[neomutt-devel] any interest in neomutt headless or as a library?

[Mike Marchywka]

Maybe just to list the uses of the hacked up neomuttl, I can see what
problems others have run into with similar tools and maybe people have
other suggestions.

1) Download entire imap foldes or accounts. I did this with libetpan without 
incident although I could not find a decent utility program for that. I can open
a larger number of threads although it looks like hotmail will only accept
about 5 connections and keep track of which uids are obtained. The "successfile"
is then used for a retry until all messages appear to be downloaded. 

2) Data exchange for systems without static IP or other limitations. Right now, I have
a laptop connected to cable modem that needs to operate unattended with for example
a motion detecting camera. I have a system now to interact via a free imap account
that produces acceptable latency on the order of minutes. I wanted to replace fetchmail
and the scripts with mostly c.c++ code including the hacked up neomutt. 

3) Data collection which may already work with some other mail processing packages.
Often the data is manually entered and subject to error. The automated email processor needs to try to parse it
and update a file or DB but maybe after user interaction via an email exchange. I don't have
this working yet but it looks like neomutt can now make that happen. with less effort  There may be parts
of this in email list processing code but I have not looked.

4) Maybe play with latex format/mime email as an alt to html. Not sure where this stands but html
is hard to read in source format and latex would let you do that while creating a somewhat structured
document allowing the reader to display as desired more easily than html. 

note new address
 Mike Marchywka 306 Charles Cox Drive Canton, GA 30115
 2295 Collinworth  Drive Marietta GA 30062.  formerly 487 Salem Woods Drive Marietta GA 30067 404-788-1216 (C)<- leave message 989-348-4796 (P)<- emergency


________________________________________
From: neomutt-devel <neomutt-devel-bounces at neomutt.org> on behalf of Mike Marchywka <marchywka at hotmail.com>
Sent: Monday, July 20, 2020 10:28 AM
To: Neomutt Developers
Subject: Re: [neomutt-devel] any interest in neomutt headless or as a library?


sorry, wrong reply address :)
It looks like the indentations got pulled from the code but
otherwise ok. I use a lot of my own macros and typedefs
in the c++ code it may not be clear.

I've been debating about dumping to github or something
but there is a lot of bad code in the overall code base :)

Curious though if people have specific issues with fetchmail
or mail processing code. I'm just writing my own because
I originally wanted to download my hotmail but now it
is nice to adapt for using email exchanges for remote data
transfer and control and monitoring.


________________________________________
From: Mike Marchywka <marchywka at hotmail.com>
Sent: Monday, July 20, 2020 9:30 AM
To: Richard Russon
Subject: Re: [neomutt-devel] any interest in neomutt headless or as a library?

On Mon, Jul 20, 2020 at 01:59:59PM +0100, Richard Russon wrote:
> Hi Mike,
>
> > I'm trying to replace "fetchmail" and integrate with my own mail processing logic.
>
> Sounds fun
>
> > any interest in neomutt headless or as a library?
>
> Not directly, but I _have_ spent over four years breaking apart
> dependencies so that this might be possible.
>
> Upstream Mutt's code has two libraries.
> NeoMutt has twenty-six :-)
>
> > I ... hacked it up to implement partial functionality without curses ...
>
> See: https://github.com/neomutt/test-library
> Especially: test_conn.c and test_smtp.c
>
> They link against the libraries, without the need for NeoMutt's main().
>
> > This was easy

I wanted to preserve main for several reasons, the library does
actually require it be renamed but IIRC main.c was the only file
to change for that build due to the rename.

I changed the observers and tacked this in when in my mode,

     struct MuttWindow *dlg = index_pager_init();
      notify_observer_add(NeoMutt->notify, mjm_dlg_index_observer, dlg);
if (mjm_init_only) return 0;
mjm_input_loop();
}
 return shutdown_loop(&cls,&mml); //   cls.rc = 0;


For external usage, I have this,

int mutt_main_server(const char * config, char * envp[])
{
int argc=4;
char ** argv =malloc(4*sizeof(char*));
argv[0]="neomutt";
argv[1]="-F";
argv[2]=config;
argv[3]="-M";
#ifdef MJM_NEOMUTT_LIB
int rc= main_lib(argc,argv,envp);
#else
int rc= main(argc,argv,envp);
#endif

free(argv);
return rc;

}


>
> You're welcome :-)
>
> > I had to hack up main() pretty significantly ...
>
> Urgh.  main() *is* still a mess.
> NeoMutt always does _some_ GUI initialisaion because there are still
> some GUI dependencies deep within the code.  We're working on it, though.

I ran into this trying to update the read flag on an imap server,

diff index.c ../orig/index.c
3958d3957
< if (!Colors) return;


>
> > ... not start curses, but then drop into my own readline based
> > command interpreter ... only good for debugging.
>
> Sounds useful
>
> > linking the shared object to my c++ code "mikemail" it appeared
> > to read and send from an imap account
>
> That's great
>
> So... Let's see the code.
>
> If it looks useful, are you prepared to help integrate it?
> (I'll happily make time to mentor)

It will take some cleanup to distribute but if I know someone is interested
I can start on that. Right now I'm letting this sit for a little but
you can see how I called it from my c++ app,
( note a lot of macro and utility classes but the flow should
be apparent ),

void cmd_mutt_proc(Cip & cip , LocalVar & lv )
{
const StrTy config=cip.p1;
const IdxTy flags=myatoi(cip.p2);
const bool unread_only=Bit(flags,0);
const bool reset_on_fail=Bit(flags,1);
const bool server_update=!Bit(flags,2);
const StrTy fn=cip.wif(3);
MM_ERR(MMPR3(config,flags,fn))
if (cip.help())
{
Ss ss;
ss<<cip.cmd()<<" cmd_mutt_proc "<<CRLF;

cip.help(ss.str());
return;
}
MyMsgMap & m=m_messages;
m.clear();
MuttInterface mi;
mutt_set_init_only(true);
int i=0;
int rc=mutt_main_server(config.c_str(),0);
if (rc!=0)
{
MM_ERR(" failed to open mutt "<<MMPR(rc))
return;
}
struct Email ** elist= mutt_email_list(unread_only?1:0);
if (elist!=0)
{
std::map<IdxTy,IdxTy> serial;
while (elist[i]!=0)
{
MyMsg msg;
// StrTy account(){ return  m_user+StrTy(";")+m_host+StrTy(";")+m_folder; }

mi.convert(msg,elist[i]);
serial[elist[i]->index]=i;
m[elist[i]->index]=msg;
++i;
} // i
IdxTy mflags=0;
/MyMsgMap & m=m_messages; // def?m_messages:m_messages_map[mbox];
MailProc & mp= m_proc_map[fn];
//proc_msg_map(mp,mmm);
MM_SZ_LOOP(k,m,szm)
{
const int idx=elist[k]->index;
const MyMsg & msg=m[idx];
//MM_ERR(MMPR2(i,msg.verbatim()))
MM_ERR(MMPR4(szm,k,idx,msg.verbatim().size()))
const IdxTy prc=mp.proc(msg,mflags);
MM_ERR(MMPR(prc))
if (server_update) {
if (prc==0) set_read_flag(elist[k]);
else if (reset_on_fail) reset_read_flag(elist[k]);
}
char * subj="subect";
char * text=" message text";
char ** lto= new char*[2];
lto[0]="marchywka at hotmail.com";
lto[1]=0;
// the reply function fails but sending new messag ok
//int rcs=mjm_send_via_mutt(lto,subj, text,0,elist[k]);
int rcs=mjm_send_via_mutt(lto,subj, text,0,0);

MM_ERR(MMPR(rcs))


}



}
MM_ERR("free "<<MMPR(i))
mutt_mjm_free(elist);
shutdown_mutt();
MM_ERR(MMPR(i))
} // cmd_mutt_proc


>
> Cheers,
>     Rich / FlatCap



--

mike marchywka
306 charles cox
canton GA 30115
USA, Earth
marchywka at hotmail.com
404-788-1216
ORCID: 0000-0001-9237-455X