_search_regex(r'''(?s)]+class\s*=\s*("|')description formatted\1[^>]*>\s*

\s*(.+?)\s*

\s*''', + webpage, 'description', default=None, group=2) + + this_page_still_has_something = False + for _, video_path, video_id in re.findall( + r'''(?s)]+\bhref\s*=\s*("|')(/videos/((?:(?!\1).)+?))\1[^>]*>''', + webpage): + video_url = urljoin(url, video_path) + if video_url: + this_page_still_has_something = True + entries.append(self.url_result(video_url, ie=SwitchTubeIE.ie_key(), video_id=video_id)) + + if not this_page_still_has_something: + break + + return self.playlist_result(entries, channel_id, channel_title, + description) From 93fdde431c0ff0cb8eb8a738f2629734b66b527e Mon Sep 17 00:00:00 2001 From: dirkf Date: Tue, 9 Aug 2022 21:05:00 +0100 Subject: [PATCH 02/21] [test, etc] Improve download test logs; also clean up some new flake8 issues (#31153) * [test] Identify testcase errors better * [test] Identify download errors better * [extractor/minds] Linter * [extractor/aes] Linter --- test/test_download.py | 7 +++++-- youtube_dl/aes.py | 2 +- youtube_dl/extractor/minds.py | 2 +- 3 files changed, 7 insertions(+), 4 deletions(-) diff --git a/test/test_download.py b/test/test_download.py index 6a6673bc2..19936969f 100644 --- a/test/test_download.py +++ b/test/test_download.py @@ -33,6 +33,7 @@ from youtube_dl.compat import ( from youtube_dl.utils import ( DownloadError, ExtractorError, + error_to_compat_str, format_bytes, UnavailableVideoError, ) @@ -108,7 +109,7 @@ def generator(test_case, tname): for tc in test_cases: info_dict = tc.get('info_dict', {}) if not (info_dict.get('id') and info_dict.get('ext')): - raise Exception('Test definition incorrect. The output file cannot be known. Are both \'id\' and \'ext\' keys present?') + raise Exception('Test definition (%s) requires both \'id\' and \'ext\' keys present to define the output file' % (tname, )) if 'skip' in test_case: print_skipping(test_case['skip']) @@ -161,7 +162,9 @@ def generator(test_case, tname): except (DownloadError, ExtractorError) as err: # Check if the exception is not a network related one if not err.exc_info[0] in (compat_urllib_error.URLError, socket.timeout, UnavailableVideoError, compat_http_client.BadStatusLine) or (err.exc_info[0] == compat_HTTPError and err.exc_info[1].code == 503): - raise + msg = getattr(err, 'msg', error_to_compat_str(err)) + err.msg = '%s (%s)' % (msg, tname, ) + raise err if try_num == RETRIES: report_warning('%s failed due to network errors, skipping...' % tname) diff --git a/youtube_dl/aes.py b/youtube_dl/aes.py index 461bb6d41..d0de2d93f 100644 --- a/youtube_dl/aes.py +++ b/youtube_dl/aes.py @@ -303,7 +303,7 @@ def xor(data1, data2): def rijndael_mul(a, b): - if(a == 0 or b == 0): + if (a == 0 or b == 0): return 0 return RIJNDAEL_EXP_TABLE[(RIJNDAEL_LOG_TABLE[a] + RIJNDAEL_LOG_TABLE[b]) % 0xFF] diff --git a/youtube_dl/extractor/minds.py b/youtube_dl/extractor/minds.py index 8e9f0f825..e8fd582aa 100644 --- a/youtube_dl/extractor/minds.py +++ b/youtube_dl/extractor/minds.py @@ -78,7 +78,7 @@ class MindsIE(MindsBaseIE): else: return self.url_result(entity['perma_url']) else: - assert(entity['subtype'] == 'video') + assert (entity['subtype'] == 'video') video_id = entity_id # 1080p and webm formats available only on the sources array video = self._call_api( From c9234b1cceafeddb10387321ecbc9328d150aaba Mon Sep 17 00:00:00 2001 From: dirkf Date: Wed, 10 Aug 2022 15:37:59 +0100 Subject: [PATCH 03/21] [core] Make `--max-downloads ...` stop immediately on reaching the limit Based on and closes #26638. --- youtube_dl/YoutubeDL.py | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/youtube_dl/YoutubeDL.py b/youtube_dl/YoutubeDL.py index 3895b408f..e77b8d50c 100755 --- a/youtube_dl/YoutubeDL.py +++ b/youtube_dl/YoutubeDL.py @@ -1779,10 +1779,9 @@ class YoutubeDL(object): assert info_dict.get('_type', 'video') == 'video' - max_downloads = self.params.get('max_downloads') - if max_downloads is not None: - if self._num_downloads >= int(max_downloads): - raise MaxDownloadsReached() + max_downloads = int_or_none(self.params.get('max_downloads')) or float('inf') + if self._num_downloads >= max_downloads: + raise MaxDownloadsReached() # TODO: backward compatibility, to be removed info_dict['fulltitle'] = info_dict['title'] @@ -2062,6 +2061,9 @@ class YoutubeDL(object): self.report_error('postprocessing: %s' % str(err)) return self.record_download_archive(info_dict) + # avoid possible nugatory search for further items (PR #26638) + if self._num_downloads >= max_downloads: + raise MaxDownloadsReached() def download(self, url_list): """Download a given list of URLs.""" From 27281eb5bb87a150f41d3c74edb589a5bf010efa Mon Sep 17 00:00:00 2001 From: dirkf Date: Sun, 14 Aug 2022 18:45:45 +0100 Subject: [PATCH 04/21] [jsinterp] Overhaul JSInterp to handle new YT players 4c3f79c5, 324f67b9 (#31170) * back-port from yt-dlp 8f53dc44a0cc1c2d98c35740b9293462c080f5d0, thanks pukkandan * also support void, improve <> precedence, improve expressions in comma-list * add more tests --- test/test_jsinterp.py | 49 ++- test/test_utils.py | 3 + test/test_youtube_signature.py | 13 + youtube_dl/compat.py | 54 ++- youtube_dl/jsinterp.py | 581 ++++++++++++++++++++------------- youtube_dl/utils.py | 47 ++- 6 files changed, 500 insertions(+), 247 deletions(-) diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py index acdabffb1..c6c931743 100644 --- a/test/test_jsinterp.py +++ b/test/test_jsinterp.py @@ -19,6 +19,9 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter('function x3(){return 42;}') self.assertEqual(jsi.call_function('x3'), 42) + jsi = JSInterpreter('function x3(){42}') + self.assertEqual(jsi.call_function('x3'), None) + jsi = JSInterpreter('var x5 = function(){return 42;}') self.assertEqual(jsi.call_function('x5'), 42) @@ -51,8 +54,11 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter('function f(){return 11 >> 2;}') self.assertEqual(jsi.call_function('f'), 2) + jsi = JSInterpreter('function f(){return []? 2+3: 4;}') + self.assertEqual(jsi.call_function('f'), 5) + def test_array_access(self): - jsi = JSInterpreter('function f(){var x = [1,2,3]; x[0] = 4; x[0] = 5; x[2] = 7; return x;}') + jsi = JSInterpreter('function f(){var x = [1,2,3]; x[0] = 4; x[0] = 5; x[2.0] = 7; return x;}') self.assertEqual(jsi.call_function('f'), [5, 2, 7]) def test_parens(self): @@ -62,6 +68,10 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter('function f(){return (1 + 2) * 3;}') self.assertEqual(jsi.call_function('f'), 9) + def test_quotes(self): + jsi = JSInterpreter(r'function f(){return "a\"\\("}') + self.assertEqual(jsi.call_function('f'), r'a"\(') + def test_assignments(self): jsi = JSInterpreter('function f(){var x = 20; x = 30 + 1; return x;}') self.assertEqual(jsi.call_function('f'), 31) @@ -104,18 +114,29 @@ class TestJSInterpreter(unittest.TestCase): }''') self.assertEqual(jsi.call_function('x'), [20, 20, 30, 40, 50]) + def test_builtins(self): + jsi = JSInterpreter(''' + function x() { return new Date('Wednesday 31 December 1969 18:01:26 MDT') - 0; } + ''') + self.assertEqual(jsi.call_function('x'), 86000) + jsi = JSInterpreter(''' + function x(dt) { return new Date(dt) - 0; } + ''') + self.assertEqual(jsi.call_function('x', 'Wednesday 31 December 1969 18:01:26 MDT'), 86000) + def test_call(self): jsi = JSInterpreter(''' function x() { return 2; } - function y(a) { return x() + a; } + function y(a) { return x() + (a?a:0); } function z() { return y(3); } ''') self.assertEqual(jsi.call_function('z'), 5) + self.assertEqual(jsi.call_function('y'), 2) def test_for_loop(self): # function x() { a=0; for (i=0; i-10; i++) {a++} a } jsi = JSInterpreter(''' - function x() { a=0; for (i=0; i-10; i = i + 1) {a++} a } + function x() { a=0; for (i=0; i-10; i++) {a++} return a } ''') self.assertEqual(jsi.call_function('x'), 10) @@ -156,19 +177,19 @@ class TestJSInterpreter(unittest.TestCase): def test_for_loop_continue(self): jsi = JSInterpreter(''' - function x() { a=0; for (i=0; i-10; i++) { continue; a++ } a } + function x() { a=0; for (i=0; i-10; i++) { continue; a++ } return a } ''') self.assertEqual(jsi.call_function('x'), 0) def test_for_loop_break(self): jsi = JSInterpreter(''' - function x() { a=0; for (i=0; i-10; i++) { break; a++ } a } + function x() { a=0; for (i=0; i-10; i++) { break; a++ } return a } ''') self.assertEqual(jsi.call_function('x'), 0) def test_literal_list(self): jsi = JSInterpreter(''' - function x() { [1, 2, "asdf", [5, 6, 7]][3] } + function x() { return [1, 2, "asdf", [5, 6, 7]][3] } ''') self.assertEqual(jsi.call_function('x'), [5, 6, 7]) @@ -177,6 +198,22 @@ class TestJSInterpreter(unittest.TestCase): function x() { a=5; a -= 1, a+=3; return a } ''') self.assertEqual(jsi.call_function('x'), 7) + jsi = JSInterpreter(''' + function x() { a=5; return (a -= 1, a+=3, a); } + ''') + self.assertEqual(jsi.call_function('x'), 7) + + def test_void(self): + jsi = JSInterpreter(''' + function x() { return void 42; } + ''') + self.assertEqual(jsi.call_function('x'), None) + + def test_return_function(self): + jsi = JSInterpreter(''' + function x() { return [1, function(){return 1}][1] } + ''') + self.assertEqual(jsi.call_function('x')([]), 1) if __name__ == '__main__': diff --git a/test/test_utils.py b/test/test_utils.py index 259c4763e..f1a748dde 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -370,6 +370,9 @@ class TestUtil(unittest.TestCase): self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540) self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140) self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363) + self.assertEqual(unified_timestamp('December 31 1969 20:00:01 EDT'), 1) + self.assertEqual(unified_timestamp('Wednesday 31 December 1969 18:01:26 MDT'), 86) + self.assertEqual(unified_timestamp('12/31/1969 20:01:18 EDT', False), 78) def test_determine_ext(self): self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4') diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py index fc5e9828e..6e955e0f0 100644 --- a/test/test_youtube_signature.py +++ b/test/test_youtube_signature.py @@ -90,12 +90,25 @@ _NSIG_TESTS = [ 'https://www.youtube.com/s/player/e06dea74/player_ias.vflset/en_US/base.js', 'AiuodmaDDYw8d3y4bf', 'ankd8eza2T6Qmw', ), + ( + 'https://www.youtube.com/s/player/5dd88d1d/player-plasma-ias-phone-en_US.vflset/base.js', + 'kSxKFLeqzv_ZyHSAt', 'n8gS8oRlHOxPFA', + ), + ( + 'https://www.youtube.com/s/player/324f67b9/player_ias.vflset/en_US/base.js', + 'xdftNy7dh9QGnhW', '22qLGxrmX8F1rA', + ), + ( + 'https://www.youtube.com/s/player/4c3f79c5/player_ias.vflset/en_US/base.js', + 'TDCstCG66tEAO5pR9o', 'dbxNtZ14c-yWyw', + ), ] class TestPlayerInfo(unittest.TestCase): def test_youtube_extract_player_info(self): PLAYER_URLS = ( + ('https://www.youtube.com/s/player/4c3f79c5/player_ias.vflset/en_US/base.js', '4c3f79c5'), ('https://www.youtube.com/s/player/64dddad9/player_ias.vflset/en_US/base.js', '64dddad9'), ('https://www.youtube.com/s/player/64dddad9/player_ias.vflset/fr_FR/base.js', '64dddad9'), ('https://www.youtube.com/s/player/64dddad9/player-plasma-ias-phone-en_US.vflset/base.js', '64dddad9'), diff --git a/youtube_dl/compat.py b/youtube_dl/compat.py index 9f5f85dae..6d2c31a61 100644 --- a/youtube_dl/compat.py +++ b/youtube_dl/compat.py @@ -2985,7 +2985,6 @@ except ImportError: except ImportError: compat_filter = filter - try: from future_builtins import zip as compat_zip except ImportError: # not 2.6+ or is 3.x @@ -2995,6 +2994,57 @@ except ImportError: # not 2.6+ or is 3.x compat_zip = zip +# method renamed between Py2/3 +try: + from itertools import zip_longest as compat_itertools_zip_longest +except ImportError: + from itertools import izip_longest as compat_itertools_zip_longest + + +# new class in collections +try: + from collections import ChainMap as compat_collections_chain_map +except ImportError: + # Py < 3.3 + class compat_collections_chain_map(compat_collections_abc.MutableMapping): + + maps = [{}] + + def __init__(self, *maps): + self.maps = list(maps) or [{}] + + def __getitem__(self, k): + for m in self.maps: + if k in m: + return m[k] + raise KeyError(k) + + def __setitem__(self, k, v): + self.maps[0].__setitem__(k, v) + return + + def __delitem__(self, k): + if k in self.maps[0]: + del self.maps[0][k] + return + raise KeyError(k) + + def __iter__(self): + return itertools.chain(*reversed(self.maps)) + + def __len__(self): + return len(iter(self)) + + def new_child(self, m=None, **kwargs): + m = m or {} + m.update(kwargs) + return compat_collections_chain_map(m, *self.maps) + + @property + def parents(self): + return compat_collections_chain_map(*(self.maps[1:])) + + if sys.version_info < (3, 3): def compat_b64decode(s, *args, **kwargs): if isinstance(s, compat_str): @@ -3031,6 +3081,7 @@ __all__ = [ 'compat_basestring', 'compat_chr', 'compat_collections_abc', + 'compat_collections_chain_map', 'compat_cookiejar', 'compat_cookiejar_Cookie', 'compat_cookies', @@ -3051,6 +3102,7 @@ __all__ = [ 'compat_input', 'compat_integer_types', 'compat_itertools_count', + 'compat_itertools_zip_longest', 'compat_kwargs', 'compat_map', 'compat_numeric_types', diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index 8eaa911cd..c60a9b3c2 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -1,42 +1,87 @@ from __future__ import unicode_literals +import itertools import json +import math import operator import re from .utils import ( + NO_DEFAULT, ExtractorError, + js_to_json, remove_quotes, + unified_timestamp, ) from .compat import ( - compat_collections_abc, + compat_collections_chain_map as ChainMap, + compat_itertools_zip_longest as zip_longest, compat_str, ) -MutableMapping = compat_collections_abc.MutableMapping +_NAME_RE = r'[a-zA-Z_$][\w$]*' -class Nonlocal: - pass +# (op, definition) in order of binding priority, tightest first +# avoid dict to maintain order +# definition None => Defined in JSInterpreter._operator +_DOT_OPERATORS = ( + ('.', None), + # TODO: ('?.', None), +) - -_OPERATORS = [ +_OPERATORS = ( ('|', operator.or_), ('^', operator.xor), ('&', operator.and_), ('>>', operator.rshift), ('<<', operator.lshift), - ('-', operator.sub), ('+', operator.add), - ('%', operator.mod), - ('/', operator.truediv), + ('-', operator.sub), ('*', operator.mul), -] -_ASSIGN_OPERATORS = [(op + '=', opfunc) for op, opfunc in _OPERATORS] -_ASSIGN_OPERATORS.append(('=', (lambda cur, right: right))) + ('/', operator.truediv), + ('%', operator.mod), +) -_NAME_RE = r'[a-zA-Z_$][a-zA-Z_$0-9]*' +_COMP_OPERATORS = ( + ('===', operator.is_), + ('==', operator.eq), + ('!==', operator.is_not), + ('!=', operator.ne), + ('<=', operator.le), + ('>=', operator.ge), + ('<', operator.lt), + ('>', operator.gt), +) + +_LOG_OPERATORS = ( + ('&', operator.and_), + ('|', operator.or_), + ('^', operator.xor), +) + +_SC_OPERATORS = ( + ('?', None), + ('||', None), + ('&&', None), + # TODO: ('??', None), +) + +_OPERATOR_RE = '|'.join(map(lambda x: re.escape(x[0]), _OPERATORS + _LOG_OPERATORS)) _MATCHING_PARENS = dict(zip(*zip('()', '{}', '[]'))) +_QUOTES = '\'"' + + +def _ternary(cndn, if_true=True, if_false=False): + """Simulate JS's ternary operator (cndn?if_true:if_false)""" + if cndn in (False, None, 0, ''): + return if_false + try: + if math.isnan(cndn): # NB: NaN cannot be checked by membership + return if_false + except TypeError: + pass + return if_true class JS_Break(ExtractorError): @@ -49,70 +94,77 @@ class JS_Continue(ExtractorError): ExtractorError.__init__(self, 'Invalid continue') -class LocalNameSpace(MutableMapping): - def __init__(self, *stack): - self.stack = tuple(stack) - - def __getitem__(self, key): - for scope in self.stack: - if key in scope: - return scope[key] - raise KeyError(key) - +class LocalNameSpace(ChainMap): def __setitem__(self, key, value): - for scope in self.stack: + for scope in self.maps: if key in scope: scope[key] = value - break - else: - self.stack[0][key] = value - return value + return + self.maps[0][key] = value def __delitem__(self, key): raise NotImplementedError('Deleting is not supported') - def __iter__(self): - for scope in self.stack: - for scope_item in iter(scope): - yield scope_item - - def __len__(self, key): - return len(iter(self)) - def __repr__(self): - return 'LocalNameSpace%s' % (self.stack, ) + return 'LocalNameSpace%s' % (self.maps, ) class JSInterpreter(object): + __named_object_counter = 0 + def __init__(self, code, objects=None): - if objects is None: - objects = {} - self.code = code - self._functions = {} - self._objects = objects - self.__named_object_counter = 0 + self.code, self._functions = code, {} + self._objects = {} if objects is None else objects + + class Exception(ExtractorError): + def __init__(self, msg, *args, **kwargs): + expr = kwargs.pop('expr', None) + if expr is not None: + msg = '{0} in: {1!r}'.format(msg.rstrip(), expr[:100]) + super(JSInterpreter.Exception, self).__init__(msg, *args, **kwargs) def _named_object(self, namespace, obj): self.__named_object_counter += 1 - name = '__youtube_dl_jsinterp_obj%s' % (self.__named_object_counter, ) + name = '__youtube_dl_jsinterp_obj%d' % (self.__named_object_counter, ) namespace[name] = obj return name @staticmethod - def _separate(expr, delim=',', max_split=None): + def _separate(expr, delim=',', max_split=None, skip_delims=None): if not expr: return counters = {k: 0 for k in _MATCHING_PARENS.values()} - start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1 + start, splits, pos, skipping, delim_len = 0, 0, 0, 0, len(delim) - 1 + in_quote, escaping = None, False for idx, char in enumerate(expr): - if char in _MATCHING_PARENS: - counters[_MATCHING_PARENS[char]] += 1 - elif char in counters: - counters[char] -= 1 - if char != delim[pos] or any(counters.values()): - pos = 0 + if not in_quote: + if char in _MATCHING_PARENS: + counters[_MATCHING_PARENS[char]] += 1 + elif char in counters: + counters[char] -= 1 + if not escaping: + if char in _QUOTES and in_quote in (char, None): + in_quote = None if in_quote else char + else: + escaping = in_quote and char == '\\' + else: + escaping = False + + if char != delim[pos] or any(counters.values()) or in_quote: + pos = skipping = 0 continue - elif pos != delim_len: + elif skipping > 0: + skipping -= 1 + continue + elif pos == 0 and skip_delims: + here = expr[idx:] + for s in skip_delims if isinstance(skip_delims, (list, tuple)) else [skip_delims]: + if here.startswith(s) and s: + skipping = len(s) - 1 + break + if skipping > 0: + continue + if pos < delim_len: pos += 1 continue yield expr[start: idx - delim_len] @@ -122,61 +174,108 @@ class JSInterpreter(object): break yield expr[start:] - @staticmethod - def _separate_at_paren(expr, delim): - separated = list(JSInterpreter._separate(expr, delim, 1)) + @classmethod + def _separate_at_paren(cls, expr, delim): + separated = list(cls._separate(expr, delim, 1)) + if len(separated) < 2: - raise ExtractorError('No terminating paren {0} in {1}'.format(delim, expr)) + raise cls.Exception('No terminating paren {delim} in {expr}'.format(**locals())) return separated[0][1:].strip(), separated[1].strip() + @staticmethod + def _all_operators(): + return itertools.chain( + _SC_OPERATORS, _LOG_OPERATORS, _COMP_OPERATORS, _OPERATORS) + + def _operator(self, op, left_val, right_expr, expr, local_vars, allow_recursion): + if op in ('||', '&&'): + if (op == '&&') ^ _ternary(left_val): + return left_val # short circuiting + elif op == '?': + right_expr = _ternary(left_val, *self._separate(right_expr, ':', 1)) + + right_val = self.interpret_expression(right_expr, local_vars, allow_recursion) + opfunc = op and next((v for k, v in self._all_operators() if k == op), None) + if not opfunc: + return right_val + + try: + return opfunc(left_val, right_val) + except Exception as e: + raise self.Exception('Failed to evaluate {left_val!r} {op} {right_val!r}'.format(**locals()), expr, cause=e) + + def _index(self, obj, idx): + if idx == 'length': + return len(obj) + try: + return obj[int(idx)] if isinstance(obj, list) else obj[idx] + except Exception as e: + raise self.Exception('Cannot get index {idx}'.format(**locals()), expr=repr(obj), cause=e) + + def _dump(self, obj, namespace): + try: + return json.dumps(obj) + except TypeError: + return self._named_object(namespace, obj) + def interpret_statement(self, stmt, local_vars, allow_recursion=100): if allow_recursion < 0: - raise ExtractorError('Recursion limit reached') + raise self.Exception('Recursion limit reached') + allow_recursion -= 1 - sub_statements = list(self._separate(stmt, ';')) - stmt = (sub_statements or ['']).pop() + should_return = False + sub_statements = list(self._separate(stmt, ';')) or [''] + expr = stmt = sub_statements.pop().strip() for sub_stmt in sub_statements: - ret, should_abort = self.interpret_statement(sub_stmt, local_vars, allow_recursion - 1) - if should_abort: - return ret + ret, should_return = self.interpret_statement(sub_stmt, local_vars, allow_recursion) + if should_return: + return ret, should_return - should_abort = False - stmt = stmt.lstrip() - stmt_m = re.match(r'var\s', stmt) - if stmt_m: - expr = stmt[len(stmt_m.group(0)):] - else: - return_m = re.match(r'return(?:\s+|$)', stmt) - if return_m: - expr = stmt[len(return_m.group(0)):] - should_abort = True + m = re.match(r'(?P(?:var|const|let)\s)|return(?:\s+|$)', stmt) + if m: + expr = stmt[len(m.group(0)):].strip() + should_return = not m.group('var') + if not expr: + return None, should_return + + if expr[0] in _QUOTES: + inner, outer = self._separate(expr, expr[0], 1) + inner = json.loads(js_to_json(inner + expr[0])) # , strict=True)) + if not outer: + return inner, should_return + expr = self._named_object(local_vars, inner) + outer + + if expr.startswith('new '): + obj = expr[4:] + if obj.startswith('Date('): + left, right = self._separate_at_paren(obj[4:], ')') + left = self.interpret_expression(left, local_vars, allow_recursion) + expr = unified_timestamp(left, False) + if not expr: + raise self.Exception('Failed to parse date {left!r}'.format(**locals()), expr=expr) + expr = self._dump(int(expr * 1000), local_vars) + right else: - # Try interpreting it as an expression - expr = stmt + raise self.Exception('Unsupported object {obj}'.format(**locals()), expr=expr) - v = self.interpret_expression(expr, local_vars, allow_recursion) - return v, should_abort - - def interpret_expression(self, expr, local_vars, allow_recursion): - expr = expr.strip() - if expr == '': # Empty expression - return None + if expr.startswith('void '): + left = self.interpret_expression(expr[5:], local_vars, allow_recursion) + return None, should_return if expr.startswith('{'): inner, outer = self._separate_at_paren(expr, '}') - inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion - 1) + inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion) if not outer or should_abort: - return inner + return inner, should_abort or should_return else: - expr = json.dumps(inner) + outer + expr = self._dump(inner, local_vars) + outer if expr.startswith('('): inner, outer = self._separate_at_paren(expr, ')') - inner = self.interpret_expression(inner, local_vars, allow_recursion) - if not outer: - return inner + inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion) + if not outer or should_abort: + return inner, should_abort or should_return else: - expr = json.dumps(inner) + outer + expr = self._dump(inner, local_vars) + outer if expr.startswith('['): inner, outer = self._separate_at_paren(expr, ']') @@ -185,57 +284,53 @@ class JSInterpreter(object): for item in self._separate(inner)]) expr = name + outer - m = re.match(r'try\s*', expr) - if m: + m = re.match(r'(?Ptry|finally)\s*|(?:(?Pcatch)|(?Pfor)|(?Pswitch))\s*\(', expr) + md = m.groupdict() if m else {} + if md.get('try'): if expr[m.end()] == '{': try_expr, expr = self._separate_at_paren(expr[m.end():], '}') else: try_expr, expr = expr[m.end() - 1:], '' - ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion - 1) + ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion) if should_abort: - return ret - return self.interpret_statement(expr, local_vars, allow_recursion - 1)[0] + return ret, True + ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) + return ret, should_abort or should_return - m = re.match(r'(?:(?Pcatch)|(?Pfor)|(?Pswitch))\s*\(', expr) - md = m.groupdict() if m else {} - if md.get('catch'): + elif md.get('catch'): # We ignore the catch block _, expr = self._separate_at_paren(expr, '}') - return self.interpret_statement(expr, local_vars, allow_recursion - 1)[0] + ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) + return ret, should_abort or should_return elif md.get('for'): - def raise_constructor_error(c): - raise ExtractorError( - 'Premature return in the initialization of a for loop in {0!r}'.format(c)) - constructor, remaining = self._separate_at_paren(expr[m.end() - 1:], ')') if remaining.startswith('{'): body, expr = self._separate_at_paren(remaining, '}') else: - m = re.match(r'switch\s*\(', remaining) # FIXME - if m: - switch_val, remaining = self._separate_at_paren(remaining[m.end() - 1:], ')') + switch_m = re.match(r'switch\s*\(', remaining) # FIXME + if switch_m: + switch_val, remaining = self._separate_at_paren(remaining[switch_m.end() - 1:], ')') body, expr = self._separate_at_paren(remaining, '}') body = 'switch(%s){%s}' % (switch_val, body) else: body, expr = remaining, '' start, cndn, increment = self._separate(constructor, ';') - if self.interpret_statement(start, local_vars, allow_recursion - 1)[1]: - raise_constructor_error(constructor) + self.interpret_expression(start, local_vars, allow_recursion) while True: - if not self.interpret_expression(cndn, local_vars, allow_recursion): + if not _ternary(self.interpret_expression(cndn, local_vars, allow_recursion)): break try: - ret, should_abort = self.interpret_statement(body, local_vars, allow_recursion - 1) + ret, should_abort = self.interpret_statement(body, local_vars, allow_recursion) if should_abort: - return ret + return ret, True except JS_Break: break except JS_Continue: pass - if self.interpret_statement(increment, local_vars, allow_recursion - 1)[1]: - raise_constructor_error(constructor) - return self.interpret_statement(expr, local_vars, allow_recursion - 1)[0] + self.interpret_expression(increment, local_vars, allow_recursion) + ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) + return ret, should_abort or should_return elif md.get('switch'): switch_val, remaining = self._separate_at_paren(expr[m.end() - 1:], ')') @@ -245,7 +340,7 @@ class JSInterpreter(object): for default in (False, True): matched = False for item in items: - case, stmt = [i.strip() for i in self._separate(item, ':', 1)] + case, stmt = (i.strip() for i in self._separate(item, ':', 1)) if default: matched = matched or case == 'default' elif not matched: @@ -254,24 +349,28 @@ class JSInterpreter(object): if not matched: continue try: - ret, should_abort = self.interpret_statement(stmt, local_vars, allow_recursion - 1) + ret, should_abort = self.interpret_statement(stmt, local_vars, allow_recursion) if should_abort: return ret except JS_Break: break if matched: break - return self.interpret_statement(expr, local_vars, allow_recursion - 1)[0] + ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) + return ret, should_abort or should_return # Comma separated statements sub_expressions = list(self._separate(expr)) - expr = sub_expressions.pop().strip() if sub_expressions else '' - for sub_expr in sub_expressions: - self.interpret_expression(sub_expr, local_vars, allow_recursion) + if len(sub_expressions) > 1: + for sub_expr in sub_expressions: + ret, should_abort = self.interpret_statement(sub_expr, local_vars, allow_recursion) + if should_abort: + return ret, True + return ret, False for m in re.finditer(r'''(?x) - (?P\+\+|--)(?P%(_NAME_RE)s)| - (?P%(_NAME_RE)s)(?P\+\+|--)''' % globals(), expr): + (?P\+\+|--)(?P{_NAME_RE})| + (?P{_NAME_RE})(?P\+\+|--)'''.format(**globals()), expr): var = m.group('var1') or m.group('var2') start, end = m.span() sign = m.group('pre_sign') or m.group('post_sign') @@ -279,85 +378,87 @@ class JSInterpreter(object): local_vars[var] += 1 if sign[0] == '+' else -1 if m.group('pre_sign'): ret = local_vars[var] - expr = expr[:start] + json.dumps(ret) + expr[end:] + expr = expr[:start] + self._dump(ret, local_vars) + expr[end:] - for op, opfunc in _ASSIGN_OPERATORS: - m = re.match(r'''(?x) - (?P%s)(?:\[(?P[^\]]+?)\])? - \s*%s - (?P.*)$''' % (_NAME_RE, re.escape(op)), expr) - if not m: - continue - right_val = self.interpret_expression(m.group('expr'), local_vars, allow_recursion) + if not expr: + return None, should_return - if m.groupdict().get('index'): - lvar = local_vars[m.group('out')] - idx = self.interpret_expression(m.group('index'), local_vars, allow_recursion) - if not isinstance(idx, int): - raise ExtractorError('List indices must be integers: %s' % (idx, )) - cur = lvar[idx] - val = opfunc(cur, right_val) - lvar[idx] = val - return val - else: - cur = local_vars.get(m.group('out')) - val = opfunc(cur, right_val) - local_vars[m.group('out')] = val - return val + m = re.match(r'''(?x) + (?P + (?P{_NAME_RE})(?:\[(?P[^\]]+?)\])?\s* + (?P{_OPERATOR_RE})? + =(?P.*)$ + )|(?P + (?!if|return|true|false|null|undefined)(?P{_NAME_RE})$ + )|(?P + (?P{_NAME_RE})\[(?P.+)\]$ + )|(?P + (?P{_NAME_RE})(?:\.(?P[^(]+)|\[(?P[^\]]+)\])\s* + )|(?P + (?P{_NAME_RE})\((?P.*)\)$ + )'''.format(**globals()), expr) + md = m.groupdict() if m else {} + if md.get('assign'): + left_val = local_vars.get(m.group('out')) - if expr.isdigit(): - return int(expr) + if not m.group('index'): + local_vars[m.group('out')] = self._operator( + m.group('op'), left_val, m.group('expr'), expr, local_vars, allow_recursion) + return local_vars[m.group('out')], should_return + elif left_val is None: + raise self.Exception('Cannot index undefined variable ' + m.group('out'), expr=expr) - if expr == 'break': + idx = self.interpret_expression(m.group('index'), local_vars, allow_recursion) + if not isinstance(idx, (int, float)): + raise self.Exception('List index %s must be integer' % (idx, ), expr=expr) + idx = int(idx) + left_val[idx] = self._operator( + m.group('op'), left_val[idx], m.group('expr'), expr, local_vars, allow_recursion) + return left_val[idx], should_return + + elif expr.isdigit(): + return int(expr), should_return + + elif expr == 'break': raise JS_Break() elif expr == 'continue': raise JS_Continue() - var_m = re.match( - r'(?!if|return|true|false|null)(?P%s)$' % _NAME_RE, - expr) - if var_m: - return local_vars[var_m.group('name')] + elif md.get('return'): + return local_vars[m.group('name')], should_return try: - return json.loads(expr) + ret = json.loads(js_to_json(expr)) # strict=True) + if not md.get('attribute'): + return ret, should_return except ValueError: pass - m = re.match( - r'(?P%s)\[(?P.+)\]$' % _NAME_RE, expr) - if m: + if md.get('indexing'): val = local_vars[m.group('in')] idx = self.interpret_expression(m.group('idx'), local_vars, allow_recursion) - return val[idx] + return self._index(val, idx), should_return - def raise_expr_error(where, op, exp): - raise ExtractorError('Premature {0} return of {1} in {2!r}'.format(where, op, exp)) - - for op, opfunc in _OPERATORS: - separated = list(self._separate(expr, op)) + for op, _ in self._all_operators(): + # hackety: have higher priority than <>, but don't confuse them + skip_delim = (op + op) if op in ('<', '>') else None + separated = list(self._separate(expr, op, skip_delims=skip_delim)) if len(separated) < 2: continue - right_val = separated.pop() - left_val = op.join(separated) - left_val, should_abort = self.interpret_statement( - left_val, local_vars, allow_recursion - 1) - if should_abort: - raise_expr_error('left-side', op, expr) - right_val, should_abort = self.interpret_statement( - right_val, local_vars, allow_recursion - 1) - if should_abort: - raise_expr_error('right-side', op, expr) - return opfunc(left_val or 0, right_val) - m = re.match( - r'(?P%s)(?:\.(?P[^(]+)|\[(?P[^]]+)\])\s*' % _NAME_RE, - expr) - if m: + right_expr = separated.pop() + while op == '-' and len(separated) > 1 and not separated[-1].strip(): + right_expr = '-' + right_expr + separated.pop() + left_val = self.interpret_expression(op.join(separated), local_vars, allow_recursion) + return self._operator(op, 0 if left_val is None else left_val, + right_expr, expr, local_vars, allow_recursion), should_return + + if md.get('attribute'): variable = m.group('var') - nl = Nonlocal() - - nl.member = remove_quotes(m.group('member') or m.group('member2')) + member = m.group('member') + if not member: + member = self.interpret_expression(m.group('member2'), local_vars, allow_recursion) arg_str = expr[m.end():] if arg_str.startswith('('): arg_str, remaining = self._separate_at_paren(arg_str, ')') @@ -367,25 +468,24 @@ class JSInterpreter(object): def assertion(cndn, msg): """ assert, but without risk of getting optimized out """ if not cndn: - raise ExtractorError('{0} {1}: {2}'.format(nl.member, msg, expr)) + raise ExtractorError('{member} {msg}'.format(**locals()), expr=expr) def eval_method(): - # nonlocal member - member = nl.member - if variable == 'String': - obj = compat_str - elif variable in local_vars: - obj = local_vars[variable] - else: + if (variable, member) == ('console', 'debug'): + return + types = { + 'String': compat_str, + 'Math': float, + } + obj = local_vars.get(variable, types.get(variable, NO_DEFAULT)) + if obj is NO_DEFAULT: if variable not in self._objects: self._objects[variable] = self.extract_object(variable) obj = self._objects[variable] + # Member access if arg_str is None: - # Member access - if member == 'length': - return len(obj) - return obj[member] + return self._index(obj, member) # Function call argvals = [ @@ -396,12 +496,17 @@ class JSInterpreter(object): if member == 'fromCharCode': assertion(argvals, 'takes one or more arguments') return ''.join(map(chr, argvals)) - raise ExtractorError('Unsupported string method %s' % (member, )) + raise self.Exception('Unsupported string method ' + member, expr=expr) + elif obj == float: + if member == 'pow': + assertion(len(argvals) == 2, 'takes two arguments') + return argvals[0] ** argvals[1] + raise self.Exception('Unsupported Math method ' + member, expr=expr) if member == 'split': assertion(argvals, 'takes one or more arguments') - assertion(argvals == [''], 'with arguments is not implemented') - return list(obj) + assertion(len(argvals) == 1, 'with limit argument is not implemented') + return obj.split(argvals[0]) if argvals[0] else list(obj) elif member == 'join': assertion(isinstance(obj, list), 'must be applied on a list') assertion(len(argvals) == 1, 'takes exactly one argument') @@ -447,7 +552,7 @@ class JSInterpreter(object): assertion(argvals, 'takes one or more arguments') assertion(len(argvals) <= 2, 'takes at-most 2 arguments') f, this = (argvals + [''])[:2] - return [f((item, idx, obj), this=this) for idx, item in enumerate(obj)] + return [f((item, idx, obj), {'this': this}, allow_recursion) for idx, item in enumerate(obj)] elif member == 'indexOf': assertion(argvals, 'takes one or more arguments') assertion(len(argvals) <= 2, 'takes at-most 2 arguments') @@ -457,32 +562,35 @@ class JSInterpreter(object): except ValueError: return -1 - if isinstance(obj, list): - member = int(member) - nl.member = member - return obj[member](argvals) + idx = int(member) if isinstance(obj, list) else member + return obj[idx](argvals, allow_recursion=allow_recursion) if remaining: - return self.interpret_expression( + ret, should_abort = self.interpret_statement( self._named_object(local_vars, eval_method()) + remaining, local_vars, allow_recursion) + return ret, should_return or should_abort else: - return eval_method() + return eval_method(), should_return - m = re.match(r'^(?P%s)\((?P[a-zA-Z0-9_$,]*)\)$' % _NAME_RE, expr) - if m: - fname = m.group('func') - argvals = tuple([ - int(v) if v.isdigit() else local_vars[v] - for v in self._separate(m.group('args'))]) + elif md.get('function'): + fname = m.group('fname') + argvals = [self.interpret_expression(v, local_vars, allow_recursion) + for v in self._separate(m.group('args'))] if fname in local_vars: - return local_vars[fname](argvals) + return local_vars[fname](argvals, allow_recursion=allow_recursion), should_return elif fname not in self._functions: self._functions[fname] = self.extract_function(fname) - return self._functions[fname](argvals) + return self._functions[fname](argvals, allow_recursion=allow_recursion), should_return - if expr: - raise ExtractorError('Unsupported JS expression %r' % expr) + raise self.Exception( + 'Unsupported JS expression ' + (expr[:40] if expr != stmt else ''), expr=stmt) + + def interpret_expression(self, expr, local_vars, allow_recursion): + ret, should_return = self.interpret_statement(expr, local_vars, allow_recursion) + if should_return: + raise self.Exception('Cannot return from an expression', expr) + return ret def extract_object(self, objname): _FUNC_NAME_RE = r'''(?:[a-zA-Z$0-9]+|"[a-zA-Z$0-9]+"|'[a-zA-Z$0-9]+')''' @@ -494,15 +602,17 @@ class JSInterpreter(object): }\s*; ''' % (re.escape(objname), _FUNC_NAME_RE), self.code) + if not obj_m: + raise self.Exception('Could not find object ' + objname) fields = obj_m.group('fields') # Currently, it only supports function definitions fields_m = re.finditer( r'''(?x) - (?P%s)\s*:\s*function\s*\((?P[a-z,]+)\){(?P[^}]+)} - ''' % _FUNC_NAME_RE, + (?P%s)\s*:\s*function\s*\((?P(?:%s|,)*)\){(?P[^}]+)} + ''' % (_FUNC_NAME_RE, _NAME_RE), fields) for f in fields_m: - argnames = f.group('args').split(',') + argnames = self.build_arglist(f.group('args')) obj[remove_quotes(f.group('key'))] = self.build_function(argnames, f.group('code')) return obj @@ -510,15 +620,19 @@ class JSInterpreter(object): def extract_function_code(self, funcname): """ @returns argnames, code """ func_m = re.search( - r'''(?x) - (?:function\s+%(f_n)s|[{;,]\s*%(f_n)s\s*=\s*function|var\s+%(f_n)s\s*=\s*function)\s* + r'''(?xs) + (?: + function\s+%(name)s| + [{;,]\s*%(name)s\s*=\s*function| + (?:var|const|let)\s+%(name)s\s*=\s*function + )\s* \((?P[^)]*)\)\s* - (?P\{(?:(?!};)[^"]|"([^"]|\\")*")+\})''' % {'f_n': re.escape(funcname), }, + (?P{.+})''' % {'name': re.escape(funcname)}, self.code) code, _ = self._separate_at_paren(func_m.group('code'), '}') # refine the match if func_m is None: - raise ExtractorError('Could not find JS function %r' % funcname) - return func_m.group('args').split(','), code + raise self.Exception('Could not find JS function "{funcname}"'.format(**locals())) + return self.build_arglist(func_m.group('args')), code def extract_function(self, funcname): return self.extract_function_from_code(*self.extract_function_code(funcname)) @@ -534,7 +648,7 @@ class JSInterpreter(object): name = self._named_object( local_vars, self.extract_function_from_code( - [x.strip() for x in mobj.group('args').split(',')], + self.build_arglist(mobj.group('args')), body, local_vars, *global_stack)) code = code[:start] + name + remaining return self.build_function(argnames, code, local_vars, *global_stack) @@ -542,17 +656,22 @@ class JSInterpreter(object): def call_function(self, funcname, *args): return self.extract_function(funcname)(args) + @classmethod + def build_arglist(cls, arg_text): + if not arg_text: + return [] + return list(filter(None, (x.strip() or None for x in cls._separate(arg_text)))) + def build_function(self, argnames, code, *global_stack): global_stack = list(global_stack) or [{}] - local_vars = global_stack.pop(0) + argnames = tuple(argnames) - def resf(args, **kwargs): - local_vars.update(dict(zip(argnames, args))) - local_vars.update(kwargs) - var_stack = LocalNameSpace(local_vars, *global_stack) - for stmt in self._separate(code.replace('\n', ''), ';'): - ret, should_abort = self.interpret_statement(stmt, var_stack) - if should_abort: - break - return ret + def resf(args, kwargs={}, allow_recursion=100): + global_stack[0].update( + zip_longest(argnames, args, fillvalue=None)) + global_stack[0].update(kwargs) + var_stack = LocalNameSpace(*global_stack) + ret, should_abort = self.interpret_statement(code.replace('\n', ''), var_stack, allow_recursion - 1) + if should_abort: + return ret return resf diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 4e00317f1..a5f584ec5 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -1696,6 +1696,17 @@ MONTH_NAMES = { 'juillet', 'août', 'septembre', 'octobre', 'novembre', 'décembre'], } +# Timezone names for RFC2822 obs-zone +# From https://github.com/python/cpython/blob/3.11/Lib/email/_parseaddr.py#L36-L42 +TIMEZONE_NAMES = { + 'UT': 0, 'UTC': 0, 'GMT': 0, 'Z': 0, + 'AST': -4, 'ADT': -3, # Atlantic (used in Canada) + 'EST': -5, 'EDT': -4, # Eastern + 'CST': -6, 'CDT': -5, # Central + 'MST': -7, 'MDT': -6, # Mountain + 'PST': -8, 'PDT': -7 # Pacific +} + KNOWN_EXTENSIONS = ( 'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'aac', 'flv', 'f4v', 'f4a', 'f4b', @@ -1735,12 +1746,17 @@ DATE_FORMATS = ( '%b %dth %Y %I:%M', '%Y %m %d', '%Y-%m-%d', + '%Y.%m.%d.', '%Y/%m/%d', '%Y/%m/%d %H:%M', '%Y/%m/%d %H:%M:%S', + '%Y%m%d%H%M', + '%Y%m%d%H%M%S', + '%Y%m%d', '%Y-%m-%d %H:%M', '%Y-%m-%d %H:%M:%S', '%Y-%m-%d %H:%M:%S.%f', + '%Y-%m-%d %H:%M:%S:%f', '%d.%m.%Y %H:%M', '%d.%m.%Y %H.%M', '%Y-%m-%dT%H:%M:%SZ', @@ -1753,6 +1769,7 @@ DATE_FORMATS = ( '%b %d %Y at %H:%M:%S', '%B %d %Y at %H:%M', '%B %d %Y at %H:%M:%S', + '%H:%M %d-%b-%Y', ) DATE_FORMATS_DAY_FIRST = list(DATE_FORMATS) @@ -1763,6 +1780,7 @@ DATE_FORMATS_DAY_FIRST.extend([ '%d/%m/%Y', '%d/%m/%y', '%d/%m/%Y %H:%M:%S', + '%d-%m-%Y %H:%M', ]) DATE_FORMATS_MONTH_FIRST = list(DATE_FORMATS) @@ -2966,10 +2984,22 @@ class YoutubeDLRedirectHandler(compat_urllib_request.HTTPRedirectHandler): def extract_timezone(date_str): m = re.search( - r'^.{8,}?(?PZ$| ?(?P\+|-)(?P[0-9]{2}):?(?P[0-9]{2})$)', - date_str) + r'''(?x) + ^.{8,}? # >=8 char non-TZ prefix, if present + (?PZ| # just the UTC Z, or + (?:(?<=.\b\d{4}|\b\d{2}:\d\d)| # preceded by 4 digits or hh:mm or + (?= 4 alpha or 2 digits + [ ]? # optional space + (?P\+|-) # +/- + (?P[0-9]{2}):?(?P[0-9]{2}) # hh[:]mm + $) + ''', date_str) if not m: - timezone = datetime.timedelta() + m = re.search(r'\d{1,2}:\d{1,2}(?:\.\d+)?(?P\s*[A-Z]+)$', date_str) + timezone = TIMEZONE_NAMES.get(m and m.group('tz').strip()) + if timezone is not None: + date_str = date_str[:-len(m.group('tz'))] + timezone = datetime.timedelta(hours=timezone or 0) else: date_str = date_str[:-len(m.group('tz'))] if not m.group('sign'): @@ -3037,7 +3067,8 @@ def unified_timestamp(date_str, day_first=True): if date_str is None: return None - date_str = re.sub(r'[,|]', '', date_str) + date_str = re.sub(r'\s+', ' ', re.sub( + r'(?i)[,|]|(mon|tues?|wed(nes)?|thu(rs)?|fri|sat(ur)?)(day)?', '', date_str)) pm_delta = 12 if re.search(r'(?i)PM', date_str) else 0 timezone, date_str = extract_timezone(date_str) @@ -3063,7 +3094,7 @@ def unified_timestamp(date_str, day_first=True): pass timetuple = email.utils.parsedate_tz(date_str) if timetuple: - return calendar.timegm(timetuple) + pm_delta * 3600 + return calendar.timegm(timetuple) + pm_delta * 3600 - timezone.total_seconds() def determine_ext(url, default_ext='unknown_video'): @@ -3673,13 +3704,11 @@ def int_or_none(v, scale=1, default=None, get_attr=None, invscale=1): if get_attr: if v is not None: v = getattr(v, get_attr, None) - if v == '': - v = None - if v is None: + if v in (None, ''): return default try: return int(v) * invscale // scale - except (ValueError, TypeError): + except (ValueError, TypeError, OverflowError): return default From 6b98eb46fcb5e5a505a2bbe0460b49b1cfc84c89 Mon Sep 17 00:00:00 2001 From: dirkf Date: Mon, 15 Aug 2022 16:45:04 +0100 Subject: [PATCH 05/21] [postprocessor] Don't replace existing value with null metadata parsed from title --- youtube_dl/postprocessor/metadatafromtitle.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/youtube_dl/postprocessor/metadatafromtitle.py b/youtube_dl/postprocessor/metadatafromtitle.py index f5c14d974..6cd5bb70f 100644 --- a/youtube_dl/postprocessor/metadatafromtitle.py +++ b/youtube_dl/postprocessor/metadatafromtitle.py @@ -40,6 +40,8 @@ class MetadataFromTitlePP(PostProcessor): % self._titleformat) return [], info for attribute, value in match.groupdict().items(): + if value is None: + continue info[attribute] = value self._downloader.to_screen( '[fromtitle] parsed %s: %s' From 1862bf568bcc1f4f18e19e46838c641066961f96 Mon Sep 17 00:00:00 2001 From: dirkf Date: Wed, 17 Aug 2022 14:22:02 +0100 Subject: [PATCH 06/21] [jsinterp] Improve JS language support (#31175) * operator ?? * operator ?. * operator ** * accurate operator functions * `undefined` handling * object literals {a: 1, "b": expr} * more tests for weird JS comparisons: see https://github.com/ytdl-org/youtube-dl/issues/31173#issuecomment-1217854397. --- test/test_jsinterp.py | 114 ++++++++++++++++++++ test/test_youtube_signature.py | 4 + youtube_dl/jsinterp.py | 189 ++++++++++++++++++++++++++------- 3 files changed, 267 insertions(+), 40 deletions(-) diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py index c6c931743..328941e09 100644 --- a/test/test_jsinterp.py +++ b/test/test_jsinterp.py @@ -8,7 +8,10 @@ import sys import unittest sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) +import math + from youtube_dl.jsinterp import JSInterpreter +undefined = JSInterpreter.undefined class TestJSInterpreter(unittest.TestCase): @@ -48,6 +51,9 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter('function f(){return 1 << 5;}') self.assertEqual(jsi.call_function('f'), 32) + jsi = JSInterpreter('function f(){return 2 ** 5}') + self.assertEqual(jsi.call_function('f'), 32) + jsi = JSInterpreter('function f(){return 19 & 21;}') self.assertEqual(jsi.call_function('f'), 17) @@ -57,6 +63,15 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter('function f(){return []? 2+3: 4;}') self.assertEqual(jsi.call_function('f'), 5) + jsi = JSInterpreter('function f(){return 1 == 2}') + self.assertEqual(jsi.call_function('f'), False) + + jsi = JSInterpreter('function f(){return 0 && 1 || 2;}') + self.assertEqual(jsi.call_function('f'), 2) + + jsi = JSInterpreter('function f(){return 0 ?? 42;}') + self.assertEqual(jsi.call_function('f'), 0) + def test_array_access(self): jsi = JSInterpreter('function f(){var x = [1,2,3]; x[0] = 4; x[0] = 5; x[2.0] = 7; return x;}') self.assertEqual(jsi.call_function('f'), [5, 2, 7]) @@ -203,6 +218,11 @@ class TestJSInterpreter(unittest.TestCase): ''') self.assertEqual(jsi.call_function('x'), 7) + jsi = JSInterpreter(''' + function x() { return (l=[0,1,2,3], function(a, b){return a+b})((l[1], l[2]), l[3]) } + ''') + self.assertEqual(jsi.call_function('x'), 5) + def test_void(self): jsi = JSInterpreter(''' function x() { return void 42; } @@ -215,6 +235,100 @@ class TestJSInterpreter(unittest.TestCase): ''') self.assertEqual(jsi.call_function('x')([]), 1) + def test_null(self): + jsi = JSInterpreter(''' + function x() { return null; } + ''') + self.assertIs(jsi.call_function('x'), None) + + jsi = JSInterpreter(''' + function x() { return [null > 0, null < 0, null == 0, null === 0]; } + ''') + self.assertEqual(jsi.call_function('x'), [False, False, False, False]) + + jsi = JSInterpreter(''' + function x() { return [null >= 0, null <= 0]; } + ''') + self.assertEqual(jsi.call_function('x'), [True, True]) + + def test_undefined(self): + jsi = JSInterpreter(''' + function x() { return undefined === undefined; } + ''') + self.assertTrue(jsi.call_function('x')) + + jsi = JSInterpreter(''' + function x() { return undefined; } + ''') + self.assertIs(jsi.call_function('x'), undefined) + + jsi = JSInterpreter(''' + function x() { let v; return v; } + ''') + self.assertIs(jsi.call_function('x'), undefined) + + jsi = JSInterpreter(''' + function x() { return [undefined === undefined, undefined == undefined, undefined < undefined, undefined > undefined]; } + ''') + self.assertEqual(jsi.call_function('x'), [True, True, False, False]) + + jsi = JSInterpreter(''' + function x() { return [undefined === 0, undefined == 0, undefined < 0, undefined > 0]; } + ''') + self.assertEqual(jsi.call_function('x'), [False, False, False, False]) + + jsi = JSInterpreter(''' + function x() { return [undefined >= 0, undefined <= 0]; } + ''') + self.assertEqual(jsi.call_function('x'), [False, False]) + + jsi = JSInterpreter(''' + function x() { return [undefined > null, undefined < null, undefined == null, undefined === null]; } + ''') + self.assertEqual(jsi.call_function('x'), [False, False, True, False]) + + jsi = JSInterpreter(''' + function x() { return [undefined === null, undefined == null, undefined < null, undefined > null]; } + ''') + self.assertEqual(jsi.call_function('x'), [False, True, False, False]) + + jsi = JSInterpreter(''' + function x() { let v; return [42+v, v+42, v**42, 42**v, 0**v]; } + ''') + for y in jsi.call_function('x'): + self.assertTrue(math.isnan(y)) + + jsi = JSInterpreter(''' + function x() { let v; return v**0; } + ''') + self.assertEqual(jsi.call_function('x'), 1) + + jsi = JSInterpreter(''' + function x() { let v; return [v>42, v<=42, v&&42, 42&&v]; } + ''') + self.assertEqual(jsi.call_function('x'), [False, False, undefined, undefined]) + + jsi = JSInterpreter('function x(){return undefined ?? 42; }') + self.assertEqual(jsi.call_function('x'), 42) + + def test_object(self): + jsi = JSInterpreter(''' + function x() { return {}; } + ''') + self.assertEqual(jsi.call_function('x'), {}) + jsi = JSInterpreter(''' + function x() { let a = {m1: 42, m2: 0 }; return [a["m1"], a.m2]; } + ''') + self.assertEqual(jsi.call_function('x'), [42, 0]) + jsi = JSInterpreter(''' + function x() { let a; return a?.qq; } + ''') + self.assertIs(jsi.call_function('x'), undefined) + jsi = JSInterpreter(''' + function x() { let a = {m1: 42, m2: 0 }; return a?.qq; } + ''') + self.assertIs(jsi.call_function('x'), undefined) + if __name__ == '__main__': unittest.main() diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py index 6e955e0f0..4d756dad3 100644 --- a/test/test_youtube_signature.py +++ b/test/test_youtube_signature.py @@ -102,6 +102,10 @@ _NSIG_TESTS = [ 'https://www.youtube.com/s/player/4c3f79c5/player_ias.vflset/en_US/base.js', 'TDCstCG66tEAO5pR9o', 'dbxNtZ14c-yWyw', ), + ( + 'https://www.youtube.com/s/player/c81bbb4a/player_ias.vflset/en_US/base.js', + 'gre3EcLurNY2vqp94', 'Z9DfGxWP115WTg', + ), ] diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index c60a9b3c2..8e119d08a 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -7,7 +7,6 @@ import operator import re from .utils import ( - NO_DEFAULT, ExtractorError, js_to_json, remove_quotes, @@ -21,6 +20,70 @@ from .compat import ( _NAME_RE = r'[a-zA-Z_$][\w$]*' +_UNDEFINED = object() + + +def _js_bit_op(op): + + def wrapped(a, b): + def zeroise(x): + return 0 if x in (None, _UNDEFINED) else x + return op(zeroise(a), zeroise(b)) + + return wrapped + + +def _js_arith_op(op): + + def wrapped(a, b): + if _UNDEFINED in (a, b): + return float('nan') + return op(a or 0, b or 0) + + return wrapped + + +def _js_div(a, b): + if _UNDEFINED in (a, b) or not (a and b): + return float('nan') + return float('inf') if not b else operator.truediv(a or 0, b) + + +def _js_mod(a, b): + if _UNDEFINED in (a, b) or not b: + return float('nan') + return (a or 0) % b + + +def _js_exp(a, b): + if not b: + # even 0 ** 0 !! + return 1 + if _UNDEFINED in (a, b): + return float('nan') + return (a or 0) ** b + + +def _js_eq_op(op): + + def wrapped(a, b): + if set((a, b)) <= set((None, _UNDEFINED)): + return op(a, a) + return op(a, b) + + return wrapped + + +def _js_comp_op(op): + + def wrapped(a, b): + if _UNDEFINED in (a, b): + return False + return op(a or 0, b or 0) + + return wrapped + + # (op, definition) in order of binding priority, tightest first # avoid dict to maintain order # definition None => Defined in JSInterpreter._operator @@ -30,40 +93,38 @@ _DOT_OPERATORS = ( ) _OPERATORS = ( - ('|', operator.or_), - ('^', operator.xor), - ('&', operator.and_), - ('>>', operator.rshift), - ('<<', operator.lshift), - ('+', operator.add), - ('-', operator.sub), - ('*', operator.mul), - ('/', operator.truediv), - ('%', operator.mod), + ('>>', _js_bit_op(operator.rshift)), + ('<<', _js_bit_op(operator.lshift)), + ('+', _js_arith_op(operator.add)), + ('-', _js_arith_op(operator.sub)), + ('*', _js_arith_op(operator.mul)), + ('/', _js_div), + ('%', _js_mod), + ('**', _js_exp), ) _COMP_OPERATORS = ( ('===', operator.is_), - ('==', operator.eq), + ('==', _js_eq_op(operator.eq)), ('!==', operator.is_not), - ('!=', operator.ne), - ('<=', operator.le), - ('>=', operator.ge), - ('<', operator.lt), - ('>', operator.gt), + ('!=', _js_eq_op(operator.ne)), + ('<=', _js_comp_op(operator.le)), + ('>=', _js_comp_op(operator.ge)), + ('<', _js_comp_op(operator.lt)), + ('>', _js_comp_op(operator.gt)), ) _LOG_OPERATORS = ( - ('&', operator.and_), - ('|', operator.or_), - ('^', operator.xor), + ('|', _js_bit_op(operator.or_)), + ('^', _js_bit_op(operator.xor)), + ('&', _js_bit_op(operator.and_)), ) _SC_OPERATORS = ( ('?', None), + ('??', None), ('||', None), ('&&', None), - # TODO: ('??', None), ) _OPERATOR_RE = '|'.join(map(lambda x: re.escape(x[0]), _OPERATORS + _LOG_OPERATORS)) @@ -74,7 +135,7 @@ _QUOTES = '\'"' def _ternary(cndn, if_true=True, if_false=False): """Simulate JS's ternary operator (cndn?if_true:if_false)""" - if cndn in (False, None, 0, ''): + if cndn in (False, None, 0, '', _UNDEFINED): return if_false try: if math.isnan(cndn): # NB: NaN cannot be checked by membership @@ -95,6 +156,12 @@ class JS_Continue(ExtractorError): class LocalNameSpace(ChainMap): + def __getitem__(self, key): + try: + return super(LocalNameSpace, self).__getitem__(key) + except KeyError: + return _UNDEFINED + def __setitem__(self, key, value): for scope in self.maps: if key in scope: @@ -105,6 +172,13 @@ class LocalNameSpace(ChainMap): def __delitem__(self, key): raise NotImplementedError('Deleting is not supported') + def __contains__(self, key): + try: + super(LocalNameSpace, self).__getitem__(key) + return True + except KeyError: + return False + def __repr__(self): return 'LocalNameSpace%s' % (self.maps, ) @@ -112,6 +186,8 @@ class LocalNameSpace(ChainMap): class JSInterpreter(object): __named_object_counter = 0 + undefined = _UNDEFINED + def __init__(self, code, objects=None): self.code, self._functions = code, {} self._objects = {} if objects is None else objects @@ -185,12 +261,16 @@ class JSInterpreter(object): @staticmethod def _all_operators(): return itertools.chain( + # Ref: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Operator_Precedence _SC_OPERATORS, _LOG_OPERATORS, _COMP_OPERATORS, _OPERATORS) def _operator(self, op, left_val, right_expr, expr, local_vars, allow_recursion): if op in ('||', '&&'): if (op == '&&') ^ _ternary(left_val): return left_val # short circuiting + elif op == '??': + if left_val not in (None, self.undefined): + return left_val elif op == '?': right_expr = _ternary(left_val, *self._separate(right_expr, ':', 1)) @@ -204,12 +284,14 @@ class JSInterpreter(object): except Exception as e: raise self.Exception('Failed to evaluate {left_val!r} {op} {right_val!r}'.format(**locals()), expr, cause=e) - def _index(self, obj, idx): + def _index(self, obj, idx, allow_undefined=False): if idx == 'length': return len(obj) try: return obj[int(idx)] if isinstance(obj, list) else obj[idx] except Exception as e: + if allow_undefined: + return self.undefined raise self.Exception('Cannot get index {idx}'.format(**locals()), expr=repr(obj), cause=e) def _dump(self, obj, namespace): @@ -249,8 +331,8 @@ class JSInterpreter(object): obj = expr[4:] if obj.startswith('Date('): left, right = self._separate_at_paren(obj[4:], ')') - left = self.interpret_expression(left, local_vars, allow_recursion) - expr = unified_timestamp(left, False) + expr = unified_timestamp( + self.interpret_expression(left, local_vars, allow_recursion), False) if not expr: raise self.Exception('Failed to parse date {left!r}'.format(**locals()), expr=expr) expr = self._dump(int(expr * 1000), local_vars) + right @@ -263,6 +345,14 @@ class JSInterpreter(object): if expr.startswith('{'): inner, outer = self._separate_at_paren(expr, '}') + # try for object expression + sub_expressions = [list(self._separate(sub_expr.strip(), ':', 1)) for sub_expr in self._separate(inner)] + if all(len(sub_expr) == 2 for sub_expr in sub_expressions): + return dict( + (key_expr if re.match(_NAME_RE, key_expr) else key_expr, + self.interpret_expression(val_expr, local_vars, allow_recursion)) + for key_expr, val_expr in sub_expressions), should_return + # or statement list inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion) if not outer or should_abort: return inner, should_abort or should_return @@ -387,13 +477,13 @@ class JSInterpreter(object): (?P (?P{_NAME_RE})(?:\[(?P[^\]]+?)\])?\s* (?P{_OPERATOR_RE})? - =(?P.*)$ + =(?!=)(?P.*)$ )|(?P (?!if|return|true|false|null|undefined)(?P{_NAME_RE})$ )|(?P (?P{_NAME_RE})\[(?P.+)\]$ )|(?P - (?P{_NAME_RE})(?:\.(?P[^(]+)|\[(?P[^\]]+)\])\s* + (?P{_NAME_RE})(?:(?P\?)?\.(?P[^(]+)|\[(?P[^\]]+)\])\s* )|(?P (?P{_NAME_RE})\((?P.*)\)$ )'''.format(**globals()), expr) @@ -405,7 +495,7 @@ class JSInterpreter(object): local_vars[m.group('out')] = self._operator( m.group('op'), left_val, m.group('expr'), expr, local_vars, allow_recursion) return local_vars[m.group('out')], should_return - elif left_val is None: + elif left_val in (None, self.undefined): raise self.Exception('Cannot index undefined variable ' + m.group('out'), expr=expr) idx = self.interpret_expression(m.group('index'), local_vars, allow_recursion) @@ -424,6 +514,9 @@ class JSInterpreter(object): elif expr == 'continue': raise JS_Continue() + elif expr == 'undefined': + return self.undefined, should_return + elif md.get('return'): return local_vars[m.group('name')], should_return @@ -441,7 +534,9 @@ class JSInterpreter(object): for op, _ in self._all_operators(): # hackety: have higher priority than <>, but don't confuse them - skip_delim = (op + op) if op in ('<', '>') else None + skip_delim = (op + op) if op in '<>*?' else None + if op == '?': + skip_delim = (skip_delim, '?.') separated = list(self._separate(expr, op, skip_delims=skip_delim)) if len(separated) < 2: continue @@ -451,12 +546,10 @@ class JSInterpreter(object): right_expr = '-' + right_expr separated.pop() left_val = self.interpret_expression(op.join(separated), local_vars, allow_recursion) - return self._operator(op, 0 if left_val is None else left_val, - right_expr, expr, local_vars, allow_recursion), should_return + return self._operator(op, left_val, right_expr, expr, local_vars, allow_recursion), should_return if md.get('attribute'): - variable = m.group('var') - member = m.group('member') + variable, member, nullish = m.group('var', 'member', 'nullish') if not member: member = self.interpret_expression(m.group('member2'), local_vars, allow_recursion) arg_str = expr[m.end():] @@ -477,15 +570,24 @@ class JSInterpreter(object): 'String': compat_str, 'Math': float, } - obj = local_vars.get(variable, types.get(variable, NO_DEFAULT)) - if obj is NO_DEFAULT: - if variable not in self._objects: - self._objects[variable] = self.extract_object(variable) - obj = self._objects[variable] + obj = local_vars.get(variable) + if obj in (self.undefined, None): + obj = types.get(variable, self.undefined) + if obj is self.undefined: + try: + if variable not in self._objects: + self._objects[variable] = self.extract_object(variable) + obj = self._objects[variable] + except self.Exception: + if not nullish: + raise + + if nullish and obj is self.undefined: + return self.undefined # Member access if arg_str is None: - return self._index(obj, member) + return self._index(obj, member, nullish) # Function call argvals = [ @@ -660,7 +762,14 @@ class JSInterpreter(object): def build_arglist(cls, arg_text): if not arg_text: return [] - return list(filter(None, (x.strip() or None for x in cls._separate(arg_text)))) + + def valid_arg(y): + y = y.strip() + if not y: + raise cls.Exception('Missing arg in "%s"' % (arg_text, )) + return y + + return [valid_arg(x) for x in cls._separate(arg_text)] def build_function(self, argnames, code, *global_stack): global_stack = list(global_stack) or [{}] From a63bbcbfcf66ce6bc9ba4b793e1dc4412a0cd559 Mon Sep 17 00:00:00 2001 From: dirkf Date: Fri, 19 Aug 2022 11:45:04 +0100 Subject: [PATCH 07/21] [jsinterp] Handle regexp literals and throw/catch execution (#31182) * based on https://github.com/yt-dlp/yt-dlp/commit/f6ca640b122239d5ab215f8c2564efb7ac3e8c65, thanks pukkandan * adds parse support for regexp flags --- test/test_jsinterp.py | 21 +++++ test/test_youtube_signature.py | 4 + youtube_dl/jsinterp.py | 136 +++++++++++++++++++++++++++------ 3 files changed, 139 insertions(+), 22 deletions(-) diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py index 328941e09..faddf00d5 100644 --- a/test/test_jsinterp.py +++ b/test/test_jsinterp.py @@ -9,6 +9,7 @@ import unittest sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) import math +import re from youtube_dl.jsinterp import JSInterpreter undefined = JSInterpreter.undefined @@ -316,19 +317,39 @@ class TestJSInterpreter(unittest.TestCase): function x() { return {}; } ''') self.assertEqual(jsi.call_function('x'), {}) + jsi = JSInterpreter(''' function x() { let a = {m1: 42, m2: 0 }; return [a["m1"], a.m2]; } ''') self.assertEqual(jsi.call_function('x'), [42, 0]) + jsi = JSInterpreter(''' function x() { let a; return a?.qq; } ''') self.assertIs(jsi.call_function('x'), undefined) + jsi = JSInterpreter(''' function x() { let a = {m1: 42, m2: 0 }; return a?.qq; } ''') self.assertIs(jsi.call_function('x'), undefined) + def test_regex(self): + jsi = JSInterpreter(''' + function x() { let a=/,,[/,913,/](,)}/; } + ''') + self.assertIs(jsi.call_function('x'), None) + + jsi = JSInterpreter(''' + function x() { let a=/,,[/,913,/](,)}/; return a; } + ''') + # Pythons disagree on the type of a pattern + self.assertTrue(isinstance(jsi.call_function('x'), type(re.compile('')))) + + jsi = JSInterpreter(''' + function x() { let a=/,,[/,913,/](,)}/i; return a; } + ''') + self.assertEqual(jsi.call_function('x').flags & re.I, re.I) + if __name__ == '__main__': unittest.main() diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py index 4d756dad3..43e22388d 100644 --- a/test/test_youtube_signature.py +++ b/test/test_youtube_signature.py @@ -106,6 +106,10 @@ _NSIG_TESTS = [ 'https://www.youtube.com/s/player/c81bbb4a/player_ias.vflset/en_US/base.js', 'gre3EcLurNY2vqp94', 'Z9DfGxWP115WTg', ), + ( + 'https://www.youtube.com/s/player/1f7d5369/player_ias.vflset/en_US/base.js', + 'batNX7sYqIJdkJ', 'IhOkL_zxbkOZBw', + ), ] diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index 8e119d08a..48c27a1c0 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -7,6 +7,7 @@ import operator import re from .utils import ( + error_to_compat_str, ExtractorError, js_to_json, remove_quotes, @@ -130,7 +131,7 @@ _SC_OPERATORS = ( _OPERATOR_RE = '|'.join(map(lambda x: re.escape(x[0]), _OPERATORS + _LOG_OPERATORS)) _MATCHING_PARENS = dict(zip(*zip('()', '{}', '[]'))) -_QUOTES = '\'"' +_QUOTES = '\'"/' def _ternary(cndn, if_true=True, if_false=False): @@ -155,6 +156,12 @@ class JS_Continue(ExtractorError): ExtractorError.__init__(self, 'Invalid continue') +class JS_Throw(ExtractorError): + def __init__(self, e): + self.error = e + ExtractorError.__init__(self, 'Uncaught exception ' + error_to_compat_str(e)) + + class LocalNameSpace(ChainMap): def __getitem__(self, key): try: @@ -172,6 +179,17 @@ class LocalNameSpace(ChainMap): def __delitem__(self, key): raise NotImplementedError('Deleting is not supported') + # except + def pop(self, key, *args): + try: + off = self.__getitem__(key) + super(LocalNameSpace, self).__delitem__(key) + return off + except KeyError: + if len(args) > 0: + return args[0] + raise + def __contains__(self, key): try: super(LocalNameSpace, self).__getitem__(key) @@ -188,9 +206,29 @@ class JSInterpreter(object): undefined = _UNDEFINED + RE_FLAGS = { + # special knowledge: Python's re flags are bitmask values, current max 128 + # invent new bitmask values well above that for literal parsing + # TODO: new pattern class to execute matches with these flags + 'd': 1024, # Generate indices for substring matches + 'g': 2048, # Global search + 'i': re.I, # Case-insensitive search + 'm': re.M, # Multi-line search + 's': re.S, # Allows . to match newline characters + 'u': re.U, # Treat a pattern as a sequence of unicode code points + 'y': 4096, # Perform a "sticky" search that matches starting at the current position in the target string + } + + _EXC_NAME = '__youtube_dl_exception__' + _OBJ_NAME = '__youtube_dl_jsinterp_obj' + + OP_CHARS = None + def __init__(self, code, objects=None): self.code, self._functions = code, {} self._objects = {} if objects is None else objects + if type(self).OP_CHARS is None: + type(self).OP_CHARS = self.OP_CHARS = self.__op_chars() class Exception(ExtractorError): def __init__(self, msg, *args, **kwargs): @@ -199,32 +237,64 @@ class JSInterpreter(object): msg = '{0} in: {1!r}'.format(msg.rstrip(), expr[:100]) super(JSInterpreter.Exception, self).__init__(msg, *args, **kwargs) + @classmethod + def __op_chars(cls): + op_chars = set(';,') + for op in cls._all_operators(): + for c in op[0]: + op_chars.add(c) + return op_chars + def _named_object(self, namespace, obj): self.__named_object_counter += 1 - name = '__youtube_dl_jsinterp_obj%d' % (self.__named_object_counter, ) + name = '%s%d' % (self._OBJ_NAME, self.__named_object_counter) namespace[name] = obj return name - @staticmethod - def _separate(expr, delim=',', max_split=None, skip_delims=None): + @classmethod + def _regex_flags(cls, expr): + flags = 0 + if not expr: + return flags, expr + for idx, ch in enumerate(expr): + if ch not in cls.RE_FLAGS: + break + flags |= cls.RE_FLAGS[ch] + return flags, expr[idx:] if idx > 0 else expr + + @classmethod + def _separate(cls, expr, delim=',', max_split=None, skip_delims=None): if not expr: return counters = {k: 0 for k in _MATCHING_PARENS.values()} - start, splits, pos, skipping, delim_len = 0, 0, 0, 0, len(delim) - 1 - in_quote, escaping = None, False + start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1 + in_quote, escaping, skipping = None, False, 0 + after_op, in_regex_char_group, skip_re = True, False, 0 + for idx, char in enumerate(expr): + if skip_re > 0: + skip_re -= 1 + continue if not in_quote: if char in _MATCHING_PARENS: counters[_MATCHING_PARENS[char]] += 1 elif char in counters: counters[char] -= 1 - if not escaping: - if char in _QUOTES and in_quote in (char, None): - in_quote = None if in_quote else char - else: - escaping = in_quote and char == '\\' - else: - escaping = False + if not escaping and char in _QUOTES and in_quote in (char, None): + if in_quote or after_op or char != '/': + in_quote = None if in_quote and not in_regex_char_group else char + if in_quote is None and char == '/' and delim != '/': + # regexp flags + n_idx = idx + 1 + while n_idx < len(expr) and expr[n_idx] in cls.RE_FLAGS: + n_idx += 1 + skip_re = n_idx - idx - 1 + if skip_re > 0: + continue + elif in_quote == '/' and char in '[]': + in_regex_char_group = char == '[' + escaping = not escaping and in_quote and char == '\\' + after_op = not in_quote and char in cls.OP_CHARS or (char == ' ' and after_op) if char != delim[pos] or any(counters.values()) or in_quote: pos = skipping = 0 @@ -313,16 +383,23 @@ class JSInterpreter(object): if should_return: return ret, should_return - m = re.match(r'(?P(?:var|const|let)\s)|return(?:\s+|$)', stmt) + m = re.match(r'(?P(?:var|const|let)\s)|return(?:\s+|(?=["\'])|$)|(?Pthrow\s+)', stmt) if m: expr = stmt[len(m.group(0)):].strip() + if m.group('throw'): + raise JS_Throw(self.interpret_expression(expr, local_vars, allow_recursion)) should_return = not m.group('var') if not expr: return None, should_return if expr[0] in _QUOTES: inner, outer = self._separate(expr, expr[0], 1) - inner = json.loads(js_to_json(inner + expr[0])) # , strict=True)) + if expr[0] == '/': + flags, _ = self._regex_flags(outer) + inner, outer = inner.replace('"', r'\"'), '' + inner = re.compile(js_to_json(inner + expr[0]), flags=flags) # , strict=True)) + else: + inner = json.loads(js_to_json(inner + expr[0])) # , strict=True)) if not outer: return inner, should_return expr = self._named_object(local_vars, inner) + outer @@ -374,22 +451,37 @@ class JSInterpreter(object): for item in self._separate(inner)]) expr = name + outer - m = re.match(r'(?Ptry|finally)\s*|(?:(?Pcatch)|(?Pfor)|(?Pswitch))\s*\(', expr) + m = re.match(r'''(?x) + (?Ptry|finally)\s*| + (?Pcatch\s*(?P\(\s*{_NAME_RE}\s*\)))| + (?Pswitch)\s*\(| + (?Pfor)\s*\(|'''.format(**globals()), expr) md = m.groupdict() if m else {} if md.get('try'): if expr[m.end()] == '{': try_expr, expr = self._separate_at_paren(expr[m.end():], '}') else: try_expr, expr = expr[m.end() - 1:], '' - ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion) - if should_abort: - return ret, True + try: + ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion) + if should_abort: + return ret, True + except JS_Throw as e: + local_vars[self._EXC_NAME] = e.error + except Exception as e: + # XXX: This works for now, but makes debugging future issues very hard + local_vars[self._EXC_NAME] = e ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) return ret, should_abort or should_return elif md.get('catch'): - # We ignore the catch block - _, expr = self._separate_at_paren(expr, '}') + catch_expr, expr = self._separate_at_paren(expr[m.end():], '}') + if self._EXC_NAME in local_vars: + catch_vars = local_vars.new_child({m.group('err'): local_vars.pop(self._EXC_NAME)}) + ret, should_abort = self.interpret_statement(catch_expr, catch_vars, allow_recursion) + if should_abort: + return ret, True + ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) return ret, should_abort or should_return @@ -503,7 +595,7 @@ class JSInterpreter(object): raise self.Exception('List index %s must be integer' % (idx, ), expr=expr) idx = int(idx) left_val[idx] = self._operator( - m.group('op'), left_val[idx], m.group('expr'), expr, local_vars, allow_recursion) + m.group('op'), self._index(left_val, idx), m.group('expr'), expr, local_vars, allow_recursion) return left_val[idx], should_return elif expr.isdigit(): From 9be9fffb25fb26c124f30c36f1a01da91d5fd9a7 Mon Sep 17 00:00:00 2001 From: dirkf Date: Fri, 19 Aug 2022 15:34:33 +0100 Subject: [PATCH 08/21] [jsinterp] Clean up and pull yt-dlp style * add compat_re_Pattern * improve compat_collections_chain_map * use class JS_Undefined * remove unused code --- test/test_jsinterp.py | 20 +++--- test/test_youtube_signature.py | 3 +- youtube_dl/compat.py | 21 +++++- youtube_dl/jsinterp.py | 123 ++++++++++++--------------------- 4 files changed, 77 insertions(+), 90 deletions(-) diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py index faddf00d5..96786a84c 100644 --- a/test/test_jsinterp.py +++ b/test/test_jsinterp.py @@ -11,8 +11,9 @@ sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) import math import re -from youtube_dl.jsinterp import JSInterpreter -undefined = JSInterpreter.undefined +from youtube_dl.compat import compat_re_Pattern + +from youtube_dl.jsinterp import JS_Undefined, JSInterpreter class TestJSInterpreter(unittest.TestCase): @@ -261,12 +262,12 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter(''' function x() { return undefined; } ''') - self.assertIs(jsi.call_function('x'), undefined) + self.assertIs(jsi.call_function('x'), JS_Undefined) jsi = JSInterpreter(''' function x() { let v; return v; } ''') - self.assertIs(jsi.call_function('x'), undefined) + self.assertIs(jsi.call_function('x'), JS_Undefined) jsi = JSInterpreter(''' function x() { return [undefined === undefined, undefined == undefined, undefined < undefined, undefined > undefined]; } @@ -307,7 +308,7 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter(''' function x() { let v; return [v>42, v<=42, v&&42, 42&&v]; } ''') - self.assertEqual(jsi.call_function('x'), [False, False, undefined, undefined]) + self.assertEqual(jsi.call_function('x'), [False, False, JS_Undefined, JS_Undefined]) jsi = JSInterpreter('function x(){return undefined ?? 42; }') self.assertEqual(jsi.call_function('x'), 42) @@ -326,12 +327,12 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter(''' function x() { let a; return a?.qq; } ''') - self.assertIs(jsi.call_function('x'), undefined) + self.assertIs(jsi.call_function('x'), JS_Undefined) jsi = JSInterpreter(''' function x() { let a = {m1: 42, m2: 0 }; return a?.qq; } ''') - self.assertIs(jsi.call_function('x'), undefined) + self.assertIs(jsi.call_function('x'), JS_Undefined) def test_regex(self): jsi = JSInterpreter(''' @@ -342,13 +343,12 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter(''' function x() { let a=/,,[/,913,/](,)}/; return a; } ''') - # Pythons disagree on the type of a pattern - self.assertTrue(isinstance(jsi.call_function('x'), type(re.compile('')))) + self.assertIsInstance(jsi.call_function('x'), compat_re_Pattern) jsi = JSInterpreter(''' function x() { let a=/,,[/,913,/](,)}/i; return a; } ''') - self.assertEqual(jsi.call_function('x').flags & re.I, re.I) + self.assertEqual(jsi.call_function('x').flags & ~re.U, re.I) if __name__ == '__main__': diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py index 43e22388d..327d4c40d 100644 --- a/test/test_youtube_signature.py +++ b/test/test_youtube_signature.py @@ -12,10 +12,11 @@ import io import re import string +from youtube_dl.compat import compat_str, compat_urlretrieve + from test.helper import FakeYDL from youtube_dl.extractor import YoutubeIE from youtube_dl.jsinterp import JSInterpreter -from youtube_dl.compat import compat_str, compat_urlretrieve _SIG_TESTS = [ ( diff --git a/youtube_dl/compat.py b/youtube_dl/compat.py index 6d2c31a61..3002109ca 100644 --- a/youtube_dl/compat.py +++ b/youtube_dl/compat.py @@ -3023,18 +3023,34 @@ except ImportError: self.maps[0].__setitem__(k, v) return - def __delitem__(self, k): + def __contains__(self, k): + return any((k in m) for m in self.maps) + + def __delitem(self, k): if k in self.maps[0]: del self.maps[0][k] return raise KeyError(k) + def __delitem__(self, k): + self.__delitem(k) + def __iter__(self): return itertools.chain(*reversed(self.maps)) def __len__(self): return len(iter(self)) + # to match Py3, don't del directly + def pop(self, k, *args): + if self.__contains__(k): + off = self.__getitem__(k) + self.__delitem(k) + return off + elif len(args) > 0: + return args[0] + raise KeyError(k) + def new_child(self, m=None, **kwargs): m = m or {} m.update(kwargs) @@ -3044,6 +3060,8 @@ except ImportError: def parents(self): return compat_collections_chain_map(*(self.maps[1:])) +# Pythons disagree on the type of a pattern (RegexObject, _sre.SRE_Pattern, Pattern, ...?) +compat_re_Pattern = type(re.compile('')) if sys.version_info < (3, 3): def compat_b64decode(s, *args, **kwargs): @@ -3110,6 +3128,7 @@ __all__ = [ 'compat_os_name', 'compat_parse_qs', 'compat_print', + 'compat_re_Pattern', 'compat_realpath', 'compat_setenv', 'compat_shlex_quote', diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index 48c27a1c0..6719d0dfd 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -19,16 +19,12 @@ from .compat import ( compat_str, ) -_NAME_RE = r'[a-zA-Z_$][\w$]*' - -_UNDEFINED = object() - def _js_bit_op(op): def wrapped(a, b): def zeroise(x): - return 0 if x in (None, _UNDEFINED) else x + return 0 if x in (None, JS_Undefined) else x return op(zeroise(a), zeroise(b)) return wrapped @@ -37,7 +33,7 @@ def _js_bit_op(op): def _js_arith_op(op): def wrapped(a, b): - if _UNDEFINED in (a, b): + if JS_Undefined in (a, b): return float('nan') return op(a or 0, b or 0) @@ -45,22 +41,21 @@ def _js_arith_op(op): def _js_div(a, b): - if _UNDEFINED in (a, b) or not (a and b): + if JS_Undefined in (a, b) or not (a and b): return float('nan') return float('inf') if not b else operator.truediv(a or 0, b) def _js_mod(a, b): - if _UNDEFINED in (a, b) or not b: + if JS_Undefined in (a, b) or not b: return float('nan') return (a or 0) % b def _js_exp(a, b): if not b: - # even 0 ** 0 !! - return 1 - if _UNDEFINED in (a, b): + return 1 # even 0 ** 0 !! + elif JS_Undefined in (a, b): return float('nan') return (a or 0) ** b @@ -68,7 +63,7 @@ def _js_exp(a, b): def _js_eq_op(op): def wrapped(a, b): - if set((a, b)) <= set((None, _UNDEFINED)): + if set((a, b)) <= set((None, JS_Undefined)): return op(a, a) return op(a, b) @@ -78,21 +73,28 @@ def _js_eq_op(op): def _js_comp_op(op): def wrapped(a, b): - if _UNDEFINED in (a, b): + if JS_Undefined in (a, b): return False return op(a or 0, b or 0) return wrapped +def _js_ternary(cndn, if_true=True, if_false=False): + """Simulate JS's ternary operator (cndn?if_true:if_false)""" + if cndn in (False, None, 0, '', JS_Undefined): + return if_false + try: + if math.isnan(cndn): # NB: NaN cannot be checked by membership + return if_false + except TypeError: + pass + return if_true + + # (op, definition) in order of binding priority, tightest first # avoid dict to maintain order # definition None => Defined in JSInterpreter._operator -_DOT_OPERATORS = ( - ('.', None), - # TODO: ('?.', None), -) - _OPERATORS = ( ('>>', _js_bit_op(operator.rshift)), ('<<', _js_bit_op(operator.lshift)), @@ -130,20 +132,13 @@ _SC_OPERATORS = ( _OPERATOR_RE = '|'.join(map(lambda x: re.escape(x[0]), _OPERATORS + _LOG_OPERATORS)) +_NAME_RE = r'[a-zA-Z_$][\w$]*' _MATCHING_PARENS = dict(zip(*zip('()', '{}', '[]'))) _QUOTES = '\'"/' -def _ternary(cndn, if_true=True, if_false=False): - """Simulate JS's ternary operator (cndn?if_true:if_false)""" - if cndn in (False, None, 0, '', _UNDEFINED): - return if_false - try: - if math.isnan(cndn): # NB: NaN cannot be checked by membership - return if_false - except TypeError: - pass - return if_true +class JS_Undefined(object): + pass class JS_Break(ExtractorError): @@ -167,7 +162,7 @@ class LocalNameSpace(ChainMap): try: return super(LocalNameSpace, self).__getitem__(key) except KeyError: - return _UNDEFINED + return JS_Undefined def __setitem__(self, key, value): for scope in self.maps: @@ -179,24 +174,6 @@ class LocalNameSpace(ChainMap): def __delitem__(self, key): raise NotImplementedError('Deleting is not supported') - # except - def pop(self, key, *args): - try: - off = self.__getitem__(key) - super(LocalNameSpace, self).__delitem__(key) - return off - except KeyError: - if len(args) > 0: - return args[0] - raise - - def __contains__(self, key): - try: - super(LocalNameSpace, self).__getitem__(key) - return True - except KeyError: - return False - def __repr__(self): return 'LocalNameSpace%s' % (self.maps, ) @@ -204,9 +181,7 @@ class LocalNameSpace(ChainMap): class JSInterpreter(object): __named_object_counter = 0 - undefined = _UNDEFINED - - RE_FLAGS = { + _RE_FLAGS = { # special knowledge: Python's re flags are bitmask values, current max 128 # invent new bitmask values well above that for literal parsing # TODO: new pattern class to execute matches with these flags @@ -257,10 +232,10 @@ class JSInterpreter(object): if not expr: return flags, expr for idx, ch in enumerate(expr): - if ch not in cls.RE_FLAGS: + if ch not in cls._RE_FLAGS: break - flags |= cls.RE_FLAGS[ch] - return flags, expr[idx:] if idx > 0 else expr + flags |= cls._RE_FLAGS[ch] + return flags, expr[idx + 1:] @classmethod def _separate(cls, expr, delim=',', max_split=None, skip_delims=None): @@ -283,14 +258,6 @@ class JSInterpreter(object): if not escaping and char in _QUOTES and in_quote in (char, None): if in_quote or after_op or char != '/': in_quote = None if in_quote and not in_regex_char_group else char - if in_quote is None and char == '/' and delim != '/': - # regexp flags - n_idx = idx + 1 - while n_idx < len(expr) and expr[n_idx] in cls.RE_FLAGS: - n_idx += 1 - skip_re = n_idx - idx - 1 - if skip_re > 0: - continue elif in_quote == '/' and char in '[]': in_regex_char_group = char == '[' escaping = not escaping and in_quote and char == '\\' @@ -336,13 +303,13 @@ class JSInterpreter(object): def _operator(self, op, left_val, right_expr, expr, local_vars, allow_recursion): if op in ('||', '&&'): - if (op == '&&') ^ _ternary(left_val): + if (op == '&&') ^ _js_ternary(left_val): return left_val # short circuiting elif op == '??': - if left_val not in (None, self.undefined): + if left_val not in (None, JS_Undefined): return left_val elif op == '?': - right_expr = _ternary(left_val, *self._separate(right_expr, ':', 1)) + right_expr = _js_ternary(left_val, *self._separate(right_expr, ':', 1)) right_val = self.interpret_expression(right_expr, local_vars, allow_recursion) opfunc = op and next((v for k, v in self._all_operators() if k == op), None) @@ -361,7 +328,7 @@ class JSInterpreter(object): return obj[int(idx)] if isinstance(obj, list) else obj[idx] except Exception as e: if allow_undefined: - return self.undefined + return JS_Undefined raise self.Exception('Cannot get index {idx}'.format(**locals()), expr=repr(obj), cause=e) def _dump(self, obj, namespace): @@ -395,9 +362,8 @@ class JSInterpreter(object): if expr[0] in _QUOTES: inner, outer = self._separate(expr, expr[0], 1) if expr[0] == '/': - flags, _ = self._regex_flags(outer) - inner, outer = inner.replace('"', r'\"'), '' - inner = re.compile(js_to_json(inner + expr[0]), flags=flags) # , strict=True)) + flags, outer = self._regex_flags(outer) + inner = re.compile(inner[1:], flags=flags) # , strict=True)) else: inner = json.loads(js_to_json(inner + expr[0])) # , strict=True)) if not outer: @@ -422,7 +388,7 @@ class JSInterpreter(object): if expr.startswith('{'): inner, outer = self._separate_at_paren(expr, '}') - # try for object expression + # try for object expression (Map) sub_expressions = [list(self._separate(sub_expr.strip(), ':', 1)) for sub_expr in self._separate(inner)] if all(len(sub_expr) == 2 for sub_expr in sub_expressions): return dict( @@ -455,7 +421,8 @@ class JSInterpreter(object): (?Ptry|finally)\s*| (?Pcatch\s*(?P\(\s*{_NAME_RE}\s*\)))| (?Pswitch)\s*\(| - (?Pfor)\s*\(|'''.format(**globals()), expr) + (?Pfor)\s*\(| + '''.format(**globals()), expr) md = m.groupdict() if m else {} if md.get('try'): if expr[m.end()] == '{': @@ -500,7 +467,7 @@ class JSInterpreter(object): start, cndn, increment = self._separate(constructor, ';') self.interpret_expression(start, local_vars, allow_recursion) while True: - if not _ternary(self.interpret_expression(cndn, local_vars, allow_recursion)): + if not _js_ternary(self.interpret_expression(cndn, local_vars, allow_recursion)): break try: ret, should_abort = self.interpret_statement(body, local_vars, allow_recursion) @@ -587,7 +554,7 @@ class JSInterpreter(object): local_vars[m.group('out')] = self._operator( m.group('op'), left_val, m.group('expr'), expr, local_vars, allow_recursion) return local_vars[m.group('out')], should_return - elif left_val in (None, self.undefined): + elif left_val in (None, JS_Undefined): raise self.Exception('Cannot index undefined variable ' + m.group('out'), expr=expr) idx = self.interpret_expression(m.group('index'), local_vars, allow_recursion) @@ -607,7 +574,7 @@ class JSInterpreter(object): raise JS_Continue() elif expr == 'undefined': - return self.undefined, should_return + return JS_Undefined, should_return elif md.get('return'): return local_vars[m.group('name')], should_return @@ -663,9 +630,9 @@ class JSInterpreter(object): 'Math': float, } obj = local_vars.get(variable) - if obj in (self.undefined, None): - obj = types.get(variable, self.undefined) - if obj is self.undefined: + if obj in (JS_Undefined, None): + obj = types.get(variable, JS_Undefined) + if obj is JS_Undefined: try: if variable not in self._objects: self._objects[variable] = self.extract_object(variable) @@ -674,8 +641,8 @@ class JSInterpreter(object): if not nullish: raise - if nullish and obj is self.undefined: - return self.undefined + if nullish and obj is JS_Undefined: + return JS_Undefined # Member access if arg_str is None: From baa75fca868b41f4ac6532a2705b427fba515be0 Mon Sep 17 00:00:00 2001 From: dirkf Date: Fri, 19 Aug 2022 19:11:08 +0100 Subject: [PATCH 09/21] [uktvplay] Support domain without .uktv --- youtube_dl/extractor/uktvplay.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/youtube_dl/extractor/uktvplay.py b/youtube_dl/extractor/uktvplay.py index f28fd514d..9ef9638cd 100644 --- a/youtube_dl/extractor/uktvplay.py +++ b/youtube_dl/extractor/uktvplay.py @@ -5,7 +5,7 @@ from .common import InfoExtractor class UKTVPlayIE(InfoExtractor): - _VALID_URL = r'https?://uktvplay\.uktv\.co\.uk/(?:.+?\?.*?\bvideo=|([^/]+/)*watch-online/)(?P\d+)' + _VALID_URL = r'https?://uktvplay\.(?:uktv\.)?co\.uk/(?:.+?\?.*?\bvideo=|([^/]+/)*watch-online/)(?P\d+)' _TESTS = [{ 'url': 'https://uktvplay.uktv.co.uk/shows/world-at-war/c/200/watch-online/?video=2117008346001', 'info_dict': { From 722e31467bf9e7ad323f428676b5e82cb8903675 Mon Sep 17 00:00:00 2001 From: gudata Date: Fri, 19 Aug 2022 23:00:21 +0300 Subject: [PATCH 10/21] [infoq] Avoid crash if the page has no `mp3Form` * proposed fix for issue #31131, aligns with yt-dlp Co-authored-by: dirkf --- youtube_dl/extractor/infoq.py | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/youtube_dl/extractor/infoq.py b/youtube_dl/extractor/infoq.py index 0a70a1fb4..60b02b699 100644 --- a/youtube_dl/extractor/infoq.py +++ b/youtube_dl/extractor/infoq.py @@ -1,6 +1,9 @@ # coding: utf-8 from __future__ import unicode_literals +from ..utils import ( + ExtractorError, +) from ..compat import ( compat_b64decode, @@ -90,7 +93,11 @@ class InfoQIE(BokeCCBaseIE): }] def _extract_http_audio(self, webpage, video_id): - fields = self._form_hidden_inputs('mp3Form', webpage) + try: + fields = self._form_hidden_inputs('mp3Form', webpage) + except ExtractorError: + fields = {} + http_audio_url = fields.get('filename') if not http_audio_url: return [] From 43aebf131383d2c94ae5007dc8926f7cb698e1a8 Mon Sep 17 00:00:00 2001 From: dirkf Date: Sun, 21 Aug 2022 00:19:19 +0100 Subject: [PATCH 11/21] [utils] Ensure RFC3986 encoding result is unicode --- youtube_dl/utils.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index a5f584ec5..fea38ed32 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -3970,7 +3970,8 @@ def escape_rfc3986(s): """Escape non-ASCII characters as suggested by RFC 3986""" if sys.version_info < (3, 0) and isinstance(s, compat_str): s = s.encode('utf-8') - return compat_urllib_parse.quote(s, b"%/;:@&=+$,!~*'()?#[]") + # ensure unicode: after quoting, it can always be converted + return compat_str(compat_urllib_parse.quote(s, b"%/;:@&=+$,!~*'()?#[]")) def escape_url(url): From 1b907b46cc33c656569b0478ba4b5309d1c6b32a Mon Sep 17 00:00:00 2001 From: dirkf Date: Sun, 21 Aug 2022 00:21:02 +0100 Subject: [PATCH 12/21] [core] Avoid processing empty format list after removing bad formats * also ensure compat encoding of error strings --- youtube_dl/YoutubeDL.py | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/youtube_dl/YoutubeDL.py b/youtube_dl/YoutubeDL.py index e77b8d50c..8e8546596 100755 --- a/youtube_dl/YoutubeDL.py +++ b/youtube_dl/YoutubeDL.py @@ -721,7 +721,7 @@ class YoutubeDL(object): filename = encodeFilename(filename, True).decode(preferredencoding()) return sanitize_path(filename) except ValueError as err: - self.report_error('Error in output template: ' + str(err) + ' (encoding: ' + repr(preferredencoding()) + ')') + self.report_error('Error in output template: ' + error_to_compat_str(err) + ' (encoding: ' + repr(preferredencoding()) + ')') return None def _match_entry(self, info_dict, incomplete): @@ -1570,9 +1570,6 @@ class YoutubeDL(object): else: formats = info_dict['formats'] - if not formats: - raise ExtractorError('No video formats found!') - def is_wellformed(f): url = f.get('url') if not url: @@ -1585,7 +1582,10 @@ class YoutubeDL(object): return True # Filter out malformed formats for better extraction robustness - formats = list(filter(is_wellformed, formats)) + formats = list(filter(is_wellformed, formats or [])) + + if not formats: + raise ExtractorError('No video formats found!') formats_dict = {} @@ -2058,7 +2058,7 @@ class YoutubeDL(object): try: self.post_process(filename, info_dict) except (PostProcessingError) as err: - self.report_error('postprocessing: %s' % str(err)) + self.report_error('postprocessing: %s' % error_to_compat_str(err)) return self.record_download_archive(info_dict) # avoid possible nugatory search for further items (PR #26638) From 5dabc11e927b3d869c588d6eb4ff10fdc3bf086d Mon Sep 17 00:00:00 2001 From: dirkf Date: Thu, 25 Aug 2022 12:14:59 +0100 Subject: [PATCH 13/21] [YouTube] Improve error check for n-sig processing --- youtube_dl/extractor/youtube.py | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-) diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 91a3b6058..3d12e2e4a 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -1500,7 +1500,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor): return lambda s: jsi.extract_function_from_code(*func_code)([s]) def _n_descramble(self, n_param, player_url, video_id): - """Compute the response to YT's "n" parameter challenge + """Compute the response to YT's "n" parameter challenge, + or None Args: n_param -- challenge string that is the value of the @@ -1518,7 +1519,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor): if player_id not in self._player_cache: self._player_cache[player_id] = self._extract_n_function(video_id, player_url) func = self._player_cache[player_id] - self._player_cache[sig_id] = func(n_param) + ret = func(n_param) + if ret.startswith('enhanced_except_'): + raise ExtractorError('Unhandled exception in decode') + self._player_cache[sig_id] = ret if self._downloader.params.get('verbose', False): self._downloader.to_screen('[debug] [%s] %s' % (self.IE_NAME, 'Decrypted nsig {0} => {1}'.format(n_param, self._player_cache[sig_id]))) return self._player_cache[sig_id] @@ -1539,10 +1543,12 @@ class YoutubeIE(YoutubeBaseInfoExtractor): continue n_param = n_param[-1] n_response = self._n_descramble(n_param, player_url, video_id) - if n_response: - qs['n'] = [n_response] - fmt['url'] = compat_urlparse.urlunparse( - parsed_fmt_url._replace(query=compat_urllib_parse_urlencode(qs, True))) + if n_response is None: + # give up if descrambling failed + break + qs['n'] = [n_response] + fmt['url'] = compat_urlparse.urlunparse( + parsed_fmt_url._replace(query=compat_urllib_parse_urlencode(qs, True))) def _mark_watched(self, video_id, player_response): playback_url = url_or_none(try_get( From 660cd68133060e577ab0c7a3a43f115749d419e3 Mon Sep 17 00:00:00 2001 From: dirkf Date: Thu, 25 Aug 2022 12:16:10 +0100 Subject: [PATCH 14/21] [jsinterp] Fix bug in operator precedence * from https://github.com/yt-dlp/yt-dlp/commit/164b03c4864b0d44cfee5e7702f7c2317164a6cf * added tests --- test/test_jsinterp.py | 25 +++++++++++++++++++++++++ test/test_youtube_signature.py | 4 ++++ youtube_dl/jsinterp.py | 7 ++++++- 3 files changed, 35 insertions(+), 1 deletion(-) diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py index 96786a84c..0a97bdbc4 100644 --- a/test/test_jsinterp.py +++ b/test/test_jsinterp.py @@ -192,6 +192,31 @@ class TestJSInterpreter(unittest.TestCase): ''') self.assertEqual(jsi.call_function('x'), 10) + def test_catch(self): + jsi = JSInterpreter(''' + function x() { try{throw 10} catch(e){return 5} } + ''') + self.assertEqual(jsi.call_function('x'), 5) + + @unittest.expectedFailure + def test_finally(self): + jsi = JSInterpreter(''' + function x() { try{throw 10} finally {return 42} } + ''') + self.assertEqual(jsi.call_function('x'), 42) + jsi = JSInterpreter(''' + function x() { try{throw 10} catch(e){return 5} finally {return 42} } + ''') + self.assertEqual(jsi.call_function('x'), 42) + + def test_nested_try(self): + jsi = JSInterpreter(''' + function x() {try { + try{throw 10} finally {throw 42} + } catch(e){return 5} } + ''') + self.assertEqual(jsi.call_function('x'), 5) + def test_for_loop_continue(self): jsi = JSInterpreter(''' function x() { a=0; for (i=0; i-10; i++) { continue; a++ } return a } diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py index 327d4c40d..4bb0a30b0 100644 --- a/test/test_youtube_signature.py +++ b/test/test_youtube_signature.py @@ -111,6 +111,10 @@ _NSIG_TESTS = [ 'https://www.youtube.com/s/player/1f7d5369/player_ias.vflset/en_US/base.js', 'batNX7sYqIJdkJ', 'IhOkL_zxbkOZBw', ), + ( + 'https://www.youtube.com/s/player/dc0c6770/player_ias.vflset/en_US/base.js', + '5EHDMgYLV6HPGk_Mu-kk', 'n9lUJLHbxUI0GQ', + ), ] diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index 6719d0dfd..a8456ec1c 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -5,6 +5,7 @@ import json import math import operator import re +from collections import Counter from .utils import ( error_to_compat_str, @@ -108,8 +109,8 @@ _OPERATORS = ( _COMP_OPERATORS = ( ('===', operator.is_), - ('==', _js_eq_op(operator.eq)), ('!==', operator.is_not), + ('==', _js_eq_op(operator.eq)), ('!=', _js_eq_op(operator.ne)), ('<=', _js_comp_op(operator.le)), ('>=', _js_comp_op(operator.ge)), @@ -241,7 +242,9 @@ class JSInterpreter(object): def _separate(cls, expr, delim=',', max_split=None, skip_delims=None): if not expr: return + # collections.Counter() is ~10% slower counters = {k: 0 for k in _MATCHING_PARENS.values()} + # counters = Counter() start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1 in_quote, escaping, skipping = None, False, 0 after_op, in_regex_char_group, skip_re = True, False, 0 @@ -442,6 +445,7 @@ class JSInterpreter(object): return ret, should_abort or should_return elif md.get('catch'): + catch_expr, expr = self._separate_at_paren(expr[m.end():], '}') if self._EXC_NAME in local_vars: catch_vars = local_vars.new_child({m.group('err'): local_vars.pop(self._EXC_NAME)}) @@ -450,6 +454,7 @@ class JSInterpreter(object): return ret, True ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) + return ret, should_abort or should_return elif md.get('for'): From 86018749387c18690c3c2745690ec9f9783e2d66 Mon Sep 17 00:00:00 2001 From: dirkf Date: Fri, 26 Aug 2022 08:17:54 +0100 Subject: [PATCH 15/21] [jsinterp] Improve try/catch/finally support --- test/test_jsinterp.py | 14 ++++++- youtube_dl/jsinterp.py | 88 +++++++++++++++++++++++------------------- 2 files changed, 61 insertions(+), 41 deletions(-) diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py index 0a97bdbc4..fb4882d00 100644 --- a/test/test_jsinterp.py +++ b/test/test_jsinterp.py @@ -74,6 +74,9 @@ class TestJSInterpreter(unittest.TestCase): jsi = JSInterpreter('function f(){return 0 ?? 42;}') self.assertEqual(jsi.call_function('f'), 0) + jsi = JSInterpreter('function f(){return "life, the universe and everything" < 42;}') + self.assertFalse(jsi.call_function('f')) + def test_array_access(self): jsi = JSInterpreter('function f(){var x = [1,2,3]; x[0] = 4; x[0] = 5; x[2.0] = 7; return x;}') self.assertEqual(jsi.call_function('f'), [5, 2, 7]) @@ -198,7 +201,6 @@ class TestJSInterpreter(unittest.TestCase): ''') self.assertEqual(jsi.call_function('x'), 5) - @unittest.expectedFailure def test_finally(self): jsi = JSInterpreter(''' function x() { try{throw 10} finally {return 42} } @@ -212,7 +214,7 @@ class TestJSInterpreter(unittest.TestCase): def test_nested_try(self): jsi = JSInterpreter(''' function x() {try { - try{throw 10} finally {throw 42} + try{throw 10} finally {throw 42} } catch(e){return 5} } ''') self.assertEqual(jsi.call_function('x'), 5) @@ -229,6 +231,14 @@ class TestJSInterpreter(unittest.TestCase): ''') self.assertEqual(jsi.call_function('x'), 0) + def test_for_loop_try(self): + jsi = JSInterpreter(''' + function x() { + for (i=0; i-10; i++) { try { if (i == 5) throw i} catch {return 10} finally {break} }; + return 42 } + ''') + self.assertEqual(jsi.call_function('x'), 42) + def test_literal_list(self): jsi = JSInterpreter(''' function x() { return [1, 2, "asdf", [5, 6, 7]][3] } diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index a8456ec1c..08726e478 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -5,7 +5,6 @@ import json import math import operator import re -from collections import Counter from .utils import ( error_to_compat_str, @@ -15,6 +14,7 @@ from .utils import ( unified_timestamp, ) from .compat import ( + compat_basestring, compat_collections_chain_map as ChainMap, compat_itertools_zip_longest as zip_longest, compat_str, @@ -76,6 +76,10 @@ def _js_comp_op(op): def wrapped(a, b): if JS_Undefined in (a, b): return False + if isinstance(a, compat_basestring): + b = compat_str(b or 0) + elif isinstance(b, compat_basestring): + a = compat_str(a or 0) return op(a or 0, b or 0) return wrapped @@ -195,7 +199,6 @@ class JSInterpreter(object): 'y': 4096, # Perform a "sticky" search that matches starting at the current position in the target string } - _EXC_NAME = '__youtube_dl_exception__' _OBJ_NAME = '__youtube_dl_jsinterp_obj' OP_CHARS = None @@ -242,9 +245,8 @@ class JSInterpreter(object): def _separate(cls, expr, delim=',', max_split=None, skip_delims=None): if not expr: return - # collections.Counter() is ~10% slower + # collections.Counter() is ~10% slower in both 2.7 and 3.9 counters = {k: 0 for k in _MATCHING_PARENS.values()} - # counters = Counter() start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1 in_quote, escaping, skipping = None, False, 0 after_op, in_regex_char_group, skip_re = True, False, 0 @@ -291,7 +293,9 @@ class JSInterpreter(object): yield expr[start:] @classmethod - def _separate_at_paren(cls, expr, delim): + def _separate_at_paren(cls, expr, delim=None): + if delim is None: + delim = expr and _MATCHING_PARENS[expr[0]] separated = list(cls._separate(expr, delim, 1)) if len(separated) < 2: @@ -376,7 +380,7 @@ class JSInterpreter(object): if expr.startswith('new '): obj = expr[4:] if obj.startswith('Date('): - left, right = self._separate_at_paren(obj[4:], ')') + left, right = self._separate_at_paren(obj[4:]) expr = unified_timestamp( self.interpret_expression(left, local_vars, allow_recursion), False) if not expr: @@ -390,7 +394,7 @@ class JSInterpreter(object): return None, should_return if expr.startswith('{'): - inner, outer = self._separate_at_paren(expr, '}') + inner, outer = self._separate_at_paren(expr) # try for object expression (Map) sub_expressions = [list(self._separate(sub_expr.strip(), ':', 1)) for sub_expr in self._separate(inner)] if all(len(sub_expr) == 2 for sub_expr in sub_expressions): @@ -406,7 +410,7 @@ class JSInterpreter(object): expr = self._dump(inner, local_vars) + outer if expr.startswith('('): - inner, outer = self._separate_at_paren(expr, ')') + inner, outer = self._separate_at_paren(expr) inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion) if not outer or should_abort: return inner, should_abort or should_return @@ -414,57 +418,63 @@ class JSInterpreter(object): expr = self._dump(inner, local_vars) + outer if expr.startswith('['): - inner, outer = self._separate_at_paren(expr, ']') + inner, outer = self._separate_at_paren(expr) name = self._named_object(local_vars, [ self.interpret_expression(item, local_vars, allow_recursion) for item in self._separate(inner)]) expr = name + outer m = re.match(r'''(?x) - (?Ptry|finally)\s*| - (?Pcatch\s*(?P\(\s*{_NAME_RE}\s*\)))| - (?Pswitch)\s*\(| - (?Pfor)\s*\(| - '''.format(**globals()), expr) + (?Ptry)\s*\{| + (?Pswitch)\s*\(| + (?Pfor)\s*\( + ''', expr) md = m.groupdict() if m else {} if md.get('try'): - if expr[m.end()] == '{': - try_expr, expr = self._separate_at_paren(expr[m.end():], '}') - else: - try_expr, expr = expr[m.end() - 1:], '' + try_expr, expr = self._separate_at_paren(expr[m.end() - 1:]) + err = None try: ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion) if should_abort: return ret, True - except JS_Throw as e: - local_vars[self._EXC_NAME] = e.error except Exception as e: # XXX: This works for now, but makes debugging future issues very hard - local_vars[self._EXC_NAME] = e - ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) - return ret, should_abort or should_return + err = e - elif md.get('catch'): + pending = (None, False) + m = re.match(r'catch\s*(?P\(\s*{_NAME_RE}\s*\))?\{{'.format(**globals()), expr) + if m: + sub_expr, expr = self._separate_at_paren(expr[m.end() - 1:]) + if err: + catch_vars = {} + if m.group('err'): + catch_vars[m.group('err')] = err.error if isinstance(err, JS_Throw) else err + catch_vars = local_vars.new_child(m=catch_vars) + err = None + pending = self.interpret_statement(sub_expr, catch_vars, allow_recursion) - catch_expr, expr = self._separate_at_paren(expr[m.end():], '}') - if self._EXC_NAME in local_vars: - catch_vars = local_vars.new_child({m.group('err'): local_vars.pop(self._EXC_NAME)}) - ret, should_abort = self.interpret_statement(catch_expr, catch_vars, allow_recursion) + m = re.match(r'finally\s*\{', expr) + if m: + sub_expr, expr = self._separate_at_paren(expr[m.end() - 1:]) + ret, should_abort = self.interpret_statement(sub_expr, local_vars, allow_recursion) if should_abort: return ret, True - ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) + ret, should_abort = pending + if should_abort: + return ret, True - return ret, should_abort or should_return + if err: + raise err elif md.get('for'): - constructor, remaining = self._separate_at_paren(expr[m.end() - 1:], ')') + constructor, remaining = self._separate_at_paren(expr[m.end() - 1:]) if remaining.startswith('{'): - body, expr = self._separate_at_paren(remaining, '}') + body, expr = self._separate_at_paren(remaining) else: switch_m = re.match(r'switch\s*\(', remaining) # FIXME if switch_m: - switch_val, remaining = self._separate_at_paren(remaining[switch_m.end() - 1:], ')') + switch_val, remaining = self._separate_at_paren(remaining[switch_m.end() - 1:]) body, expr = self._separate_at_paren(remaining, '}') body = 'switch(%s){%s}' % (switch_val, body) else: @@ -483,11 +493,9 @@ class JSInterpreter(object): except JS_Continue: pass self.interpret_expression(increment, local_vars, allow_recursion) - ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) - return ret, should_abort or should_return elif md.get('switch'): - switch_val, remaining = self._separate_at_paren(expr[m.end() - 1:], ')') + switch_val, remaining = self._separate_at_paren(expr[m.end() - 1:]) switch_val = self.interpret_expression(switch_val, local_vars, allow_recursion) body, expr = self._separate_at_paren(remaining, '}') items = body.replace('default:', 'case default:').split('case ')[1:] @@ -510,6 +518,8 @@ class JSInterpreter(object): break if matched: break + + if md: ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion) return ret, should_abort or should_return @@ -618,7 +628,7 @@ class JSInterpreter(object): member = self.interpret_expression(m.group('member2'), local_vars, allow_recursion) arg_str = expr[m.end():] if arg_str.startswith('('): - arg_str, remaining = self._separate_at_paren(arg_str, ')') + arg_str, remaining = self._separate_at_paren(arg_str) else: arg_str, remaining = None, arg_str @@ -795,7 +805,7 @@ class JSInterpreter(object): \((?P[^)]*)\)\s* (?P{.+})''' % {'name': re.escape(funcname)}, self.code) - code, _ = self._separate_at_paren(func_m.group('code'), '}') # refine the match + code, _ = self._separate_at_paren(func_m.group('code')) # refine the match if func_m is None: raise self.Exception('Could not find JS function "{funcname}"'.format(**locals())) return self.build_arglist(func_m.group('args')), code @@ -810,7 +820,7 @@ class JSInterpreter(object): if mobj is None: break start, body_start = mobj.span() - body, remaining = self._separate_at_paren(code[body_start - 1:], '}') + body, remaining = self._separate_at_paren(code[body_start - 1:]) name = self._named_object( local_vars, self.extract_function_from_code( From cdfc866e6453c5d437335fd19e70997e5eacfe4b Mon Sep 17 00:00:00 2001 From: dirkf Date: Fri, 26 Aug 2022 10:17:56 +0100 Subject: [PATCH 16/21] [compat] Replace deficient ChainMap class in Py3.3 and earlier --- youtube_dl/compat.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/youtube_dl/compat.py b/youtube_dl/compat.py index 3002109ca..366a93924 100644 --- a/youtube_dl/compat.py +++ b/youtube_dl/compat.py @@ -3004,8 +3004,11 @@ except ImportError: # new class in collections try: from collections import ChainMap as compat_collections_chain_map + # Py3.3's ChainMap is deficient + if sys.version_info <= (3, 3): + raise ImportError except ImportError: - # Py < 3.3 + # Py <= 3.3 class compat_collections_chain_map(compat_collections_abc.MutableMapping): maps = [{}] @@ -3060,6 +3063,7 @@ except ImportError: def parents(self): return compat_collections_chain_map(*(self.maps[1:])) + # Pythons disagree on the type of a pattern (RegexObject, _sre.SRE_Pattern, Pattern, ...?) compat_re_Pattern = type(re.compile('')) From 7b4e2290bebddb4e2504596c92030e9eae93043e Mon Sep 17 00:00:00 2001 From: dirkf Date: Fri, 26 Aug 2022 12:22:01 +0100 Subject: [PATCH 17/21] [compat] Replace deficient ChainMap class in Py3.3 and earlier * fix version check --- youtube_dl/compat.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/youtube_dl/compat.py b/youtube_dl/compat.py index 366a93924..eca6d63de 100644 --- a/youtube_dl/compat.py +++ b/youtube_dl/compat.py @@ -3005,7 +3005,7 @@ except ImportError: try: from collections import ChainMap as compat_collections_chain_map # Py3.3's ChainMap is deficient - if sys.version_info <= (3, 3): + if sys.version_info < (3, 4): raise ImportError except ImportError: # Py <= 3.3 From bd41d33fc564d1d53000253e1ff0ae2fa954ff38 Mon Sep 17 00:00:00 2001 From: dirkf Date: Mon, 29 Aug 2022 13:02:17 +0100 Subject: [PATCH 18/21] [options] Document that postprocessing is not forced by --postprocessor-args Resolves #30307 --- youtube_dl/options.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/youtube_dl/options.py b/youtube_dl/options.py index f6621ef91..f6d2b0898 100644 --- a/youtube_dl/options.py +++ b/youtube_dl/options.py @@ -801,7 +801,7 @@ def parseOpts(overrideArguments=None): postproc.add_option( '--postprocessor-args', dest='postprocessor_args', metavar='ARGS', - help='Give these arguments to the postprocessor') + help='Give these arguments to the postprocessor (if postprocessing is required)') postproc.add_option( '-k', '--keep-video', action='store_true', dest='keepvideo', default=False, From b35aee2f97379113efcc8fc50f07af0c122594ad Mon Sep 17 00:00:00 2001 From: dirkf Date: Wed, 31 Aug 2022 23:22:48 +0100 Subject: [PATCH 19/21] [jsinterp] Handle new YT players 113ca41c, c57c113c * add NaN * allow any white-space character for `after_op` * align with yt-dlp f26af78a8ac11d9d617ed31ea5282cfaa5bcbcfa (charcodeAt and bitwise overflow) * allow escaping in regex, fixing player c57c113c --- test/test_jsinterp.py | 21 ++++++++++++++++ test/test_youtube_signature.py | 16 ++++++++++++ youtube_dl/jsinterp.py | 46 +++++++++++++++++++++------------- 3 files changed, 65 insertions(+), 18 deletions(-) diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py index fb4882d00..5121c8cf8 100644 --- a/test/test_jsinterp.py +++ b/test/test_jsinterp.py @@ -135,6 +135,11 @@ class TestJSInterpreter(unittest.TestCase): self.assertEqual(jsi.call_function('x'), [20, 20, 30, 40, 50]) def test_builtins(self): + jsi = JSInterpreter(''' + function x() { return NaN } + ''') + self.assertTrue(math.isnan(jsi.call_function('x'))) + jsi = JSInterpreter(''' function x() { return new Date('Wednesday 31 December 1969 18:01:26 MDT') - 0; } ''') @@ -385,6 +390,22 @@ class TestJSInterpreter(unittest.TestCase): ''') self.assertEqual(jsi.call_function('x').flags & ~re.U, re.I) + def test_char_code_at(self): + jsi = JSInterpreter('function x(i){return "test".charCodeAt(i)}') + self.assertEqual(jsi.call_function('x', 0), 116) + self.assertEqual(jsi.call_function('x', 1), 101) + self.assertEqual(jsi.call_function('x', 2), 115) + self.assertEqual(jsi.call_function('x', 3), 116) + self.assertEqual(jsi.call_function('x', 4), None) + self.assertEqual(jsi.call_function('x', 'not_a_number'), 116) + + def test_bitwise_operators_overflow(self): + jsi = JSInterpreter('function x(){return -524999584 << 5}') + self.assertEqual(jsi.call_function('x'), 379882496) + + jsi = JSInterpreter('function x(){return 1236566549 << 5}') + self.assertEqual(jsi.call_function('x'), 915423904) + if __name__ == '__main__': unittest.main() diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py index 4bb0a30b0..ec914a871 100644 --- a/test/test_youtube_signature.py +++ b/test/test_youtube_signature.py @@ -111,10 +111,26 @@ _NSIG_TESTS = [ 'https://www.youtube.com/s/player/1f7d5369/player_ias.vflset/en_US/base.js', 'batNX7sYqIJdkJ', 'IhOkL_zxbkOZBw', ), + ( + 'https://www.youtube.com/s/player/009f1d77/player_ias.vflset/en_US/base.js', + '5dwFHw8aFWQUQtffRq', 'audescmLUzI3jw', + ), ( 'https://www.youtube.com/s/player/dc0c6770/player_ias.vflset/en_US/base.js', '5EHDMgYLV6HPGk_Mu-kk', 'n9lUJLHbxUI0GQ', ), + ( + 'https://www.youtube.com/s/player/c2199353/player_ias.vflset/en_US/base.js', + '5EHDMgYLV6HPGk_Mu-kk', 'AD5rgS85EkrE7', + ), + ( + 'https://www.youtube.com/s/player/113ca41c/player_ias.vflset/en_US/base.js', + 'cgYl-tlYkhjT7A', 'hI7BBr2zUgcmMg', + ), + ( + 'https://www.youtube.com/s/player/c57c113c/player_ias.vflset/en_US/base.js', + '-Txvy6bT5R6LqgnQNx', 'dcklJCnRUHbgSg', + ), ] diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index 08726e478..d13329396 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -23,10 +23,11 @@ from .compat import ( def _js_bit_op(op): + def zeroise(x): + return 0 if x in (None, JS_Undefined) else x + def wrapped(a, b): - def zeroise(x): - return 0 if x in (None, JS_Undefined) else x - return op(zeroise(a), zeroise(b)) + return op(zeroise(a), zeroise(b)) & 0xffffffff return wrapped @@ -44,7 +45,7 @@ def _js_arith_op(op): def _js_div(a, b): if JS_Undefined in (a, b) or not (a and b): return float('nan') - return float('inf') if not b else operator.truediv(a or 0, b) + return operator.truediv(a or 0, b) if b else float('inf') def _js_mod(a, b): @@ -260,13 +261,14 @@ class JSInterpreter(object): counters[_MATCHING_PARENS[char]] += 1 elif char in counters: counters[char] -= 1 - if not escaping and char in _QUOTES and in_quote in (char, None): - if in_quote or after_op or char != '/': - in_quote = None if in_quote and not in_regex_char_group else char - elif in_quote == '/' and char in '[]': - in_regex_char_group = char == '[' + if not escaping: + if char in _QUOTES and in_quote in (char, None): + if in_quote or after_op or char != '/': + in_quote = None if in_quote and not in_regex_char_group else char + elif in_quote == '/' and char in '[]': + in_regex_char_group = char == '[' escaping = not escaping and in_quote and char == '\\' - after_op = not in_quote and char in cls.OP_CHARS or (char == ' ' and after_op) + after_op = not in_quote and (char in cls.OP_CHARS or (char.isspace() and after_op)) if char != delim[pos] or any(counters.values()) or in_quote: pos = skipping = 0 @@ -590,6 +592,8 @@ class JSInterpreter(object): elif expr == 'undefined': return JS_Undefined, should_return + elif expr == 'NaN': + return float('NaN'), should_return elif md.get('return'): return local_vars[m.group('name')], should_return @@ -635,7 +639,8 @@ class JSInterpreter(object): def assertion(cndn, msg): """ assert, but without risk of getting optimized out """ if not cndn: - raise ExtractorError('{member} {msg}'.format(**locals()), expr=expr) + memb = member + raise self.Exception('{member} {msg}'.format(**locals()), expr=expr) def eval_method(): if (variable, member) == ('console', 'debug'): @@ -737,6 +742,13 @@ class JSInterpreter(object): return obj.index(idx, start) except ValueError: return -1 + elif member == 'charCodeAt': + assertion(isinstance(obj, compat_str), 'must be applied on a string') + # assertion(len(argvals) == 1, 'takes exactly one argument') # but not enforced + idx = argvals[0] if isinstance(argvals[0], int) else 0 + if idx >= len(obj): + return None + return ord(obj[idx]) idx = int(member) if isinstance(obj, list) else member return obj[idx](argvals, allow_recursion=allow_recursion) @@ -820,12 +832,10 @@ class JSInterpreter(object): if mobj is None: break start, body_start = mobj.span() - body, remaining = self._separate_at_paren(code[body_start - 1:]) - name = self._named_object( - local_vars, - self.extract_function_from_code( - self.build_arglist(mobj.group('args')), - body, local_vars, *global_stack)) + body, remaining = self._separate_at_paren(code[body_start - 1:], '}') + name = self._named_object(local_vars, self.extract_function_from_code( + [x.strip() for x in mobj.group('args').split(',')], + body, local_vars, *global_stack)) code = code[:start] + name + remaining return self.build_function(argnames, code, local_vars, *global_stack) @@ -854,7 +864,7 @@ class JSInterpreter(object): zip_longest(argnames, args, fillvalue=None)) global_stack[0].update(kwargs) var_stack = LocalNameSpace(*global_stack) - ret, should_abort = self.interpret_statement(code.replace('\n', ''), var_stack, allow_recursion - 1) + ret, should_abort = self.interpret_statement(code.replace('\n', ' '), var_stack, allow_recursion - 1) if should_abort: return ret return resf From c342d50ce8b321437439ccb37db344c07c6c6e45 Mon Sep 17 00:00:00 2001 From: dirkf Date: Thu, 1 Sep 2022 13:28:30 +0100 Subject: [PATCH 20/21] [cache] Add cache validation by program version, based on yt-dlp --- test/test_cache.py | 16 ++++++++++++++-- youtube_dl/cache.py | 28 +++++++++++++++++++++++----- 2 files changed, 37 insertions(+), 7 deletions(-) diff --git a/test/test_cache.py b/test/test_cache.py index a16160142..931074aa1 100644 --- a/test/test_cache.py +++ b/test/test_cache.py @@ -3,17 +3,18 @@ from __future__ import unicode_literals -import shutil - # Allow direct execution import os import sys import unittest sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) +import shutil from test.helper import FakeYDL from youtube_dl.cache import Cache +from youtube_dl.utils import version_tuple +from youtube_dl.version import __version__ def _is_empty(d): @@ -54,6 +55,17 @@ class TestCache(unittest.TestCase): self.assertFalse(os.path.exists(self.test_dir)) self.assertEqual(c.load('test_cache', 'k.'), None) + def test_cache_validation(self): + ydl = FakeYDL({ + 'cachedir': self.test_dir, + }) + c = Cache(ydl) + obj = {'x': 1, 'y': ['ä', '\\a', True]} + c.store('test_cache', 'k.', obj) + self.assertEqual(c.load('test_cache', 'k.', min_ver='1970.01.01'), obj) + new_version = '.'.join(('%d' % ((v + 1) if i == 0 else v, )) for i, v in enumerate(version_tuple(__version__))) + self.assertIs(c.load('test_cache', 'k.', min_ver=new_version), None) + if __name__ == '__main__': unittest.main() diff --git a/youtube_dl/cache.py b/youtube_dl/cache.py index 7bdade1bd..4822439d0 100644 --- a/youtube_dl/cache.py +++ b/youtube_dl/cache.py @@ -10,12 +10,21 @@ import traceback from .compat import compat_getenv from .utils import ( + error_to_compat_str, expand_path, + is_outdated_version, + try_get, write_json_file, ) +from .version import __version__ class Cache(object): + + _YTDL_DIR = 'youtube-dl' + _VERSION_KEY = _YTDL_DIR + '_version' + _DEFAULT_VERSION = '2021.12.17' + def __init__(self, ydl): self._ydl = ydl @@ -23,7 +32,7 @@ class Cache(object): res = self._ydl.params.get('cachedir') if res is None: cache_root = compat_getenv('XDG_CACHE_HOME', '~/.cache') - res = os.path.join(cache_root, 'youtube-dl') + res = os.path.join(cache_root, self._YTDL_DIR) return expand_path(res) def _get_cache_fn(self, section, key, dtype): @@ -50,13 +59,22 @@ class Cache(object): except OSError as ose: if ose.errno != errno.EEXIST: raise - write_json_file(data, fn) + write_json_file({self._VERSION_KEY: __version__, 'data': data}, fn) except Exception: tb = traceback.format_exc() self._ydl.report_warning( 'Writing cache to %r failed: %s' % (fn, tb)) - def load(self, section, key, dtype='json', default=None): + def _validate(self, data, min_ver): + version = try_get(data, lambda x: x[self._VERSION_KEY]) + if not version: # Backward compatibility + data, version = {'data': data}, self._DEFAULT_VERSION + if not is_outdated_version(version, min_ver or '0', assume_new=False): + return data['data'] + self._ydl.to_screen( + 'Discarding old cache from version {version} (needs {min_ver})'.format(**locals())) + + def load(self, section, key, dtype='json', default=None, min_ver=None): assert dtype in ('json',) if not self.enabled: @@ -66,12 +84,12 @@ class Cache(object): try: try: with io.open(cache_fn, 'r', encoding='utf-8') as cachef: - return json.load(cachef) + return self._validate(json.load(cachef), min_ver) except ValueError: try: file_size = os.path.getsize(cache_fn) except (OSError, IOError) as oe: - file_size = str(oe) + file_size = error_to_compat_str(oe) self._ydl.report_warning( 'Cache retrieval from %s failed (%s)' % (cache_fn, file_size)) except IOError: From 5662dad552f3797c63d1ef980fca83f527e6c8c7 Mon Sep 17 00:00:00 2001 From: pukkandan Date: Fri, 2 Sep 2022 20:41:39 +0530 Subject: [PATCH 21/21] [jsinterp] Workaround operator associativity issue * temporary fix for player 5a3b6271 [1] 1. https://github.com/yt-dlp/yt-dlp/issues/4635#issuecomment-1235384480 --- test/test_youtube_signature.py | 4 ++++ youtube_dl/jsinterp.py | 2 +- 2 files changed, 5 insertions(+), 1 deletion(-) diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py index ec914a871..4e678cae0 100644 --- a/test/test_youtube_signature.py +++ b/test/test_youtube_signature.py @@ -131,6 +131,10 @@ _NSIG_TESTS = [ 'https://www.youtube.com/s/player/c57c113c/player_ias.vflset/en_US/base.js', '-Txvy6bT5R6LqgnQNx', 'dcklJCnRUHbgSg', ), + ( + 'https://www.youtube.com/s/player/5a3b6271/player_ias.vflset/en_US/base.js', + 'B2j7f_UPT4rfje85Lu_e', 'm5DmNymaGQ5RdQ', + ), ] diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index d13329396..99dd98435 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -107,8 +107,8 @@ _OPERATORS = ( ('+', _js_arith_op(operator.add)), ('-', _js_arith_op(operator.sub)), ('*', _js_arith_op(operator.mul)), - ('/', _js_div), ('%', _js_mod), + ('/', _js_div), ('**', _js_exp), )