Commit graph

268 commits

Author SHA1 Message Date
jesopo
ee6360be22 don't check already-read data when checking for too-large requests
this check was here because the first read will return empty if it was an
invalid byte sequence for e.g. gzip because we needed to receive more data. the
second read will always return data (not decoded) so regardless of what the
already-read data is, the second read is the only criteria we need.
2019-09-17 17:33:23 +01:00
jesopo
1ac7f2697e log which URL caused an error in request_many 2019-09-17 17:09:19 +01:00
jesopo
98545a9fb4 only decode content-types in DECODE_CONTENT_TYPES 2019-09-17 16:12:03 +01:00
jesopo
8ca0d30fef Response.__init__() needs encoding now 2019-09-17 14:11:12 +01:00
jesopo
b7dd78ef1a restore 5 second (instead of default 10) deadline for http.request 2019-09-17 13:44:14 +01:00
jesopo
94c3ff962b use utils.deadline_process() in utils.http._request() so background threads can
call _request()
2019-09-17 13:41:11 +01:00
jesopo
fa95eaa9eb add .get() to CaseInsensitiveDict 2019-09-17 13:40:37 +01:00
jesopo
d454f9b732 use Queue.get() with timeout, not Process.join() for timeout
this was because the threads spawned by multiprocessing.Queue seemed to be
making Process.join() believe the subprocess had not exited.
2019-09-17 13:39:23 +01:00
jesopo
1ed14f9a17 first draft of multiprocess.Process deadline system 2019-09-17 11:56:30 +01:00
jesopo
334d580c57 'seperate_hostmask()' -> 'parse_hostmask()' 2019-09-16 18:43:57 +01:00
jesopo
47735421b8 add json_body arg to Request to json-encode body, only return from body if
not null
2019-09-16 10:57:18 +01:00
jesopo
f9d13dc373 support '0' as an IntSetting value 2019-09-15 22:22:30 +01:00
jesopo
d950eb3660 add utils.SensitiveSetting, to .format() hide value 2019-09-12 12:17:31 +01:00
jesopo
ba0911f2e7 add utils.Setting.format() so subtypes can format differently 2019-09-12 12:17:09 +01:00
jesopo
9d33354d16 translate INVITE from [channel_name, target] to [target, channel_name] 2019-09-12 11:24:25 +01:00
jesopo
540c7b8c44 Revert "INVITE should be [channel_name, target]"
This reverts commit f3d8ffad2c.
2019-09-12 11:23:29 +01:00
jesopo
f3d8ffad2c INVITE should be [channel_name, target] 2019-09-12 11:21:29 +01:00
jesopo
77f50187c5 allow Requests to specify a useragent 2019-09-12 10:41:50 +01:00
jesopo
9d6a3982ed add a helper utils.http.Client static object 2019-09-11 17:53:49 +01:00
jesopo
51dc26d113 add proxy to Request objects 2019-09-11 17:53:37 +01:00
jesopo
4a97c9eb0d refactor utils.http.requests to support a Request object 2019-09-11 17:44:07 +01:00
jesopo
8f8cf92ae2 automatically decode certain http content types 2019-09-11 15:28:13 +01:00
jesopo
a9b106c6be Don't try to .decode non-html things, default iso-lat-1 for non-html too 2019-09-09 16:17:26 +01:00
jesopo
b83f5d9e30 add flag to disable encoding detection 2019-09-09 14:59:08 +01:00
jesopo
5ef2b7af27 'str.split' -> 's.split' 2019-09-09 14:53:11 +01:00
jesopo
1df82c1cb2 still default to iso-latin-1 if no on-page or in-header content-type is present 2019-09-09 14:48:26 +01:00
jesopo
0a67659637 only look for <meta>-related tags when there are meta tags 2019-09-09 14:39:19 +01:00
jesopo
0a1077c5cd add explicit None return for _find_encoding (mypy) 2019-09-09 14:25:01 +01:00
jesopo
ff9c82bf67 change utils.http.request to best-effort detect on-page encoding
closes #113
2019-09-09 14:11:18 +01:00
jesopo
007bb78d30 make utils.from_pretty_time() format much stricter 2019-09-04 11:22:56 +01:00
jesopo
397cfa8e7e correctly qualify DeadlineExceededException namespace 2019-09-03 14:54:59 +01:00
jesopo
b7b2f31c1c use utils.deadline() in utils.http.request, not raw sigalrm 2019-09-02 15:50:21 +01:00
jesopo
d42d694e64 move deadline alarm time check inside try/finally 2019-09-02 15:50:12 +01:00
jesopo
9cc1ee98eb Pass the content of a webpage to HTTPParsingException 2019-09-02 13:27:44 +01:00
jesopo
408b89aeb7 use \S+ for url regex (for non-ascii chars), use url_sanitize to catch <> 2019-09-02 13:25:48 +01:00
jesopo
9e43ea1599 add per-user fediverse setting 2019-09-02 08:51:56 +01:00
jesopo
722d6418c8 switch to using signal.setitimer so we can restore previous timer 2019-08-30 18:36:48 +01:00
jesopo
a57a06b1cc switch utils.deadline to a context manager 2019-08-30 18:13:12 +01:00
jesopo
e97e3c8f30 add utils.deadline(func, seconds) to put hard limits on code exec time 2019-08-30 17:17:03 +01:00
jesopo
4bfb4c3200 Add IntRangeSetting, reorder OptionsSetting arg order 2019-08-30 14:40:54 +01:00
jesopo
58a4fd7e74 round microseconds up to a whole number and zero-pad to 3 chars 2019-08-30 13:46:00 +01:00
jesopo
d3ed08e336 Don't use 'Z' for "utc timezone" - actually parse timezone out of datetimes 2019-08-30 13:25:19 +01:00
jesopo
b6361ba4e3 Remove extraneous space 2019-08-07 14:00:56 +01:00
jesopo
20042edfd9 Allow bypass of content-type check in utils.http.request 2019-08-05 15:41:02 +01:00
jesopo
d093027431 not all HTTP responses have content-type 2019-08-02 17:33:16 +01:00
jesopo
77dfc76591 switch to function/module magic being a single object 2019-07-26 11:58:06 +01:00
jesopo
c19c6c0e14 asyncio.gather -> asyncio.wait (with timeout) 2019-07-08 14:50:11 +01:00
jesopo
469c725675 tell asyncio.gather which loop to use 2019-07-08 14:41:12 +01:00
jesopo
a1438abf66 close event loop when we're done with it (request_many()) 2019-07-08 13:59:48 +01:00
jesopo
81c7af8ab5 Don't try/except async http exceptions 2019-07-08 13:51:02 +01:00