yt_dlp/plugins.py

   1 import contextlib
   2 import importlib
   3 import importlib.abc
   4 import importlib.machinery
   5 import importlib.util
   6 import inspect
   7 import itertools
   8 import pkgutil
   9 import sys
  10 import traceback
  11 import zipimport
  12 from pathlib import Path
  13 from zipfile import ZipFile
  14
  15 from .compat import functools  # isort: split
  16 from .utils import (
  17     get_executable_path,
  18     get_system_config_dirs,
  19     get_user_config_dirs,
  20     orderedSet,
  21     write_string,
  22 )
  23
  24 PACKAGE_NAME = 'yt_dlp_plugins'
  25 COMPAT_PACKAGE_NAME = 'ytdlp_plugins'
  26
  27
  28 class PluginLoader(importlib.abc.Loader):
  29     """Dummy loader for virtual namespace packages"""
  30
  31     def exec_module(self, module):
  32         return None
  33
  34
  35 @functools.cache
  36 def dirs_in_zip(archive):
  37     try:
  38         with ZipFile(archive) as zip_:
  39             return set(itertools.chain.from_iterable(
  40                 Path(file).parents for file in zip_.namelist()))
  41     except FileNotFoundError:
  42         pass
  43     except Exception as e:
  44         write_string(f'WARNING: Could not read zip file {archive}: {e}\n')
  45     return set()
  46
  47
  48 class PluginFinder(importlib.abc.MetaPathFinder):
  49     """
  50     This class provides one or multiple namespace packages.
  51     It searches in sys.path and yt-dlp config folders for
  52     the existing subdirectories from which the modules can be imported
  53     """
  54
  55     def __init__(self, *packages):
  56         self._zip_content_cache = {}
  57         self.packages = set(itertools.chain.from_iterable(
  58             itertools.accumulate(name.split('.'), lambda a, b: '.'.join((a, b)))
  59             for name in packages))
  60
  61     def search_locations(self, fullname):
  62         candidate_locations = []
  63
  64         def _get_package_paths(*root_paths, containing_folder='plugins'):
  65             for config_dir in orderedSet(map(Path, root_paths), lazy=True):
  66                 with contextlib.suppress(OSError):
  67                     yield from (config_dir / containing_folder).iterdir()
  68
  69         # Load from yt-dlp config folders
  70         candidate_locations.extend(_get_package_paths(
  71             *get_user_config_dirs('yt-dlp'),
  72             *get_system_config_dirs('yt-dlp'),
  73             containing_folder='plugins'))
  74
  75         # Load from yt-dlp-plugins folders
  76         candidate_locations.extend(_get_package_paths(
  77             get_executable_path(),
  78             *get_user_config_dirs(''),
  79             *get_system_config_dirs(''),
  80             containing_folder='yt-dlp-plugins'))
  81
  82         candidate_locations.extend(map(Path, sys.path))  # PYTHONPATH
  83         with contextlib.suppress(ValueError):  # Added when running __main__.py directly
  84             candidate_locations.remove(Path(__file__).parent)
  85
  86         parts = Path(*fullname.split('.'))
  87         for path in orderedSet(candidate_locations, lazy=True):
  88             candidate = path / parts
  89             if candidate.is_dir():
  90                 yield candidate
  91             elif path.suffix in ('.zip', '.egg', '.whl') and path.is_file():
  92                 if parts in dirs_in_zip(path):
  93                     yield candidate
  94
  95     def find_spec(self, fullname, path=None, target=None):
  96         if fullname not in self.packages:
  97             return None
  98
  99         search_locations = list(map(str, self.search_locations(fullname)))
 100         if not search_locations:
 101             return None
 102
 103         spec = importlib.machinery.ModuleSpec(fullname, PluginLoader(), is_package=True)
 104         spec.submodule_search_locations = search_locations
 105         return spec
 106
 107     def invalidate_caches(self):
 108         dirs_in_zip.cache_clear()
 109         for package in self.packages:
 110             if package in sys.modules:
 111                 del sys.modules[package]
 112
 113
 114 def directories():
 115     spec = importlib.util.find_spec(PACKAGE_NAME)
 116     return spec.submodule_search_locations if spec else []
 117
 118
 119 def iter_modules(subpackage):
 120     fullname = f'{PACKAGE_NAME}.{subpackage}'
 121     with contextlib.suppress(ModuleNotFoundError):
 122         pkg = importlib.import_module(fullname)
 123         yield from pkgutil.iter_modules(path=pkg.__path__, prefix=f'{fullname}.')
 124
 125
 126 def load_module(module, module_name, suffix):
 127     return inspect.getmembers(module, lambda obj: (
 128         inspect.isclass(obj)
 129         and obj.__name__.endswith(suffix)
 130         and obj.__module__.startswith(module_name)
 131         and not obj.__name__.startswith('_')
 132         and obj.__name__ in getattr(module, '__all__', [obj.__name__])))
 133
 134
 135 def load_plugins(name, suffix):
 136     classes = {}
 137
 138     for finder, module_name, _ in iter_modules(name):
 139         if any(x.startswith('_') for x in module_name.split('.')):
 140             continue
 141         try:
 142             if sys.version_info < (3, 10) and isinstance(finder, zipimport.zipimporter):
 143                 # zipimporter.load_module() is deprecated in 3.10 and removed in 3.12
 144                 # The exec_module branch below is the replacement for >= 3.10
 145                 # See: https://docs.python.org/3/library/zipimport.html#zipimport.zipimporter.exec_module
 146                 module = finder.load_module(module_name)
 147             else:
 148                 spec = finder.find_spec(module_name)
 149                 module = importlib.util.module_from_spec(spec)
 150                 sys.modules[module_name] = module
 151                 spec.loader.exec_module(module)
 152         except Exception:
 153             write_string(f'Error while importing module {module_name!r}\n{traceback.format_exc(limit=-1)}')
 154             continue
 155         classes.update(load_module(module, module_name, suffix))
 156
 157     # Compat: old plugin system using __init__.py
 158     # Note: plugins imported this way do not show up in directories()
 159     # nor are considered part of the yt_dlp_plugins namespace package
 160     with contextlib.suppress(FileNotFoundError):
 161         spec = importlib.util.spec_from_file_location(
 162             name, Path(get_executable_path(), COMPAT_PACKAGE_NAME, name, '__init__.py'))
 163         plugins = importlib.util.module_from_spec(spec)
 164         sys.modules[spec.name] = plugins
 165         spec.loader.exec_module(plugins)
 166         classes.update(load_module(plugins, spec.name, suffix))
 167
 168     return classes
 169
 170
 171 sys.meta_path.insert(0, PluginFinder(f'{PACKAGE_NAME}.extractor', f'{PACKAGE_NAME}.postprocessor'))
 172
 173 __all__ = ['directories', 'load_plugins', 'PACKAGE_NAME', 'COMPAT_PACKAGE_NAME']