ok, everything fetches ok.. except bad utf8 in subjects

master
Vitaliy Filippov 2016-06-27 22:11:34 +03:00
parent 84e801d31e
commit c152ffa122
3 changed files with 75 additions and 26 deletions

2
db.sql
View File

@ -38,5 +38,7 @@ create table messages (
headers text not null,
body text not null,
time timestamptz not null,
flags smallint not null,
foreign key (folder_id) references folders (id) on delete cascade on update cascade
);
create index messages_messageid on messages (messageid);

View File

@ -11,22 +11,32 @@ var inspect = require('util').inspect;
var pg = require('pg-bricks').configure('postgresql://'+cfg.pg.user+':'+cfg.pg.password+'@'+(cfg.pg.host||'')+':'+cfg.pg.port+'/'+cfg.pg.database);
var flagNum = {
'\\recent': 1,
'\\flagged': 2,
'\\answered': 4,
'\\seen': 8,
};
function splitEmails(s)
{
var re = /(((['"])((?:[^'"\\]+|\\["'\\])*)\3|[^"'<,]+)\s*)?<([^>,]*)>/g, m;
var r = [];
var re = /^[\s,]*(?:(?:["'](.*?)["']|([^<]+))\s*<([^>]+)>|<?([^<>]+)>?)/; // '
var m, r = [];
while (m = re.exec(s))
r.push({ name: m[4]||m[2], email: m[5] });
{
s = s.substr(m[0].length);
r.push({ name: (m[1]||m[2]||'').trim(), email: (m[3]||m[4]||'').trim() });
}
return r;
}
function* main(thread, account)
{
var accountId;
var [ err, row ] = yield pg.select('id').from('accounts').where({ email: account.email }).row(thread);
var [ err, rows ] = yield pg.select('id').from('accounts').where({ email: account.email }).rows(thread);
if (err) throw err;
if (row && row.id)
accountId = row.id;
if (rows[0] && rows[0].id)
accountId = rows[0].id;
else
{
var [ err, row ] = yield pg.insert('accounts', {
@ -46,17 +56,30 @@ function* main(thread, account)
for (var k in boxes)
{
var [ err, box ] = yield srv.openBox(k, true, thread);
var boxrow = {
name: box.name,
uidvalidity: box.uidvalidity,
account_id: 1
};
console.log(box);
if (err) throw err;
var boxId;
var [ err, rows ] = yield pg.update('folders', { uidvalidity: box.uidvalidity, unread_count: box.messages.new })
.where({ account_id: accountId, name: box.name }).returning('id').rows(thread);
if (err) throw err;
if (rows[0] && rows[0].id)
boxId = rows[0].id;
else
{
var [ err, row ] = yield pg.insert('folders', {
name: box.name,
uidvalidity: box.uidvalidity,
account_id: accountId,
unread_count: box.messages.new,
// total_count: box.messages.count
}).returning('id').row(thread);
if (err) throw err;
boxId = row.id;
}
var f = srv.fetch('1:*', {
size: true,
bodies: 'HEADER.FIELDS (FROM TO CC SUBJECT DATE MESSAGE-ID IN-REPLY-TO)'
bodies: 'HEADER'
});
f.on('message', function(msg)
f.on('message', function(msg, seqnum)
{
gen.run(function*(thread)
{
@ -68,20 +91,43 @@ function* main(thread, account)
buffer += chunk.toString('utf8');
});
yield stream.once('end', thread);
var header = Imap.parseHeader(buffer);
header.from = splitEmails(header.from && header.from[0]);
msgrow.from = header.from && header.from[0];
msgrow.to = header.to && header.to[0];
msgrow.cc = header.cc && header.cc[0];
msgrow.subject = header.subject && header.subject[0];
msgrow.date = header.date && header.date[0];
msgrow.messageid = header['message-id'] && header['message-id'][0];
msgrow.inreplyto = header['in-reply-to'] && header['in-reply-to'][0];
msgrow.headers = buffer;
msgrow.body = '';
var header = Imap.parseHeader(msgrow.headers);
for (var i in header)
for (var k = 0; k < header[i].length; k++)
header[i][k] = header[i][k].replace(/\x00/g, '');
header.from = header.from && splitEmails(header.from[0])[0];
msgrow.from_email = header.from && header.from.email || '';
msgrow.from_name = header.from && header.from.name || '';
header.replyto = header['reply-to'] && splitEmails(header['reply-to'][0])[0];
msgrow.replyto_email = header.replyto && header.replyto.email || '';
msgrow.replyto_name = header.replyto && header.replyto.name || '';
msgrow.to_list = header.to && header.to[0] || '';
msgrow.cc_list = header.cc && header.cc[0] || '';
msgrow.bcc_list = header.bcc && header.bcc[0] || '';
msgrow.subject = header.subject && header.subject[0] || '';
msgrow.messageid = header['message-id'] && header['message-id'][0] || '';
msgrow.inreplyto = header['in-reply-to'] && header['in-reply-to'][0] || '';
var [ attrs ] = yield msg.once('attributes', thread);
msgrow.date = msgrow.date || attrs.date;
if (header.date)
{
var t = Date.parse(header.date[0]);
if (!isNaN(t))
msgrow.time = new Date(t);
}
if (!msgrow.time)
msgrow.time = new Date(attrs.date);
msgrow.uid = attrs.uid;
msgrow.folder_id = boxId;
msgrow.flags = 0;
for (var i = 0; i < attrs.flags.length; i++)
msgrow.flags = msgrow.flags || flagNum[attrs.flags[i].toLowerCase()];
msgrow.flags = (msgrow.flags & ~8) | (msgrow.flags & 8 ? 0 : 8); // invert "\seen" (unread) flag
yield msg.once('end', thread);
console.log(msgrow);
console.log(msgrow.time+' '+msgrow.from_email+' '+msgrow.subject);
var [ err, row ] = yield pg.insert('messages', msgrow).run(thread);
if (err) throw err;
});
});
yield f.on('end', thread);

3
run.sh
View File

@ -1 +1,2 @@
node_modules/.bin/babel operetta.js | nodejs
node_modules/.bin/babel operetta.js > operetta.c.js
nodejs operetta.c.js