Aldo Cortesi
def93ea8ca
websockets: remove validation
...
We don't really need this any more. The interface is much less error prone
because bit flags are no longer integers, we have a range check on opcode on
header instantiation, and we've deferred length code calculation and so forth
into the byte render methods.
2015-04-24 15:23:00 +12:00
Aldo Cortesi
f22bc0b4c7
websocket: interface refactoring
...
- Separate out FrameHeader. We need to deal with this separately in many circumstances.
- Simpler equality scheme.
- Bits are now specified by truthiness - we don't care about the integer value.
This means lots of validation is not needed any more.
2015-04-24 15:09:21 +12:00
Aldo Cortesi
3519871f34
websockets: refactor to avoid rundantly specifying payloads and payload lengths
2015-04-24 09:21:04 +12:00
Aldo Cortesi
bdd52fead3
websockets: extract frame header creation into a function
2015-04-24 08:47:09 +12:00
Aldo Cortesi
42a87a1d8b
websockets: handshake checks only take headers
2015-04-23 08:23:51 +12:00
Aldo Cortesi
4fb49c8e55
websockets: (very) slightly nicer is_valid constraints
2015-04-21 23:49:27 +12:00
Aldo Cortesi
176e29fc09
websockets: constants, variable names, refactoring
2015-04-21 23:13:42 +12:00
Aldo Cortesi
1b509d5aea
Whitespace, interface simplification
...
- safe_tobytes doesn't buy us much
- move masking key generation inline
2015-04-21 22:51:01 +12:00
Aldo Cortesi
3e0a71ea34
websockets: refactor to use http and header functions in http.py
2015-04-21 22:39:45 +12:00
Aldo Cortesi
e5f1264838
Whitespace, indentation, nounce -> nonce
2015-04-21 13:39:00 +12:00
Aldo Cortesi
7d83e388aa
Whitespace, pep8, mixed indentation
2015-04-21 11:19:00 +12:00
Aldo Cortesi
dd7ea896f2
Return a named tuple from read_response
2015-04-21 11:11:16 +12:00
Aldo Cortesi
2c660d7633
Migrate requeset reading from mitmproxy to netlib
2015-04-21 11:05:12 +12:00
Chandler Abraham
4ea1ccb638
fixing test coverage, adding to_file/from_file reader writes to match socks.py
2015-04-19 22:18:30 -07:00
Aldo Cortesi
74389ef04a
Websockets: reorganise
...
- websockets.py to top-level
- implementations into test suite
2015-04-20 09:38:09 +12:00
Maximilian Hils
08ba987a84
Merge branch 'master' of github.com:mitmproxy/netlib
2015-04-17 16:29:25 +02:00
Maximilian Hils
0c2ad1edb1
fix socket_close on Windows, refs mitmproxy/mitmproxy#527
2015-04-17 16:29:09 +02:00
Aldo Cortesi
7defb5be86
websockets: more whitespace, WebSocketFrame -> Frame
2015-04-17 14:29:20 +12:00
Aldo Cortesi
488c25d812
websockets: whitespace, PEP8
2015-04-17 13:57:39 +12:00
Aldo Cortesi
3bbafa24bd
Merge pull request #54 from Chandler/websockets
...
Netlib WebSockets take 1
2015-04-17 13:46:51 +12:00
Aldo Cortesi
c53d89fd7f
Improve flexibility of http_cookies._format_pairs
2015-04-16 08:30:54 +12:00
Aldo Cortesi
0c85c72dc4
ODict improvements
...
- Setting values now tries to preserve the existing order, rather than
just appending to the end.
- __repr__ now returns a repr of the tuple list. The old repr becomes a
.format() method. This is clearer, makes troubleshooting easier, and
doesn't assume all data in ODicts are header-like
2015-04-15 10:28:17 +12:00
Aldo Cortesi
aeebf31927
odict: don't convert values to strings when added
2015-04-14 16:20:02 +12:00
Aldo Cortesi
d739882bf2
Add an .extend method for ODicts
2015-04-14 13:50:57 +12:00
Aldo Cortesi
6db5e0a4a1
Remove old-style set-cookie cruft, unit tests to 100%
2015-04-14 10:13:03 +12:00
Aldo Cortesi
de9e741125
Firm up cookie parsing and formatting API
...
Make a tough call: we won't support old-style comma-separated set-cookie
headers. Real world testing has shown that the latest rfc (6265) is
often violated in ways that make the parsing problem indeterminate.
Since this is much more common than the old style deprecated set-cookie
variant, we focus on the most useful case.
2015-04-14 10:02:10 +12:00
Chandler Abraham
2d72a1b6b5
100% test coverage, though still need plenty more
2015-04-13 13:36:09 -07:00
Chandler Abraham
f131f9b855
handshake tests, serialization test
2015-04-11 17:26:59 -07:00
Aldo Cortesi
1a79ef8b6c
Merge branch 'master' of https://github.com/mitmproxy/netlib
2015-04-12 11:32:27 +12:00
Aldo Cortesi
2630da7263
cookies: Cater for special values, fix some bugs found in real-world testing
2015-04-12 11:30:35 +12:00
Aldo Cortesi
73ce169e3d
Initial outline of a cookie parsing and serialization module.
2015-04-12 10:26:09 +12:00
Chandler Abraham
0edc04814e
small cleanups, working on tests
2015-04-11 11:35:15 -07:00
Chandler Abraham
e41e5cbfdd
netlib websockets
2015-04-10 18:37:41 -07:00
Maximilian Hils
e58f76aec1
fix code smell
2015-04-09 02:09:33 +02:00
Maximilian Hils
7f7ccd3a18
100% test coverage
2015-04-09 00:57:37 +02:00
Maximilian Hils
6fbe3006af
fail gracefully if we cannot start a new thread
2015-04-09 00:13:01 +02:00
Maximilian Hils
d5eff70b6e
fix tests on Windows
2015-03-07 01:31:31 +01:00
Maximilian Hils
dbadc1b613
clean up cert handling, fix mitmproxy/mitmproxy#472
2015-03-07 01:22:02 +01:00
Maximilian Hils
d71f3b68fd
make tests more robust, fix coveralls
2015-02-27 22:27:23 +01:00
Maximilian Hils
da1eb94ccd
100% test coverage 🎉
2015-02-27 22:02:52 +01:00
Maximilian Hils
63fb433690
fix #53
2015-02-27 20:40:17 +01:00
Maximilian Hils
738a7b34a2
Merge branch 'master' of github.com:mitmproxy/netlib
2015-02-27 17:00:19 +01:00
Maximilian Hils
224f737646
add option to log ssl keys
...
refs mitmproxy/mitmproxy#475
2015-02-27 16:59:29 +01:00
Maximilian Hils
2a2402dfff
...two years is not enough.
2015-02-17 00:10:10 +01:00
Aldo Cortesi
7e5bb74e72
5 years is enough...
2015-02-17 12:03:52 +13:00
Aldo Cortesi
c9de3e770b
By popular demand, bump dummy cert expiry to 5 years
...
fixes #52
2015-02-17 11:59:07 +13:00
Aldo Cortesi
3c919631d4
Bump version
2014-12-28 22:46:19 +13:00
Maximilian Hils
438c1fbc7d
TCPClient: Use TLS1.1+ where available, BaseHandler: disable SSLv2
2014-12-15 12:32:36 +01:00
Aldo Cortesi
7098c90a6d
Bump version to 0.11.1
2014-11-15 12:45:06 +13:00
Aldo Cortesi
c56e7a90d8
Fix tracebacks in connection finish
2014-11-15 12:31:13 +13:00
