Coverage for an_website / patches / __init__.py: 89.412%

1# This program is free software: you can redistribute it and/or modify 

2# it under the terms of the GNU Affero General Public License as 

3# published by the Free Software Foundation, either version 3 of the 

4# License, or (at your option) any later version. 

5# 

6# This program is distributed in the hope that it will be useful, 

7# but WITHOUT ANY WARRANTY; without even the implied warranty of 

8# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 

9# GNU Affero General Public License for more details. 

10# 

11# You should have received a copy of the GNU Affero General Public License 

12# along with this program. If not, see <https://www.gnu.org/licenses/>. 

13# pylint: disable=protected-access 

14 

15"""Patches that improve everything.""" 

16 

17from __future__ import annotations 

18 

19import asyncio 

20import http.client 

21import json as stdlib_json # pylint: disable=preferred-module 

22import logging 

23import os 

24import sys 

25from collections.abc import Callable 

26from configparser import RawConfigParser 

27from contextlib import suppress 

28from importlib import import_module 

29from pathlib import Path 

30from threading import Thread 

31from types import MethodType 

32from typing import Any 

33from urllib.parse import urlsplit 

34from warnings import catch_warnings, simplefilter 

35 

36import certifi 

37import defusedxml # type: ignore[import-untyped] 

38import jsonpickle # type: ignore[import-untyped] 

39import orjson 

40import pycurl 

41import tornado.httputil 

42import yaml 

43from emoji import EMOJI_DATA 

44from pillow_jxl import JpegXLImagePlugin # noqa: F401 

45from setproctitle import setthreadtitle 

46from tornado.httpclient import AsyncHTTPClient, HTTPRequest 

47from tornado.httputil import HTTPFile, HTTPHeaders, HTTPServerRequest 

48from tornado.log import gen_log 

49from tornado.web import GZipContentEncoding, RedirectHandler, RequestHandler 

50 

51from .. import CA_BUNDLE_PATH, MEDIA_TYPES 

52from . import braille, json # noqa: F401 # pylint: disable=reimported 

53 

54 

55def apply() -> None: 

56 """Improve.""" 

57 patch_asyncio() 

58 patch_certifi() 

59 patch_configparser() 

60 patch_emoji() 

61 patch_http() 

62 patch_json() 

63 patch_jsonpickle() 

64 patch_threading() 

65 patch_xml() 

66 

67 patch_tornado_418() 

68 patch_tornado_arguments() 

69 patch_tornado_gzip() 

70 patch_tornado_httpclient() 

71 patch_tornado_logs() 

72 patch_tornado_redirect() 

73 

74 

75def patch_asyncio() -> None: 

76 """Make stuff faster.""" 

77 if os.environ.get("DISABLE_UVLOOP") not in { 

78 "y", "yes", "t", "true", "on", "1" # fmt: skip 

79 }: 

80 with catch_warnings(): 

81 simplefilter("ignore", DeprecationWarning) 

82 with suppress(ModuleNotFoundError): 

83 asyncio.set_event_loop_policy( 

84 import_module("uvloop").EventLoopPolicy() 

85 ) 

86 

87 

88def patch_certifi() -> None: 

89 """Make everything use our CA bundle.""" 

90 certifi.where = lambda: CA_BUNDLE_PATH 

91 certifi.contents = lambda: Path(certifi.where()).read_text("ASCII") 

92 

93 

94def patch_configparser() -> None: 

95 """Make configparser funky.""" 

96 RawConfigParser.BOOLEAN_STATES.update( # type: ignore[attr-defined] 

97 { 

98 "sure": True, 

99 "nope": False, 

100 "accept": True, 

101 "reject": False, 

102 "enabled": True, 

103 "disabled": False, 

104 } 

105 ) 

106 

107 

108def patch_emoji() -> None: 

109 """Add cool new emoji.""" 

110 EMOJI_DATA["🐱\u200D💻"] = { 

111 "de": ":hacker_katze:", 

112 "en": ":hacker_cat:", 

113 "status": 2, 

114 "E": 1, 

115 } 

116 for de_name, en_name, rect in ( 

117 ("rot", "red", "🟥"), 

118 ("blau", "blue", "🟦"), 

119 ("orang", "orange", "🟧"), 

120 ("gelb", "yellow", "🟨"), 

121 ("grün", "green", "🟩"), 

122 ("lilan", "purple", "🟪"), 

123 ("braun", "brown", "🟫"), 

124 ): 

125 EMOJI_DATA[f"🫙\u200D{rect}"] = { 

126 "de": f":{de_name}es_glas:", 

127 "en": f":{en_name}_jar:", 

128 "status": 2, 

129 "E": 14, 

130 } 

131 EMOJI_DATA[f"🏳\uFE0F\u200D{rect}"] = { 

132 "de": f":{de_name}e_flagge:", 

133 "en": f":{en_name}_flag:", 

134 "status": 2, 

135 "E": 11, 

136 } 

137 EMOJI_DATA[f"\u2691\uFE0F\u200D{rect}"] = { 

138 "de": f":tief{de_name}e_flagge:", 

139 "en": f":deep_{en_name}_flag:", 

140 "status": 2, 

141 "E": 11, 

142 } 

143 

144 

145def patch_http() -> None: 

146 """Add response code 420.""" 

147 http.client.responses[420] = "Enhance Your Calm" 

148 

149 

150def patch_json() -> None: 

151 """Replace json with orjson.""" 

152 if getattr(stdlib_json, "_omegajson", False) or sys.version_info < (3, 12): 

153 return 

154 stdlib_json.dumps = json.dumps 

155 stdlib_json.dump = json.dump # type: ignore[assignment] 

156 stdlib_json.loads = json.loads # type: ignore[assignment] 

157 stdlib_json.load = json.load 

158 

159 

160def patch_jsonpickle() -> None: 

161 """Make jsonpickle return bytes.""" 

162 jsonpickle.load_backend("orjson") 

163 jsonpickle.set_preferred_backend("orjson") 

164 jsonpickle.enable_fallthrough(False) 

165 

166 

167def patch_threading() -> None: 

168 """Set thread names.""" 

169 _bootstrap = Thread._bootstrap # type: ignore[attr-defined] 

170 

171 def bootstrap(self: Thread) -> None: 

172 with suppress(Exception): 

173 setthreadtitle(self.name) 

174 _bootstrap(self) 

175 

176 Thread._bootstrap = bootstrap # type: ignore[attr-defined] 

177 

178 

179def patch_tornado_418() -> None: 

180 """Add support for RFC 7168.""" 

181 RequestHandler.SUPPORTED_METHODS += ( 

182 "PROPFIND", 

183 "BREW", 

184 "WHEN", 

185 ) 

186 _ = RequestHandler._unimplemented_method 

187 RequestHandler.propfind = _ # type: ignore[attr-defined] 

188 RequestHandler.brew = _ # type: ignore[attr-defined] 

189 RequestHandler.when = _ # type: ignore[attr-defined] 

190 

191 

192def patch_tornado_arguments() -> None: # noqa: C901 

193 """Improve argument parsing.""" 

194 # pylint: disable=too-complex 

195 

196 def ensure_bytes(value: Any) -> bytes: 

197 """Return the value as bytes.""" 

198 if isinstance(value, bool): 

199 return b"true" if value else b"false" 

200 if isinstance(value, bytes): 

201 return value 

202 return str(value).encode("UTF-8") 

203 

204 def parse_body_arguments( 

205 content_type: str, 

206 body: bytes, 

207 arguments: dict[str, list[bytes]], 

208 files: dict[str, list[HTTPFile]], 

209 headers: None | HTTPHeaders = None, 

210 *, 

211 _: Callable[..., None] = tornado.httputil.parse_body_arguments, 

212 ) -> None: 

213 # pylint: disable=too-many-branches 

214 if content_type.startswith("application/json"): 

215 if headers and "Content-Encoding" in headers: 

216 gen_log.warning( 

217 "Unsupported Content-Encoding: %s", 

218 headers["Content-Encoding"], 

219 ) 

220 return 

221 try: 

222 spam = orjson.loads(body) 

223 except Exception as exc: # pylint: disable=broad-except 

224 gen_log.warning("Invalid JSON body: %s", exc) 

225 else: 

226 if not isinstance(spam, dict): 

227 return 

228 for key, value in spam.items(): 

229 if value is not None: 

230 arguments.setdefault(key, []).append( 

231 ensure_bytes(value) 

232 ) 

233 elif content_type.startswith("application/yaml"): 

234 if headers and "Content-Encoding" in headers: 

235 gen_log.warning( 

236 "Unsupported Content-Encoding: %s", 

237 headers["Content-Encoding"], 

238 ) 

239 return 

240 try: 

241 spam = yaml.safe_load(body) 

242 except Exception as exc: # pylint: disable=broad-except 

243 gen_log.warning("Invalid YAML body: %s", exc) 

244 else: 

245 if not isinstance(spam, dict): 

246 return 

247 for key, value in spam.items(): 

248 if value is not None: 

249 arguments.setdefault(key, []).append( 

250 ensure_bytes(value) 

251 ) 

252 else: 

253 _(content_type, body, arguments, files, headers) 

254 

255 parse_body_arguments.__doc__ = tornado.httputil.parse_body_arguments.__doc__ 

256 

257 tornado.httputil.parse_body_arguments = parse_body_arguments 

258 

259 

260def patch_tornado_gzip() -> None: 

261 """Use gzip for more content types.""" 

262 GZipContentEncoding.CONTENT_TYPES = { 

263 type for type, data in MEDIA_TYPES.items() if data.get("compressible") 

264 } 

265 

266 

267def patch_tornado_httpclient() -> None: # fmt: off 

268 """Make requests quick.""" 

269 BACON = 0x75800 # noqa: N806 # pylint: disable=invalid-name 

270 EGGS = 1 << 25 # noqa: N806 # pylint: disable=invalid-name 

271 

272 AsyncHTTPClient.configure("tornado.curl_httpclient.CurlAsyncHTTPClient") 

273 

274 def prepare_curl_callback(self: HTTPRequest, curl: pycurl.Curl) -> None: 

275 # pylint: disable=c-extension-no-member, useless-suppression 

276 if urlsplit(self.url).scheme == "https": # noqa: SIM102 

277 if (ver := pycurl.version_info())[2] >= BACON and ver[4] & EGGS: 

278 curl.setopt(pycurl.HTTP_VERSION, pycurl.CURL_HTTP_VERSION_3) 

279 

280 original_request_init = HTTPRequest.__init__ 

281 

282 def request_init(self: HTTPRequest, *args: Any, **kwargs: Any) -> None: 

283 if len(args) < 18: # there are too many positional arguments here 

284 prepare_curl_method = MethodType(prepare_curl_callback, self) 

285 kwargs.setdefault("prepare_curl_callback", prepare_curl_method) 

286 original_request_init(self, *args, **kwargs) 

287 

288 request_init.__doc__ = HTTPRequest.__init__.__doc__ 

289 

290 HTTPRequest.__init__ = request_init # type: ignore[method-assign] 

291 

292 

293def patch_tornado_logs() -> None: 

294 """Anonymize Tornado logs.""" 

295 # pylint: disable=import-outside-toplevel 

296 from ..utils.utils import SUS_PATHS, anonymize_ip 

297 

298 RequestHandler._request_summary = ( # type: ignore[method-assign] 

299 lambda self: "%s %s (%s)" # pylint: disable=consider-using-f-string 

300 % ( 

301 self.request.method, 

302 self.request.uri, 

303 ( 

304 self.request.remote_ip 

305 if self.request.path == "/robots.txt" 

306 or self.request.path.lower() in SUS_PATHS 

307 else anonymize_ip(self.request.remote_ip, ignore_invalid=True) 

308 ), 

309 ) 

310 ) 

311 

312 HTTPServerRequest.__repr__ = ( # type: ignore[method-assign] 

313 lambda self: "%s(%s)" # pylint: disable=consider-using-f-string 

314 % ( 

315 self.__class__.__name__, 

316 ", ".join( 

317 [ 

318 "%s=%r" # pylint: disable=consider-using-f-string 

319 % ( 

320 n, 

321 getattr(self, n), 

322 ) 

323 for n in ("protocol", "host", "method", "uri", "version") 

324 ] 

325 ), 

326 ) 

327 ) 

328 

329 

330def patch_tornado_redirect() -> None: 

331 """Use modern redirect codes and support HEAD requests.""" 

332 

333 def redirect( 

334 self: RequestHandler, 

335 url: str, 

336 permanent: bool = False, 

337 status: None | int = None, 

338 ) -> None: 

339 if url == self.request.full_url(): 

340 logging.getLogger( 

341 f"{self.__class__.__module__}.{self.__class__.__qualname__}" 

342 ).critical("Infinite redirect to %r detected", url) 

343 if self._headers_written: 

344 # pylint: disable=broad-exception-raised 

345 raise Exception("Cannot redirect after headers have been written") 

346 if status is None: 

347 status = 308 if permanent else 307 

348 else: 

349 assert isinstance(status, int) and 300 <= status <= 399 # type: ignore[redundant-expr] # noqa: B950 

350 self.set_status(status) 

351 self.set_header("Location", url) 

352 self.finish() # type: ignore[unused-awaitable] 

353 

354 if RequestHandler.redirect.__doc__: 

355 # fmt: off 

356 redirect.__doc__ = ( 

357 RequestHandler.redirect.__doc__ 

358 .replace("301", "308") 

359 .replace("302", "307") 

360 ) 

361 # fmt: on 

362 

363 RequestHandler.redirect = redirect # type: ignore[method-assign] 

364 

365 RedirectHandler.head = RedirectHandler.get 

366 

367 

368def patch_xml() -> None: 

369 """Make XML safer.""" 

370 defusedxml.defuse_stdlib() 

371 defusedxml.xmlrpc.monkey_patch()