X-Git-Url: https://jfr.im/git/yt-dlp.git/blobdiff_plain/7dbce2c53266ebedc824d4263b08911770234c35..bef9a9e5361fd7a72e21d0f1a8c8afb70d89e8c5:/test/test_YoutubeDL.py diff --git a/test/test_YoutubeDL.py b/test/test_YoutubeDL.py index bacab60a4..5242cf88f 100644 --- a/test/test_YoutubeDL.py +++ b/test/test_YoutubeDL.py @@ -1,39 +1,49 @@ -#!/usr/bin/env python -# coding: utf-8 - -from __future__ import unicode_literals +#!/usr/bin/env python3 # Allow direct execution import os import sys import unittest + sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) -import copy -from test.helper import FakeYDL, assertRegexpMatches -from youtube_dlc import YoutubeDL -from youtube_dlc.compat import compat_str, compat_urllib_error -from youtube_dlc.extractor import YoutubeIE -from youtube_dlc.extractor.common import InfoExtractor -from youtube_dlc.postprocessor.common import PostProcessor -from youtube_dlc.utils import ExtractorError, match_filter_func +import copy +import json + +from test.helper import FakeYDL, assertRegexpMatches, try_rm +from yt_dlp import YoutubeDL +from yt_dlp.compat import compat_os_name +from yt_dlp.extractor import YoutubeIE +from yt_dlp.extractor.common import InfoExtractor +from yt_dlp.postprocessor.common import PostProcessor +from yt_dlp.utils import ( + ExtractorError, + LazyList, + OnDemandPagedList, + int_or_none, + match_filter_func, +) +from yt_dlp.utils.traversal import traverse_obj TEST_URL = 'http://localhost/sample.mp4' class YDL(FakeYDL): def __init__(self, *args, **kwargs): - super(YDL, self).__init__(*args, **kwargs) + super().__init__(*args, **kwargs) self.downloaded_info_dicts = [] self.msgs = [] def process_info(self, info_dict): - self.downloaded_info_dicts.append(info_dict) + self.downloaded_info_dicts.append(info_dict.copy()) - def to_screen(self, msg): + def to_screen(self, msg, *args, **kwargs): self.msgs.append(msg) + def dl(self, *args, **kwargs): + assert False, 'Downloader must not be invoked for test_YoutubeDL' + def _make_result(formats, **kwargs): res = { @@ -58,8 +68,7 @@ def test_prefer_free_formats(self): {'ext': 'mp4', 'height': 460, 'url': TEST_URL}, ] info_dict = _make_result(formats) - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['ext'], 'webm') @@ -72,8 +81,7 @@ def test_prefer_free_formats(self): {'ext': 'mp4', 'height': 1080, 'url': TEST_URL}, ] info_dict['formats'] = formats - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['ext'], 'mp4') @@ -87,8 +95,7 @@ def test_prefer_free_formats(self): {'ext': 'flv', 'height': 720, 'url': TEST_URL}, ] info_dict['formats'] = formats - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['ext'], 'mp4') @@ -100,15 +107,14 @@ def test_prefer_free_formats(self): {'ext': 'webm', 'height': 720, 'url': TEST_URL}, ] info_dict['formats'] = formats - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['ext'], 'webm') def test_format_selection(self): formats = [ - {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL}, + {'format_id': '35', 'ext': 'mp4', 'preference': 0, 'url': TEST_URL}, {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL}, {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL}, {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL}, @@ -116,35 +122,26 @@ def test_format_selection(self): ] info_dict = _make_result(formats) - ydl = YDL({'format': '20/47'}) - ydl.process_ie_result(info_dict.copy()) - downloaded = ydl.downloaded_info_dicts[0] - self.assertEqual(downloaded['format_id'], '47') - - ydl = YDL({'format': '20/71/worst'}) - ydl.process_ie_result(info_dict.copy()) - downloaded = ydl.downloaded_info_dicts[0] - self.assertEqual(downloaded['format_id'], '35') - - ydl = YDL() - ydl.process_ie_result(info_dict.copy()) - downloaded = ydl.downloaded_info_dicts[0] - self.assertEqual(downloaded['format_id'], '2') - - ydl = YDL({'format': 'webm/mp4'}) - ydl.process_ie_result(info_dict.copy()) - downloaded = ydl.downloaded_info_dicts[0] - self.assertEqual(downloaded['format_id'], '47') - - ydl = YDL({'format': '3gp/40/mp4'}) - ydl.process_ie_result(info_dict.copy()) - downloaded = ydl.downloaded_info_dicts[0] - self.assertEqual(downloaded['format_id'], '35') - - ydl = YDL({'format': 'example-with-dashes'}) - ydl.process_ie_result(info_dict.copy()) - downloaded = ydl.downloaded_info_dicts[0] - self.assertEqual(downloaded['format_id'], 'example-with-dashes') + def test(inp, *expected, multi=False): + ydl = YDL({ + 'format': inp, + 'allow_multiple_video_streams': multi, + 'allow_multiple_audio_streams': multi, + }) + ydl.process_ie_result(info_dict.copy()) + downloaded = map(lambda x: x['format_id'], ydl.downloaded_info_dicts) + self.assertEqual(list(downloaded), list(expected)) + + test('20/47', '47') + test('20/71/worst', '35') + test(None, '2') + test('webm/mp4', '47') + test('3gp/40/mp4', '35') + test('example-with-dashes', 'example-with-dashes') + test('all', '2', '47', '45', 'example-with-dashes', '35') + test('mergeall', '2+47+45+example-with-dashes+35', multi=True) + # See: https://github.com/yt-dlp/yt-dlp/pulls/8797 + test('7_a/worst', '35') def test_format_selection_audio(self): formats = [ @@ -186,23 +183,20 @@ def test_format_selection_audio_exts(self): ] info_dict = _make_result(formats) - ydl = YDL({'format': 'best'}) - ie = YoutubeIE(ydl) - ie._sort_formats(info_dict['formats']) + ydl = YDL({'format': 'best', 'format_sort': ['abr', 'ext']}) + ydl.sort_formats(info_dict) ydl.process_ie_result(copy.deepcopy(info_dict)) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['format_id'], 'aac-64') ydl = YDL({'format': 'mp3'}) - ie = YoutubeIE(ydl) - ie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(copy.deepcopy(info_dict)) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['format_id'], 'mp3-64') - ydl = YDL({'prefer_free_formats': True}) - ie = YoutubeIE(ydl) - ie._sort_formats(info_dict['formats']) + ydl = YDL({'prefer_free_formats': True, 'format_sort': ['abr', 'ext']}) + ydl.sort_formats(info_dict) ydl.process_ie_result(copy.deepcopy(info_dict)) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['format_id'], 'ogg-64') @@ -311,8 +305,8 @@ def test_format_selection_string_ops(self): self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy()) def test_youtube_format_selection(self): + # FIXME: Rewrite in accordance with the new format sorting options return - # disabled for now - this needs some changes order = [ '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13', @@ -347,8 +341,7 @@ def format_info(f_id): info_dict = _make_result(list(formats_order), extractor='youtube') ydl = YDL({'format': 'bestvideo+bestaudio'}) - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['format_id'], '248+172') @@ -356,40 +349,35 @@ def format_info(f_id): info_dict = _make_result(list(formats_order), extractor='youtube') ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'}) - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['format_id'], '38') info_dict = _make_result(list(formats_order), extractor='youtube') ydl = YDL({'format': 'bestvideo/best,bestaudio'}) - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts] self.assertEqual(downloaded_ids, ['137', '141']) info_dict = _make_result(list(formats_order), extractor='youtube') ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'}) - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts] self.assertEqual(downloaded_ids, ['137+141', '248+141']) info_dict = _make_result(list(formats_order), extractor='youtube') ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'}) - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts] self.assertEqual(downloaded_ids, ['136+141', '247+141']) info_dict = _make_result(list(formats_order), extractor='youtube') ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'}) - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts] self.assertEqual(downloaded_ids, ['248+141']) @@ -397,16 +385,14 @@ def format_info(f_id): for f1, f2 in zip(formats_order, formats_order[1:]): info_dict = _make_result([f1, f2], extractor='youtube') ydl = YDL({'format': 'best/bestvideo'}) - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['format_id'], f1['format_id']) info_dict = _make_result([f2, f1], extractor='youtube') ydl = YDL({'format': 'best/bestvideo'}) - yie = YoutubeIE(ydl) - yie._sort_formats(info_dict['formats']) + ydl.sort_formats(info_dict) ydl.process_ie_result(info_dict) downloaded = ydl.downloaded_info_dicts[0] self.assertEqual(downloaded['format_id'], f1['format_id']) @@ -460,14 +446,13 @@ def test_format_selection_issue_10083(self): def test_invalid_format_specs(self): def assert_syntax_error(format_spec): - ydl = YDL({'format': format_spec}) - info_dict = _make_result([{'format_id': 'foo', 'url': TEST_URL}]) - self.assertRaises(SyntaxError, ydl.process_ie_result, info_dict) + self.assertRaises(SyntaxError, YDL, {'format': format_spec}) assert_syntax_error('bestvideo,,best') assert_syntax_error('+bestaudio') assert_syntax_error('bestvideo+') assert_syntax_error('/') + assert_syntax_error('[720=400][width<=600]'}) ydl.process_ie_result(info_dict) downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts] - self.assertEqual(downloaded_ids, ['B', 'C', 'D']) + self.assertEqual(downloaded_ids, ['D', 'C', 'B']) ydl = YDL({'format': 'best[height<40]'}) try: @@ -559,11 +544,11 @@ def test_subtitles(self): def s_formats(lang, autocaption=False): return [{ 'ext': ext, - 'url': 'http://localhost/video.%s.%s' % (lang, ext), + 'url': f'http://localhost/video.{lang}.{ext}', '_auto': autocaption, } for ext in ['vtt', 'srt', 'ass']] - subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es']) - auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es']) + subtitles = {l: s_formats(l) for l in ['en', 'fr', 'es']} + auto_captions = {l: s_formats(l, True) for l in ['it', 'pt', 'es']} info_dict = { 'id': 'test', 'title': 'Test', @@ -588,7 +573,7 @@ def get_info(params={}): result = get_info({'writesubtitles': True}) subs = result['requested_subtitles'] self.assertTrue(subs) - self.assertEqual(set(subs.keys()), set(['en'])) + self.assertEqual(set(subs.keys()), {'en'}) self.assertTrue(subs['en'].get('data') is None) self.assertEqual(subs['en']['ext'], 'ass') @@ -599,19 +584,39 @@ def get_info(params={}): result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']}) subs = result['requested_subtitles'] self.assertTrue(subs) - self.assertEqual(set(subs.keys()), set(['es', 'fr'])) + self.assertEqual(set(subs.keys()), {'es', 'fr'}) + + result = get_info({'writesubtitles': True, 'subtitleslangs': ['all', '-en']}) + subs = result['requested_subtitles'] + self.assertTrue(subs) + self.assertEqual(set(subs.keys()), {'es', 'fr'}) + + result = get_info({'writesubtitles': True, 'subtitleslangs': ['en', 'fr', '-en']}) + subs = result['requested_subtitles'] + self.assertTrue(subs) + self.assertEqual(set(subs.keys()), {'fr'}) + + result = get_info({'writesubtitles': True, 'subtitleslangs': ['-en', 'en']}) + subs = result['requested_subtitles'] + self.assertTrue(subs) + self.assertEqual(set(subs.keys()), {'en'}) + + result = get_info({'writesubtitles': True, 'subtitleslangs': ['e.+']}) + subs = result['requested_subtitles'] + self.assertTrue(subs) + self.assertEqual(set(subs.keys()), {'es', 'en'}) result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']}) subs = result['requested_subtitles'] self.assertTrue(subs) - self.assertEqual(set(subs.keys()), set(['es', 'pt'])) + self.assertEqual(set(subs.keys()), {'es', 'pt'}) self.assertFalse(subs['es']['_auto']) self.assertTrue(subs['pt']['_auto']) result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']}) subs = result['requested_subtitles'] self.assertTrue(subs) - self.assertEqual(set(subs.keys()), set(['es', 'pt'])) + self.assertEqual(set(subs.keys()), {'es', 'pt'}) self.assertTrue(subs['es']['_auto']) self.assertTrue(subs['pt']['_auto']) @@ -627,40 +632,236 @@ def test_add_extra_info(self): self.assertEqual(test_dict['extractor'], 'Foo') self.assertEqual(test_dict['playlist'], 'funny videos') - def test_prepare_filename(self): - info = { - 'id': '1234', - 'ext': 'mp4', - 'width': None, - 'height': 1080, - 'title1': '$PATH', - 'title2': '%PATH%', - } + outtmpl_info = { + 'id': '1234', + 'ext': 'mp4', + 'width': None, + 'height': 1080, + 'filesize': 1024, + 'title1': '$PATH', + 'title2': '%PATH%', + 'title3': 'foo/bar\\test', + 'title4': 'foo "bar" test', + 'title5': 'áéí 𝐀', + 'timestamp': 1618488000, + 'duration': 100000, + 'playlist_index': 1, + 'playlist_autonumber': 2, + '__last_playlist_index': 100, + 'n_entries': 10, + 'formats': [ + {'id': 'id 1', 'height': 1080, 'width': 1920}, + {'id': 'id 2', 'height': 720}, + {'id': 'id 3'} + ] + } - def fname(templ): - ydl = YoutubeDL({'outtmpl': templ}) - return ydl.prepare_filename(info) - self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4') - self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4') - # Replace missing fields with 'NA' - self.assertEqual(fname('%(uploader_date)s-%(id)s.%(ext)s'), 'NA-1234.mp4') - self.assertEqual(fname('%(height)d.%(ext)s'), '1080.mp4') - self.assertEqual(fname('%(height)6d.%(ext)s'), ' 1080.mp4') - self.assertEqual(fname('%(height)-6d.%(ext)s'), '1080 .mp4') - self.assertEqual(fname('%(height)06d.%(ext)s'), '001080.mp4') - self.assertEqual(fname('%(height) 06d.%(ext)s'), ' 01080.mp4') - self.assertEqual(fname('%(height) 06d.%(ext)s'), ' 01080.mp4') - self.assertEqual(fname('%(height)0 6d.%(ext)s'), ' 01080.mp4') - self.assertEqual(fname('%(height)0 6d.%(ext)s'), ' 01080.mp4') - self.assertEqual(fname('%(height) 0 6d.%(ext)s'), ' 01080.mp4') - self.assertEqual(fname('%%'), '%') - self.assertEqual(fname('%%%%'), '%%') - self.assertEqual(fname('%%(height)06d.%(ext)s'), '%(height)06d.mp4') - self.assertEqual(fname('%(width)06d.%(ext)s'), 'NA.mp4') - self.assertEqual(fname('%(width)06d.%%(ext)s'), 'NA.%(ext)s') - self.assertEqual(fname('%%(width)06d.%(ext)s'), '%(width)06d.mp4') - self.assertEqual(fname('Hello %(title1)s'), 'Hello $PATH') - self.assertEqual(fname('Hello %(title2)s'), 'Hello %PATH%') + def test_prepare_outtmpl_and_filename(self): + def test(tmpl, expected, *, info=None, **params): + params['outtmpl'] = tmpl + ydl = FakeYDL(params) + ydl._num_downloads = 1 + self.assertEqual(ydl.validate_outtmpl(tmpl), None) + + out = ydl.evaluate_outtmpl(tmpl, info or self.outtmpl_info) + fname = ydl.prepare_filename(info or self.outtmpl_info) + + if not isinstance(expected, (list, tuple)): + expected = (expected, expected) + for (name, got), expect in zip((('outtmpl', out), ('filename', fname)), expected): + if callable(expect): + self.assertTrue(expect(got), f'Wrong {name} from {tmpl}') + elif expect is not None: + self.assertEqual(got, expect, f'Wrong {name} from {tmpl}') + + # Side-effects + original_infodict = dict(self.outtmpl_info) + test('foo.bar', 'foo.bar') + original_infodict['epoch'] = self.outtmpl_info.get('epoch') + self.assertTrue(isinstance(original_infodict['epoch'], int)) + test('%(epoch)d', int_or_none) + self.assertEqual(original_infodict, self.outtmpl_info) + + # Auto-generated fields + test('%(id)s.%(ext)s', '1234.mp4') + test('%(duration_string)s', ('27:46:40', '27-46-40')) + test('%(resolution)s', '1080p') + test('%(playlist_index|)s', '001') + test('%(playlist_index&{}!)s', '1!') + test('%(playlist_autonumber)s', '02') + test('%(autonumber)s', '00001') + test('%(autonumber+2)03d', '005', autonumber_start=3) + test('%(autonumber)s', '001', autonumber_size=3) + + # Escaping % + test('%', '%') + test('%%', '%') + test('%%%%', '%%') + test('%s', '%s') + test('%%%s', '%%s') + test('%d', '%d') + test('%abc%', '%abc%') + test('%%(width)06d.%(ext)s', '%(width)06d.mp4') + test('%%%(height)s', '%1080') + test('%(width)06d.%(ext)s', 'NA.mp4') + test('%(width)06d.%%(ext)s', 'NA.%(ext)s') + test('%%(width)06d.%(ext)s', '%(width)06d.mp4') + + # ID sanitization + test('%(id)s', '_abcd', info={'id': '_abcd'}) + test('%(some_id)s', '_abcd', info={'some_id': '_abcd'}) + test('%(formats.0.id)s', '_abcd', info={'formats': [{'id': '_abcd'}]}) + test('%(id)s', '-abcd', info={'id': '-abcd'}) + test('%(id)s', '.abcd', info={'id': '.abcd'}) + test('%(id)s', 'ab__cd', info={'id': 'ab__cd'}) + test('%(id)s', ('ab:cd', 'ab：cd'), info={'id': 'ab:cd'}) + test('%(id.0)s', '-', info={'id': '--'}) + + # Invalid templates + self.assertTrue(isinstance(YoutubeDL.validate_outtmpl('%(title)'), ValueError)) + test('%(invalid@tmpl|def)s', 'none', outtmpl_na_placeholder='none') + test('%(..)s', 'NA') + test('%(formats.{id)s', 'NA') + + # Entire info_dict + def expect_same_infodict(out): + got_dict = json.loads(out) + for info_field, expected in self.outtmpl_info.items(): + self.assertEqual(got_dict.get(info_field), expected, info_field) + return True + + test('%()j', (expect_same_infodict, None)) + + # NA placeholder + NA_TEST_OUTTMPL = '%(uploader_date)s-%(width)d-%(x|def)s-%(id)s.%(ext)s' + test(NA_TEST_OUTTMPL, 'NA-NA-def-1234.mp4') + test(NA_TEST_OUTTMPL, 'none-none-def-1234.mp4', outtmpl_na_placeholder='none') + test(NA_TEST_OUTTMPL, '--def-1234.mp4', outtmpl_na_placeholder='') + test('%(non_existent.0)s', 'NA') + + # String formatting + FMT_TEST_OUTTMPL = '%%(height)%s.%%(ext)s' + test(FMT_TEST_OUTTMPL % 's', '1080.mp4') + test(FMT_TEST_OUTTMPL % 'd', '1080.mp4') + test(FMT_TEST_OUTTMPL % '6d', ' 1080.mp4') + test(FMT_TEST_OUTTMPL % '-6d', '1080 .mp4') + test(FMT_TEST_OUTTMPL % '06d', '001080.mp4') + test(FMT_TEST_OUTTMPL % ' 06d', ' 01080.mp4') + test(FMT_TEST_OUTTMPL % ' 06d', ' 01080.mp4') + test(FMT_TEST_OUTTMPL % '0 6d', ' 01080.mp4') + test(FMT_TEST_OUTTMPL % '0 6d', ' 01080.mp4') + test(FMT_TEST_OUTTMPL % ' 0 6d', ' 01080.mp4') + + # Type casting + test('%(id)d', '1234') + test('%(height)c', '1') + test('%(ext)c', 'm') + test('%(id)d %(id)r', "1234 '1234'") + test('%(id)r %(height)r', "'1234' 1080") + test('%(title5)a %(height)a', (R"'\xe1\xe9\xed \U0001d400' 1080", None)) + test('%(ext)s-%(ext|def)d', 'mp4-def') + test('%(width|0)04d', '0') + test('a%(width|b)d', 'ab', outtmpl_na_placeholder='none') + + FORMATS = self.outtmpl_info['formats'] + + # Custom type casting + test('%(formats.:.id)l', 'id 1, id 2, id 3') + test('%(formats.:.id)#l', ('id 1\nid 2\nid 3', 'id 1 id 2 id 3')) + test('%(ext)l', 'mp4') + test('%(formats.:.id) 18l', ' id 1, id 2, id 3') + test('%(formats)j', (json.dumps(FORMATS), None)) + test('%(formats)#j', ( + json.dumps(FORMATS, indent=4), + json.dumps(FORMATS, indent=4).replace(':', '：').replace('"', "＂").replace('\n', ' ') + )) + test('%(title5).3B', 'á') + test('%(title5)U', 'áéí 𝐀') + test('%(title5)#U', 'a\u0301e\u0301i\u0301 𝐀') + test('%(title5)+U', 'áéí A') + test('%(title5)+#U', 'a\u0301e\u0301i\u0301 A') + test('%(height)D', '1k') + test('%(filesize)#D', '1Ki') + test('%(height)5.2D', ' 1.08k') + test('%(title4)#S', 'foo_bar_test') + test('%(title4).10S', ('foo ＂bar＂ ', 'foo ＂bar＂' + ('#' if compat_os_name == 'nt' else ' '))) + if compat_os_name == 'nt': + test('%(title4)q', ('"foo ""bar"" test"', None)) + test('%(formats.:.id)#q', ('"id 1" "id 2" "id 3"', None)) + test('%(formats.0.id)#q', ('"id 1"', None)) + else: + test('%(title4)q', ('\'foo "bar" test\'', '\'foo ＂bar＂ test\'')) + test('%(formats.:.id)#q', "'id 1' 'id 2' 'id 3'") + test('%(formats.0.id)#q', "'id 1'") + + # Internal formatting + test('%(timestamp-1000>%H-%M-%S)s', '11-43-20') + test('%(title|%)s %(title|%%)s', '% %%') + test('%(id+1-height+3)05d', '00158') + test('%(width+100)05d', 'NA') + test('%(filesize*8)d', '8192') + test('%(formats.0) 15s', ('% 15s' % FORMATS[0], None)) + test('%(formats.0)r', (repr(FORMATS[0]), None)) + test('%(height.0)03d', '001') + test('%(-height.0)04d', '-001') + test('%(formats.-1.id)s', FORMATS[-1]['id']) + test('%(formats.0.id.-1)d', FORMATS[0]['id'][-1]) + test('%(formats.3)s', 'NA') + test('%(formats.:2:-1)r', repr(FORMATS[:2:-1])) + test('%(formats.0.id.-1+id)f', '1235.000000') + test('%(formats.0.id.-1+formats.1.id.-1)d', '3') + out = json.dumps([{'id': f['id'], 'height.:2': str(f['height'])[:2]} + if 'height' in f else {'id': f['id']} + for f in FORMATS]) + test('%(formats.:.{id,height.:2})j', (out, None)) + test('%(formats.:.{id,height}.id)l', ', '.join(f['id'] for f in FORMATS)) + test('%(.{id,title})j', ('{"id": "1234"}', '{＂id＂： ＂1234＂}')) + + # Alternates + test('%(title,id)s', '1234') + test('%(width-100,height+20|def)d', '1100') + test('%(width-100,height+width|def)s', 'def') + test('%(timestamp-x>%H\\,%M\\,%S,timestamp>%H\\,%M\\,%S)s', '12,00,00') + + # Replacement + test('%(id&foo)s.bar', 'foo.bar') + test('%(title&foo)s.bar', 'NA.bar') + test('%(title&foo|baz)s.bar', 'baz.bar') + test('%(x,id&foo|baz)s.bar', 'foo.bar') + test('%(x,title&foo|baz)s.bar', 'baz.bar') + test('%(id&a\nb|)s', ('a\nb', 'a b')) + test('%(id&hi {:>10} {}|)s', 'hi 1234 1234') + test(R'%(id&{0} {}|)s', 'NA') + test(R'%(id&{0.1}|)s', 'NA') + test('%(height&{:,d})S', '1,080') + + # Laziness + def gen(): + yield from range(5) + raise self.assertTrue(False, 'LazyList should not be evaluated till here') + test('%(key.4)s', '4', info={'key': LazyList(gen())}) + + # Empty filename + test('%(foo|)s-%(bar|)s.%(ext)s', '-.mp4') + # test('%(foo|)s.%(ext)s', ('.mp4', '_.mp4')) # fixme + # test('%(foo|)s', ('', '_')) # fixme + + # Environment variable expansion for prepare_filename + os.environ['__yt_dlp_var'] = 'expanded' + envvar = '%__yt_dlp_var%' if compat_os_name == 'nt' else '$__yt_dlp_var' + test(envvar, (envvar, 'expanded')) + if compat_os_name == 'nt': + test('%s%', ('%s%', '%s%')) + os.environ['s'] = 'expanded' + test('%s%', ('%s%', 'expanded')) # %s% should be expanded before escaping %s + os.environ['(test)s'] = 'expanded' + test('%(test)s%', ('NA%', 'expanded')) # Environment should take priority over template + + # Path expansion and escaping + test('Hello %(title1)s', 'Hello $PATH') + test('Hello %(title2)s', 'Hello %PATH%') + test('%(title3)s', ('foo/bar\\test', 'foo⧸bar⧹test')) + test('folder/%(title3)s', ('folder/foo/bar\\test', 'folder%sfoo⧸bar⧹test' % os.path.sep)) def test_format_note(self): ydl = YoutubeDL() @@ -678,12 +879,12 @@ def test_postprocessors(self): class SimplePP(PostProcessor): def run(self, info): - with open(audiofile, 'wt') as f: + with open(audiofile, 'w') as f: f.write('EXAMPLE') return [info['filepath']], info def run_pp(params, PP): - with open(filename, 'wt') as f: + with open(filename, 'w') as f: f.write('EXAMPLE') ydl = YoutubeDL(params) ydl.add_post_processor(PP()) @@ -702,7 +903,7 @@ def run_pp(params, PP): class ModifierPP(PostProcessor): def run(self, info): - with open(info['filepath'], 'wt') as f: + with open(info['filepath'], 'w') as f: f.write('MODIFIED') return [], info @@ -711,20 +912,6 @@ def run(self, info): os.unlink(filename) def test_match_filter(self): - class FilterYDL(YDL): - def __init__(self, *args, **kwargs): - super(FilterYDL, self).__init__(*args, **kwargs) - self.params['simulate'] = True - - def process_info(self, info_dict): - super(YDL, self).process_info(info_dict) - - def _match_entry(self, info_dict, incomplete): - res = super(FilterYDL, self)._match_entry(info_dict, incomplete) - if res is None: - self.downloaded_info_dicts.append(info_dict) - return res - first = { 'id': '1', 'url': TEST_URL, @@ -752,15 +939,15 @@ def _match_entry(self, info_dict, incomplete): videos = [first, second] def get_videos(filter_=None): - ydl = FilterYDL({'match_filter': filter_}) + ydl = YDL({'match_filter': filter_, 'simulate': True}) for v in videos: - ydl.process_ie_result(v, download=True) + ydl.process_ie_result(v.copy(), download=True) return [v['id'] for v in ydl.downloaded_info_dicts] res = get_videos() self.assertEqual(res, ['1', '2']) - def f(v): + def f(v, incomplete): if v['id'] == '1': return None else: @@ -809,80 +996,114 @@ def f(v): self.assertEqual(res, []) def test_playlist_items_selection(self): - entries = [{ - 'id': compat_str(i), - 'title': compat_str(i), - 'url': TEST_URL, - } for i in range(1, 5)] - playlist = { - '_type': 'playlist', - 'id': 'test', - 'entries': entries, - 'extractor': 'test:playlist', - 'extractor_key': 'test:playlist', - 'webpage_url': 'http://example.com', - } - - def get_downloaded_info_dicts(params): - ydl = YDL(params) - # make a deep copy because the dictionary and nested entries - # can be modified - ydl.process_ie_result(copy.deepcopy(playlist)) - return ydl.downloaded_info_dicts - - def get_ids(params): - return [int(v['id']) for v in get_downloaded_info_dicts(params)] + INDICES, PAGE_SIZE = list(range(1, 11)), 3 - result = get_ids({}) - self.assertEqual(result, [1, 2, 3, 4]) + def entry(i, evaluated): + evaluated.append(i) + return { + 'id': str(i), + 'title': str(i), + 'url': TEST_URL, + } - result = get_ids({'playlistend': 10}) - self.assertEqual(result, [1, 2, 3, 4]) + def pagedlist_entries(evaluated): + def page_func(n): + start = PAGE_SIZE * n + for i in INDICES[start: start + PAGE_SIZE]: + yield entry(i, evaluated) + return OnDemandPagedList(page_func, PAGE_SIZE) - result = get_ids({'playlistend': 2}) - self.assertEqual(result, [1, 2]) + def page_num(i): + return (i + PAGE_SIZE - 1) // PAGE_SIZE - result = get_ids({'playliststart': 10}) - self.assertEqual(result, []) + def generator_entries(evaluated): + for i in INDICES: + yield entry(i, evaluated) - result = get_ids({'playliststart': 2}) - self.assertEqual(result, [2, 3, 4]) + def list_entries(evaluated): + return list(generator_entries(evaluated)) - result = get_ids({'playlist_items': '2-4'}) - self.assertEqual(result, [2, 3, 4]) + def lazylist_entries(evaluated): + return LazyList(generator_entries(evaluated)) - result = get_ids({'playlist_items': '2,4'}) - self.assertEqual(result, [2, 4]) - - result = get_ids({'playlist_items': '10'}) - self.assertEqual(result, []) - - result = get_ids({'playlist_items': '3-10'}) - self.assertEqual(result, [3, 4]) + def get_downloaded_info_dicts(params, entries): + ydl = YDL(params) + ydl.process_ie_result({ + '_type': 'playlist', + 'id': 'test', + 'extractor': 'test:playlist', + 'extractor_key': 'test:playlist', + 'webpage_url': 'http://example.com', + 'entries': entries, + }) + return ydl.downloaded_info_dicts - result = get_ids({'playlist_items': '2-4,3-4,3'}) - self.assertEqual(result, [2, 3, 4]) + def test_selection(params, expected_ids, evaluate_all=False): + expected_ids = list(expected_ids) + if evaluate_all: + generator_eval = pagedlist_eval = INDICES + elif not expected_ids: + generator_eval = pagedlist_eval = [] + else: + generator_eval = INDICES[0: max(expected_ids)] + pagedlist_eval = INDICES[PAGE_SIZE * page_num(min(expected_ids)) - PAGE_SIZE: + PAGE_SIZE * page_num(max(expected_ids))] + + for name, func, expected_eval in ( + ('list', list_entries, INDICES), + ('Generator', generator_entries, generator_eval), + # ('LazyList', lazylist_entries, generator_eval), # Generator and LazyList follow the exact same code path + ('PagedList', pagedlist_entries, pagedlist_eval), + ): + evaluated = [] + entries = func(evaluated) + results = [(v['playlist_autonumber'] - 1, (int(v['id']), v['playlist_index'])) + for v in get_downloaded_info_dicts(params, entries)] + self.assertEqual(results, list(enumerate(zip(expected_ids, expected_ids))), f'Entries of {name} for {params}') + self.assertEqual(sorted(evaluated), expected_eval, f'Evaluation of {name} for {params}') + + test_selection({}, INDICES) + test_selection({'playlistend': 20}, INDICES, True) + test_selection({'playlistend': 2}, INDICES[:2]) + test_selection({'playliststart': 11}, [], True) + test_selection({'playliststart': 2}, INDICES[1:]) + test_selection({'playlist_items': '2-4'}, INDICES[1:4]) + test_selection({'playlist_items': '2,4'}, [2, 4]) + test_selection({'playlist_items': '20'}, [], True) + test_selection({'playlist_items': '0'}, []) # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591 - # @{ - result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'}) - self.assertEqual(result[0]['playlist_index'], 2) - self.assertEqual(result[1]['playlist_index'], 3) - - result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'}) - self.assertEqual(result[0]['playlist_index'], 2) - self.assertEqual(result[1]['playlist_index'], 3) - self.assertEqual(result[2]['playlist_index'], 4) - - result = get_downloaded_info_dicts({'playlist_items': '4,2'}) - self.assertEqual(result[0]['playlist_index'], 4) - self.assertEqual(result[1]['playlist_index'], 2) - # @} - - def test_urlopen_no_file_protocol(self): - # see https://github.com/ytdl-org/youtube-dl/issues/8227 - ydl = YDL() - self.assertRaises(compat_urllib_error.URLError, ydl.urlopen, 'file:///etc/passwd') + test_selection({'playlist_items': '2-4,3-4,3'}, [2, 3, 4]) + test_selection({'playlist_items': '4,2'}, [4, 2]) + + # Tests for https://github.com/yt-dlp/yt-dlp/issues/720 + # https://github.com/yt-dlp/yt-dlp/issues/302 + test_selection({'playlistreverse': True}, INDICES[::-1]) + test_selection({'playliststart': 2, 'playlistreverse': True}, INDICES[:0:-1]) + test_selection({'playlist_items': '2,4', 'playlistreverse': True}, [4, 2]) + test_selection({'playlist_items': '4,2'}, [4, 2]) + + # Tests for --playlist-items start:end:step + test_selection({'playlist_items': ':'}, INDICES, True) + test_selection({'playlist_items': '::1'}, INDICES, True) + test_selection({'playlist_items': '::-1'}, INDICES[::-1], True) + test_selection({'playlist_items': ':6'}, INDICES[:6]) + test_selection({'playlist_items': ':-6'}, INDICES[:-5], True) + test_selection({'playlist_items': '-1:6:-2'}, INDICES[:4:-2], True) + test_selection({'playlist_items': '9:-6:-2'}, INDICES[8:3:-2], True) + + test_selection({'playlist_items': '1:inf:2'}, INDICES[::2], True) + test_selection({'playlist_items': '-2:inf'}, INDICES[-2:], True) + test_selection({'playlist_items': ':inf:-1'}, [], True) + test_selection({'playlist_items': '0-2:2'}, [2]) + test_selection({'playlist_items': '1-:2'}, INDICES[::2], True) + test_selection({'playlist_items': '0--2:2'}, INDICES[1:-1:2], True) + + test_selection({'playlist_items': '10::3'}, [10], True) + test_selection({'playlist_items': '-1::3'}, [10], True) + test_selection({'playlist_items': '11::3'}, [], True) + test_selection({'playlist_items': '-15::2'}, INDICES[1::2], True) + test_selection({'playlist_items': '-15::15'}, [], True) def test_do_not_override_ie_key_in_url_transparent(self): ydl = YDL() @@ -931,7 +1152,7 @@ def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self): class _YDL(YDL): def __init__(self, *args, **kwargs): - super(_YDL, self).__init__(*args, **kwargs) + super().__init__(*args, **kwargs) def trouble(self, s, tb=None): pass @@ -968,7 +1189,7 @@ class PlaylistIE(InfoExtractor): def _entries(self): for n in range(3): - video_id = compat_str(n) + video_id = str(n) yield { '_type': 'url_transparent', 'ie_key': VideoIE.ie_key(), @@ -989,6 +1210,7 @@ def _real_extract(self, url): self.assertTrue(entries[1] is None) self.assertEqual(len(ydl.downloaded_info_dicts), 1) downloaded = ydl.downloaded_info_dicts[0] + entries[2].pop('requested_downloads', None) self.assertEqual(entries[2], downloaded) self.assertEqual(downloaded['url'], TEST_URL) self.assertEqual(downloaded['title'], 'Video Transparent 2') @@ -996,6 +1218,129 @@ def _real_extract(self, url): self.assertEqual(downloaded['extractor'], 'Video') self.assertEqual(downloaded['extractor_key'], 'Video') + def test_header_cookies(self): + from http.cookiejar import Cookie + + ydl = FakeYDL() + ydl.report_warning = lambda *_, **__: None + + def cookie(name, value, version=None, domain='', path='', secure=False, expires=None): + return Cookie( + version or 0, name, value, None, False, + domain, bool(domain), bool(domain), path, bool(path), + secure, expires, False, None, None, rest={}) + + _test_url = 'https://yt.dlp/test' + + def test(encoded_cookies, cookies, *, headers=False, round_trip=None, error_re=None): + def _test(): + ydl.cookiejar.clear() + ydl._load_cookies(encoded_cookies, autoscope=headers) + if headers: + ydl._apply_header_cookies(_test_url) + data = {'url': _test_url} + ydl._calc_headers(data) + self.assertCountEqual( + map(vars, ydl.cookiejar), map(vars, cookies), + 'Extracted cookiejar.Cookie is not the same') + if not headers: + self.assertEqual( + data.get('cookies'), round_trip or encoded_cookies, + 'Cookie is not the same as round trip') + ydl.__dict__['_YoutubeDL__header_cookies'] = [] + + with self.subTest(msg=encoded_cookies): + if not error_re: + _test() + return + with self.assertRaisesRegex(Exception, error_re): + _test() + + test('test=value; Domain=.yt.dlp', [cookie('test', 'value', domain='.yt.dlp')]) + test('test=value', [cookie('test', 'value')], error_re=r'Unscoped cookies are not allowed') + test('cookie1=value1; Domain=.yt.dlp; Path=/test; cookie2=value2; Domain=.yt.dlp; Path=/', [ + cookie('cookie1', 'value1', domain='.yt.dlp', path='/test'), + cookie('cookie2', 'value2', domain='.yt.dlp', path='/')]) + test('test=value; Domain=.yt.dlp; Path=/test; Secure; Expires=9999999999', [ + cookie('test', 'value', domain='.yt.dlp', path='/test', secure=True, expires=9999999999)]) + test('test="value; "; path=/test; domain=.yt.dlp', [ + cookie('test', 'value; ', domain='.yt.dlp', path='/test')], + round_trip='test="value\\073 "; Domain=.yt.dlp; Path=/test') + test('name=; Domain=.yt.dlp', [cookie('name', '', domain='.yt.dlp')], + round_trip='name=""; Domain=.yt.dlp') + + test('test=value', [cookie('test', 'value', domain='.yt.dlp')], headers=True) + test('cookie1=value; Domain=.yt.dlp; cookie2=value', [], headers=True, error_re=r'Invalid syntax') + ydl.deprecated_feature = ydl.report_error + test('test=value', [], headers=True, error_re=r'Passing cookies as a header is a potential security risk') + + def test_infojson_cookies(self): + TEST_FILE = 'test_infojson_cookies.info.json' + TEST_URL = 'https://example.com/example.mp4' + COOKIES = 'a=b; Domain=.example.com; c=d; Domain=.example.com' + COOKIE_HEADER = {'Cookie': 'a=b; c=d'} + + ydl = FakeYDL() + ydl.process_info = lambda x: ydl._write_info_json('test', x, TEST_FILE) + + def make_info(info_header_cookies=False, fmts_header_cookies=False, cookies_field=False): + fmt = {'url': TEST_URL} + if fmts_header_cookies: + fmt['http_headers'] = COOKIE_HEADER + if cookies_field: + fmt['cookies'] = COOKIES + return _make_result([fmt], http_headers=COOKIE_HEADER if info_header_cookies else None) + + def test(initial_info, note): + result = {} + result['processed'] = ydl.process_ie_result(initial_info) + self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL), + msg=f'No cookies set in cookiejar after initial process when {note}') + ydl.cookiejar.clear() + with open(TEST_FILE) as infojson: + result['loaded'] = ydl.sanitize_info(json.load(infojson), True) + result['final'] = ydl.process_ie_result(result['loaded'].copy(), download=False) + self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL), + msg=f'No cookies set in cookiejar after final process when {note}') + ydl.cookiejar.clear() + for key in ('processed', 'loaded', 'final'): + info = result[key] + self.assertIsNone( + traverse_obj(info, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False), + msg=f'Cookie header not removed in {key} result when {note}') + self.assertEqual( + traverse_obj(info, ((None, ('formats', 0)), 'cookies'), get_all=False), COOKIES, + msg=f'No cookies field found in {key} result when {note}') + + test({'url': TEST_URL, 'http_headers': COOKIE_HEADER, 'id': '1', 'title': 'x'}, 'no formats field') + test(make_info(info_header_cookies=True), 'info_dict header cokies') + test(make_info(fmts_header_cookies=True), 'format header cookies') + test(make_info(info_header_cookies=True, fmts_header_cookies=True), 'info_dict and format header cookies') + test(make_info(info_header_cookies=True, fmts_header_cookies=True, cookies_field=True), 'all cookies fields') + test(make_info(cookies_field=True), 'cookies format field') + test({'url': TEST_URL, 'cookies': COOKIES, 'id': '1', 'title': 'x'}, 'info_dict cookies field only') + + try_rm(TEST_FILE) + + def test_add_headers_cookie(self): + def check_for_cookie_header(result): + return traverse_obj(result, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False) + + ydl = FakeYDL({'http_headers': {'Cookie': 'a=b'}}) + ydl._apply_header_cookies(_make_result([])['webpage_url']) # Scope to input webpage URL: .example.com + + fmt = {'url': 'https://example.com/video.mp4'} + result = ydl.process_ie_result(_make_result([fmt]), download=False) + self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies in result info_dict') + self.assertEqual(result.get('cookies'), 'a=b; Domain=.example.com', msg='No cookies were set in cookies field') + self.assertIn('a=b', ydl.cookiejar.get_cookie_header(fmt['url']), msg='No cookies were set in cookiejar') + + fmt = {'url': 'https://wrong.com/video.mp4'} + result = ydl.process_ie_result(_make_result([fmt]), download=False) + self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies for wrong domain') + self.assertFalse(result.get('cookies'), msg='Cookies set in cookies field for wrong domain') + self.assertFalse(ydl.cookiejar.get_cookie_header(fmt['url']), msg='Cookies set in cookiejar for wrong domain') + if __name__ == '__main__': unittest.main()