Dmitry D. Chernov
45d0ba9e2f
Initial implementation of MTProxy support ( #1107 )
2019-02-11 00:16:46 +01:00
Dmitry D. Chernov
8c428e8566
Fix broken connection establishment in ConnectionTcpObfuscated
...
This regression was introduced in ebde3be820
.
2019-02-11 07:54:59 +10:00
Dmitry D. Chernov
c39cc06908
Fix incorrect check for reserved data prefix in ConnectionTcpObfuscated
2019-02-11 05:47:09 +10:00
Dmitry D. Chernov
fc2977fc0d
Tiny style fixes
2019-02-07 06:55:34 +10:00
Dmitry D. Chernov
c8f16a4e89
Fix a couple of inconsistencies in the public interface ( #1102 )
...
* Create `_NOT_A_REQUEST` when needed. Currently, modifications
in the raised exception would be "global".
* `retries` parameters were actually attempts. This has been fixed
to actually be the amount of retries, so 0 now means don't retry.
* Helper function to deal with retries instead of using a range with
different styles every time.
2019-02-06 19:41:45 +01:00
Lonami Exo
ae4d4ba3ef
Fix-up missing loggers from f271316
2019-01-12 12:15:29 +01:00
Tulir Asokan
f271316d7d
Make logger fully configurable ( #1087 )
2019-01-11 15:52:30 +01:00
Lonami Exo
95cf873bad
Don't mark the user as connected until successfull connection
...
The idea behind distinguishing between the user being connected and
the actual connection having been made was to support retries while
the user decided to connect, even if the connection wasn't made yet.
The problem is that _user_connected is used directly to tell whether
the sender is connected or not which will be wrong if an exception
occurs on the initial connection. Since the logic retry isn't used
there we can simply mark as connected once a successfull connection
is made.
2019-01-02 10:53:10 +01:00
Lonami Exo
d2ac7e5b0a
Actually perform all checks in 2fa
2018-12-24 14:16:50 +01:00
Sister Midnight
f3013c6817
Fix ConnectionHttp SSL socket wrap ( #1064 )
2018-12-10 14:43:48 +01:00
Lonami Exo
c48d41d99d
Expect ConnectionError in the send loop
2018-12-06 16:27:09 +01:00
Lonami Exo
10b9b4b969
Handle TimeoutError on automatic reconnect
2018-11-29 13:11:34 +01:00
Lonami Exo
d2e995ef95
Change code to recv and handle disconnections
2018-11-24 20:53:28 +01:00
Lonami Exo
f046d1f0a3
Fix deadlock on unexpected disconnections
2018-11-10 11:23:19 +01:00
Lonami Exo
cdbd1f6193
Fix valid auth_key never being saved after switching DC
2018-11-03 18:53:26 +01:00
Lonami Exo
67c5572d7b
Possibly fix possible deadlock
2018-10-28 11:53:28 +01:00
painor
45999001be
Added retry_delay parameter for auto-reconnection ( #1031 )
2018-10-28 10:55:58 +01:00
Lonami Exo
b0e587c03d
Except expected ConnectionError in the receive loop
2018-10-26 10:36:33 +02:00
Lonami Exo
091180b32d
Except CancelledError in MTProtoSender send/recv loop
...
This way, the tasks should not end with the cancelled exception
which should get rid of the warning if any.
2018-10-25 15:54:25 +02:00
Lonami Exo
4562ba9ccf
Fix auth_key not actually being saved
...
This would make the user have to login every time.
2018-10-22 20:58:07 +02:00
Lonami Exo
7dece209a0
Cancel tasks on reconnect instead of awaiting them
...
This prevents us from locking forever on any task that doesn't
rely on cancellation tokens, in this case, Connection.recv()'s
_recv_queue.get() would never complete after the server closed
the connection.
Additionally, working with cancellation tokens in asyncio is
somewhat annoying to do.
Last but not least removing the Connection._disconnected future
avoids the need to use its state (if an exception was set it
should be retrieved) to prevent asyncio from complaining, which
it was before.
2018-10-21 16:20:05 +02:00
Lonami Exo
f2e77f4030
Fix using the wrong logger in Connection
2018-10-21 16:10:09 +02:00
Lonami Exo
0686ec4440
Implement after_msg_id back
2018-10-19 16:53:50 +02:00
Lonami Exo
9dc4009152
Handle the right errors
2018-10-19 14:41:50 +02:00
Lonami Exo
b93b01cb02
Fix manual BadMessageError instantiation
2018-10-19 14:02:20 +02:00
Lonami Exo
6b280dc3bd
Clarify MTProtoSender._disconnected
2018-10-19 14:01:03 +02:00
Lonami Exo
945b34b103
Set MTProtoSender.auth_key on its creation
2018-10-19 13:56:11 +02:00
Lonami Exo
8563b9560d
Fix auth_key not being set
2018-10-19 13:35:29 +02:00
Lonami Exo
939854a0dd
Fix-up e2fe3eb
(packer was OK, encryption was not made)
2018-10-19 13:33:24 +02:00
Lonami Exo
f5bc952309
Don't set the disconnected flag twice
2018-10-19 13:25:00 +02:00
Lonami Exo
e2fe3eb503
Use new broken MessagePacker
2018-10-19 13:24:52 +02:00
Lonami Exo
6d30a38316
Let Connection._disconnected be a proper Future
...
This means that awaiting on disconnect will properly raise errors,
allowing to differentiate clean disconnects from faulty ones.
2018-10-19 10:46:34 +02:00
Lonami Exo
542d0f539b
Revisit connection and properly detail the class' intent
2018-10-19 10:35:22 +02:00
Lonami Exo
9cbc088b76
Make disconnect synchronous
...
This also fixes a bug when auto-reconnecting in MTProtoSender.
2018-10-16 11:56:17 +02:00
Lonami Exo
99129daeee
Show the request that caused RPC errors in their messages
2018-10-15 18:35:51 +02:00
Lonami Exo
6e77f583f1
Detect arbitrary negative HTTP error codes
2018-10-12 19:47:40 +02:00
Lonami Exo
ded24db3dd
Fix proxy support by removing conflicting arguments
2018-10-06 20:56:09 +02:00
Lonami Exo
fb9660afe0
Add missing awaits
2018-10-06 12:39:07 +02:00
josephbiko
3dd8b7c6d1
Support async def in sessions ( #1013 )
2018-10-05 20:25:49 +02:00
Lonami Exo
dc77136453
Don't expect responses from ack, log send errors, remove TODOs
2018-10-05 13:26:44 +02:00
Lonami Exo
ef60ade647
Rewrite container packing to support arbitrary sizes
2018-10-05 13:26:44 +02:00
Lonami Exo
7e7bbcf4c0
Remove irrelevant TODOs and add more logging
2018-10-05 13:26:44 +02:00
Lonami Exo
a5d4e97922
Fix alternative connection modes
2018-10-05 13:26:44 +02:00
Lonami Exo
ebde3be820
Add support for proxy again
2018-10-05 13:26:44 +02:00
Lonami Exo
db83709c6b
Support connection timeout
2018-10-05 13:26:44 +02:00
Lonami Exo
1b9d6aac06
Gzip only content related data
2018-10-05 13:26:44 +02:00
Lonami Exo
37b9922f64
Handle cancellation on the receive loop
2018-10-05 13:26:44 +02:00
Lonami Exo
e319fa3aa9
Handle IncompleteReadError and InvalidChecksumError
2018-10-05 13:26:44 +02:00
Lonami Exo
3fd7c33127
Fix automatic reconnect (e.g. on bad auth key)
...
This took more time than it should have to debug.
2018-10-05 13:26:44 +02:00
Lonami Exo
bc1fd9039d
Handle receiving errors
2018-10-05 13:26:44 +02:00
Lonami Exo
cf7e5d5592
Set auth_key on connection
2018-10-05 13:26:44 +02:00
Lonami Exo
21ffa2f26b
Fix DC migration and seqno
2018-10-05 13:26:44 +02:00
Lonami Exo
ac567ebf1d
Handle bad salt/msg not resending containers
2018-10-05 13:26:44 +02:00
Lonami Exo
3b1142aaca
Add back auth key generation process
2018-10-05 13:26:44 +02:00
Lonami Exo
5edc2216c7
Handle initial connection if network is down correctly
2018-10-05 13:26:44 +02:00
Lonami Exo
2d275989cb
Properly handle cancellation in _ReadyQueue
2018-10-05 13:26:44 +02:00
Lonami Exo
105bd52eee
Abstract the send queue off MTProtoSender
2018-10-05 13:26:44 +02:00
Lonami Exo
b02ebcb69b
Stop waiting for send items on disconnection
2018-10-05 13:26:44 +02:00
Lonami Exo
470fb9f5df
Make use of the MTProtoLayer in MTProtoSender
2018-10-05 13:26:44 +02:00
Lonami Exo
9402b4a26d
Create a new layer to lift encryption off the MTProtoSender
2018-10-05 13:26:44 +02:00
Lonami Exo
5daad2aaab
Actually use the new connection class
2018-10-05 13:26:44 +02:00
Lonami Exo
daf94e416b
Fix connection never receiving and missing clone method
2018-10-05 13:26:44 +02:00
Lonami Exo
2fd51b8582
Let all connection modes implement the new Connection
2018-10-05 13:26:44 +02:00
Lonami Exo
096424ea78
Create a new Connection class to work through queues
2018-10-05 13:26:44 +02:00
Lonami Exo
fbc46bd388
Use loop.create_future instead asyncio.Future
2018-08-21 12:22:06 +02:00
Lonami Exo
47190d7d55
Fix event loop not being passed into many asyncio calls
2018-08-21 11:31:14 +02:00
Lonami Exo
f0cd1fdd6e
Don't disconnect sender on cancellation ( #958 )
...
This was causing a "race condition" where the sender would
be disconnected after cancellation when another was created.
2018-08-20 14:59:17 +02:00
Lonami Exo
171e13daa9
Special case File responses without parent request ( #658 )
2018-08-20 11:42:51 +02:00
Lonami Exo
f4b9c9d6d4
Remove critical code from assert statements
2018-08-14 19:14:13 +02:00
Lonami Exo
a688c8a4ce
Fix exception ignored warnings and invalid state #935
2018-08-04 15:06:11 +02:00
Lonami Exo
7729a2a78f
More logging for bad messages ( #907 )
2018-07-25 12:33:12 +02:00
Lonami Exo
b3990546eb
Fix RPCError may occur for no parent message ( #908 )
2018-07-25 12:19:28 +02:00
josephbiko
e6981e7676
Remove empty except ( #887 )
2018-07-09 20:54:43 +02:00
Lonami Exo
7b6e65a7a5
Re-export ConnectionHttp from the network package
2018-07-08 17:48:38 +02:00
Lonami
128053750d
Implement HTTP(S) mode ( closes #112 ) ( #883 )
2018-07-08 17:45:49 +02:00
Lonami Exo
393e1966c7
Avoid exceeding maximum container size
...
This issue would likely be triggered when automatically
merging multiple requests into a single one while having
their size exceed 1044456 bytes like SaveFilePartRequest.
This commit avoids such issue by keeping track of the
current size, and if it exceeds the limit, avoid merge.
2018-07-07 11:58:48 +02:00
Lonami Exo
33ce702ab9
Pre-pack outgoing TLMessage
...
This has several benefits:
- The message can be resent without re-calling bytes(),
which for some requests may be expensive.
- Converting requests to bytes early lets us detect
errors early, such as OverflowError on bad requests.
- Containers can't exceed 1044456 bytes so knowing their
length is important. This can now be done in O(1).
But also several drawbacks:
- If the object is modified the bytes won't reflect this.
This isn't an issue because it's only done for in msgs.
- Incoming messages can no longer be reconverted into
bytes but this was never needed anyway.
2018-07-07 11:46:21 +02:00
Lonami Exo
fb40e7b508
Update mtprotoplainsender.py asserts
2018-06-28 09:48:03 +02:00
Lonami Exo
a6782ac6ea
Send getState upon successful reconnection
2018-06-27 19:40:32 +02:00
Lonami Exo
18f06f1a25
Fix bad notification due to wrong system clock never ending
2018-06-27 19:04:33 +02:00
Lonami Exo
4be248932d
Fix non-ending awaits due to requests not being resent on reconnect
2018-06-27 10:26:26 +02:00
Lonami Exo
58aadce5ba
Fix-up 2b090f8
not actually resaving new auth_keys
2018-06-27 10:15:59 +02:00
Lonami Exo
19900a373a
More useful messages from authenticator.py
2018-06-26 11:32:09 +02:00
Lonami Exo
59f6b75391
Stop showing "data left after" warning
2018-06-25 12:54:33 +02:00
Lonami Exo
184424957d
Create a separate method to start reconnections
2018-06-24 10:44:31 +02:00
Lonami Exo
a99fce471a
Fix cancelling a None handle on first connection+error
2018-06-21 09:48:11 +02:00
Lonami Exo
4ab5c7d92d
Fix reconnect sentinel when not reconnecting in send_loop
...
It would cause issues with the debug logs, since it assumes it's
a TLMessage, which have an .obj attribute. Second, the send_loop
is also smarter regarding reconnects (since OSError often occur
due to not being connected at all, namely ConnectionResetError).
2018-06-20 18:19:57 +02:00
Lonami Exo
ad1ab64415
Save last ack to resend in case of bad salt
2018-06-20 11:12:04 +02:00
Lonami Exo
463847ad50
Revisit and remove a few TODOs
2018-06-18 18:46:06 +02:00
Lonami Exo
ebfe8ebf40
Configurable per-client request/connection retries
2018-06-18 18:13:00 +02:00
Lonami Exo
09ea1179ca
Except msg_id KeyError on bad salt/msg
2018-06-18 17:20:31 +02:00
Lonami Exo
3648f7c756
Log outgoing message types and incoming message IDs
2018-06-18 17:14:04 +02:00
Lonami Exo
51d8ea0fa8
Shield disconnected Future in the MTProtoSender instead
2018-06-18 15:55:24 +02:00
Lonami Exo
7641ba2d32
Fix sleep on the wrong indent and handle CancelledError
...
This was causing a huge slow-down every time a request was
received delaying all the subsequent incoming messages too
2018-06-18 13:29:12 +02:00
Lonami Exo
374f5e2aab
Periodically send ping/getState
2018-06-18 13:22:25 +02:00
Lonami Exo
94d6c339c4
Add more logging in the MTProtoSender
2018-06-18 11:29:54 +02:00
Lonami Exo
e36aff51dd
Raise ConnectionError when sending without connection
...
Otherwise the program could potentially block on await client(...)
since this has no timeout (unless asyncio.wait_for() was used) and
would give no further hint that the program was halted there.
2018-06-17 20:25:22 +02:00
Dan Elkouby
d5b349e031
Implement a mechanism to notify of connection failures ( #852 )
2018-06-17 19:29:41 +02:00
Lonami Exo
d9d586171f
Fix MTProtoSender should not send content-related queries
2018-06-17 16:23:22 +02:00
Lonami Exo
226c35ff8f
Pump up default timeout from 5 to 10s
2018-06-17 11:46:56 +02:00