annotate libervia/backend/memory/cache.py @ 4177:0f1a4ffcd419

doc (cli/blog): Document new `--alt-link` option
author Goffi <goffi@goffi.org>
date Tue, 05 Dec 2023 13:14:55 +0100
parents 4b842c1fb686
children 5f2d496c633f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
1 #!/usr/bin/env python3
3137
559a625a236b fixed shebangs
Goffi <goffi@goffi.org>
parents: 3136
diff changeset
2
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
3
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
4 # SAT: a jabber client
3479
be6d91572633 date update
Goffi <goffi@goffi.org>
parents: 3210
diff changeset
5 # Copyright (C) 2009-2021 Jérôme Poisson (goffi@goffi.org)
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
6
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
7 # This program is free software: you can redistribute it and/or modify
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
8 # it under the terms of the GNU Affero General Public License as published by
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
9 # the Free Software Foundation, either version 3 of the License, or
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
10 # (at your option) any later version.
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
11
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
12 # This program is distributed in the hope that it will be useful,
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
13 # but WITHOUT ANY WARRANTY; without even the implied warranty of
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
14 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
15 # GNU Affero General Public License for more details.
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
16
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
17 # You should have received a copy of the GNU Affero General Public License
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
18 # along with this program. If not, see <http://www.gnu.org/licenses/>.
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
19
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
20 from io import BufferedIOBase
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
21 import mimetypes
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
22 from pathlib import Path
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
23 import pickle as pickle
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
24 import time
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
25 from typing import Any, Dict, Optional
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
26
4071
4b842c1fb686 refactoring: renamed `sat` package to `libervia.backend`
Goffi <goffi@goffi.org>
parents: 4037
diff changeset
27 from libervia.backend.core import exceptions
4b842c1fb686 refactoring: renamed `sat` package to `libervia.backend`
Goffi <goffi@goffi.org>
parents: 4037
diff changeset
28 from libervia.backend.core.constants import Const as C
4b842c1fb686 refactoring: renamed `sat` package to `libervia.backend`
Goffi <goffi@goffi.org>
parents: 4037
diff changeset
29 from libervia.backend.core.i18n import _
4b842c1fb686 refactoring: renamed `sat` package to `libervia.backend`
Goffi <goffi@goffi.org>
parents: 4037
diff changeset
30 from libervia.backend.core.log import getLogger
4b842c1fb686 refactoring: renamed `sat` package to `libervia.backend`
Goffi <goffi@goffi.org>
parents: 4037
diff changeset
31 from libervia.backend.tools.common import regex
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
32
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
33
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
34 log = getLogger(__name__)
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
35
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
36 DEFAULT_EXT = ".raw"
2509
d485e9416493 core (memory/cache): common cache:
Goffi <goffi@goffi.org>
parents: 2506
diff changeset
37
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
38
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
39 class Cache(object):
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
40 """generic file caching"""
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
41
2509
d485e9416493 core (memory/cache): common cache:
Goffi <goffi@goffi.org>
parents: 2506
diff changeset
42 def __init__(self, host, profile):
d485e9416493 core (memory/cache): common cache:
Goffi <goffi@goffi.org>
parents: 2506
diff changeset
43 """
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
44 @param profile(unicode, None): name of the profile to set the cache for
2509
d485e9416493 core (memory/cache): common cache:
Goffi <goffi@goffi.org>
parents: 2506
diff changeset
45 if None, the cache will be common for all profiles
d485e9416493 core (memory/cache): common cache:
Goffi <goffi@goffi.org>
parents: 2506
diff changeset
46 """
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
47 self.profile = profile
4037
524856bd7b19 massive refactoring to switch from camelCase to snake_case:
Goffi <goffi@goffi.org>
parents: 3818
diff changeset
48 path_elts = [host.memory.config_get("", "local_dir"), C.CACHE_DIR]
2509
d485e9416493 core (memory/cache): common cache:
Goffi <goffi@goffi.org>
parents: 2506
diff changeset
49 if profile:
4037
524856bd7b19 massive refactoring to switch from camelCase to snake_case:
Goffi <goffi@goffi.org>
parents: 3818
diff changeset
50 path_elts.extend(["profiles", regex.path_escape(profile)])
2509
d485e9416493 core (memory/cache): common cache:
Goffi <goffi@goffi.org>
parents: 2506
diff changeset
51 else:
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
52 path_elts.append("common")
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
53 self.cache_dir = Path(*path_elts)
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
54
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
55 self.cache_dir.mkdir(0o700, parents=True, exist_ok=True)
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
56 self.purge()
2509
d485e9416493 core (memory/cache): common cache:
Goffi <goffi@goffi.org>
parents: 2506
diff changeset
57
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
58 def purge(self):
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
59 # remove expired files from cache
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
60 # TODO: this should not be called only on startup, but at regular interval
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
61 # (e.g. once a day)
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
62 purged = set()
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
63 # we sort files to have metadata files first
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
64 for cache_file in sorted(self.cache_dir.iterdir()):
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
65 if cache_file in purged:
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
66 continue
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
67 try:
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
68 with cache_file.open('rb') as f:
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
69 cache_data = pickle.load(f)
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
70 except IOError:
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
71 log.warning(
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
72 _("Can't read metadata file at {path}")
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
73 .format(path=cache_file))
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
74 continue
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
75 except (pickle.UnpicklingError, EOFError):
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
76 log.debug(f"File at {cache_file} is not a metadata file")
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
77 continue
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
78 try:
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
79 eol = cache_data['eol']
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
80 filename = cache_data['filename']
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
81 except KeyError:
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
82 log.warning(
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
83 _("Invalid cache metadata at {path}")
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
84 .format(path=cache_file))
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
85 continue
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
86
3209
f14eb24328d0 core (memory/cache): purge cache metadata when the referenced file doesn't exist
Goffi <goffi@goffi.org>
parents: 3198
diff changeset
87 filepath = self.getPath(filename)
f14eb24328d0 core (memory/cache): purge cache metadata when the referenced file doesn't exist
Goffi <goffi@goffi.org>
parents: 3198
diff changeset
88
f14eb24328d0 core (memory/cache): purge cache metadata when the referenced file doesn't exist
Goffi <goffi@goffi.org>
parents: 3198
diff changeset
89 if not filepath.exists():
f14eb24328d0 core (memory/cache): purge cache metadata when the referenced file doesn't exist
Goffi <goffi@goffi.org>
parents: 3198
diff changeset
90 log.warning(_(
f14eb24328d0 core (memory/cache): purge cache metadata when the referenced file doesn't exist
Goffi <goffi@goffi.org>
parents: 3198
diff changeset
91 "cache {cache_file!r} references an inexisting file: {filepath!r}"
f14eb24328d0 core (memory/cache): purge cache metadata when the referenced file doesn't exist
Goffi <goffi@goffi.org>
parents: 3198
diff changeset
92 ).format(cache_file=str(cache_file), filepath=str(filepath)))
f14eb24328d0 core (memory/cache): purge cache metadata when the referenced file doesn't exist
Goffi <goffi@goffi.org>
parents: 3198
diff changeset
93 log.debug("purging cache with missing file")
f14eb24328d0 core (memory/cache): purge cache metadata when the referenced file doesn't exist
Goffi <goffi@goffi.org>
parents: 3198
diff changeset
94 cache_file.unlink()
f14eb24328d0 core (memory/cache): purge cache metadata when the referenced file doesn't exist
Goffi <goffi@goffi.org>
parents: 3198
diff changeset
95 elif eol < time.time():
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
96 log.debug(
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
97 "purging expired cache {filepath!r} (expired for {time}s)"
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
98 .format(filepath=str(filepath), time=int(time.time() - eol))
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
99 )
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
100 cache_file.unlink()
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
101 try:
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
102 filepath.unlink()
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
103 except FileNotFoundError:
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
104 log.warning(
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
105 _("following file is missing while purging cache: {path}")
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
106 .format(path=filepath)
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
107 )
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
108 purged.add(cache_file)
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
109 purged.add(filepath)
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
110
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
111 def getPath(self, filename: str) -> Path:
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
112 """return cached file URL
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
113
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
114 @param filename: cached file name (cache data or actual file)
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
115 @return: path to the cached file
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
116 """
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
117 if not filename or "/" in filename:
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
118 log.error(
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
119 "invalid char found in file name, hack attempt? name:{}".format(filename)
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
120 )
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
121 raise exceptions.DataError("Invalid char found")
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
122 return self.cache_dir / filename
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
123
4037
524856bd7b19 massive refactoring to switch from camelCase to snake_case:
Goffi <goffi@goffi.org>
parents: 3818
diff changeset
124 def get_metadata(self, uid: str, update_eol: bool = True) -> Optional[Dict[str, Any]]:
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
125 """Retrieve metadata for cached data
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
126
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
127 @param uid(unicode): unique identifier of file
3188
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
128 @param update_eol(bool): True if eol must extended
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
129 if True, max_age will be added to eol (only if it is not already expired)
2517
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
130 @return (dict, None): metadata with following keys:
4037
524856bd7b19 massive refactoring to switch from camelCase to snake_case:
Goffi <goffi@goffi.org>
parents: 3818
diff changeset
131 see [cache_data] for data details, an additional "path" key is the full path to
3188
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
132 cached file.
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
133 None if file is not in cache (or cache is invalid)
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
134 """
2517
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
135
2116
766dbbec56f2 core (memory/cache): geFilePath now return None when uid is empty
Goffi <goffi@goffi.org>
parents: 2109
diff changeset
136 uid = uid.strip()
766dbbec56f2 core (memory/cache): geFilePath now return None when uid is empty
Goffi <goffi@goffi.org>
parents: 2109
diff changeset
137 if not uid:
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
138 raise exceptions.InternalError("uid must not be empty")
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
139 cache_url = self.getPath(uid)
3188
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
140 if not cache_url.exists():
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
141 return None
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
142
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
143 try:
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
144 with cache_url.open("rb") as f:
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
145 cache_data = pickle.load(f)
3572
b3fa179417e7 core (memory/cache): don't crash on EOFError in getMetadata
Goffi <goffi@goffi.org>
parents: 3479
diff changeset
146 except (IOError, EOFError) as e:
3188
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
147 log.warning(f"can't read cache at {cache_url}: {e}")
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
148 return None
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
149 except pickle.UnpicklingError:
3188
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
150 log.warning(f"invalid cache found at {cache_url}")
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
151 return None
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
152
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
153 try:
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
154 eol = cache_data["eol"]
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
155 except KeyError:
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
156 log.warning("no End Of Life found for cached file {}".format(uid))
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
157 eol = 0
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
158 if eol < time.time():
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
159 log.debug(
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
160 "removing expired cache (expired for {}s)".format(time.time() - eol)
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
161 )
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
162 return None
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
163
3188
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
164 if update_eol:
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
165 try:
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
166 max_age = cache_data["max_age"]
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
167 except KeyError:
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
168 log.warning(f"no max_age found for cache at {cache_url}, using default")
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
169 max_age = cache_data["max_age"] = C.DEFAULT_MAX_AGE
3198
08151c103636 core (memory/cache): added some metadata:
Goffi <goffi@goffi.org>
parents: 3188
diff changeset
170 now = int(time.time())
08151c103636 core (memory/cache): added some metadata:
Goffi <goffi@goffi.org>
parents: 3188
diff changeset
171 cache_data["last_access"] = now
08151c103636 core (memory/cache): added some metadata:
Goffi <goffi@goffi.org>
parents: 3188
diff changeset
172 cache_data["eol"] = now + max_age
3188
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
173 with cache_url.open("wb") as f:
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
174 pickle.dump(cache_data, f, protocol=2)
a15773c6c273 memory(cache): extend EOL when a file metadata is retrieved
Goffi <goffi@goffi.org>
parents: 3185
diff changeset
175
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
176 cache_data["path"] = self.getPath(cache_data["filename"])
2517
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
177 return cache_data
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
178
4037
524856bd7b19 massive refactoring to switch from camelCase to snake_case:
Goffi <goffi@goffi.org>
parents: 3818
diff changeset
179 def get_file_path(self, uid: str) -> Path:
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
180 """Retrieve absolute path to file
2517
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
181
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
182 @param uid(unicode): unique identifier of file
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
183 @return (unicode, None): absolute path to cached file
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
184 None if file is not in cache (or cache is invalid)
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
185 """
4037
524856bd7b19 massive refactoring to switch from camelCase to snake_case:
Goffi <goffi@goffi.org>
parents: 3818
diff changeset
186 metadata = self.get_metadata(uid)
2517
cd7a53c31eb6 core (memory/cache): new getMetadata method to retrieve metadata without opening the file
Goffi <goffi@goffi.org>
parents: 2509
diff changeset
187 if metadata is not None:
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
188 return metadata["path"]
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
189
4037
524856bd7b19 massive refactoring to switch from camelCase to snake_case:
Goffi <goffi@goffi.org>
parents: 3818
diff changeset
190 def remove_from_cache(self, uid, metadata=None):
3210
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
191 """Remove data from cache
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
192
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
193 @param uid(unicode): unique identifier cache file
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
194 """
4037
524856bd7b19 massive refactoring to switch from camelCase to snake_case:
Goffi <goffi@goffi.org>
parents: 3818
diff changeset
195 cache_data = self.get_metadata(uid, update_eol=False)
3210
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
196 if cache_data is None:
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
197 log.debug(f"cache with uid {uid!r} has already expired or been removed")
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
198 return
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
199
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
200 try:
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
201 filename = cache_data['filename']
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
202 except KeyError:
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
203 log.warning(_("missing filename for cache {uid!r}") .format(uid=uid))
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
204 else:
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
205 filepath = self.getPath(filename)
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
206 try:
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
207 filepath.unlink()
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
208 except FileNotFoundError:
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
209 log.warning(
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
210 _("missing file referenced in cache {uid!r}: {filename}")
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
211 .format(uid=uid, filename=filename)
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
212 )
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
213
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
214 cache_file = self.getPath(uid)
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
215 cache_file.unlink()
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
216 log.debug(f"cache with uid {uid!r} has been removed")
fedec192a83f core (memory/cache): new removeFromCache method
Goffi <goffi@goffi.org>
parents: 3209
diff changeset
217
4037
524856bd7b19 massive refactoring to switch from camelCase to snake_case:
Goffi <goffi@goffi.org>
parents: 3818
diff changeset
218 def cache_data(
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
219 self,
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
220 source: str,
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
221 uid: str,
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
222 mime_type: Optional[str] = None,
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
223 max_age: Optional[int] = None,
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
224 original_filename: Optional[str] = None
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
225 ) -> BufferedIOBase:
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
226 """create cache metadata and file object to use for actual data
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
227
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
228 @param source: source of the cache (should be plugin's import_name)
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
229 @param uid: an identifier of the file which must be unique
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
230 @param mime_type: MIME type of the file to cache
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
231 it will be used notably to guess file extension
3198
08151c103636 core (memory/cache): added some metadata:
Goffi <goffi@goffi.org>
parents: 3188
diff changeset
232 It may be autogenerated if filename is specified
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
233 @param max_age: maximum age in seconds
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
234 the cache metadata will have an "eol" (end of life)
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
235 None to use default value
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
236 0 to ignore cache (file will be re-downloaded on each access)
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
237 @param original_filename: if not None, will be used to retrieve file extension and
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
238 guess
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
239 mime type, and stored in "original_filename"
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
240 @return: file object opened in write mode
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
241 you have to close it yourself (hint: use ``with`` statement)
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
242 """
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
243 if max_age is None:
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
244 max_age = C.DEFAULT_MAX_AGE
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
245 cache_data = {
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
246 "source": source,
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
247 # we also store max_age for updating eol
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
248 "max_age": max_age,
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
249 }
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
250 cache_url = self.getPath(uid)
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
251 if original_filename is not None:
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
252 cache_data["original_filename"] = original_filename
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
253 if mime_type is None:
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
254 # we have original_filename but not MIME type, we try to guess the later
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
255 mime_type = mimetypes.guess_type(original_filename, strict=False)[0]
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
256 if mime_type:
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
257 ext = mimetypes.guess_extension(mime_type, strict=False)
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
258 if ext is None:
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
259 log.warning(
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
260 "can't find extension for MIME type {}".format(mime_type)
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
261 )
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
262 ext = DEFAULT_EXT
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
263 elif ext == ".jpe":
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
264 ext = ".jpg"
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
265 else:
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
266 ext = DEFAULT_EXT
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
267 mime_type = None
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
268 filename = uid + ext
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
269 now = int(time.time())
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
270 cache_data.update({
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
271 "filename": filename,
3198
08151c103636 core (memory/cache): added some metadata:
Goffi <goffi@goffi.org>
parents: 3188
diff changeset
272 "creation": now,
08151c103636 core (memory/cache): added some metadata:
Goffi <goffi@goffi.org>
parents: 3188
diff changeset
273 "eol": now + max_age,
3028
ab2696e34d29 Python 3 port:
Goffi <goffi@goffi.org>
parents: 2771
diff changeset
274 "mime_type": mime_type,
3818
2863345c9bbb core (memory/cache): type hints + filename fix:
Goffi <goffi@goffi.org>
parents: 3572
diff changeset
275 })
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
276 file_path = self.getPath(filename)
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
277
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
278 with open(cache_url, "wb") as f:
2109
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
279 pickle.dump(cache_data, f, protocol=2)
85f3e12e984d core (memory/cache): file caching handling, first draft:
Goffi <goffi@goffi.org>
parents:
diff changeset
280
3185
554b3b632378 memory (cache): purge + pathlib:
Goffi <goffi@goffi.org>
parents: 3137
diff changeset
281 return file_path.open("wb")