yt_dlp/cache.py

   1 import contextlib
   2 import json
   3 import os
   4 import re
   5 import shutil
   6 import traceback
   7 import urllib.parse
   8
   9 from .utils import expand_path, traverse_obj, version_tuple, write_json_file
  10 from .version import __version__
  11
  12
  13 class Cache:
  14     def __init__(self, ydl):
  15         self._ydl = ydl
  16
  17     def _get_root_dir(self):
  18         res = self._ydl.params.get('cachedir')
  19         if res is None:
  20             cache_root = os.getenv('XDG_CACHE_HOME', '~/.cache')
  21             res = os.path.join(cache_root, 'yt-dlp')
  22         return expand_path(res)
  23
  24     def _get_cache_fn(self, section, key, dtype):
  25         assert re.match(r'^[\w.-]+$', section), f'invalid section {section!r}'
  26         key = urllib.parse.quote(key, safe='').replace('%', ',')  # encode non-ascii characters
  27         return os.path.join(self._get_root_dir(), section, f'{key}.{dtype}')
  28
  29     @property
  30     def enabled(self):
  31         return self._ydl.params.get('cachedir') is not False
  32
  33     def store(self, section, key, data, dtype='json'):
  34         assert dtype in ('json',)
  35
  36         if not self.enabled:
  37             return
  38
  39         fn = self._get_cache_fn(section, key, dtype)
  40         try:
  41             os.makedirs(os.path.dirname(fn), exist_ok=True)
  42             self._ydl.write_debug(f'Saving {section}.{key} to cache')
  43             write_json_file({'yt-dlp_version': __version__, 'data': data}, fn)
  44         except Exception:
  45             tb = traceback.format_exc()
  46             self._ydl.report_warning(f'Writing cache to {fn!r} failed: {tb}')
  47
  48     def _validate(self, data, min_ver):
  49         version = traverse_obj(data, 'yt-dlp_version')
  50         if not version:  # Backward compatibility
  51             data, version = {'data': data}, '2022.08.19'
  52         if not min_ver or version_tuple(version) >= version_tuple(min_ver):
  53             return data['data']
  54         self._ydl.write_debug(f'Discarding old cache from version {version} (needs {min_ver})')
  55
  56     def load(self, section, key, dtype='json', default=None, *, min_ver=None):
  57         assert dtype in ('json',)
  58
  59         if not self.enabled:
  60             return default
  61
  62         cache_fn = self._get_cache_fn(section, key, dtype)
  63         with contextlib.suppress(OSError):
  64             try:
  65                 with open(cache_fn, encoding='utf-8') as cachef:
  66                     self._ydl.write_debug(f'Loading {section}.{key} from cache')
  67                     return self._validate(json.load(cachef), min_ver)
  68             except (ValueError, KeyError):
  69                 try:
  70                     file_size = os.path.getsize(cache_fn)
  71                 except OSError as oe:
  72                     file_size = str(oe)
  73                 self._ydl.report_warning(f'Cache retrieval from {cache_fn} failed ({file_size})')
  74
  75         return default
  76
  77     def remove(self):
  78         if not self.enabled:
  79             self._ydl.to_screen('Cache is disabled (Did you combine --no-cache-dir and --rm-cache-dir?)')
  80             return
  81
  82         cachedir = self._get_root_dir()
  83         if not any((term in cachedir) for term in ('cache', 'tmp')):
  84             raise Exception(f'Not removing directory {cachedir} - this does not look like a cache dir')
  85
  86         self._ydl.to_screen(
  87             f'Removing cache dir {cachedir} .', skip_eol=True)
  88         if os.path.exists(cachedir):
  89             self._ydl.to_screen('.', skip_eol=True)
  90             shutil.rmtree(cachedir)
  91         self._ydl.to_screen('.')