test/helper.py

   1 import errno
   2 import hashlib
   3 import json
   4 import os.path
   5 import re
   6 import ssl
   7 import sys
   8 import types
   9
  10 import yt_dlp.extractor
  11 from yt_dlp import YoutubeDL
  12 from yt_dlp.compat import compat_os_name
  13 from yt_dlp.utils import preferredencoding, try_call, write_string
  14
  15 if 'pytest' in sys.modules:
  16     import pytest
  17     is_download_test = pytest.mark.download
  18 else:
  19     def is_download_test(testClass):
  20         return testClass
  21
  22
  23 def get_params(override=None):
  24     PARAMETERS_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)),
  25                                    'parameters.json')
  26     LOCAL_PARAMETERS_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)),
  27                                          'local_parameters.json')
  28     with open(PARAMETERS_FILE, encoding='utf-8') as pf:
  29         parameters = json.load(pf)
  30     if os.path.exists(LOCAL_PARAMETERS_FILE):
  31         with open(LOCAL_PARAMETERS_FILE, encoding='utf-8') as pf:
  32             parameters.update(json.load(pf))
  33     if override:
  34         parameters.update(override)
  35     return parameters
  36
  37
  38 def try_rm(filename):
  39     """ Remove a file if it exists """
  40     try:
  41         os.remove(filename)
  42     except OSError as ose:
  43         if ose.errno != errno.ENOENT:
  44             raise
  45
  46
  47 def report_warning(message, *args, **kwargs):
  48     '''
  49     Print the message to stderr, it will be prefixed with 'WARNING:'
  50     If stderr is a tty file the 'WARNING:' will be colored
  51     '''
  52     if sys.stderr.isatty() and compat_os_name != 'nt':
  53         _msg_header = '\033[0;33mWARNING:\033[0m'
  54     else:
  55         _msg_header = 'WARNING:'
  56     output = f'{_msg_header} {message}\n'
  57     if 'b' in getattr(sys.stderr, 'mode', ''):
  58         output = output.encode(preferredencoding())
  59     sys.stderr.write(output)
  60
  61
  62 class FakeYDL(YoutubeDL):
  63     def __init__(self, override=None):
  64         # Different instances of the downloader can't share the same dictionary
  65         # some test set the "sublang" parameter, which would break the md5 checks.
  66         params = get_params(override=override)
  67         super().__init__(params, auto_init=False)
  68         self.result = []
  69
  70     def to_screen(self, s, *args, **kwargs):
  71         print(s)
  72
  73     def trouble(self, s, *args, **kwargs):
  74         raise Exception(s)
  75
  76     def download(self, x):
  77         self.result.append(x)
  78
  79     def expect_warning(self, regex):
  80         # Silence an expected warning matching a regex
  81         old_report_warning = self.report_warning
  82
  83         def report_warning(self, message, *args, **kwargs):
  84             if re.match(regex, message):
  85                 return
  86             old_report_warning(message, *args, **kwargs)
  87         self.report_warning = types.MethodType(report_warning, self)
  88
  89
  90 def gettestcases(include_onlymatching=False):
  91     for ie in yt_dlp.extractor.gen_extractors():
  92         yield from ie.get_testcases(include_onlymatching)
  93
  94
  95 def getwebpagetestcases():
  96     for ie in yt_dlp.extractor.gen_extractors():
  97         for tc in ie.get_webpage_testcases():
  98             tc.setdefault('add_ie', []).append('Generic')
  99             yield tc
 100
 101
 102 md5 = lambda s: hashlib.md5(s.encode()).hexdigest()
 103
 104
 105 def expect_value(self, got, expected, field):
 106     if isinstance(expected, str) and expected.startswith('re:'):
 107         match_str = expected[len('re:'):]
 108         match_rex = re.compile(match_str)
 109
 110         self.assertTrue(
 111             isinstance(got, str),
 112             f'Expected a {str.__name__} object, but got {type(got).__name__} for field {field}')
 113         self.assertTrue(
 114             match_rex.match(got),
 115             f'field {field} (value: {got!r}) should match {match_str!r}')
 116     elif isinstance(expected, str) and expected.startswith('startswith:'):
 117         start_str = expected[len('startswith:'):]
 118         self.assertTrue(
 119             isinstance(got, str),
 120             f'Expected a {str.__name__} object, but got {type(got).__name__} for field {field}')
 121         self.assertTrue(
 122             got.startswith(start_str),
 123             f'field {field} (value: {got!r}) should start with {start_str!r}')
 124     elif isinstance(expected, str) and expected.startswith('contains:'):
 125         contains_str = expected[len('contains:'):]
 126         self.assertTrue(
 127             isinstance(got, str),
 128             f'Expected a {str.__name__} object, but got {type(got).__name__} for field {field}')
 129         self.assertTrue(
 130             contains_str in got,
 131             f'field {field} (value: {got!r}) should contain {contains_str!r}')
 132     elif isinstance(expected, type):
 133         self.assertTrue(
 134             isinstance(got, expected),
 135             f'Expected type {expected!r} for field {field}, but got value {got!r} of type {type(got)!r}')
 136     elif isinstance(expected, dict) and isinstance(got, dict):
 137         expect_dict(self, got, expected)
 138     elif isinstance(expected, list) and isinstance(got, list):
 139         self.assertEqual(
 140             len(expected), len(got),
 141             'Expect a list of length %d, but got a list of length %d for field %s' % (
 142                 len(expected), len(got), field))
 143         for index, (item_got, item_expected) in enumerate(zip(got, expected)):
 144             type_got = type(item_got)
 145             type_expected = type(item_expected)
 146             self.assertEqual(
 147                 type_expected, type_got,
 148                 'Type mismatch for list item at index %d for field %s, expected %r, got %r' % (
 149                     index, field, type_expected, type_got))
 150             expect_value(self, item_got, item_expected, field)
 151     else:
 152         if isinstance(expected, str) and expected.startswith('md5:'):
 153             self.assertTrue(
 154                 isinstance(got, str),
 155                 f'Expected field {field} to be a unicode object, but got value {got!r} of type {type(got)!r}')
 156             got = 'md5:' + md5(got)
 157         elif isinstance(expected, str) and re.match(r'^(?:min|max)?count:\d+', expected):
 158             self.assertTrue(
 159                 isinstance(got, (list, dict)),
 160                 f'Expected field {field} to be a list or a dict, but it is of type {type(got).__name__}')
 161             op, _, expected_num = expected.partition(':')
 162             expected_num = int(expected_num)
 163             if op == 'mincount':
 164                 assert_func = assertGreaterEqual
 165                 msg_tmpl = 'Expected %d items in field %s, but only got %d'
 166             elif op == 'maxcount':
 167                 assert_func = assertLessEqual
 168                 msg_tmpl = 'Expected maximum %d items in field %s, but got %d'
 169             elif op == 'count':
 170                 assert_func = assertEqual
 171                 msg_tmpl = 'Expected exactly %d items in field %s, but got %d'
 172             else:
 173                 assert False
 174             assert_func(
 175                 self, len(got), expected_num,
 176                 msg_tmpl % (expected_num, field, len(got)))
 177             return
 178         self.assertEqual(
 179             expected, got,
 180             f'Invalid value for field {field}, expected {expected!r}, got {got!r}')
 181
 182
 183 def expect_dict(self, got_dict, expected_dict):
 184     for info_field, expected in expected_dict.items():
 185         got = got_dict.get(info_field)
 186         expect_value(self, got, expected, info_field)
 187
 188
 189 def sanitize_got_info_dict(got_dict):
 190     IGNORED_FIELDS = (
 191         *YoutubeDL._format_fields,
 192
 193         # Lists
 194         'formats', 'thumbnails', 'subtitles', 'automatic_captions', 'comments', 'entries',
 195
 196         # Auto-generated
 197         'autonumber', 'playlist', 'format_index', 'video_ext', 'audio_ext', 'duration_string', 'epoch', 'n_entries',
 198         'fulltitle', 'extractor', 'extractor_key', 'filename', 'filepath', 'infojson_filename', 'original_url',
 199
 200         # Only live_status needs to be checked
 201         'is_live', 'was_live',
 202     )
 203
 204     IGNORED_PREFIXES = ('', 'playlist', 'requested', 'webpage')
 205
 206     def sanitize(key, value):
 207         if isinstance(value, str) and len(value) > 100 and key != 'thumbnail':
 208             return f'md5:{md5(value)}'
 209         elif isinstance(value, list) and len(value) > 10:
 210             return f'count:{len(value)}'
 211         elif key.endswith('_count') and isinstance(value, int):
 212             return int
 213         return value
 214
 215     test_info_dict = {
 216         key: sanitize(key, value) for key, value in got_dict.items()
 217         if value is not None and key not in IGNORED_FIELDS and (
 218             not any(key.startswith(f'{prefix}_') for prefix in IGNORED_PREFIXES)
 219             or key == '_old_archive_ids')
 220     }
 221
 222     # display_id may be generated from id
 223     if test_info_dict.get('display_id') == test_info_dict.get('id'):
 224         test_info_dict.pop('display_id')
 225
 226     # release_year may be generated from release_date
 227     if try_call(lambda: test_info_dict['release_year'] == int(test_info_dict['release_date'][:4])):
 228         test_info_dict.pop('release_year')
 229
 230     # Check url for flat entries
 231     if got_dict.get('_type', 'video') != 'video' and got_dict.get('url'):
 232         test_info_dict['url'] = got_dict['url']
 233
 234     return test_info_dict
 235
 236
 237 def expect_info_dict(self, got_dict, expected_dict):
 238     expect_dict(self, got_dict, expected_dict)
 239     # Check for the presence of mandatory fields
 240     if got_dict.get('_type') not in ('playlist', 'multi_video'):
 241         mandatory_fields = ['id', 'title']
 242         if expected_dict.get('ext'):
 243             mandatory_fields.extend(('url', 'ext'))
 244         for key in mandatory_fields:
 245             self.assertTrue(got_dict.get(key), 'Missing mandatory field %s' % key)
 246     # Check for mandatory fields that are automatically set by YoutubeDL
 247     if got_dict.get('_type', 'video') == 'video':
 248         for key in ['webpage_url', 'extractor', 'extractor_key']:
 249             self.assertTrue(got_dict.get(key), 'Missing field: %s' % key)
 250
 251     test_info_dict = sanitize_got_info_dict(got_dict)
 252
 253     missing_keys = set(test_info_dict.keys()) - set(expected_dict.keys())
 254     if missing_keys:
 255         def _repr(v):
 256             if isinstance(v, str):
 257                 return "'%s'" % v.replace('\\', '\\\\').replace("'", "\\'").replace('\n', '\\n')
 258             elif isinstance(v, type):
 259                 return v.__name__
 260             else:
 261                 return repr(v)
 262         info_dict_str = ''.join(
 263             f'    {_repr(k)}: {_repr(v)},\n'
 264             for k, v in test_info_dict.items() if k not in missing_keys)
 265         if info_dict_str:
 266             info_dict_str += '\n'
 267         info_dict_str += ''.join(
 268             f'    {_repr(k)}: {_repr(test_info_dict[k])},\n'
 269             for k in missing_keys)
 270         info_dict_str = '\n\'info_dict\': {\n' + info_dict_str + '},\n'
 271         write_string(info_dict_str.replace('\n', '\n        '), out=sys.stderr)
 272         self.assertFalse(
 273             missing_keys,
 274             'Missing keys in test definition: %s' % (
 275                 ', '.join(sorted(missing_keys))))
 276
 277
 278 def assertRegexpMatches(self, text, regexp, msg=None):
 279     if hasattr(self, 'assertRegexp'):
 280         return self.assertRegexp(text, regexp, msg)
 281     else:
 282         m = re.match(regexp, text)
 283         if not m:
 284             note = 'Regexp didn\'t match: %r not found' % (regexp)
 285             if len(text) < 1000:
 286                 note += ' in %r' % text
 287             if msg is None:
 288                 msg = note
 289             else:
 290                 msg = note + ', ' + msg
 291             self.assertTrue(m, msg)
 292
 293
 294 def assertGreaterEqual(self, got, expected, msg=None):
 295     if not (got >= expected):
 296         if msg is None:
 297             msg = f'{got!r} not greater than or equal to {expected!r}'
 298         self.assertTrue(got >= expected, msg)
 299
 300
 301 def assertLessEqual(self, got, expected, msg=None):
 302     if not (got <= expected):
 303         if msg is None:
 304             msg = f'{got!r} not less than or equal to {expected!r}'
 305         self.assertTrue(got <= expected, msg)
 306
 307
 308 def assertEqual(self, got, expected, msg=None):
 309     if not (got == expected):
 310         if msg is None:
 311             msg = f'{got!r} not equal to {expected!r}'
 312         self.assertTrue(got == expected, msg)
 313
 314
 315 def expect_warnings(ydl, warnings_re):
 316     real_warning = ydl.report_warning
 317
 318     def _report_warning(w, *args, **kwargs):
 319         if not any(re.search(w_re, w) for w_re in warnings_re):
 320             real_warning(w, *args, **kwargs)
 321
 322     ydl.report_warning = _report_warning
 323
 324
 325 def http_server_port(httpd):
 326     if os.name == 'java' and isinstance(httpd.socket, ssl.SSLSocket):
 327         # In Jython SSLSocket is not a subclass of socket.socket
 328         sock = httpd.socket.sock
 329     else:
 330         sock = httpd.socket
 331     return sock.getsockname()[1]