[extractor] Use classmethod/property where possible

and refactor lazy extractors accordingly.

This reduces the need to create extractor instances
This commit is contained in:
pukkandan 2022-05-11 21:24:44 +05:30
parent 7ddbf09c25
commit 82d020804d
No known key found for this signature in database
GPG Key ID: 7EEE9E1E817D0A39
11 changed files with 188 additions and 167 deletions

View File

@ -1,30 +1,28 @@
import importlib
import random
import re
from ..utils import bug_reports_message, write_string
from ..utils import bug_reports_message, classproperty, write_string
class LazyLoadMetaClass(type):
def __getattr__(cls, name):
if '_real_class' not in cls.__dict__:
# "is_suitable" requires "_TESTS". However, they bloat the lazy_extractors
if '_real_class' not in cls.__dict__ and name not in ('is_suitable', 'get_testcases'):
write_string(
'WARNING: Falling back to normal extractor since lazy extractor '
f'{cls.__name__} does not have attribute {name}{bug_reports_message()}')
return getattr(cls._get_real_class(), name)
f'{cls.__name__} does not have attribute {name}{bug_reports_message()}\n')
return getattr(cls.real_class, name)
class LazyLoadExtractor(metaclass=LazyLoadMetaClass):
_module = None
_WORKING = True
@classmethod
def _get_real_class(cls):
@classproperty
def real_class(cls):
if '_real_class' not in cls.__dict__:
mod = __import__(cls._module, fromlist=(cls.__name__,))
cls._real_class = getattr(mod, cls.__name__)
cls._real_class = getattr(importlib.import_module(cls._module), cls.__name__)
return cls._real_class
def __new__(cls, *args, **kwargs):
real_cls = cls._get_real_class()
instance = real_cls.__new__(real_cls)
instance = cls.real_class.__new__(cls.real_class)
instance.__init__(*args, **kwargs)
return instance

View File

@ -1,101 +1,125 @@
#!/usr/bin/env python3
import os
import optparse
import sys
from inspect import getsource
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
lazy_extractors_filename = sys.argv[1] if len(sys.argv) > 1 else 'yt_dlp/extractor/lazy_extractors.py'
if os.path.exists(lazy_extractors_filename):
os.remove(lazy_extractors_filename)
# Block plugins from loading
plugins_dirname = 'ytdlp_plugins'
plugins_blocked_dirname = 'ytdlp_plugins_blocked'
if os.path.exists(plugins_dirname):
os.rename(plugins_dirname, plugins_blocked_dirname)
from yt_dlp.extractor import _ALL_CLASSES
from yt_dlp.extractor.common import InfoExtractor, SearchInfoExtractor
if os.path.exists(plugins_blocked_dirname):
os.rename(plugins_blocked_dirname, plugins_dirname)
with open('devscripts/lazy_load_template.py', encoding='utf-8') as f:
module_template = f.read()
CLASS_PROPERTIES = ['ie_key', 'working', '_match_valid_url', 'suitable', '_match_id', 'get_temp_id']
module_contents = [
module_template,
*[getsource(getattr(InfoExtractor, k)) for k in CLASS_PROPERTIES],
'\nclass LazyLoadSearchExtractor(LazyLoadExtractor):\n pass\n']
ie_template = '''
NO_ATTR = object()
STATIC_CLASS_PROPERTIES = ['IE_NAME', 'IE_DESC', 'SEARCH_KEY', '_WORKING', '_NETRC_MACHINE']
CLASS_METHODS = [
'ie_key', 'working', 'description', 'suitable', '_match_valid_url', '_match_id', 'get_temp_id',
]
IE_TEMPLATE = '''
class {name}({bases}):
_module = '{module}'
_module = {module!r}
'''
with open('devscripts/lazy_load_template.py', encoding='utf-8') as f:
MODULE_TEMPLATE = f.read()
def get_base_name(base):
if base is InfoExtractor:
return 'LazyLoadExtractor'
elif base is SearchInfoExtractor:
return 'LazyLoadSearchExtractor'
else:
return base.__name__
def main():
parser = optparse.OptionParser(usage='%prog [OUTFILE.py]')
args = parser.parse_args()[1] or ['yt_dlp/extractor/lazy_extractors.py']
if len(args) != 1:
parser.error('Expected only an output filename')
lazy_extractors_filename = args[0]
if os.path.exists(lazy_extractors_filename):
os.remove(lazy_extractors_filename)
_ALL_CLASSES = get_all_ies() # Must be before import
from yt_dlp.extractor.common import InfoExtractor, SearchInfoExtractor
DummyInfoExtractor = type('InfoExtractor', (InfoExtractor,), {'IE_NAME': NO_ATTR})
module_src = '\n'.join((
MODULE_TEMPLATE,
' _module = None',
*extra_ie_code(DummyInfoExtractor),
'\nclass LazyLoadSearchExtractor(LazyLoadExtractor):\n pass\n',
*build_ies(_ALL_CLASSES, (InfoExtractor, SearchInfoExtractor), DummyInfoExtractor),
))
with open(lazy_extractors_filename, 'wt', encoding='utf-8') as f:
f.write(f'{module_src}\n')
def build_lazy_ie(ie, name):
s = ie_template.format(
name=name,
bases=', '.join(map(get_base_name, ie.__bases__)),
module=ie.__module__)
def get_all_ies():
PLUGINS_DIRNAME = 'ytdlp_plugins'
BLOCKED_DIRNAME = f'{PLUGINS_DIRNAME}_blocked'
if os.path.exists(PLUGINS_DIRNAME):
os.rename(PLUGINS_DIRNAME, BLOCKED_DIRNAME)
try:
from yt_dlp.extractor import _ALL_CLASSES
finally:
if os.path.exists(BLOCKED_DIRNAME):
os.rename(BLOCKED_DIRNAME, PLUGINS_DIRNAME)
return _ALL_CLASSES
def extra_ie_code(ie, base=None):
for var in STATIC_CLASS_PROPERTIES:
val = getattr(ie, var)
if val != (getattr(base, var) if base else NO_ATTR):
yield f' {var} = {val!r}'
yield ''
for name in CLASS_METHODS:
f = getattr(ie, name)
if not base or f.__func__ != getattr(base, name).__func__:
yield getsource(f)
def build_ies(ies, bases, attr_base):
names = []
for ie in sort_ies(ies, bases):
yield build_lazy_ie(ie, ie.__name__, attr_base)
if ie in ies:
names.append(ie.__name__)
yield f'\n_ALL_CLASSES = [{", ".join(names)}]'
def sort_ies(ies, ignored_bases):
"""find the correct sorting and add the required base classes so that subclasses can be correctly created"""
classes, returned_classes = ies[:-1], set()
assert ies[-1].__name__ == 'GenericIE', 'Last IE must be GenericIE'
while classes:
for c in classes[:]:
bases = set(c.__bases__) - {object, *ignored_bases}
restart = False
for b in bases:
if b not in classes and b not in returned_classes:
assert b.__name__ != 'GenericIE', 'Cannot inherit from GenericIE'
classes.insert(0, b)
restart = True
if restart:
break
if bases <= returned_classes:
yield c
returned_classes.add(c)
classes.remove(c)
break
yield ies[-1]
def build_lazy_ie(ie, name, attr_base):
bases = ', '.join({
'InfoExtractor': 'LazyLoadExtractor',
'SearchInfoExtractor': 'LazyLoadSearchExtractor',
}.get(base.__name__, base.__name__) for base in ie.__bases__)
s = IE_TEMPLATE.format(name=name, module=ie.__module__, bases=bases)
valid_url = getattr(ie, '_VALID_URL', None)
if not valid_url and hasattr(ie, '_make_valid_url'):
valid_url = ie._make_valid_url()
if valid_url:
s += f' _VALID_URL = {valid_url!r}\n'
if not ie._WORKING:
s += ' _WORKING = False\n'
if ie.suitable.__func__ is not InfoExtractor.suitable.__func__:
s += f'\n{getsource(ie.suitable)}'
return s
return s + '\n'.join(extra_ie_code(ie, attr_base))
# find the correct sorting and add the required base classes so that subclasses
# can be correctly created
classes = _ALL_CLASSES[:-1]
ordered_cls = []
while classes:
for c in classes[:]:
bases = set(c.__bases__) - {object, InfoExtractor, SearchInfoExtractor}
stop = False
for b in bases:
if b not in classes and b not in ordered_cls:
if b.__name__ == 'GenericIE':
exit()
classes.insert(0, b)
stop = True
if stop:
break
if all(b in ordered_cls for b in bases):
ordered_cls.append(c)
classes.remove(c)
break
ordered_cls.append(_ALL_CLASSES[-1])
names = []
for ie in ordered_cls:
name = ie.__name__
src = build_lazy_ie(ie, name)
module_contents.append(src)
if ie in _ALL_CLASSES:
names.append(name)
module_contents.append(
'\n_ALL_CLASSES = [{}]'.format(', '.join(names)))
module_src = '\n'.join(module_contents) + '\n'
with open(lazy_extractors_filename, 'wt', encoding='utf-8') as f:
f.write(module_src)
if __name__ == '__main__':
main()

View File

@ -5,7 +5,7 @@
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from yt_dlp.extractor import list_extractors
from yt_dlp.extractor import list_extractor_classes
def main():
@ -14,7 +14,7 @@ def main():
if len(args) != 1:
parser.error('Expected an output filename')
out = '\n'.join(ie.description() for ie in list_extractors(None) if ie.IE_DESC is not False)
out = '\n'.join(ie.description() for ie in list_extractor_classes() if ie.IE_DESC is not False)
with open(args[0], 'w', encoding='utf-8') as outf:
outf.write(f'# Supported sites\n{out}\n')

View File

@ -431,7 +431,6 @@ # Supported sites
- **gem.cbc.ca**: [<abbr title="netrc machine"><em>cbcgem</em></abbr>]
- **gem.cbc.ca:live**
- **gem.cbc.ca:playlist**
- **generic**: Generic downloader that works on some sites
- **Gettr**
- **GettrStreaming**
- **Gfycat**
@ -1553,3 +1552,4 @@ # Supported sites
- **zingmp3:album**
- **zoom**
- **Zype**
- **generic**: Generic downloader that works on some sites

View File

@ -11,7 +11,7 @@
from .compat import compat_getpass, compat_os_name, compat_shlex_quote
from .cookies import SUPPORTED_BROWSERS, SUPPORTED_KEYRINGS
from .downloader import FileDownloader
from .extractor import list_extractors
from .extractor import GenericIE, list_extractor_classes
from .extractor.adobepass import MSO_INFO
from .extractor.common import InfoExtractor
from .options import parseOpts
@ -76,14 +76,20 @@ def get_urls(urls, batchfile, verbose):
def print_extractor_information(opts, urls):
out = ''
if opts.list_extractors:
for ie in list_extractors(opts.age_limit):
urls = dict.fromkeys(urls, False)
for ie in list_extractor_classes(opts.age_limit):
out += ie.IE_NAME + (' (CURRENTLY BROKEN)' if not ie.working() else '') + '\n'
out += ''.join(f' {url}\n' for url in filter(ie.suitable, urls))
if ie == GenericIE:
matched_urls = [url for url, matched in urls.items() if not matched]
else:
matched_urls = tuple(filter(ie.suitable, urls.keys()))
urls.update(dict.fromkeys(matched_urls, True))
out += ''.join(f' {url}\n' for url in matched_urls)
elif opts.list_extractor_descriptions:
_SEARCHES = ('cute kittens', 'slithering pythons', 'falling cat', 'angry poodle', 'purple fish', 'running tortoise', 'sleeping bunny', 'burping cow')
out = '\n'.join(
ie.description(markdown=False, search_examples=_SEARCHES)
for ie in list_extractors(opts.age_limit) if ie.working() and ie.IE_DESC is not False) + '\n'
for ie in list_extractor_classes(opts.age_limit) if ie.working() and ie.IE_DESC is not False)
elif opts.ap_list_mso:
out = 'Supported TV Providers:\n%s\n' % render_table(
['mso', 'mso name'],
@ -862,7 +868,7 @@ def main(argv=None):
sys.exit(f'\nERROR: {e}')
from .extractor import gen_extractors
from .extractor import gen_extractors, list_extractors
__all__ = [
'main',
'YoutubeDL',

View File

@ -37,11 +37,17 @@ def gen_extractors():
return [klass() for klass in gen_extractor_classes()]
def list_extractors(age_limit):
def list_extractor_classes(age_limit=None):
"""Return a list of extractors that are suitable for the given age, sorted by extractor name"""
return sorted(filter(
lambda ie: ie.is_suitable(age_limit),
gen_extractors()), key=lambda ie: ie.IE_NAME.lower())
yield from sorted(filter(
lambda ie: ie.is_suitable(age_limit) and ie != GenericIE, # noqa: F405
gen_extractor_classes()), key=lambda ie: ie.IE_NAME.lower())
yield GenericIE # noqa: F405
def list_extractors(age_limit=None):
"""Return a list of extractor instances that are suitable for the given age, sorted by extractor name"""
return [ie() for ie in list_extractor_classes(age_limit)]
def get_info_extractor(ie_name):

View File

@ -40,6 +40,7 @@
age_restricted,
base_url,
bug_reports_message,
classproperty,
clean_html,
determine_ext,
determine_protocol,
@ -710,9 +711,9 @@ def ie_key(cls):
"""A string for getting the InfoExtractor with get_info_extractor"""
return cls.__name__[:-2]
@property
def IE_NAME(self):
return type(self).__name__[:-2]
@classproperty
def IE_NAME(cls):
return cls.__name__[:-2]
@staticmethod
def __can_accept_status_code(err, expected_status):
@ -3624,56 +3625,57 @@ def _apply_first_set_cookie_header(self, url_handle, cookie):
self._set_cookie(domain, cookie, value)
break
def get_testcases(self, include_onlymatching=False):
t = getattr(self, '_TEST', None)
@classmethod
def get_testcases(cls, include_onlymatching=False):
t = getattr(cls, '_TEST', None)
if t:
assert not hasattr(self, '_TESTS'), \
'%s has _TEST and _TESTS' % type(self).__name__
assert not hasattr(cls, '_TESTS'), f'{cls.ie_key()}IE has _TEST and _TESTS'
tests = [t]
else:
tests = getattr(self, '_TESTS', [])
tests = getattr(cls, '_TESTS', [])
for t in tests:
if not include_onlymatching and t.get('only_matching', False):
continue
t['name'] = type(self).__name__[:-len('IE')]
t['name'] = cls.ie_key()
yield t
def is_suitable(self, age_limit):
@classmethod
def is_suitable(cls, age_limit):
""" Test whether the extractor is generally suitable for the given
age limit (i.e. pornographic sites are not, all others usually are) """
any_restricted = False
for tc in self.get_testcases(include_onlymatching=False):
for tc in cls.get_testcases(include_onlymatching=False):
if tc.get('playlist', []):
tc = tc['playlist'][0]
is_restricted = age_restricted(
tc.get('info_dict', {}).get('age_limit'), age_limit)
is_restricted = age_restricted(tc.get('info_dict', {}).get('age_limit'), age_limit)
if not is_restricted:
return True
any_restricted = any_restricted or is_restricted
return not any_restricted
def description(self, *, markdown=True, search_examples=None):
@classmethod
def description(cls, *, markdown=True, search_examples=None):
"""Description of the extractor"""
desc = ''
if self._NETRC_MACHINE:
if cls._NETRC_MACHINE:
if markdown:
desc += f' [<abbr title="netrc machine"><em>{self._NETRC_MACHINE}</em></abbr>]'
desc += f' [<abbr title="netrc machine"><em>{cls._NETRC_MACHINE}</em></abbr>]'
else:
desc += f' [{self._NETRC_MACHINE}]'
if self.IE_DESC is False:
desc += f' [{cls._NETRC_MACHINE}]'
if cls.IE_DESC is False:
desc += ' [HIDDEN]'
elif self.IE_DESC:
desc += f' {self.IE_DESC}'
if self.SEARCH_KEY:
desc += f'; "{self.SEARCH_KEY}:" prefix'
elif cls.IE_DESC:
desc += f' {cls.IE_DESC}'
if cls.SEARCH_KEY:
desc += f'; "{cls.SEARCH_KEY}:" prefix'
if search_examples:
_COUNTS = ('', '5', '10', 'all')
desc += f' (Example: "{self.SEARCH_KEY}{random.choice(_COUNTS)}:{random.choice(search_examples)}")'
if not self.working():
desc += f' (Example: "{cls.SEARCH_KEY}{random.choice(_COUNTS)}:{random.choice(search_examples)}")'
if not cls.working():
desc += ' (**Currently broken**)' if markdown else ' (Currently broken)'
name = f' - **{self.IE_NAME}**' if markdown else self.IE_NAME
name = f' - **{cls.IE_NAME}**' if markdown else cls.IE_NAME
return f'{name}:{desc}' if desc else name
def extract_subtitles(self, *args, **kwargs):
@ -3849,6 +3851,6 @@ def _search_results(self, query):
"""Returns an iterator of search results"""
raise NotImplementedError('This method must be implemented by subclasses')
@property
def SEARCH_KEY(self):
return self._SEARCH_KEY
@classproperty
def SEARCH_KEY(cls):
return cls._SEARCH_KEY

View File

@ -18,6 +18,7 @@
url_or_none,
)
class DRTVIE(InfoExtractor):
_VALID_URL = r'''(?x)
https?://

View File

@ -8,55 +8,36 @@ class TestURLIE(InfoExtractor):
""" Allows addressing of the test cases as test:yout.*be_1 """
IE_DESC = False # Do not list
_VALID_URL = r'test(?:url)?:(?P<id>(?P<extractor>.+?)(?:_(?P<num>[0-9]+))?)$'
_VALID_URL = r'test(?:url)?:(?P<extractor>.+?)(?:_(?P<num>[0-9]+))?$'
def _real_extract(self, url):
from ..extractor import gen_extractors
from ..extractor import gen_extractor_classes
mobj = self._match_valid_url(url)
video_id = mobj.group('id')
extractor_id = mobj.group('extractor')
all_extractors = gen_extractors()
extractor_id, num = self._match_valid_url(url).group('extractor', 'num')
rex = re.compile(extractor_id, flags=re.IGNORECASE)
matching_extractors = [
e for e in all_extractors if rex.search(e.IE_NAME)]
matching_extractors = [e for e in gen_extractor_classes() if rex.search(e.IE_NAME)]
if len(matching_extractors) == 0:
raise ExtractorError(
'No extractors matching %r found' % extractor_id,
expected=True)
raise ExtractorError('No extractors matching {extractor_id!r} found', expected=True)
elif len(matching_extractors) > 1:
# Is it obvious which one to pick?
try:
try: # Check for exact match
extractor = next(
ie for ie in matching_extractors
if ie.IE_NAME.lower() == extractor_id.lower())
except StopIteration:
raise ExtractorError(
('Found multiple matching extractors: %s' %
' '.join(ie.IE_NAME for ie in matching_extractors)),
'Found multiple matching extractors: %s' % ' '.join(ie.IE_NAME for ie in matching_extractors),
expected=True)
else:
extractor = matching_extractors[0]
num_str = mobj.group('num')
num = int(num_str) if num_str else 0
testcases = []
t = getattr(extractor, '_TEST', None)
if t:
testcases.append(t)
testcases.extend(getattr(extractor, '_TESTS', []))
testcases = tuple(extractor.get_testcases(True))
try:
tc = testcases[num]
tc = testcases[int(num or 0)]
except IndexError:
raise ExtractorError(
('Test case %d not found, got only %d tests' %
(num, len(testcases))),
expected=True)
f'Test case {num or 0} not found, got only {len(testcases)} tests', expected=True)
self.to_screen('Test URL: %s' % tc['url'])
return self.url_result(tc['url'], video_id=video_id)
self.to_screen(f'Test URL: {tc["url"]}')
return self.url_result(tc['url'])

View File

@ -31,6 +31,7 @@
NO_DEFAULT,
ExtractorError,
bug_reports_message,
classproperty,
clean_html,
datetime_from_str,
dict_get,
@ -5781,16 +5782,17 @@ def _real_extract(self, url):
class YoutubeFeedsInfoExtractor(InfoExtractor):
"""
Base class for feed extractors
Subclasses must define the _FEED_NAME property.
Subclasses must re-define the _FEED_NAME property.
"""
_LOGIN_REQUIRED = True
_FEED_NAME = 'feeds'
def _real_initialize(self):
YoutubeBaseInfoExtractor._check_login_required(self)
@property
@classproperty
def IE_NAME(self):
return 'youtube:%s' % self._FEED_NAME
return f'youtube:{self._FEED_NAME}'
def _real_extract(self, url):
return self.url_result(

View File

@ -5321,6 +5321,7 @@ def merge_headers(*dicts):
class classproperty:
def __init__(self, f):
functools.update_wrapper(self, f)
self.f = f
def __get__(self, _, cls):