X-Git-Url: https://jfr.im/git/yt-dlp.git/blobdiff_plain/ec9311c41b111110bc52cfbd6ea682c6fb23f77a..42ded0a429c20ec13dc006825e1508d9a02f0ad4:/test/test_YoutubeDL.py diff --git a/test/test_YoutubeDL.py b/test/test_YoutubeDL.py index 3c26bd7c6..b4f770ca5 100644 --- a/test/test_YoutubeDL.py +++ b/test/test_YoutubeDL.py @@ -10,9 +10,8 @@ import copy import json -import urllib.error -from test.helper import FakeYDL, assertRegexpMatches +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 @@ -25,6 +24,8 @@ int_or_none, match_filter_func, ) +from yt_dlp.utils.traversal import traverse_obj + TEST_URL = 'http://localhost/sample.mp4' @@ -631,6 +632,7 @@ def test_add_extra_info(self): self.assertEqual(test_dict['playlist'], 'funny videos') outtmpl_info = { + 'id': '1234', 'id': '1234', 'ext': 'mp4', 'width': None, @@ -669,7 +671,7 @@ def test(tmpl, expected, *, info=None, **params): for (name, got), expect in zip((('outtmpl', out), ('filename', fname)), expected): if callable(expect): self.assertTrue(expect(got), f'Wrong {name} from {tmpl}') - else: + elif expect is not None: self.assertEqual(got, expect, f'Wrong {name} from {tmpl}') # Side-effects @@ -684,7 +686,7 @@ def test(tmpl, expected, *, info=None, **params): 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', '001') test('%(playlist_autonumber)s', '02') test('%(autonumber)s', '00001') test('%(autonumber+2)03d', '005', autonumber_start=3) @@ -755,20 +757,23 @@ def expect_same_infodict(out): 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', '0000') - test('a%(width|)d', 'a', outtmpl_na_placeholder='none') + test('%(width|0)04d', '0') + test('a%(width|b)d', 'ab', outtmpl_na_placeholder='none') FORMATS = self.outtmpl_info['formats'] - sanitize = lambda x: x.replace(':', '：').replace('"', "＂").replace('\n', ' ') # 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), sanitize(json.dumps(FORMATS)))) - test('%(formats)#j', (json.dumps(FORMATS, indent=4), sanitize(json.dumps(FORMATS, indent=4)))) + 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') @@ -793,8 +798,8 @@ def expect_same_infodict(out): test('%(title|%)s %(title|%%)s', '% %%') test('%(id+1-height+3)05d', '00158') test('%(width+100)05d', 'NA') - test('%(formats.0) 15s', ('% 15s' % FORMATS[0], '% 15s' % sanitize(str(FORMATS[0])))) - test('%(formats.0)r', (repr(FORMATS[0]), sanitize(repr(FORMATS[0])))) + 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']) @@ -806,7 +811,7 @@ def expect_same_infodict(out): 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, sanitize(out))) + 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＂}')) @@ -871,12 +876,12 @@ 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()) @@ -895,7 +900,7 @@ 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 @@ -1097,11 +1102,6 @@ def test_selection(params, expected_ids, evaluate_all=False): test_selection({'playlist_items': '-15::2'}, INDICES[1::2], True) test_selection({'playlist_items': '-15::15'}, [], True) - def test_urlopen_no_file_protocol(self): - # see https://github.com/ytdl-org/youtube-dl/issues/8227 - ydl = YDL() - self.assertRaises(urllib.error.URLError, ydl.urlopen, 'file:///etc/passwd') - def test_do_not_override_ie_key_in_url_transparent(self): ydl = YDL() @@ -1215,6 +1215,129 @@ 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()