Attention! v0.166.2 has some important changes, check them out.
It is a pure-Golang MTProto client for Telegram API.
Initially based on https://github.com/sdidyk/mtproto and https://github.com/ronaksoft/mtproto.
TL layer: 192
Consists of two parts:
You may use tg-history-dumper repo as an example, in particular this file: https://github.com/3bl3gamer/tg-history-dumper/blob/master/tg.go#L67
First, TGClient
object should be created with tgclient.NewTGClient
or tgclient.NewTGClientExt
for example:
tg := tgclient.NewTGClient(appID, appHash, &mtproto.SimpleLogHandler{})
And if you need more customization:
// Telegram client info
cfg := &mtproto.AppConfig{
AppID: appID,
AppHash: appHash,
AppVersion: "0.0.1",
DeviceModel: "Unknown",
SystemVersion: runtime.GOOS + "/" + runtime.GOARCH,
SystemLangCode: "en",
LangPack: "",
LangCode: "en",
}
// store that will save/read session data
sessStore := &mtproto.SessFileStore{FPath: "tg_session.json"}
// optional dialer (may be nil) for proxying requests
dialer, err := proxy.SOCKS5("tcp", "127.0.0.1:9050", nil, proxy.Direct)
tg := tgclient.NewTGClientExt(cfg, sessStore, logHandler, dialer)
Then, the connection should be opened:
err := tg.InitAndConnect()
And client should authenticate with AuthAndInitEvents
or AuthExt
, for example:
// object that will... provide auth data:
// phone number, confirmation code and optionally password.
// ScanfAuthDataProvider will just prompt all this from terminal.
authDataProvider := mtproto.ScanfAuthDataProvider{}
err := tgclient.AuthAndInitEvents(authDataProvider)
While authing, AuthAndInitEvents
sends mtproto.TL_updates_getState
request. Same request will also be sent after each reconnection. It makes TG server send updates to client (like new incoming messages). If you do not need those (maybe you just want to dump your chats history), you may send something different:
authDataProvider := mtproto.ScanfAuthDataProvider{}
payload := mtproto.TL_users_getUsers{ID: []mtproto.TL{mtproto.TL_inputUserSelf{}}}
resp, err := tg.AuthExt(authDataProvider, payload)
// here resp should be a mtproto.VectorObject with one mtproto.TL_user item.
With tg.SendSync
or tg.SendSyncRetry
. First one just sends request and returns response whatever it will be, so you generally should check if you got what you expected. For example:
res := c.tg.SendSync(mtproto.TL_contacts_resolveUsername{Username: "some chat name"})
peer, ok := res.(mtproto.TL_contacts_resolvedPeer)
if !ok {
// res will likely be TL_rpc_error with message USERNAME_NOT_OCCUPIED or RPC_CALL_FAIL or other
return mtproto.WrongRespError(res)
}
for _, chat := range peer.Chats {
fmt.Printf("%#v\n", chat)
}
Often you will receive temporary errors like RPC_CALL_FAIL
of FOOLD_WAIT_123
and want to re-send same request after little delay. This is done by
res := tg.SendSyncRetry(request, time.Second, 0, 30*time.Second)
Pass callback func:
tg.SetUpdateHandler(func(updateTL mtproto.TL) {
switch update := updateTL.(type) {
case mtproto.TL_updateUserStatus:
fmt.Printf("U#%d %T\n", update.UserID, update.Status)
case mtproto.TL_updateNewChannelMessage:
...
case mtproto.TL_updateEditChannelMessage:
...
...
default:
return mtproto.WrongRespError(updateTL)
}
}
})
Get new schema from https://core.telegram.org/schema (remove definitions for int
, long
, double
, string
, vector
, int128
and int256
: they are hard-coded and must not be generated). If it is still outdated check other repos (like official ones), some useful links are at the top of generate_tl_schema.go.
Place new .tl
file to mtproto/scheme
folder.
Update //go:generate
command in mtproto/mtproto.go
. It should be
//go:generate go run scheme/generate_tl_schema.go <layer> scheme/<file>.tl tl_schema.go
Run go generate
in mtproto
folder.
Update this README.
TL_invokeWithLayer
in Connect()
, two TL_invokeWithLayer
may be sent. Nothing bad happens though.