yt-dlp/test/helper.py

337 lines
12 KiB
Python
Raw Normal View History

2013-10-05 21:47:17 -06:00
import errno
import hashlib
2013-06-26 16:09:05 -06:00
import json
import os.path
import re
import ssl
import sys
import types
2013-06-26 16:09:05 -06:00
import yt_dlp.extractor
from yt_dlp import YoutubeDL
from yt_dlp.compat import compat_os_name
from yt_dlp.utils import preferredencoding, try_call, write_string, find_available_port
2013-06-26 16:09:05 -06:00
2021-10-08 18:23:15 -06:00
if 'pytest' in sys.modules:
import pytest
is_download_test = pytest.mark.download
else:
def is_download_test(testClass):
return testClass
def get_params(override=None):
PARAMETERS_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)),
2021-10-08 18:23:15 -06:00
'parameters.json')
LOCAL_PARAMETERS_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)),
2021-10-08 18:23:15 -06:00
'local_parameters.json')
with open(PARAMETERS_FILE, encoding='utf-8') as pf:
parameters = json.load(pf)
if os.path.exists(LOCAL_PARAMETERS_FILE):
with open(LOCAL_PARAMETERS_FILE, encoding='utf-8') as pf:
parameters.update(json.load(pf))
if override:
parameters.update(override)
return parameters
2013-06-26 16:09:05 -06:00
2013-10-05 21:47:17 -06:00
def try_rm(filename):
""" Remove a file if it exists """
try:
os.remove(filename)
except OSError as ose:
if ose.errno != errno.ENOENT:
raise
def report_warning(message, *args, **kwargs):
'''
Print the message to stderr, it will be prefixed with 'WARNING:'
If stderr is a tty file the 'WARNING:' will be colored
'''
if sys.stderr.isatty() and compat_os_name != 'nt':
2014-09-29 14:11:24 -06:00
_msg_header = '\033[0;33mWARNING:\033[0m'
else:
2014-09-29 14:11:24 -06:00
_msg_header = 'WARNING:'
output = f'{_msg_header} {message}\n'
if 'b' in getattr(sys.stderr, 'mode', ''):
output = output.encode(preferredencoding())
sys.stderr.write(output)
2013-06-26 16:09:05 -06:00
class FakeYDL(YoutubeDL):
2013-10-17 16:46:35 -06:00
def __init__(self, override=None):
2013-06-26 16:09:05 -06:00
# Different instances of the downloader can't share the same dictionary
# some test set the "sublang" parameter, which would break the md5 checks.
2013-10-17 16:46:35 -06:00
params = get_params(override=override)
super().__init__(params, auto_init=False)
2013-10-05 21:47:17 -06:00
self.result = []
2014-11-23 12:41:03 -07:00
def to_screen(self, s, *args, **kwargs):
2013-06-26 16:09:05 -06:00
print(s)
2013-10-05 21:47:17 -06:00
def trouble(self, s, *args, **kwargs):
2013-06-26 16:09:05 -06:00
raise Exception(s)
2013-10-05 21:47:17 -06:00
2013-06-26 16:09:05 -06:00
def download(self, x):
2013-06-27 13:15:16 -06:00
self.result.append(x)
2013-10-05 21:47:17 -06:00
def expect_warning(self, regex):
# Silence an expected warning matching a regex
old_report_warning = self.report_warning
2014-11-23 12:41:03 -07:00
def report_warning(self, message, *args, **kwargs):
2014-11-23 12:41:03 -07:00
if re.match(regex, message):
return
old_report_warning(message, *args, **kwargs)
self.report_warning = types.MethodType(report_warning, self)
2013-06-27 13:15:16 -06:00
def gettestcases(include_onlymatching=False):
for ie in yt_dlp.extractor.gen_extractors():
yield from ie.get_testcases(include_onlymatching)
def getwebpagetestcases():
for ie in yt_dlp.extractor.gen_extractors():
for tc in ie.get_webpage_testcases():
tc.setdefault('add_ie', []).append('Generic')
yield tc
md5 = lambda s: hashlib.md5(s.encode()).hexdigest()
def expect_value(self, got, expected, field):
if isinstance(expected, str) and expected.startswith('re:'):
match_str = expected[len('re:'):]
match_rex = re.compile(match_str)
self.assertTrue(
isinstance(got, str),
f'Expected a {str.__name__} object, but got {type(got).__name__} for field {field}')
self.assertTrue(
match_rex.match(got),
f'field {field} (value: {got!r}) should match {match_str!r}')
elif isinstance(expected, str) and expected.startswith('startswith:'):
start_str = expected[len('startswith:'):]
self.assertTrue(
isinstance(got, str),
f'Expected a {str.__name__} object, but got {type(got).__name__} for field {field}')
self.assertTrue(
got.startswith(start_str),
f'field {field} (value: {got!r}) should start with {start_str!r}')
elif isinstance(expected, str) and expected.startswith('contains:'):
contains_str = expected[len('contains:'):]
self.assertTrue(
isinstance(got, str),
f'Expected a {str.__name__} object, but got {type(got).__name__} for field {field}')
self.assertTrue(
contains_str in got,
f'field {field} (value: {got!r}) should contain {contains_str!r}')
elif isinstance(expected, type):
2015-09-30 08:21:01 -06:00
self.assertTrue(
isinstance(got, expected),
f'Expected type {expected!r} for field {field}, but got value {got!r} of type {type(got)!r}')
elif isinstance(expected, dict) and isinstance(got, dict):
expect_dict(self, got, expected)
elif isinstance(expected, list) and isinstance(got, list):
2015-09-30 08:21:01 -06:00
self.assertEqual(
len(expected), len(got),
'Expect a list of length %d, but got a list of length %d for field %s' % (
len(expected), len(got), field))
2015-09-30 02:30:04 -06:00
for index, (item_got, item_expected) in enumerate(zip(got, expected)):
type_got = type(item_got)
type_expected = type(item_expected)
2015-09-30 08:21:01 -06:00
self.assertEqual(
type_expected, type_got,
2015-09-30 08:26:42 -06:00
'Type mismatch for list item at index %d for field %s, expected %r, got %r' % (
index, field, type_expected, type_got))
2015-09-30 02:30:04 -06:00
expect_value(self, item_got, item_expected, field)
else:
if isinstance(expected, str) and expected.startswith('md5:'):
self.assertTrue(
isinstance(got, str),
f'Expected field {field} to be a unicode object, but got value {got!r} of type {type(got)!r}')
got = 'md5:' + md5(got)
elif isinstance(expected, str) and re.match(r'^(?:min|max)?count:\d+', expected):
2015-02-18 11:58:41 -07:00
self.assertTrue(
isinstance(got, (list, dict)),
f'Expected field {field} to be a list or a dict, but it is of type {type(got).__name__}')
op, _, expected_num = expected.partition(':')
expected_num = int(expected_num)
if op == 'mincount':
assert_func = assertGreaterEqual
msg_tmpl = 'Expected %d items in field %s, but only got %d'
elif op == 'maxcount':
assert_func = assertLessEqual
msg_tmpl = 'Expected maximum %d items in field %s, but got %d'
elif op == 'count':
assert_func = assertEqual
msg_tmpl = 'Expected exactly %d items in field %s, but got %d'
else:
assert False
assert_func(
self, len(got), expected_num,
msg_tmpl % (expected_num, field, len(got)))
return
2015-09-30 08:21:01 -06:00
self.assertEqual(
expected, got,
f'Invalid value for field {field}, expected {expected!r}, got {got!r}')
def expect_dict(self, got_dict, expected_dict):
for info_field, expected in expected_dict.items():
got = got_dict.get(info_field)
expect_value(self, got, expected, info_field)
def sanitize_got_info_dict(got_dict):
IGNORED_FIELDS = (
*YoutubeDL._format_fields,
2021-12-14 15:02:40 -07:00
# Lists
'formats', 'thumbnails', 'subtitles', 'automatic_captions', 'comments', 'entries',
# Auto-generated
'autonumber', 'playlist', 'format_index', 'video_ext', 'audio_ext', 'duration_string', 'epoch', 'n_entries',
'fulltitle', 'extractor', 'extractor_key', 'filename', 'filepath', 'infojson_filename', 'original_url',
2021-12-14 15:02:40 -07:00
# Only live_status needs to be checked
'is_live', 'was_live',
)
IGNORED_PREFIXES = ('', 'playlist', 'requested', 'webpage')
2021-12-14 15:02:40 -07:00
def sanitize(key, value):
if isinstance(value, str) and len(value) > 100 and key != 'thumbnail':
2021-12-14 15:02:40 -07:00
return f'md5:{md5(value)}'
elif isinstance(value, list) and len(value) > 10:
return f'count:{len(value)}'
2022-01-07 04:54:57 -07:00
elif key.endswith('_count') and isinstance(value, int):
return int
2021-12-14 15:02:40 -07:00
return value
test_info_dict = {
key: sanitize(key, value) for key, value in got_dict.items()
if value is not None and key not in IGNORED_FIELDS and (
not any(key.startswith(f'{prefix}_') for prefix in IGNORED_PREFIXES)
or key == '_old_archive_ids')
2021-12-14 15:02:40 -07:00
}
# display_id may be generated from id
2022-01-19 15:57:36 -07:00
if test_info_dict.get('display_id') == test_info_dict.get('id'):
2021-12-14 15:02:40 -07:00
test_info_dict.pop('display_id')
# release_year may be generated from release_date
if try_call(lambda: test_info_dict['release_year'] == int(test_info_dict['release_date'][:4])):
test_info_dict.pop('release_year')
# Check url for flat entries
if got_dict.get('_type', 'video') != 'video' and got_dict.get('url'):
test_info_dict['url'] = got_dict['url']
return test_info_dict
def expect_info_dict(self, got_dict, expected_dict):
expect_dict(self, got_dict, expected_dict)
# Check for the presence of mandatory fields
if got_dict.get('_type') not in ('playlist', 'multi_video'):
mandatory_fields = ['id', 'title']
if expected_dict.get('ext'):
mandatory_fields.extend(('url', 'ext'))
for key in mandatory_fields:
self.assertTrue(got_dict.get(key), 'Missing mandatory field %s' % key)
# Check for mandatory fields that are automatically set by YoutubeDL
if got_dict.get('_type', 'video') == 'video':
for key in ['webpage_url', 'extractor', 'extractor_key']:
self.assertTrue(got_dict.get(key), 'Missing field: %s' % key)
test_info_dict = sanitize_got_info_dict(got_dict)
missing_keys = set(test_info_dict.keys()) - set(expected_dict.keys())
if missing_keys:
def _repr(v):
if isinstance(v, str):
return "'%s'" % v.replace('\\', '\\\\').replace("'", "\\'").replace('\n', '\\n')
2022-01-07 04:54:57 -07:00
elif isinstance(v, type):
return v.__name__
else:
return repr(v)
info_dict_str = ''.join(
f' {_repr(k)}: {_repr(v)},\n'
for k, v in test_info_dict.items() if k not in missing_keys)
if info_dict_str:
info_dict_str += '\n'
info_dict_str += ''.join(
f' {_repr(k)}: {_repr(test_info_dict[k])},\n'
for k in missing_keys)
2022-11-06 13:59:58 -07:00
info_dict_str = '\n\'info_dict\': {\n' + info_dict_str + '},\n'
write_string(info_dict_str.replace('\n', '\n '), out=sys.stderr)
self.assertFalse(
missing_keys,
'Missing keys in test definition: %s' % (
', '.join(sorted(missing_keys))))
def assertRegexpMatches(self, text, regexp, msg=None):
if hasattr(self, 'assertRegexp'):
return self.assertRegexp(text, regexp, msg)
else:
m = re.match(regexp, text)
if not m:
2014-12-12 09:06:52 -07:00
note = 'Regexp didn\'t match: %r not found' % (regexp)
if len(text) < 1000:
note += ' in %r' % text
if msg is None:
msg = note
else:
msg = note + ', ' + msg
self.assertTrue(m, msg)
def assertGreaterEqual(self, got, expected, msg=None):
if not (got >= expected):
if msg is None:
msg = f'{got!r} not greater than or equal to {expected!r}'
self.assertTrue(got >= expected, msg)
def assertLessEqual(self, got, expected, msg=None):
if not (got <= expected):
if msg is None:
msg = f'{got!r} not less than or equal to {expected!r}'
self.assertTrue(got <= expected, msg)
def assertEqual(self, got, expected, msg=None):
if not (got == expected):
if msg is None:
msg = f'{got!r} not equal to {expected!r}'
self.assertTrue(got == expected, msg)
def expect_warnings(ydl, warnings_re):
real_warning = ydl.report_warning
def _report_warning(w, *args, **kwargs):
if not any(re.search(w_re, w) for w_re in warnings_re):
real_warning(w, *args, **kwargs)
ydl.report_warning = _report_warning
def http_server_port(httpd):
if os.name == 'java' and isinstance(httpd.socket, ssl.SSLSocket):
# In Jython SSLSocket is not a subclass of socket.socket
sock = httpd.socket.sock
else:
sock = httpd.socket
return sock.getsockname()[1]
def verify_address_availability(address):
if find_available_port(address) is None:
pytest.skip(f'Unable to bind to source address {address} (address may not exist)')