You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 

1437 lines
49 KiB

  1. # Copyright 2018 New Vector Ltd
  2. # Copyright 2021 The Matrix.org Foundation C.I.C.
  3. #
  4. # Licensed under the Apache License, Version 2.0 (the "License");
  5. # you may not use this file except in compliance with the License.
  6. # You may obtain a copy of the License at
  7. #
  8. # http://www.apache.org/licenses/LICENSE-2.0
  9. #
  10. # Unless required by applicable law or agreed to in writing, software
  11. # distributed under the License is distributed on an "AS IS" BASIS,
  12. # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13. # See the License for the specific language governing permissions and
  14. # limitations under the License.
  15. import base64
  16. import json
  17. import os
  18. import re
  19. from typing import Any, Dict, Optional, Sequence, Tuple, Type
  20. from urllib.parse import quote, urlencode
  21. from twisted.internet._resolver import HostResolution
  22. from twisted.internet.address import IPv4Address, IPv6Address
  23. from twisted.internet.error import DNSLookupError
  24. from twisted.internet.interfaces import IAddress, IResolutionReceiver
  25. from twisted.test.proto_helpers import AccumulatingProtocol, MemoryReactor
  26. from twisted.web.resource import Resource
  27. from synapse.config.oembed import OEmbedEndpointConfig
  28. from synapse.media.url_previewer import IMAGE_CACHE_EXPIRY_MS
  29. from synapse.server import HomeServer
  30. from synapse.types import JsonDict
  31. from synapse.util import Clock
  32. from synapse.util.stringutils import parse_and_validate_mxc_uri
  33. from tests import unittest
  34. from tests.server import FakeTransport
  35. from tests.test_utils import SMALL_PNG
  36. try:
  37. import lxml
  38. except ImportError:
  39. lxml = None # type: ignore[assignment]
  40. class URLPreviewTests(unittest.HomeserverTestCase):
  41. if not lxml:
  42. skip = "url preview feature requires lxml"
  43. hijack_auth = True
  44. user_id = "@test:user"
  45. end_content = (
  46. b"<html><head>"
  47. b'<meta property="og:title" content="~matrix~" />'
  48. b'<meta property="og:description" content="hi" />'
  49. b"</head></html>"
  50. )
  51. def make_homeserver(self, reactor: MemoryReactor, clock: Clock) -> HomeServer:
  52. config = self.default_config()
  53. config["url_preview_enabled"] = True
  54. config["max_spider_size"] = 9999999
  55. config["url_preview_ip_range_blacklist"] = (
  56. "192.168.1.1",
  57. "1.0.0.0/8",
  58. "3fff:ffff:ffff:ffff:ffff:ffff:ffff:ffff",
  59. "2001:800::/21",
  60. )
  61. config["url_preview_ip_range_whitelist"] = ("1.1.1.1",)
  62. config["url_preview_accept_language"] = [
  63. "en-UK",
  64. "en-US;q=0.9",
  65. "fr;q=0.8",
  66. "*;q=0.7",
  67. ]
  68. self.storage_path = self.mktemp()
  69. self.media_store_path = self.mktemp()
  70. os.mkdir(self.storage_path)
  71. os.mkdir(self.media_store_path)
  72. config["media_store_path"] = self.media_store_path
  73. provider_config = {
  74. "module": "synapse.media.storage_provider.FileStorageProviderBackend",
  75. "store_local": True,
  76. "store_synchronous": False,
  77. "store_remote": True,
  78. "config": {"directory": self.storage_path},
  79. }
  80. config["media_storage_providers"] = [provider_config]
  81. hs = self.setup_test_homeserver(config=config)
  82. # After the hs is created, modify the parsed oEmbed config (to avoid
  83. # messing with files).
  84. #
  85. # Note that HTTP URLs are used to avoid having to deal with TLS in tests.
  86. hs.config.oembed.oembed_patterns = [
  87. OEmbedEndpointConfig(
  88. api_endpoint="http://publish.twitter.com/oembed",
  89. url_patterns=[
  90. re.compile(r"http://twitter\.com/.+/status/.+"),
  91. ],
  92. formats=None,
  93. ),
  94. OEmbedEndpointConfig(
  95. api_endpoint="http://www.hulu.com/api/oembed.{format}",
  96. url_patterns=[
  97. re.compile(r"http://www\.hulu\.com/watch/.+"),
  98. ],
  99. formats=["json"],
  100. ),
  101. ]
  102. return hs
  103. def prepare(self, reactor: MemoryReactor, clock: Clock, hs: HomeServer) -> None:
  104. self.media_repo = hs.get_media_repository()
  105. assert self.media_repo.url_previewer is not None
  106. self.url_previewer = self.media_repo.url_previewer
  107. self.lookups: Dict[str, Any] = {}
  108. class Resolver:
  109. def resolveHostName(
  110. _self,
  111. resolutionReceiver: IResolutionReceiver,
  112. hostName: str,
  113. portNumber: int = 0,
  114. addressTypes: Optional[Sequence[Type[IAddress]]] = None,
  115. transportSemantics: str = "TCP",
  116. ) -> IResolutionReceiver:
  117. resolution = HostResolution(hostName)
  118. resolutionReceiver.resolutionBegan(resolution)
  119. if hostName not in self.lookups:
  120. raise DNSLookupError("OH NO")
  121. for i in self.lookups[hostName]:
  122. resolutionReceiver.addressResolved(i[0]("TCP", i[1], portNumber))
  123. resolutionReceiver.resolutionComplete()
  124. return resolutionReceiver
  125. self.reactor.nameResolver = Resolver() # type: ignore[assignment]
  126. def create_resource_dict(self) -> Dict[str, Resource]:
  127. """Create a resource tree for the test server
  128. A resource tree is a mapping from path to twisted.web.resource.
  129. The default implementation creates a JsonResource and calls each function in
  130. `servlets` to register servlets against it.
  131. """
  132. return {"/_matrix/media": self.hs.get_media_repository_resource()}
  133. def _assert_small_png(self, json_body: JsonDict) -> None:
  134. """Assert properties from the SMALL_PNG test image."""
  135. self.assertTrue(json_body["og:image"].startswith("mxc://"))
  136. self.assertEqual(json_body["og:image:height"], 1)
  137. self.assertEqual(json_body["og:image:width"], 1)
  138. self.assertEqual(json_body["og:image:type"], "image/png")
  139. self.assertEqual(json_body["matrix:image:size"], 67)
  140. def test_cache_returns_correct_type(self) -> None:
  141. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  142. channel = self.make_request(
  143. "GET",
  144. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  145. shorthand=False,
  146. await_result=False,
  147. )
  148. self.pump()
  149. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  150. server = AccumulatingProtocol()
  151. server.makeConnection(FakeTransport(client, self.reactor))
  152. client.makeConnection(FakeTransport(server, self.reactor))
  153. client.dataReceived(
  154. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\nContent-Type: text/html\r\n\r\n"
  155. % (len(self.end_content),)
  156. + self.end_content
  157. )
  158. self.pump()
  159. self.assertEqual(channel.code, 200)
  160. self.assertEqual(
  161. channel.json_body, {"og:title": "~matrix~", "og:description": "hi"}
  162. )
  163. # Check the cache returns the correct response
  164. channel = self.make_request(
  165. "GET",
  166. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  167. shorthand=False,
  168. )
  169. # Check the cache response has the same content
  170. self.assertEqual(channel.code, 200)
  171. self.assertEqual(
  172. channel.json_body, {"og:title": "~matrix~", "og:description": "hi"}
  173. )
  174. # Clear the in-memory cache
  175. self.assertIn("http://matrix.org", self.url_previewer._cache)
  176. self.url_previewer._cache.pop("http://matrix.org")
  177. self.assertNotIn("http://matrix.org", self.url_previewer._cache)
  178. # Check the database cache returns the correct response
  179. channel = self.make_request(
  180. "GET",
  181. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  182. shorthand=False,
  183. )
  184. # Check the cache response has the same content
  185. self.assertEqual(channel.code, 200)
  186. self.assertEqual(
  187. channel.json_body, {"og:title": "~matrix~", "og:description": "hi"}
  188. )
  189. def test_non_ascii_preview_httpequiv(self) -> None:
  190. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  191. end_content = (
  192. b"<html><head>"
  193. b'<meta http-equiv="Content-Type" content="text/html; charset=windows-1251"/>'
  194. b'<meta property="og:title" content="\xe4\xea\xe0" />'
  195. b'<meta property="og:description" content="hi" />'
  196. b"</head></html>"
  197. )
  198. channel = self.make_request(
  199. "GET",
  200. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  201. shorthand=False,
  202. await_result=False,
  203. )
  204. self.pump()
  205. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  206. server = AccumulatingProtocol()
  207. server.makeConnection(FakeTransport(client, self.reactor))
  208. client.makeConnection(FakeTransport(server, self.reactor))
  209. client.dataReceived(
  210. (
  211. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  212. b'Content-Type: text/html; charset="utf8"\r\n\r\n'
  213. )
  214. % (len(end_content),)
  215. + end_content
  216. )
  217. self.pump()
  218. self.assertEqual(channel.code, 200)
  219. self.assertEqual(channel.json_body["og:title"], "\u0434\u043a\u0430")
  220. def test_video_rejected(self) -> None:
  221. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  222. end_content = b"anything"
  223. channel = self.make_request(
  224. "GET",
  225. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  226. shorthand=False,
  227. await_result=False,
  228. )
  229. self.pump()
  230. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  231. server = AccumulatingProtocol()
  232. server.makeConnection(FakeTransport(client, self.reactor))
  233. client.makeConnection(FakeTransport(server, self.reactor))
  234. client.dataReceived(
  235. (
  236. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  237. b"Content-Type: video/mp4\r\n\r\n"
  238. )
  239. % (len(end_content))
  240. + end_content
  241. )
  242. self.pump()
  243. self.assertEqual(channel.code, 502)
  244. self.assertEqual(
  245. channel.json_body,
  246. {
  247. "errcode": "M_UNKNOWN",
  248. "error": "Requested file's content type not allowed for this operation: video/mp4",
  249. },
  250. )
  251. def test_audio_rejected(self) -> None:
  252. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  253. end_content = b"anything"
  254. channel = self.make_request(
  255. "GET",
  256. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  257. shorthand=False,
  258. await_result=False,
  259. )
  260. self.pump()
  261. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  262. server = AccumulatingProtocol()
  263. server.makeConnection(FakeTransport(client, self.reactor))
  264. client.makeConnection(FakeTransport(server, self.reactor))
  265. client.dataReceived(
  266. (
  267. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  268. b"Content-Type: audio/aac\r\n\r\n"
  269. )
  270. % (len(end_content))
  271. + end_content
  272. )
  273. self.pump()
  274. self.assertEqual(channel.code, 502)
  275. self.assertEqual(
  276. channel.json_body,
  277. {
  278. "errcode": "M_UNKNOWN",
  279. "error": "Requested file's content type not allowed for this operation: audio/aac",
  280. },
  281. )
  282. def test_non_ascii_preview_content_type(self) -> None:
  283. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  284. end_content = (
  285. b"<html><head>"
  286. b'<meta property="og:title" content="\xe4\xea\xe0" />'
  287. b'<meta property="og:description" content="hi" />'
  288. b"</head></html>"
  289. )
  290. channel = self.make_request(
  291. "GET",
  292. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  293. shorthand=False,
  294. await_result=False,
  295. )
  296. self.pump()
  297. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  298. server = AccumulatingProtocol()
  299. server.makeConnection(FakeTransport(client, self.reactor))
  300. client.makeConnection(FakeTransport(server, self.reactor))
  301. client.dataReceived(
  302. (
  303. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  304. b'Content-Type: text/html; charset="windows-1251"\r\n\r\n'
  305. )
  306. % (len(end_content),)
  307. + end_content
  308. )
  309. self.pump()
  310. self.assertEqual(channel.code, 200)
  311. self.assertEqual(channel.json_body["og:title"], "\u0434\u043a\u0430")
  312. def test_overlong_title(self) -> None:
  313. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  314. end_content = (
  315. b"<html><head>"
  316. b"<title>" + b"x" * 2000 + b"</title>"
  317. b'<meta property="og:description" content="hi" />'
  318. b"</head></html>"
  319. )
  320. channel = self.make_request(
  321. "GET",
  322. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  323. shorthand=False,
  324. await_result=False,
  325. )
  326. self.pump()
  327. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  328. server = AccumulatingProtocol()
  329. server.makeConnection(FakeTransport(client, self.reactor))
  330. client.makeConnection(FakeTransport(server, self.reactor))
  331. client.dataReceived(
  332. (
  333. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  334. b'Content-Type: text/html; charset="windows-1251"\r\n\r\n'
  335. )
  336. % (len(end_content),)
  337. + end_content
  338. )
  339. self.pump()
  340. self.assertEqual(channel.code, 200)
  341. res = channel.json_body
  342. # We should only see the `og:description` field, as `title` is too long and should be stripped out
  343. self.assertCountEqual(["og:description"], res.keys())
  344. def test_ipaddr(self) -> None:
  345. """
  346. IP addresses can be previewed directly.
  347. """
  348. self.lookups["example.com"] = [(IPv4Address, "10.1.2.3")]
  349. channel = self.make_request(
  350. "GET",
  351. "/_matrix/media/v3/preview_url?url=http://example.com",
  352. shorthand=False,
  353. await_result=False,
  354. )
  355. self.pump()
  356. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  357. server = AccumulatingProtocol()
  358. server.makeConnection(FakeTransport(client, self.reactor))
  359. client.makeConnection(FakeTransport(server, self.reactor))
  360. client.dataReceived(
  361. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\nContent-Type: text/html\r\n\r\n"
  362. % (len(self.end_content),)
  363. + self.end_content
  364. )
  365. self.pump()
  366. self.assertEqual(channel.code, 200)
  367. self.assertEqual(
  368. channel.json_body, {"og:title": "~matrix~", "og:description": "hi"}
  369. )
  370. def test_blocked_ip_specific(self) -> None:
  371. """
  372. Blocked IP addresses, found via DNS, are not spidered.
  373. """
  374. self.lookups["example.com"] = [(IPv4Address, "192.168.1.1")]
  375. channel = self.make_request(
  376. "GET",
  377. "/_matrix/media/v3/preview_url?url=http://example.com",
  378. shorthand=False,
  379. )
  380. # No requests made.
  381. self.assertEqual(len(self.reactor.tcpClients), 0)
  382. self.assertEqual(channel.code, 502)
  383. self.assertEqual(
  384. channel.json_body,
  385. {
  386. "errcode": "M_UNKNOWN",
  387. "error": "DNS resolution failure during URL preview generation",
  388. },
  389. )
  390. def test_blocked_ip_range(self) -> None:
  391. """
  392. Blocked IP ranges, IPs found over DNS, are not spidered.
  393. """
  394. self.lookups["example.com"] = [(IPv4Address, "1.1.1.2")]
  395. channel = self.make_request(
  396. "GET",
  397. "/_matrix/media/v3/preview_url?url=http://example.com",
  398. shorthand=False,
  399. )
  400. self.assertEqual(channel.code, 502)
  401. self.assertEqual(
  402. channel.json_body,
  403. {
  404. "errcode": "M_UNKNOWN",
  405. "error": "DNS resolution failure during URL preview generation",
  406. },
  407. )
  408. def test_blocked_ip_specific_direct(self) -> None:
  409. """
  410. Blocked IP addresses, accessed directly, are not spidered.
  411. """
  412. channel = self.make_request(
  413. "GET",
  414. "/_matrix/media/v3/preview_url?url=http://192.168.1.1",
  415. shorthand=False,
  416. )
  417. # No requests made.
  418. self.assertEqual(len(self.reactor.tcpClients), 0)
  419. self.assertEqual(
  420. channel.json_body,
  421. {"errcode": "M_UNKNOWN", "error": "IP address blocked"},
  422. )
  423. self.assertEqual(channel.code, 403)
  424. def test_blocked_ip_range_direct(self) -> None:
  425. """
  426. Blocked IP ranges, accessed directly, are not spidered.
  427. """
  428. channel = self.make_request(
  429. "GET", "/_matrix/media/v3/preview_url?url=http://1.1.1.2", shorthand=False
  430. )
  431. self.assertEqual(channel.code, 403)
  432. self.assertEqual(
  433. channel.json_body,
  434. {"errcode": "M_UNKNOWN", "error": "IP address blocked"},
  435. )
  436. def test_blocked_ip_range_whitelisted_ip(self) -> None:
  437. """
  438. Blocked but then subsequently whitelisted IP addresses can be
  439. spidered.
  440. """
  441. self.lookups["example.com"] = [(IPv4Address, "1.1.1.1")]
  442. channel = self.make_request(
  443. "GET",
  444. "/_matrix/media/v3/preview_url?url=http://example.com",
  445. shorthand=False,
  446. await_result=False,
  447. )
  448. self.pump()
  449. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  450. server = AccumulatingProtocol()
  451. server.makeConnection(FakeTransport(client, self.reactor))
  452. client.makeConnection(FakeTransport(server, self.reactor))
  453. client.dataReceived(
  454. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\nContent-Type: text/html\r\n\r\n"
  455. % (len(self.end_content),)
  456. + self.end_content
  457. )
  458. self.pump()
  459. self.assertEqual(channel.code, 200)
  460. self.assertEqual(
  461. channel.json_body, {"og:title": "~matrix~", "og:description": "hi"}
  462. )
  463. def test_blocked_ip_with_external_ip(self) -> None:
  464. """
  465. If a hostname resolves a blocked IP, even if there's a non-blocked one,
  466. it will be rejected.
  467. """
  468. # Hardcode the URL resolving to the IP we want.
  469. self.lookups["example.com"] = [
  470. (IPv4Address, "1.1.1.2"),
  471. (IPv4Address, "10.1.2.3"),
  472. ]
  473. channel = self.make_request(
  474. "GET",
  475. "/_matrix/media/v3/preview_url?url=http://example.com",
  476. shorthand=False,
  477. )
  478. self.assertEqual(channel.code, 502)
  479. self.assertEqual(
  480. channel.json_body,
  481. {
  482. "errcode": "M_UNKNOWN",
  483. "error": "DNS resolution failure during URL preview generation",
  484. },
  485. )
  486. def test_blocked_ipv6_specific(self) -> None:
  487. """
  488. Blocked IP addresses, found via DNS, are not spidered.
  489. """
  490. self.lookups["example.com"] = [
  491. (IPv6Address, "3fff:ffff:ffff:ffff:ffff:ffff:ffff:ffff")
  492. ]
  493. channel = self.make_request(
  494. "GET",
  495. "/_matrix/media/v3/preview_url?url=http://example.com",
  496. shorthand=False,
  497. )
  498. # No requests made.
  499. self.assertEqual(len(self.reactor.tcpClients), 0)
  500. self.assertEqual(channel.code, 502)
  501. self.assertEqual(
  502. channel.json_body,
  503. {
  504. "errcode": "M_UNKNOWN",
  505. "error": "DNS resolution failure during URL preview generation",
  506. },
  507. )
  508. def test_blocked_ipv6_range(self) -> None:
  509. """
  510. Blocked IP ranges, IPs found over DNS, are not spidered.
  511. """
  512. self.lookups["example.com"] = [(IPv6Address, "2001:800::1")]
  513. channel = self.make_request(
  514. "GET",
  515. "/_matrix/media/v3/preview_url?url=http://example.com",
  516. shorthand=False,
  517. )
  518. self.assertEqual(channel.code, 502)
  519. self.assertEqual(
  520. channel.json_body,
  521. {
  522. "errcode": "M_UNKNOWN",
  523. "error": "DNS resolution failure during URL preview generation",
  524. },
  525. )
  526. def test_OPTIONS(self) -> None:
  527. """
  528. OPTIONS returns the OPTIONS.
  529. """
  530. channel = self.make_request(
  531. "OPTIONS",
  532. "/_matrix/media/v3/preview_url?url=http://example.com",
  533. shorthand=False,
  534. )
  535. self.assertEqual(channel.code, 204)
  536. def test_accept_language_config_option(self) -> None:
  537. """
  538. Accept-Language header is sent to the remote server
  539. """
  540. self.lookups["example.com"] = [(IPv4Address, "10.1.2.3")]
  541. # Build and make a request to the server
  542. channel = self.make_request(
  543. "GET",
  544. "/_matrix/media/v3/preview_url?url=http://example.com",
  545. shorthand=False,
  546. await_result=False,
  547. )
  548. self.pump()
  549. # Extract Synapse's tcp client
  550. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  551. # Build a fake remote server to reply with
  552. server = AccumulatingProtocol()
  553. # Connect the two together
  554. server.makeConnection(FakeTransport(client, self.reactor))
  555. client.makeConnection(FakeTransport(server, self.reactor))
  556. # Tell Synapse that it has received some data from the remote server
  557. client.dataReceived(
  558. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\nContent-Type: text/html\r\n\r\n"
  559. % (len(self.end_content),)
  560. + self.end_content
  561. )
  562. # Move the reactor along until we get a response on our original channel
  563. self.pump()
  564. self.assertEqual(channel.code, 200)
  565. self.assertEqual(
  566. channel.json_body, {"og:title": "~matrix~", "og:description": "hi"}
  567. )
  568. # Check that the server received the Accept-Language header as part
  569. # of the request from Synapse
  570. self.assertIn(
  571. (
  572. b"Accept-Language: en-UK\r\n"
  573. b"Accept-Language: en-US;q=0.9\r\n"
  574. b"Accept-Language: fr;q=0.8\r\n"
  575. b"Accept-Language: *;q=0.7"
  576. ),
  577. server.data,
  578. )
  579. def test_image(self) -> None:
  580. """An image should be precached if mentioned in the HTML."""
  581. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  582. self.lookups["cdn.matrix.org"] = [(IPv4Address, "10.1.2.4")]
  583. result = (
  584. b"""<html><body><img src="http://cdn.matrix.org/foo.png"></body></html>"""
  585. )
  586. channel = self.make_request(
  587. "GET",
  588. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  589. shorthand=False,
  590. await_result=False,
  591. )
  592. self.pump()
  593. # Respond with the HTML.
  594. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  595. server = AccumulatingProtocol()
  596. server.makeConnection(FakeTransport(client, self.reactor))
  597. client.makeConnection(FakeTransport(server, self.reactor))
  598. client.dataReceived(
  599. (
  600. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  601. b'Content-Type: text/html; charset="utf8"\r\n\r\n'
  602. )
  603. % (len(result),)
  604. + result
  605. )
  606. self.pump()
  607. # Respond with the photo.
  608. client = self.reactor.tcpClients[1][2].buildProtocol(None)
  609. server = AccumulatingProtocol()
  610. server.makeConnection(FakeTransport(client, self.reactor))
  611. client.makeConnection(FakeTransport(server, self.reactor))
  612. client.dataReceived(
  613. (
  614. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  615. b"Content-Type: image/png\r\n\r\n"
  616. )
  617. % (len(SMALL_PNG),)
  618. + SMALL_PNG
  619. )
  620. self.pump()
  621. # The image should be in the result.
  622. self.assertEqual(channel.code, 200)
  623. self._assert_small_png(channel.json_body)
  624. def test_nonexistent_image(self) -> None:
  625. """If the preview image doesn't exist, ensure some data is returned."""
  626. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  627. result = (
  628. b"""<html><body><img src="http://cdn.matrix.org/foo.jpg"></body></html>"""
  629. )
  630. channel = self.make_request(
  631. "GET",
  632. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  633. shorthand=False,
  634. await_result=False,
  635. )
  636. self.pump()
  637. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  638. server = AccumulatingProtocol()
  639. server.makeConnection(FakeTransport(client, self.reactor))
  640. client.makeConnection(FakeTransport(server, self.reactor))
  641. client.dataReceived(
  642. (
  643. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  644. b'Content-Type: text/html; charset="utf8"\r\n\r\n'
  645. )
  646. % (len(result),)
  647. + result
  648. )
  649. self.pump()
  650. # There should not be a second connection.
  651. self.assertEqual(len(self.reactor.tcpClients), 1)
  652. # The image should not be in the result.
  653. self.assertEqual(channel.code, 200)
  654. self.assertNotIn("og:image", channel.json_body)
  655. @unittest.override_config(
  656. {"url_preview_url_blacklist": [{"netloc": "cdn.matrix.org"}]}
  657. )
  658. def test_image_blocked(self) -> None:
  659. """If the preview image doesn't exist, ensure some data is returned."""
  660. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  661. self.lookups["cdn.matrix.org"] = [(IPv4Address, "10.1.2.4")]
  662. result = (
  663. b"""<html><body><img src="http://cdn.matrix.org/foo.jpg"></body></html>"""
  664. )
  665. channel = self.make_request(
  666. "GET",
  667. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  668. shorthand=False,
  669. await_result=False,
  670. )
  671. self.pump()
  672. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  673. server = AccumulatingProtocol()
  674. server.makeConnection(FakeTransport(client, self.reactor))
  675. client.makeConnection(FakeTransport(server, self.reactor))
  676. client.dataReceived(
  677. (
  678. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  679. b'Content-Type: text/html; charset="utf8"\r\n\r\n'
  680. )
  681. % (len(result),)
  682. + result
  683. )
  684. self.pump()
  685. # There should not be a second connection.
  686. self.assertEqual(len(self.reactor.tcpClients), 1)
  687. # The image should not be in the result.
  688. self.assertEqual(channel.code, 200)
  689. self.assertNotIn("og:image", channel.json_body)
  690. def test_oembed_failure(self) -> None:
  691. """If the autodiscovered oEmbed URL fails, ensure some data is returned."""
  692. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  693. result = b"""
  694. <title>oEmbed Autodiscovery Fail</title>
  695. <link rel="alternate" type="application/json+oembed"
  696. href="http://example.com/oembed?url=http%3A%2F%2Fmatrix.org&format=json"
  697. title="matrixdotorg" />
  698. """
  699. channel = self.make_request(
  700. "GET",
  701. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  702. shorthand=False,
  703. await_result=False,
  704. )
  705. self.pump()
  706. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  707. server = AccumulatingProtocol()
  708. server.makeConnection(FakeTransport(client, self.reactor))
  709. client.makeConnection(FakeTransport(server, self.reactor))
  710. client.dataReceived(
  711. (
  712. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  713. b'Content-Type: text/html; charset="utf8"\r\n\r\n'
  714. )
  715. % (len(result),)
  716. + result
  717. )
  718. self.pump()
  719. self.assertEqual(channel.code, 200)
  720. # The image should not be in the result.
  721. self.assertEqual(channel.json_body["og:title"], "oEmbed Autodiscovery Fail")
  722. def test_data_url(self) -> None:
  723. """
  724. Requesting to preview a data URL is not supported.
  725. """
  726. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  727. data = base64.b64encode(SMALL_PNG).decode()
  728. query_params = urlencode(
  729. {
  730. "url": f'<html><head><img src="data:image/png;base64,{data}" /></head></html>'
  731. }
  732. )
  733. channel = self.make_request(
  734. "GET",
  735. f"/_matrix/media/v3/preview_url?{query_params}",
  736. shorthand=False,
  737. )
  738. self.pump()
  739. self.assertEqual(channel.code, 500)
  740. def test_inline_data_url(self) -> None:
  741. """
  742. An inline image (as a data URL) should be parsed properly.
  743. """
  744. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  745. data = base64.b64encode(SMALL_PNG)
  746. end_content = (
  747. b"<html><head>" b'<img src="data:image/png;base64,%s" />' b"</head></html>"
  748. ) % (data,)
  749. channel = self.make_request(
  750. "GET",
  751. "/_matrix/media/v3/preview_url?url=http://matrix.org",
  752. shorthand=False,
  753. await_result=False,
  754. )
  755. self.pump()
  756. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  757. server = AccumulatingProtocol()
  758. server.makeConnection(FakeTransport(client, self.reactor))
  759. client.makeConnection(FakeTransport(server, self.reactor))
  760. client.dataReceived(
  761. (
  762. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  763. b'Content-Type: text/html; charset="utf8"\r\n\r\n'
  764. )
  765. % (len(end_content),)
  766. + end_content
  767. )
  768. self.pump()
  769. self.assertEqual(channel.code, 200)
  770. self._assert_small_png(channel.json_body)
  771. def test_oembed_photo(self) -> None:
  772. """Test an oEmbed endpoint which returns a 'photo' type which redirects the preview to a new URL."""
  773. self.lookups["publish.twitter.com"] = [(IPv4Address, "10.1.2.3")]
  774. self.lookups["cdn.twitter.com"] = [(IPv4Address, "10.1.2.3")]
  775. result = {
  776. "version": "1.0",
  777. "type": "photo",
  778. "url": "http://cdn.twitter.com/matrixdotorg",
  779. }
  780. oembed_content = json.dumps(result).encode("utf-8")
  781. channel = self.make_request(
  782. "GET",
  783. "/_matrix/media/v3/preview_url?url=http://twitter.com/matrixdotorg/status/12345",
  784. shorthand=False,
  785. await_result=False,
  786. )
  787. self.pump()
  788. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  789. server = AccumulatingProtocol()
  790. server.makeConnection(FakeTransport(client, self.reactor))
  791. client.makeConnection(FakeTransport(server, self.reactor))
  792. client.dataReceived(
  793. (
  794. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  795. b'Content-Type: application/json; charset="utf8"\r\n\r\n'
  796. )
  797. % (len(oembed_content),)
  798. + oembed_content
  799. )
  800. self.pump()
  801. # Ensure a second request is made to the photo URL.
  802. client = self.reactor.tcpClients[1][2].buildProtocol(None)
  803. server = AccumulatingProtocol()
  804. server.makeConnection(FakeTransport(client, self.reactor))
  805. client.makeConnection(FakeTransport(server, self.reactor))
  806. client.dataReceived(
  807. (
  808. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  809. b"Content-Type: image/png\r\n\r\n"
  810. )
  811. % (len(SMALL_PNG),)
  812. + SMALL_PNG
  813. )
  814. self.pump()
  815. # Ensure the URL is what was requested.
  816. self.assertIn(b"/matrixdotorg", server.data)
  817. self.assertEqual(channel.code, 200)
  818. body = channel.json_body
  819. self.assertEqual(body["og:url"], "http://twitter.com/matrixdotorg/status/12345")
  820. self._assert_small_png(body)
  821. def test_oembed_rich(self) -> None:
  822. """Test an oEmbed endpoint which returns HTML content via the 'rich' type."""
  823. self.lookups["publish.twitter.com"] = [(IPv4Address, "10.1.2.3")]
  824. result = {
  825. "version": "1.0",
  826. "type": "rich",
  827. # Note that this provides the author, not the title.
  828. "author_name": "Alice",
  829. "html": "<div>Content Preview</div>",
  830. }
  831. end_content = json.dumps(result).encode("utf-8")
  832. channel = self.make_request(
  833. "GET",
  834. "/_matrix/media/v3/preview_url?url=http://twitter.com/matrixdotorg/status/12345",
  835. shorthand=False,
  836. await_result=False,
  837. )
  838. self.pump()
  839. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  840. server = AccumulatingProtocol()
  841. server.makeConnection(FakeTransport(client, self.reactor))
  842. client.makeConnection(FakeTransport(server, self.reactor))
  843. client.dataReceived(
  844. (
  845. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  846. b'Content-Type: application/json; charset="utf8"\r\n\r\n'
  847. )
  848. % (len(end_content),)
  849. + end_content
  850. )
  851. self.pump()
  852. # Double check that the proper host is being connected to. (Note that
  853. # twitter.com can't be resolved so this is already implicitly checked.)
  854. self.assertIn(b"\r\nHost: publish.twitter.com\r\n", server.data)
  855. self.assertEqual(channel.code, 200)
  856. body = channel.json_body
  857. self.assertEqual(
  858. body,
  859. {
  860. "og:url": "http://twitter.com/matrixdotorg/status/12345",
  861. "og:title": "Alice",
  862. "og:description": "Content Preview",
  863. },
  864. )
  865. def test_oembed_format(self) -> None:
  866. """Test an oEmbed endpoint which requires the format in the URL."""
  867. self.lookups["www.hulu.com"] = [(IPv4Address, "10.1.2.3")]
  868. result = {
  869. "version": "1.0",
  870. "type": "rich",
  871. "html": "<div>Content Preview</div>",
  872. }
  873. end_content = json.dumps(result).encode("utf-8")
  874. channel = self.make_request(
  875. "GET",
  876. "/_matrix/media/v3/preview_url?url=http://www.hulu.com/watch/12345",
  877. shorthand=False,
  878. await_result=False,
  879. )
  880. self.pump()
  881. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  882. server = AccumulatingProtocol()
  883. server.makeConnection(FakeTransport(client, self.reactor))
  884. client.makeConnection(FakeTransport(server, self.reactor))
  885. client.dataReceived(
  886. (
  887. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  888. b'Content-Type: application/json; charset="utf8"\r\n\r\n'
  889. )
  890. % (len(end_content),)
  891. + end_content
  892. )
  893. self.pump()
  894. # The {format} should have been turned into json.
  895. self.assertIn(b"/api/oembed.json", server.data)
  896. # A URL parameter of format=json should be provided.
  897. self.assertIn(b"format=json", server.data)
  898. self.assertEqual(channel.code, 200)
  899. body = channel.json_body
  900. self.assertEqual(
  901. body,
  902. {
  903. "og:url": "http://www.hulu.com/watch/12345",
  904. "og:description": "Content Preview",
  905. },
  906. )
  907. @unittest.override_config(
  908. {"url_preview_url_blacklist": [{"netloc": "publish.twitter.com"}]}
  909. )
  910. def test_oembed_blocked(self) -> None:
  911. """The oEmbed URL should not be downloaded if the oEmbed URL is blocked."""
  912. self.lookups["twitter.com"] = [(IPv4Address, "10.1.2.3")]
  913. channel = self.make_request(
  914. "GET",
  915. "/_matrix/media/v3/preview_url?url=http://twitter.com/matrixdotorg/status/12345",
  916. shorthand=False,
  917. await_result=False,
  918. )
  919. self.pump()
  920. self.assertEqual(channel.code, 403, channel.result)
  921. def test_oembed_autodiscovery(self) -> None:
  922. """
  923. Autodiscovery works by finding the link in the HTML response and then requesting an oEmbed URL.
  924. 1. Request a preview of a URL which is not known to the oEmbed code.
  925. 2. It returns HTML including a link to an oEmbed preview.
  926. 3. The oEmbed preview is requested and returns a URL for an image.
  927. 4. The image is requested for thumbnailing.
  928. """
  929. # This is a little cheesy in that we use the www subdomain (which isn't the
  930. # list of oEmbed patterns) to get "raw" HTML response.
  931. self.lookups["www.twitter.com"] = [(IPv4Address, "10.1.2.3")]
  932. self.lookups["publish.twitter.com"] = [(IPv4Address, "10.1.2.3")]
  933. self.lookups["cdn.twitter.com"] = [(IPv4Address, "10.1.2.3")]
  934. result = b"""
  935. <link rel="alternate" type="application/json+oembed"
  936. href="http://publish.twitter.com/oembed?url=http%3A%2F%2Fcdn.twitter.com%2Fmatrixdotorg%2Fstatus%2F12345&format=json"
  937. title="matrixdotorg" />
  938. """
  939. channel = self.make_request(
  940. "GET",
  941. "/_matrix/media/v3/preview_url?url=http://www.twitter.com/matrixdotorg/status/12345",
  942. shorthand=False,
  943. await_result=False,
  944. )
  945. self.pump()
  946. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  947. server = AccumulatingProtocol()
  948. server.makeConnection(FakeTransport(client, self.reactor))
  949. client.makeConnection(FakeTransport(server, self.reactor))
  950. client.dataReceived(
  951. (
  952. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  953. b'Content-Type: text/html; charset="utf8"\r\n\r\n'
  954. )
  955. % (len(result),)
  956. + result
  957. )
  958. self.pump()
  959. # The oEmbed response.
  960. result2 = {
  961. "version": "1.0",
  962. "type": "photo",
  963. "url": "http://cdn.twitter.com/matrixdotorg",
  964. }
  965. oembed_content = json.dumps(result2).encode("utf-8")
  966. # Ensure a second request is made to the oEmbed URL.
  967. client = self.reactor.tcpClients[1][2].buildProtocol(None)
  968. server = AccumulatingProtocol()
  969. server.makeConnection(FakeTransport(client, self.reactor))
  970. client.makeConnection(FakeTransport(server, self.reactor))
  971. client.dataReceived(
  972. (
  973. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  974. b'Content-Type: application/json; charset="utf8"\r\n\r\n'
  975. )
  976. % (len(oembed_content),)
  977. + oembed_content
  978. )
  979. self.pump()
  980. # Ensure the URL is what was requested.
  981. self.assertIn(b"/oembed?", server.data)
  982. # Ensure a third request is made to the photo URL.
  983. client = self.reactor.tcpClients[2][2].buildProtocol(None)
  984. server = AccumulatingProtocol()
  985. server.makeConnection(FakeTransport(client, self.reactor))
  986. client.makeConnection(FakeTransport(server, self.reactor))
  987. client.dataReceived(
  988. (
  989. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  990. b"Content-Type: image/png\r\n\r\n"
  991. )
  992. % (len(SMALL_PNG),)
  993. + SMALL_PNG
  994. )
  995. self.pump()
  996. # Ensure the URL is what was requested.
  997. self.assertIn(b"/matrixdotorg", server.data)
  998. self.assertEqual(channel.code, 200)
  999. body = channel.json_body
  1000. self.assertEqual(
  1001. body["og:url"], "http://www.twitter.com/matrixdotorg/status/12345"
  1002. )
  1003. self._assert_small_png(body)
  1004. @unittest.override_config(
  1005. {"url_preview_url_blacklist": [{"netloc": "publish.twitter.com"}]}
  1006. )
  1007. def test_oembed_autodiscovery_blocked(self) -> None:
  1008. """
  1009. If the discovered oEmbed URL is blocked, it should be discarded.
  1010. """
  1011. # This is a little cheesy in that we use the www subdomain (which isn't the
  1012. # list of oEmbed patterns) to get "raw" HTML response.
  1013. self.lookups["www.twitter.com"] = [(IPv4Address, "10.1.2.3")]
  1014. self.lookups["publish.twitter.com"] = [(IPv4Address, "10.1.2.4")]
  1015. result = b"""
  1016. <title>Test</title>
  1017. <link rel="alternate" type="application/json+oembed"
  1018. href="http://publish.twitter.com/oembed?url=http%3A%2F%2Fcdn.twitter.com%2Fmatrixdotorg%2Fstatus%2F12345&format=json"
  1019. title="matrixdotorg" />
  1020. """
  1021. channel = self.make_request(
  1022. "GET",
  1023. "/_matrix/media/v3/preview_url?url=http://www.twitter.com/matrixdotorg/status/12345",
  1024. shorthand=False,
  1025. await_result=False,
  1026. )
  1027. self.pump()
  1028. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  1029. server = AccumulatingProtocol()
  1030. server.makeConnection(FakeTransport(client, self.reactor))
  1031. client.makeConnection(FakeTransport(server, self.reactor))
  1032. client.dataReceived(
  1033. (
  1034. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n"
  1035. b'Content-Type: text/html; charset="utf8"\r\n\r\n'
  1036. )
  1037. % (len(result),)
  1038. + result
  1039. )
  1040. self.pump()
  1041. # Ensure there's no additional connections.
  1042. self.assertEqual(len(self.reactor.tcpClients), 1)
  1043. # Ensure the URL is what was requested.
  1044. self.assertIn(b"\r\nHost: www.twitter.com\r\n", server.data)
  1045. self.assertEqual(channel.code, 200)
  1046. body = channel.json_body
  1047. self.assertEqual(body["og:title"], "Test")
  1048. self.assertNotIn("og:image", body)
  1049. def _download_image(self) -> Tuple[str, str]:
  1050. """Downloads an image into the URL cache.
  1051. Returns:
  1052. A (host, media_id) tuple representing the MXC URI of the image.
  1053. """
  1054. self.lookups["cdn.twitter.com"] = [(IPv4Address, "10.1.2.3")]
  1055. channel = self.make_request(
  1056. "GET",
  1057. "/_matrix/media/v3/preview_url?url=http://cdn.twitter.com/matrixdotorg",
  1058. shorthand=False,
  1059. await_result=False,
  1060. )
  1061. self.pump()
  1062. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  1063. server = AccumulatingProtocol()
  1064. server.makeConnection(FakeTransport(client, self.reactor))
  1065. client.makeConnection(FakeTransport(server, self.reactor))
  1066. client.dataReceived(
  1067. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\nContent-Type: image/png\r\n\r\n"
  1068. % (len(SMALL_PNG),)
  1069. + SMALL_PNG
  1070. )
  1071. self.pump()
  1072. self.assertEqual(channel.code, 200)
  1073. body = channel.json_body
  1074. mxc_uri = body["og:image"]
  1075. host, _port, media_id = parse_and_validate_mxc_uri(mxc_uri)
  1076. self.assertIsNone(_port)
  1077. return host, media_id
  1078. def test_storage_providers_exclude_files(self) -> None:
  1079. """Test that files are not stored in or fetched from storage providers."""
  1080. host, media_id = self._download_image()
  1081. rel_file_path = self.media_repo.filepaths.url_cache_filepath_rel(media_id)
  1082. media_store_path = os.path.join(self.media_store_path, rel_file_path)
  1083. storage_provider_path = os.path.join(self.storage_path, rel_file_path)
  1084. # Check storage
  1085. self.assertTrue(os.path.isfile(media_store_path))
  1086. self.assertFalse(
  1087. os.path.isfile(storage_provider_path),
  1088. "URL cache file was unexpectedly stored in a storage provider",
  1089. )
  1090. # Check fetching
  1091. channel = self.make_request(
  1092. "GET",
  1093. f"/_matrix/media/v3/download/{host}/{media_id}",
  1094. shorthand=False,
  1095. await_result=False,
  1096. )
  1097. self.pump()
  1098. self.assertEqual(channel.code, 200)
  1099. # Move cached file into the storage provider
  1100. os.makedirs(os.path.dirname(storage_provider_path), exist_ok=True)
  1101. os.rename(media_store_path, storage_provider_path)
  1102. channel = self.make_request(
  1103. "GET",
  1104. f"/_matrix/media/v3/download/{host}/{media_id}",
  1105. shorthand=False,
  1106. await_result=False,
  1107. )
  1108. self.pump()
  1109. self.assertEqual(
  1110. channel.code,
  1111. 404,
  1112. "URL cache file was unexpectedly retrieved from a storage provider",
  1113. )
  1114. def test_storage_providers_exclude_thumbnails(self) -> None:
  1115. """Test that thumbnails are not stored in or fetched from storage providers."""
  1116. host, media_id = self._download_image()
  1117. rel_thumbnail_path = (
  1118. self.media_repo.filepaths.url_cache_thumbnail_directory_rel(media_id)
  1119. )
  1120. media_store_thumbnail_path = os.path.join(
  1121. self.media_store_path, rel_thumbnail_path
  1122. )
  1123. storage_provider_thumbnail_path = os.path.join(
  1124. self.storage_path, rel_thumbnail_path
  1125. )
  1126. # Check storage
  1127. self.assertTrue(os.path.isdir(media_store_thumbnail_path))
  1128. self.assertFalse(
  1129. os.path.isdir(storage_provider_thumbnail_path),
  1130. "URL cache thumbnails were unexpectedly stored in a storage provider",
  1131. )
  1132. # Check fetching
  1133. channel = self.make_request(
  1134. "GET",
  1135. f"/_matrix/media/v3/thumbnail/{host}/{media_id}?width=32&height=32&method=scale",
  1136. shorthand=False,
  1137. await_result=False,
  1138. )
  1139. self.pump()
  1140. self.assertEqual(channel.code, 200)
  1141. # Remove the original, otherwise thumbnails will regenerate
  1142. rel_file_path = self.media_repo.filepaths.url_cache_filepath_rel(media_id)
  1143. media_store_path = os.path.join(self.media_store_path, rel_file_path)
  1144. os.remove(media_store_path)
  1145. # Move cached thumbnails into the storage provider
  1146. os.makedirs(os.path.dirname(storage_provider_thumbnail_path), exist_ok=True)
  1147. os.rename(media_store_thumbnail_path, storage_provider_thumbnail_path)
  1148. channel = self.make_request(
  1149. "GET",
  1150. f"/_matrix/media/v3/thumbnail/{host}/{media_id}?width=32&height=32&method=scale",
  1151. shorthand=False,
  1152. await_result=False,
  1153. )
  1154. self.pump()
  1155. self.assertEqual(
  1156. channel.code,
  1157. 404,
  1158. "URL cache thumbnail was unexpectedly retrieved from a storage provider",
  1159. )
  1160. def test_cache_expiry(self) -> None:
  1161. """Test that URL cache files and thumbnails are cleaned up properly on expiry."""
  1162. _host, media_id = self._download_image()
  1163. file_path = self.media_repo.filepaths.url_cache_filepath(media_id)
  1164. file_dirs = self.media_repo.filepaths.url_cache_filepath_dirs_to_delete(
  1165. media_id
  1166. )
  1167. thumbnail_dir = self.media_repo.filepaths.url_cache_thumbnail_directory(
  1168. media_id
  1169. )
  1170. thumbnail_dirs = self.media_repo.filepaths.url_cache_thumbnail_dirs_to_delete(
  1171. media_id
  1172. )
  1173. self.assertTrue(os.path.isfile(file_path))
  1174. self.assertTrue(os.path.isdir(thumbnail_dir))
  1175. self.reactor.advance(IMAGE_CACHE_EXPIRY_MS * 1000 + 1)
  1176. self.get_success(self.url_previewer._expire_url_cache_data())
  1177. for path in [file_path] + file_dirs + [thumbnail_dir] + thumbnail_dirs:
  1178. self.assertFalse(
  1179. os.path.exists(path),
  1180. f"{os.path.relpath(path, self.media_store_path)} was not deleted",
  1181. )
  1182. @unittest.override_config({"url_preview_url_blacklist": [{"port": "*"}]})
  1183. def test_blocked_port(self) -> None:
  1184. """Tests that blocking URLs with a port makes previewing such URLs
  1185. fail with a 403 error and doesn't impact other previews.
  1186. """
  1187. self.lookups["matrix.org"] = [(IPv4Address, "10.1.2.3")]
  1188. bad_url = quote("http://matrix.org:8888/foo")
  1189. good_url = quote("http://matrix.org/foo")
  1190. channel = self.make_request(
  1191. "GET",
  1192. "/_matrix/media/v3/preview_url?url=" + bad_url,
  1193. shorthand=False,
  1194. await_result=False,
  1195. )
  1196. self.pump()
  1197. self.assertEqual(channel.code, 403, channel.result)
  1198. channel = self.make_request(
  1199. "GET",
  1200. "/_matrix/media/v3/preview_url?url=" + good_url,
  1201. shorthand=False,
  1202. await_result=False,
  1203. )
  1204. self.pump()
  1205. client = self.reactor.tcpClients[0][2].buildProtocol(None)
  1206. server = AccumulatingProtocol()
  1207. server.makeConnection(FakeTransport(client, self.reactor))
  1208. client.makeConnection(FakeTransport(server, self.reactor))
  1209. client.dataReceived(
  1210. b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\nContent-Type: text/html\r\n\r\n"
  1211. % (len(self.end_content),)
  1212. + self.end_content
  1213. )
  1214. self.pump()
  1215. self.assertEqual(channel.code, 200)
  1216. @unittest.override_config(
  1217. {"url_preview_url_blacklist": [{"netloc": "example.com"}]}
  1218. )
  1219. def test_blocked_url(self) -> None:
  1220. """Tests that blocking URLs with a host makes previewing such URLs
  1221. fail with a 403 error.
  1222. """
  1223. self.lookups["example.com"] = [(IPv4Address, "10.1.2.3")]
  1224. bad_url = quote("http://example.com/foo")
  1225. channel = self.make_request(
  1226. "GET",
  1227. "/_matrix/media/v3/preview_url?url=" + bad_url,
  1228. shorthand=False,
  1229. await_result=False,
  1230. )
  1231. self.pump()
  1232. self.assertEqual(channel.code, 403, channel.result)