test/helper.py

   1 import errno
   2 import hashlib
   3 import json
   4 import os.path
   5 import re
   6 import types
   7 import ssl
   8 import sys
   9
  10 import yt_dlp.extractor
  11 from yt_dlp import YoutubeDL
  12 from yt_dlp.compat import (
  13     compat_os_name,
  14     compat_str,
  15 )
  16 from yt_dlp.utils import (
  17     preferredencoding,
  18     write_string,
  19 )
  20
  21
  22 if 'pytest' in sys.modules:
  23     import pytest
  24     is_download_test = pytest.mark.download
  25 else:
  26     def is_download_test(testClass):
  27         return testClass
  28
  29
  30 def get_params(override=None):
  31     PARAMETERS_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)),
  32                                    'parameters.json')
  33     LOCAL_PARAMETERS_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)),
  34                                          'local_parameters.json')
  35     with open(PARAMETERS_FILE, encoding='utf-8') as pf:
  36         parameters = json.load(pf)
  37     if os.path.exists(LOCAL_PARAMETERS_FILE):
  38         with open(LOCAL_PARAMETERS_FILE, encoding='utf-8') as pf:
  39             parameters.update(json.load(pf))
  40     if override:
  41         parameters.update(override)
  42     return parameters
  43
  44
  45 def try_rm(filename):
  46     """ Remove a file if it exists """
  47     try:
  48         os.remove(filename)
  49     except OSError as ose:
  50         if ose.errno != errno.ENOENT:
  51             raise
  52
  53
  54 def report_warning(message):
  55     '''
  56     Print the message to stderr, it will be prefixed with 'WARNING:'
  57     If stderr is a tty file the 'WARNING:' will be colored
  58     '''
  59     if sys.stderr.isatty() and compat_os_name != 'nt':
  60         _msg_header = '\033[0;33mWARNING:\033[0m'
  61     else:
  62         _msg_header = 'WARNING:'
  63     output = f'{_msg_header} {message}\n'
  64     if 'b' in getattr(sys.stderr, 'mode', ''):
  65         output = output.encode(preferredencoding())
  66     sys.stderr.write(output)
  67
  68
  69 class FakeYDL(YoutubeDL):
  70     def __init__(self, override=None):
  71         # Different instances of the downloader can't share the same dictionary
  72         # some test set the "sublang" parameter, which would break the md5 checks.
  73         params = get_params(override=override)
  74         super().__init__(params, auto_init=False)
  75         self.result = []
  76
  77     def to_screen(self, s, skip_eol=None):
  78         print(s)
  79
  80     def trouble(self, s, tb=None):
  81         raise Exception(s)
  82
  83     def download(self, x):
  84         self.result.append(x)
  85
  86     def expect_warning(self, regex):
  87         # Silence an expected warning matching a regex
  88         old_report_warning = self.report_warning
  89
  90         def report_warning(self, message):
  91             if re.match(regex, message):
  92                 return
  93             old_report_warning(message)
  94         self.report_warning = types.MethodType(report_warning, self)
  95
  96
  97 def gettestcases(include_onlymatching=False):
  98     for ie in yt_dlp.extractor.gen_extractors():
  99         yield from ie.get_testcases(include_onlymatching)
 100
 101
 102 md5 = lambda s: hashlib.md5(s.encode('utf-8')).hexdigest()
 103
 104
 105 def expect_value(self, got, expected, field):
 106     if isinstance(expected, compat_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, compat_str),
 112             f'Expected a {compat_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, compat_str) and expected.startswith('startswith:'):
 117         start_str = expected[len('startswith:'):]
 118         self.assertTrue(
 119             isinstance(got, compat_str),
 120             f'Expected a {compat_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, compat_str) and expected.startswith('contains:'):
 125         contains_str = expected[len('contains:'):]
 126         self.assertTrue(
 127             isinstance(got, compat_str),
 128             f'Expected a {compat_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, compat_str) and expected.startswith('md5:'):
 153             self.assertTrue(
 154                 isinstance(got, compat_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, compat_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',
 198         'fulltitle', 'extractor', 'extractor_key', 'filepath', 'infojson_filename', 'original_url', 'n_entries',
 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 not any(
 218             key.startswith(f'{prefix}_') for prefix in IGNORED_PREFIXES)
 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     return test_info_dict
 226
 227
 228 def expect_info_dict(self, got_dict, expected_dict):
 229     expect_dict(self, got_dict, expected_dict)
 230     # Check for the presence of mandatory fields
 231     if got_dict.get('_type') not in ('playlist', 'multi_video'):
 232         mandatory_fields = ['id', 'title']
 233         if expected_dict.get('ext'):
 234             mandatory_fields.extend(('url', 'ext'))
 235         for key in mandatory_fields:
 236             self.assertTrue(got_dict.get(key), 'Missing mandatory field %s' % key)
 237     # Check for mandatory fields that are automatically set by YoutubeDL
 238     for key in ['webpage_url', 'extractor', 'extractor_key']:
 239         self.assertTrue(got_dict.get(key), 'Missing field: %s' % key)
 240
 241     test_info_dict = sanitize_got_info_dict(got_dict)
 242
 243     missing_keys = set(test_info_dict.keys()) - set(expected_dict.keys())
 244     if missing_keys:
 245         def _repr(v):
 246             if isinstance(v, compat_str):
 247                 return "'%s'" % v.replace('\\', '\\\\').replace("'", "\\'").replace('\n', '\\n')
 248             elif isinstance(v, type):
 249                 return v.__name__
 250             else:
 251                 return repr(v)
 252         info_dict_str = ''
 253         if len(missing_keys) != len(expected_dict):
 254             info_dict_str += ''.join(
 255                 f'    {_repr(k)}: {_repr(v)},\n'
 256                 for k, v in test_info_dict.items() if k not in missing_keys)
 257
 258             if info_dict_str:
 259                 info_dict_str += '\n'
 260         info_dict_str += ''.join(
 261             f'    {_repr(k)}: {_repr(test_info_dict[k])},\n'
 262             for k in missing_keys)
 263         write_string(
 264             '\n\'info_dict\': {\n' + info_dict_str + '},\n', out=sys.stderr)
 265         self.assertFalse(
 266             missing_keys,
 267             'Missing keys in test definition: %s' % (
 268                 ', '.join(sorted(missing_keys))))
 269
 270
 271 def assertRegexpMatches(self, text, regexp, msg=None):
 272     if hasattr(self, 'assertRegexp'):
 273         return self.assertRegexp(text, regexp, msg)
 274     else:
 275         m = re.match(regexp, text)
 276         if not m:
 277             note = 'Regexp didn\'t match: %r not found' % (regexp)
 278             if len(text) < 1000:
 279                 note += ' in %r' % text
 280             if msg is None:
 281                 msg = note
 282             else:
 283                 msg = note + ', ' + msg
 284             self.assertTrue(m, msg)
 285
 286
 287 def assertGreaterEqual(self, got, expected, msg=None):
 288     if not (got >= expected):
 289         if msg is None:
 290             msg = f'{got!r} not greater than or equal to {expected!r}'
 291         self.assertTrue(got >= expected, msg)
 292
 293
 294 def assertLessEqual(self, got, expected, msg=None):
 295     if not (got <= expected):
 296         if msg is None:
 297             msg = f'{got!r} not less than or equal to {expected!r}'
 298         self.assertTrue(got <= expected, msg)
 299
 300
 301 def assertEqual(self, got, expected, msg=None):
 302     if not (got == expected):
 303         if msg is None:
 304             msg = f'{got!r} not equal to {expected!r}'
 305         self.assertTrue(got == expected, msg)
 306
 307
 308 def expect_warnings(ydl, warnings_re):
 309     real_warning = ydl.report_warning
 310
 311     def _report_warning(w):
 312         if not any(re.search(w_re, w) for w_re in warnings_re):
 313             real_warning(w)
 314
 315     ydl.report_warning = _report_warning
 316
 317
 318 def http_server_port(httpd):
 319     if os.name == 'java' and isinstance(httpd.socket, ssl.SSLSocket):
 320         # In Jython SSLSocket is not a subclass of socket.socket
 321         sock = httpd.socket.sock
 322     else:
 323         sock = httpd.socket
 324     return sock.getsockname()[1]