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, find_available_port
  14
  15 if 'pytest' in sys.modules:
  16     import pytest
  17     is_download_test = pytest.mark.download
  18 else:
  19     def is_download_test(test_class):
  20         return test_class
  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             f'Expect a list of length {len(expected)}, but got a list of length {len(got)} for field {field}')
 142         for index, (item_got, item_expected) in enumerate(zip(got, expected)):
 143             type_got = type(item_got)
 144             type_expected = type(item_expected)
 145             self.assertEqual(
 146                 type_expected, type_got,
 147                 f'Type mismatch for list item at index {index} for field {field}, '
 148                 f'expected {type_expected!r}, got {type_got!r}')
 149             expect_value(self, item_got, item_expected, field)
 150     else:
 151         if isinstance(expected, str) and expected.startswith('md5:'):
 152             self.assertTrue(
 153                 isinstance(got, str),
 154                 f'Expected field {field} to be a unicode object, but got value {got!r} of type {type(got)!r}')
 155             got = 'md5:' + md5(got)
 156         elif isinstance(expected, str) and re.match(r'^(?:min|max)?count:\d+', expected):
 157             self.assertTrue(
 158                 isinstance(got, (list, dict)),
 159                 f'Expected field {field} to be a list or a dict, but it is of type {type(got).__name__}')
 160             op, _, expected_num = expected.partition(':')
 161             expected_num = int(expected_num)
 162             if op == 'mincount':
 163                 assert_func = assertGreaterEqual
 164                 msg_tmpl = 'Expected %d items in field %s, but only got %d'
 165             elif op == 'maxcount':
 166                 assert_func = assertLessEqual
 167                 msg_tmpl = 'Expected maximum %d items in field %s, but got %d'
 168             elif op == 'count':
 169                 assert_func = assertEqual
 170                 msg_tmpl = 'Expected exactly %d items in field %s, but got %d'
 171             else:
 172                 assert False
 173             assert_func(
 174                 self, len(got), expected_num,
 175                 msg_tmpl % (expected_num, field, len(got)))
 176             return
 177         self.assertEqual(
 178             expected, got,
 179             f'Invalid value for field {field}, expected {expected!r}, got {got!r}')
 180
 181
 182 def expect_dict(self, got_dict, expected_dict):
 183     for info_field, expected in expected_dict.items():
 184         got = got_dict.get(info_field)
 185         expect_value(self, got, expected, info_field)
 186
 187
 188 def sanitize_got_info_dict(got_dict):
 189     IGNORED_FIELDS = (
 190         *YoutubeDL._format_fields,
 191
 192         # Lists
 193         'formats', 'thumbnails', 'subtitles', 'automatic_captions', 'comments', 'entries',
 194
 195         # Auto-generated
 196         'autonumber', 'playlist', 'format_index', 'video_ext', 'audio_ext', 'duration_string', 'epoch', 'n_entries',
 197         'fulltitle', 'extractor', 'extractor_key', 'filename', 'filepath', 'infojson_filename', 'original_url',
 198
 199         # Only live_status needs to be checked
 200         'is_live', 'was_live',
 201     )
 202
 203     IGNORED_PREFIXES = ('', 'playlist', 'requested', 'webpage')
 204
 205     def sanitize(key, value):
 206         if isinstance(value, str) and len(value) > 100 and key != 'thumbnail':
 207             return f'md5:{md5(value)}'
 208         elif isinstance(value, list) and len(value) > 10:
 209             return f'count:{len(value)}'
 210         elif key.endswith('_count') and isinstance(value, int):
 211             return int
 212         return value
 213
 214     test_info_dict = {
 215         key: sanitize(key, value) for key, value in got_dict.items()
 216         if value is not None and key not in IGNORED_FIELDS and (
 217             not any(key.startswith(f'{prefix}_') for prefix in IGNORED_PREFIXES)
 218             or key == '_old_archive_ids')
 219     }
 220
 221     # display_id may be generated from id
 222     if test_info_dict.get('display_id') == test_info_dict.get('id'):
 223         test_info_dict.pop('display_id')
 224
 225     # Remove deprecated fields
 226     for old in YoutubeDL._deprecated_multivalue_fields:
 227         test_info_dict.pop(old, None)
 228
 229     # release_year may be generated from release_date
 230     if try_call(lambda: test_info_dict['release_year'] == int(test_info_dict['release_date'][:4])):
 231         test_info_dict.pop('release_year')
 232
 233     # Check url for flat entries
 234     if got_dict.get('_type', 'video') != 'video' and got_dict.get('url'):
 235         test_info_dict['url'] = got_dict['url']
 236
 237     return test_info_dict
 238
 239
 240 def expect_info_dict(self, got_dict, expected_dict):
 241     expect_dict(self, got_dict, expected_dict)
 242     # Check for the presence of mandatory fields
 243     if got_dict.get('_type') not in ('playlist', 'multi_video'):
 244         mandatory_fields = ['id', 'title']
 245         if expected_dict.get('ext'):
 246             mandatory_fields.extend(('url', 'ext'))
 247         for key in mandatory_fields:
 248             self.assertTrue(got_dict.get(key), f'Missing mandatory field {key}')
 249     # Check for mandatory fields that are automatically set by YoutubeDL
 250     if got_dict.get('_type', 'video') == 'video':
 251         for key in ['webpage_url', 'extractor', 'extractor_key']:
 252             self.assertTrue(got_dict.get(key), f'Missing field: {key}')
 253
 254     test_info_dict = sanitize_got_info_dict(got_dict)
 255
 256     missing_keys = set(test_info_dict.keys()) - set(expected_dict.keys())
 257     if missing_keys:
 258         def _repr(v):
 259             if isinstance(v, str):
 260                 return "'{}'".format(v.replace('\\', '\\\\').replace("'", "\\'").replace('\n', '\\n'))
 261             elif isinstance(v, type):
 262                 return v.__name__
 263             else:
 264                 return repr(v)
 265         info_dict_str = ''.join(
 266             f'    {_repr(k)}: {_repr(v)},\n'
 267             for k, v in test_info_dict.items() if k not in missing_keys)
 268         if info_dict_str:
 269             info_dict_str += '\n'
 270         info_dict_str += ''.join(
 271             f'    {_repr(k)}: {_repr(test_info_dict[k])},\n'
 272             for k in missing_keys)
 273         info_dict_str = '\n\'info_dict\': {\n' + info_dict_str + '},\n'
 274         write_string(info_dict_str.replace('\n', '\n        '), out=sys.stderr)
 275         self.assertFalse(
 276             missing_keys,
 277             'Missing keys in test definition: {}'.format(', '.join(sorted(missing_keys))))
 278
 279
 280 def assertRegexpMatches(self, text, regexp, msg=None):
 281     if hasattr(self, 'assertRegexp'):
 282         return self.assertRegexp(text, regexp, msg)
 283     else:
 284         m = re.match(regexp, text)
 285         if not m:
 286             note = f'Regexp didn\'t match: {regexp!r} not found'
 287             if len(text) < 1000:
 288                 note += f' in {text!r}'
 289             if msg is None:
 290                 msg = note
 291             else:
 292                 msg = note + ', ' + msg
 293             self.assertTrue(m, msg)
 294
 295
 296 def assertGreaterEqual(self, got, expected, msg=None):
 297     if not (got >= expected):
 298         if msg is None:
 299             msg = f'{got!r} not greater than or equal to {expected!r}'
 300         self.assertTrue(got >= expected, msg)
 301
 302
 303 def assertLessEqual(self, got, expected, msg=None):
 304     if not (got <= expected):
 305         if msg is None:
 306             msg = f'{got!r} not less than or equal to {expected!r}'
 307         self.assertTrue(got <= expected, msg)
 308
 309
 310 def assertEqual(self, got, expected, msg=None):
 311     if got != expected:
 312         if msg is None:
 313             msg = f'{got!r} not equal to {expected!r}'
 314         self.assertTrue(got == expected, msg)
 315
 316
 317 def expect_warnings(ydl, warnings_re):
 318     real_warning = ydl.report_warning
 319
 320     def _report_warning(w, *args, **kwargs):
 321         if not any(re.search(w_re, w) for w_re in warnings_re):
 322             real_warning(w, *args, **kwargs)
 323
 324     ydl.report_warning = _report_warning
 325
 326
 327 def http_server_port(httpd):
 328     if os.name == 'java' and isinstance(httpd.socket, ssl.SSLSocket):
 329         # In Jython SSLSocket is not a subclass of socket.socket
 330         sock = httpd.socket.sock
 331     else:
 332         sock = httpd.socket
 333     return sock.getsockname()[1]
 334
 335
 336 def verify_address_availability(address):
 337     if find_available_port(address) is None:
 338         pytest.skip(f'Unable to bind to source address {address} (address may not exist)')
 339
 340
 341 def validate_and_send(rh, req):
 342     rh.validate(req)
 343     return rh.send(req)