Library
Module
Module type
Parameter
Class
Class type
type error = Sendmail.error
val sendmail :
?encoder:(unit -> bytes) ->
?decoder:(unit -> bytes) ->
hostname:'a Domain_name.t ->
?port:int ->
domain:Colombe.Domain.t ->
authenticator:X509.Authenticator.t ->
?authentication:Sendmail.authentication ->
Colombe.Reverse_path.t ->
Colombe.Forward_path.t list ->
(unit -> (string * int * int) option Lwt.t) ->
(unit, error) Stdlib.result Lwt.t
sendmail ~hostname ?port ~domain ~authenticator ?authentication sender
recipients mail
where:
hostname
is the hostname of the peerport
the port of the SMTP peerdomain
is the domain of the sender (probably localhost
)authenticator
is the TLS authenticatorauthentication
is the username and the password of the usersender
is the senderrecipients
are recipients of the emailmail
stream of the mailThe connection already start a TLS connection to the peer. The peer is probably available on *:465
(the default of port
argument). The mail
stream must emit for each chunk a CRLF at the end (a line). As an user of GMail, the call of sendmail
looks like:
open Mrmime
let my_domain = Colombe.Domain.of_string_exn (Unix.gethostname ())
let my_authentication =
{
Sendmail.username = "my_login";
Sendmail.password = "my_password";
Sendmail.mechanism = Sendmail.PLAIN;
}
let sender =
let open Mrmime.Mailbox in
let v =
Local.[ "my"; "address"; "mail" ] @ Domain.(domain, [ "gmail"; "com" ])
in
Result.get_ok (Colombe_emile.to_reverse_path v)
(* "my.address.mail@gmail.com" *)
let destination =
let open Mrmime.Mailbox in
let v = Local.[ "to"; "joe" ] @ Domain.(domain, [ "gmail"; "com" ]) in
Result.get_ok (Colombe_emile.to_forward_path v)
(* "to.joe@gmail.com" *)
let run () =
sendmail
~hostname:Domain_name.(host_exe (of_string_exn "gmail.com"))
~domain:my_domain ~authentictor ~authentication sender [ destination ]
mail
let () =
match Lwt_main.run (run ()) with
| Ok () -> ()
| Error err -> Format.eprintf "%a.\n%!" Sendmail.pp_error err
sendmail
does not strictly depend on mrmime
or emile
. However, we advise to use them to produce typed and well-formed mails. sendmail
does not handle properly contents of mails as are. It just emits the stream to the pipeline directly without any changes if the line does not start with a dot ("."
). Otherwise, it prepends the line with a new dot (which has a signification in terms of SMTP).
We assume that each call of mail ()
gives to us a line - something which ends up with CRLF ("\r\n"
). By this way, we can sanitize the dot character - and only on this way.
mrmime
ensures to make on its way a stream which emits line per line. A non-user of mrmime
should be aware about this assumption.
sendmail
starts by itself a TLS connection with the SMTP server.
The user is able to re-use pre-allocated Colombe.Encoder.encoder
and Colombe.Decoder.decoder
if it wants - note that these resources can not be shared concurrently. These resources can be huge (see Colombe.Encoder.io_buffer_size
/Colombe.Decoder.io_buffer_size
) and in a server context, it can be more appropriate to pre-allocate these resources and give them then to sendmail
. By this way, the whole process will allocate only minor words.
val sendmail_with_starttls :
?encoder:(unit -> bytes) ->
?decoder:(unit -> bytes) ->
?queue:(unit -> (char, Stdlib.Bigarray.int8_unsigned_elt) Ke.Rke.t) ->
hostname:'a Domain_name.t ->
?port:int ->
domain:Colombe.Domain.t ->
authenticator:X509.Authenticator.t ->
?authentication:Sendmail.authentication ->
Colombe.Reverse_path.t ->
Colombe.Forward_path.t list ->
(unit -> (string * int * int) option Lwt.t) ->
(unit, Sendmail_with_starttls.error) Stdlib.result Lwt.t
sendmail_with_starttls
is sendmail
but a part of the communication is insecure. Usually, a SMTP service provides 2 submission services:
*:465
.STARTTLS
) on *:587
.The user should use the first one but in the context of the non-existence of it, the second one is available. Usage and arguments are the same. However, default value of port
is the default value of your operating system (see Unix.getprotobyname
).