test/helper.py

   1 from __future__ import unicode_literals
   2
   3 import errno
   4 import io
   5 import hashlib
   6 import json
   7 import os.path
   8 import re
   9 import types
  10 import ssl
  11 import sys
  12
  13 import yt_dlp.extractor
  14 from yt_dlp import YoutubeDL
  15 from yt_dlp.compat import (
  16     compat_os_name,
  17     compat_str,
  18 )
  19 from yt_dlp.utils import (
  20     preferredencoding,
  21     write_string,
  22 )
  23
  24
  25 if 'pytest' in sys.modules:
  26     import pytest
  27     is_download_test = pytest.mark.download
  28 else:
  29     def is_download_test(testClass):
  30         return testClass
  31
  32
  33 def get_params(override=None):
  34     PARAMETERS_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)),
  35                                    'parameters.json')
  36     LOCAL_PARAMETERS_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)),
  37                                          'local_parameters.json')
  38     with io.open(PARAMETERS_FILE, encoding='utf-8') as pf:
  39         parameters = json.load(pf)
  40     if os.path.exists(LOCAL_PARAMETERS_FILE):
  41         with io.open(LOCAL_PARAMETERS_FILE, encoding='utf-8') as pf:
  42             parameters.update(json.load(pf))
  43     if override:
  44         parameters.update(override)
  45     return parameters
  46
  47
  48 def try_rm(filename):
  49     """ Remove a file if it exists """
  50     try:
  51         os.remove(filename)
  52     except OSError as ose:
  53         if ose.errno != errno.ENOENT:
  54             raise
  55
  56
  57 def report_warning(message):
  58     '''
  59     Print the message to stderr, it will be prefixed with 'WARNING:'
  60     If stderr is a tty file the 'WARNING:' will be colored
  61     '''
  62     if sys.stderr.isatty() and compat_os_name != 'nt':
  63         _msg_header = '\033[0;33mWARNING:\033[0m'
  64     else:
  65         _msg_header = 'WARNING:'
  66     output = '%s %s\n' % (_msg_header, message)
  67     if 'b' in getattr(sys.stderr, 'mode', '') or sys.version_info[0] < 3:
  68         output = output.encode(preferredencoding())
  69     sys.stderr.write(output)
  70
  71
  72 class FakeYDL(YoutubeDL):
  73     def __init__(self, override=None):
  74         # Different instances of the downloader can't share the same dictionary
  75         # some test set the "sublang" parameter, which would break the md5 checks.
  76         params = get_params(override=override)
  77         super(FakeYDL, self).__init__(params, auto_init=False)
  78         self.result = []
  79
  80     def to_screen(self, s, skip_eol=None):
  81         print(s)
  82
  83     def trouble(self, s, tb=None):
  84         raise Exception(s)
  85
  86     def download(self, x):
  87         self.result.append(x)
  88
  89     def expect_warning(self, regex):
  90         # Silence an expected warning matching a regex
  91         old_report_warning = self.report_warning
  92
  93         def report_warning(self, message):
  94             if re.match(regex, message):
  95                 return
  96             old_report_warning(message)
  97         self.report_warning = types.MethodType(report_warning, self)
  98
  99
 100 def gettestcases(include_onlymatching=False):
 101     for ie in yt_dlp.extractor.gen_extractors():
 102         for tc in ie.get_testcases(include_onlymatching):
 103             yield tc
 104
 105
 106 md5 = lambda s: hashlib.md5(s.encode('utf-8')).hexdigest()
 107
 108
 109 def expect_value(self, got, expected, field):
 110     if isinstance(expected, compat_str) and expected.startswith('re:'):
 111         match_str = expected[len('re:'):]
 112         match_rex = re.compile(match_str)
 113
 114         self.assertTrue(
 115             isinstance(got, compat_str),
 116             'Expected a %s object, but got %s for field %s' % (
 117                 compat_str.__name__, type(got).__name__, field))
 118         self.assertTrue(
 119             match_rex.match(got),
 120             'field %s (value: %r) should match %r' % (field, got, match_str))
 121     elif isinstance(expected, compat_str) and expected.startswith('startswith:'):
 122         start_str = expected[len('startswith:'):]
 123         self.assertTrue(
 124             isinstance(got, compat_str),
 125             'Expected a %s object, but got %s for field %s' % (
 126                 compat_str.__name__, type(got).__name__, field))
 127         self.assertTrue(
 128             got.startswith(start_str),
 129             'field %s (value: %r) should start with %r' % (field, got, start_str))
 130     elif isinstance(expected, compat_str) and expected.startswith('contains:'):
 131         contains_str = expected[len('contains:'):]
 132         self.assertTrue(
 133             isinstance(got, compat_str),
 134             'Expected a %s object, but got %s for field %s' % (
 135                 compat_str.__name__, type(got).__name__, field))
 136         self.assertTrue(
 137             contains_str in got,
 138             'field %s (value: %r) should contain %r' % (field, got, contains_str))
 139     elif isinstance(expected, type):
 140         self.assertTrue(
 141             isinstance(got, expected),
 142             'Expected type %r for field %s, but got value %r of type %r' % (expected, field, got, type(got)))
 143     elif isinstance(expected, dict) and isinstance(got, dict):
 144         expect_dict(self, got, expected)
 145     elif isinstance(expected, list) and isinstance(got, list):
 146         self.assertEqual(
 147             len(expected), len(got),
 148             'Expect a list of length %d, but got a list of length %d for field %s' % (
 149                 len(expected), len(got), field))
 150         for index, (item_got, item_expected) in enumerate(zip(got, expected)):
 151             type_got = type(item_got)
 152             type_expected = type(item_expected)
 153             self.assertEqual(
 154                 type_expected, type_got,
 155                 'Type mismatch for list item at index %d for field %s, expected %r, got %r' % (
 156                     index, field, type_expected, type_got))
 157             expect_value(self, item_got, item_expected, field)
 158     else:
 159         if isinstance(expected, compat_str) and expected.startswith('md5:'):
 160             self.assertTrue(
 161                 isinstance(got, compat_str),
 162                 'Expected field %s to be a unicode object, but got value %r of type %r' % (field, got, type(got)))
 163             got = 'md5:' + md5(got)
 164         elif isinstance(expected, compat_str) and re.match(r'^(?:min|max)?count:\d+', expected):
 165             self.assertTrue(
 166                 isinstance(got, (list, dict)),
 167                 'Expected field %s to be a list or a dict, but it is of type %s' % (
 168                     field, type(got).__name__))
 169             op, _, expected_num = expected.partition(':')
 170             expected_num = int(expected_num)
 171             if op == 'mincount':
 172                 assert_func = assertGreaterEqual
 173                 msg_tmpl = 'Expected %d items in field %s, but only got %d'
 174             elif op == 'maxcount':
 175                 assert_func = assertLessEqual
 176                 msg_tmpl = 'Expected maximum %d items in field %s, but got %d'
 177             elif op == 'count':
 178                 assert_func = assertEqual
 179                 msg_tmpl = 'Expected exactly %d items in field %s, but got %d'
 180             else:
 181                 assert False
 182             assert_func(
 183                 self, len(got), expected_num,
 184                 msg_tmpl % (expected_num, field, len(got)))
 185             return
 186         self.assertEqual(
 187             expected, got,
 188             'Invalid value for field %s, expected %r, got %r' % (field, expected, got))
 189
 190
 191 def expect_dict(self, got_dict, expected_dict):
 192     for info_field, expected in expected_dict.items():
 193         got = got_dict.get(info_field)
 194         expect_value(self, got, expected, info_field)
 195
 196
 197 def sanitize_got_info_dict(got_dict):
 198     IGNORED_FIELDS = (
 199         # Format keys
 200         'url', 'manifest_url', 'format', 'format_id', 'format_note', 'width', 'height', 'resolution',
 201         'dynamic_range', 'tbr', 'abr', 'acodec', 'asr', 'vbr', 'fps', 'vcodec', 'container', 'filesize',
 202         'filesize_approx', 'player_url', 'protocol', 'fragment_base_url', 'fragments', 'preference',
 203         'language', 'language_preference', 'quality', 'source_preference', 'http_headers',
 204         'stretched_ratio', 'no_resume', 'has_drm', 'downloader_options',
 205
 206         # RTMP formats
 207         'page_url', 'app', 'play_path', 'tc_url', 'flash_version', 'rtmp_live', 'rtmp_conn', 'rtmp_protocol', 'rtmp_real_time',
 208
 209         # Lists
 210         'formats', 'thumbnails', 'subtitles', 'automatic_captions', 'comments', 'entries',
 211
 212         # Auto-generated
 213         'autonumber', 'playlist', 'format_index', 'video_ext', 'audio_ext', 'duration_string', 'epoch',
 214         'fulltitle', 'extractor', 'extractor_key', 'filepath', 'infojson_filename', 'original_url', 'n_entries',
 215
 216         # Only live_status needs to be checked
 217         'is_live', 'was_live',
 218     )
 219
 220     IGNORED_PREFIXES = ('', 'playlist', 'requested', 'webpage')
 221
 222     def sanitize(key, value):
 223         if isinstance(value, str) and len(value) > 100:
 224             return f'md5:{md5(value)}'
 225         elif isinstance(value, list) and len(value) > 10:
 226             return f'count:{len(value)}'
 227         elif key.endswith('_count') and isinstance(value, int):
 228             return int
 229         return value
 230
 231     test_info_dict = {
 232         key: sanitize(key, value) for key, value in got_dict.items()
 233         if value is not None and key not in IGNORED_FIELDS and not any(
 234             key.startswith(f'{prefix}_') for prefix in IGNORED_PREFIXES)
 235     }
 236
 237     # display_id may be generated from id
 238     if test_info_dict.get('display_id') == test_info_dict.get('id'):
 239         test_info_dict.pop('display_id')
 240
 241     return test_info_dict
 242
 243
 244 def expect_info_dict(self, got_dict, expected_dict):
 245     expect_dict(self, got_dict, expected_dict)
 246     # Check for the presence of mandatory fields
 247     if got_dict.get('_type') not in ('playlist', 'multi_video'):
 248         mandatory_fields = ['id', 'title']
 249         if expected_dict.get('ext'):
 250             mandatory_fields.extend(('url', 'ext'))
 251         for key in mandatory_fields:
 252             self.assertTrue(got_dict.get(key), 'Missing mandatory field %s' % key)
 253     # Check for mandatory fields that are automatically set by YoutubeDL
 254     for key in ['webpage_url', 'extractor', 'extractor_key']:
 255         self.assertTrue(got_dict.get(key), 'Missing field: %s' % key)
 256
 257     test_info_dict = sanitize_got_info_dict(got_dict)
 258
 259     missing_keys = set(test_info_dict.keys()) - set(expected_dict.keys())
 260     if missing_keys:
 261         def _repr(v):
 262             if isinstance(v, compat_str):
 263                 return "'%s'" % v.replace('\\', '\\\\').replace("'", "\\'").replace('\n', '\\n')
 264             elif isinstance(v, type):
 265                 return v.__name__
 266             else:
 267                 return repr(v)
 268         info_dict_str = ''
 269         if len(missing_keys) != len(expected_dict):
 270             info_dict_str += ''.join(
 271                 '    %s: %s,\n' % (_repr(k), _repr(v))
 272                 for k, v in test_info_dict.items() if k not in missing_keys)
 273
 274             if info_dict_str:
 275                 info_dict_str += '\n'
 276         info_dict_str += ''.join(
 277             '    %s: %s,\n' % (_repr(k), _repr(test_info_dict[k]))
 278             for k in missing_keys)
 279         write_string(
 280             '\n\'info_dict\': {\n' + info_dict_str + '},\n', out=sys.stderr)
 281         self.assertFalse(
 282             missing_keys,
 283             'Missing keys in test definition: %s' % (
 284                 ', '.join(sorted(missing_keys))))
 285
 286
 287 def assertRegexpMatches(self, text, regexp, msg=None):
 288     if hasattr(self, 'assertRegexp'):
 289         return self.assertRegexp(text, regexp, msg)
 290     else:
 291         m = re.match(regexp, text)
 292         if not m:
 293             note = 'Regexp didn\'t match: %r not found' % (regexp)
 294             if len(text) < 1000:
 295                 note += ' in %r' % text
 296             if msg is None:
 297                 msg = note
 298             else:
 299                 msg = note + ', ' + msg
 300             self.assertTrue(m, msg)
 301
 302
 303 def assertGreaterEqual(self, got, expected, msg=None):
 304     if not (got >= expected):
 305         if msg is None:
 306             msg = '%r not greater than or equal to %r' % (got, expected)
 307         self.assertTrue(got >= expected, msg)
 308
 309
 310 def assertLessEqual(self, got, expected, msg=None):
 311     if not (got <= expected):
 312         if msg is None:
 313             msg = '%r not less than or equal to %r' % (got, expected)
 314         self.assertTrue(got <= expected, msg)
 315
 316
 317 def assertEqual(self, got, expected, msg=None):
 318     if not (got == expected):
 319         if msg is None:
 320             msg = '%r not equal to %r' % (got, expected)
 321         self.assertTrue(got == expected, msg)
 322
 323
 324 def expect_warnings(ydl, warnings_re):
 325     real_warning = ydl.report_warning
 326
 327     def _report_warning(w):
 328         if not any(re.search(w_re, w) for w_re in warnings_re):
 329             real_warning(w)
 330
 331     ydl.report_warning = _report_warning
 332
 333
 334 def http_server_port(httpd):
 335     if os.name == 'java' and isinstance(httpd.socket, ssl.SSLSocket):
 336         # In Jython SSLSocket is not a subclass of socket.socket
 337         sock = httpd.socket.sock
 338     else:
 339         sock = httpd.socket
 340     return sock.getsockname()[1]