]> jfr.im git - yt-dlp.git/blobdiff - devscripts/make_lazy_extractors.py
[cleanup] Add more ruff rules (#10149)
[yt-dlp.git] / devscripts / make_lazy_extractors.py
index e6de72b33a2ad4e5bf3b8a2153b16119bf106634..d74ea202f08b44b709a73c372d9c42d2004d01e9 100644 (file)
-from __future__ import unicode_literals, print_function
+#!/usr/bin/env python3
 
-from inspect import getsource
-import io
+# Allow direct execution
 import os
-from os.path import dirname as dirn
+import shutil
 import sys
 
-print('WARNING: Lazy loading extractors is an experimental feature that may not always work', file=sys.stderr)
-
-sys.path.insert(0, dirn(dirn((os.path.abspath(__file__)))))
-
-lazy_extractors_filename = sys.argv[1]
-if os.path.exists(lazy_extractors_filename):
-    os.remove(lazy_extractors_filename)
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 
-from youtube_dlc.extractor import _ALL_CLASSES
-from youtube_dlc.extractor.common import InfoExtractor, SearchInfoExtractor
 
-with open('devscripts/lazy_load_template.py', 'rt') as f:
-    module_template = f.read()
-
-module_contents = [
-    module_template + '\n' + getsource(InfoExtractor.suitable) + '\n',
-    'class LazyLoadSearchExtractor(LazyLoadExtractor):\n    pass\n']
+from inspect import getsource
 
-ie_template = '''
+from devscripts.utils import get_filename_args, read_file, write_file
+
+NO_ATTR = object()
+STATIC_CLASS_PROPERTIES = [
+    'IE_NAME', '_ENABLED', '_VALID_URL',  # Used for URL matching
+    '_WORKING', 'IE_DESC', '_NETRC_MACHINE', 'SEARCH_KEY',  # Used for --extractor-descriptions
+    'age_limit',  # Used for --age-limit (evaluated)
+    '_RETURN_TYPE',  # Accessed in CLI only with instance (evaluated)
+]
+CLASS_METHODS = [
+    'ie_key', 'suitable', '_match_valid_url',  # Used for URL matching
+    'working', 'get_temp_id', '_match_id',  # Accessed just before instance creation
+    'description',  # Used for --extractor-descriptions
+    'is_suitable',  # Used for --age-limit
+    'supports_login', 'is_single_video',  # Accessed in CLI only with instance
+]
+IE_TEMPLATE = '''
 class {name}({bases}):
-    _VALID_URL = {valid_url!r}
-    _module = '{module}'
+    _module = {module!r}
 '''
-
-make_valid_template = '''
-    @classmethod
-    def _make_valid_url(cls):
-        return {valid_url!r}
-'''
-
-
-def get_base_name(base):
-    if base is InfoExtractor:
-        return 'LazyLoadExtractor'
-    elif base is SearchInfoExtractor:
-        return 'LazyLoadSearchExtractor'
-    else:
-        return base.__name__
-
-
-def build_lazy_ie(ie, name):
-    valid_url = getattr(ie, '_VALID_URL', None)
-    s = ie_template.format(
-        name=name,
-        bases=', '.join(map(get_base_name, ie.__bases__)),
-        valid_url=valid_url,
-        module=ie.__module__)
-    if ie.suitable.__func__ is not InfoExtractor.suitable.__func__:
-        s += '\n' + getsource(ie.suitable)
-    if hasattr(ie, '_make_valid_url'):
-        # search extractors
-        s += make_valid_template.format(valid_url=ie._make_valid_url())
-    return s
-
-
-# find the correct sorting and add the required base classes so that sublcasses
-# can be correctly created
-classes = _ALL_CLASSES[:-1]
-ordered_cls = []
-while classes:
-    for c in classes[:]:
-        bases = set(c.__bases__) - set((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(
-    '_ALL_CLASSES = [{0}]'.format(', '.join(names)))
-
-module_src = '\n'.join(module_contents) + '\n'
-
-with io.open(lazy_extractors_filename, 'wt', encoding='utf-8') as f:
-    f.write(module_src)
+MODULE_TEMPLATE = read_file('devscripts/lazy_load_template.py')
+
+
+def main():
+    lazy_extractors_filename = get_filename_args(default_outfile='yt_dlp/extractor/lazy_extractors.py')
+    if os.path.exists(lazy_extractors_filename):
+        os.remove(lazy_extractors_filename)
+
+    _ALL_CLASSES = get_all_ies()  # Must be before import
+
+    import yt_dlp.plugins
+    from yt_dlp.extractor.common import InfoExtractor, SearchInfoExtractor
+
+    # Filter out plugins
+    _ALL_CLASSES = [cls for cls in _ALL_CLASSES if not cls.__module__.startswith(f'{yt_dlp.plugins.PACKAGE_NAME}.')]
+
+    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),
+    ))
+
+    write_file(lazy_extractors_filename, f'{module_src}\n')
+
+
+def get_all_ies():
+    PLUGINS_DIRNAME = 'ytdlp_plugins'
+    BLOCKED_DIRNAME = f'{PLUGINS_DIRNAME}_blocked'
+    if os.path.exists(PLUGINS_DIRNAME):
+        # os.rename cannot be used, e.g. in Docker. See https://github.com/yt-dlp/yt-dlp/pull/4958
+        shutil.move(PLUGINS_DIRNAME, BLOCKED_DIRNAME)
+    try:
+        from yt_dlp.extractor.extractors import _ALL_CLASSES
+    finally:
+        if os.path.exists(BLOCKED_DIRNAME):
+            shutil.move(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 sorted(bases, key=lambda x: x.__name__):
+                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)
+    return s + '\n'.join(extra_ie_code(ie, attr_base))
+
+
+if __name__ == '__main__':
+    main()