trabajo-afectivo/app/models/channel/email_parser.rb

647 lines
19 KiB
Ruby
Raw Normal View History

2014-02-03 19:23:00 +00:00
# Copyright (C) 2012-2014 Zammad Foundation, http://zammad-foundation.org/
2012-06-15 11:10:23 +00:00
# encoding: utf-8
require 'mail'
2012-12-05 01:38:30 +00:00
require 'encode'
2012-12-05 00:28:04 +00:00
class Channel::EmailParser
=begin
2015-12-14 09:23:14 +00:00
mail = parse(msg_as_string)
mail = {
from: 'Some Name <some@example.com>',
from_email: 'some@example.com',
from_local: 'some',
from_domain: 'example.com',
from_display_name: 'Some Name',
message_id: 'some_message_id@example.com',
to: 'Some System <system@example.com>',
cc: 'Somebody <somebody@example.com>',
subject: 'some message subject',
body: 'some message body',
attachments: [
2012-10-04 07:09:27 +00:00
{
data: 'binary of attachment',
filename: 'file_name_of_attachment.txt',
preferences: {
content-alternative: true,
Mime-Type: 'text/plain',
Charset: 'iso-8859-1',
2012-10-04 07:09:27 +00:00
},
},
],
2012-10-04 07:09:27 +00:00
# ignore email header
x-zammad-ignore: 'false',
2012-10-04 07:09:27 +00:00
# customer headers
x-zammad-customer-login: '',
x-zammad-customer-email: '',
x-zammad-customer-firstname: '',
x-zammad-customer-lastname: '',
2012-10-04 07:09:27 +00:00
# ticket headers
x-zammad-ticket-group: 'some_group',
x-zammad-ticket-state: 'some_state',
x-zammad-ticket-priority: 'some_priority',
x-zammad-ticket-owner: 'some_owner_login',
2012-10-04 07:09:27 +00:00
# article headers
x-zammad-article-internal: false,
x-zammad-article-type: 'agent',
x-zammad-article-sender: 'customer',
2012-10-04 07:09:27 +00:00
# all other email headers
some-header: 'some_value',
}
=end
2015-12-14 09:23:14 +00:00
def parse(msg)
data = {}
2015-12-14 09:23:14 +00:00
mail = Mail.new(msg)
# set all headers
mail.header.fields.each { |field|
2015-01-12 10:55:13 +00:00
next if !field.name
2015-01-12 10:55:13 +00:00
# full line, encode, ready for storage
2015-12-14 09:23:14 +00:00
data[field.name.to_s.downcase.to_sym] = Encode.conv('utf8', field.to_s)
# if we need to access the lines by objects later again
data[ "raw-#{field.name.downcase}".to_sym ] = field
}
2013-10-31 22:56:08 +00:00
# get sender
from = nil
['from', 'reply-to', 'return-path'].each { |item|
next if !mail[ item.to_sym ]
from = mail[ item.to_sym ].value
break if from
2013-10-31 22:56:08 +00:00
}
# set x-any-recipient
data['x-any-recipient'.to_sym] = ''
['to', 'cc', 'delivered-to', 'x-original-to', 'envelope-to'].each { |item|
next if !mail[item.to_sym]
if data['x-any-recipient'.to_sym] != ''
data['x-any-recipient'.to_sym] += ', '
end
data['x-any-recipient'.to_sym] += mail[item.to_sym].to_s
}
# set extra headers
2013-12-01 17:51:35 +00:00
begin
2015-12-14 09:23:14 +00:00
data[:from_email] = Mail::Address.new(from).address
data[:from_local] = Mail::Address.new(from).local
data[:from_domain] = Mail::Address.new(from).domain
data[:from_display_name] = Mail::Address.new(from).display_name ||
(Mail::Address.new(from).comments && Mail::Address.new(from).comments[0])
2013-12-01 17:51:35 +00:00
rescue
data[:from_email] = from
data[:from_local] = from
data[:from_domain] = from
2013-12-01 17:51:35 +00:00
end
# do extra decoding because we needed to use field.value
2015-12-14 09:23:14 +00:00
data[:from_display_name] = Mail::Field.new('X-From', data[:from_display_name]).to_s
# compat headers
data[:message_id] = data['message-id'.to_sym]
# body
# plain_part = mail.multipart? ? (mail.text_part ? mail.text_part.body.decoded : nil) : mail.body.decoded
# html_part = message.html_part ? message.html_part.body.decoded : nil
2012-05-07 22:37:07 +00:00
data[:attachments] = []
2012-10-05 06:42:12 +00:00
2012-07-02 18:52:27 +00:00
# multi part email
2012-05-04 19:30:22 +00:00
if mail.multipart?
2012-10-05 06:42:12 +00:00
2012-07-02 18:52:27 +00:00
# text attachment/body exists
2012-06-15 11:10:23 +00:00
if mail.text_part
data[:body] = mail.text_part.body.decoded
2015-12-14 09:23:14 +00:00
data[:body] = Encode.conv(mail.text_part.charset, data[:body])
2012-10-05 06:42:12 +00:00
2014-08-06 11:41:10 +00:00
if !data[:body].valid_encoding?
data[:body] = data[:body].encode('utf-8', 'binary', invalid: :replace, undef: :replace, replace: '?')
2014-08-06 11:41:10 +00:00
end
end
2014-08-06 11:41:10 +00:00
# html attachment/body may exists and will be converted to text
if !mail.text_part || !data[:body] || data[:body] == ''
2012-06-15 11:10:23 +00:00
filename = '-no name-'
2015-01-08 08:26:40 +00:00
if mail.html_part && mail.html_part.body
filename = 'message.html'
data[:body] = mail.html_part.body.to_s
2015-12-14 09:23:14 +00:00
data[:body] = Encode.conv(mail.html_part.charset.to_s, data[:body])
data[:body] = data[:body].html2text.to_s.force_encoding('utf-8')
2014-08-06 11:41:10 +00:00
if !data[:body].force_encoding('UTF-8').valid_encoding?
data[:body] = data[:body].encode('utf-8', 'binary', invalid: :replace, undef: :replace, replace: '?')
2014-08-06 11:41:10 +00:00
end
2012-06-15 11:10:23 +00:00
# any other attachments
2012-06-15 11:10:23 +00:00
else
data[:body] = 'no visible content'
2012-06-15 11:10:23 +00:00
end
2012-07-02 18:52:27 +00:00
end
2012-06-15 11:10:23 +00:00
2012-07-02 18:52:27 +00:00
# add html attachment/body as real attachment
if mail.html_part
filename = 'message.html'
headers_store = {
'content-alternative' => true,
}
2012-06-15 11:10:23 +00:00
if mail.mime_type
headers_store['Mime-Type'] = mail.html_part.mime_type
end
if mail.charset
headers_store['Charset'] = mail.html_part.charset
end
attachment = {
data: mail.html_part.body.to_s,
filename: mail.html_part.filename || filename,
preferences: headers_store
2012-06-15 11:10:23 +00:00
}
data[:attachments].push attachment
end
2012-07-02 18:52:27 +00:00
# get attachments
if mail.parts
mail.parts.each { |part|
# protect process to work fine with spam emails, see test/fixtures/mail15.box
begin
2015-12-14 09:23:14 +00:00
attachs = _get_attachment(part, data[:attachments], mail)
data[:attachments].concat(attachs)
rescue
2015-12-14 09:23:14 +00:00
attachs = _get_attachment(part, data[:attachments], mail)
data[:attachments].concat(attachs)
2012-07-02 18:52:27 +00:00
end
}
end
# not multipart email
2012-05-07 22:37:07 +00:00
# text part only
elsif !mail.mime_type || mail.mime_type.to_s == '' || mail.mime_type.to_s.casecmp('text/plain').zero?
data[:body] = mail.body.decoded
data[:body] = Encode.conv(mail.charset, data[:body])
if !data[:body].force_encoding('UTF-8').valid_encoding?
data[:body] = data[:body].encode('utf-8', 'binary', invalid: :replace, undef: :replace, replace: '?')
end
2016-01-15 20:20:48 +00:00
# html part only, convert to text and add it as attachment
else
filename = '-no name-'
if mail.mime_type.to_s.casecmp('text/html').zero?
filename = 'message.html'
data[:body] = mail.body.decoded
2015-12-14 09:23:14 +00:00
data[:body] = Encode.conv(mail.charset, data[:body])
data[:body] = data[:body].html2text.to_s.force_encoding('utf-8')
2012-05-07 22:37:07 +00:00
if !data[:body].valid_encoding?
data[:body] = data[:body].encode('utf-8', 'binary', invalid: :replace, undef: :replace, replace: '?')
2014-08-06 11:41:10 +00:00
end
# any other attachments
2012-07-02 18:52:27 +00:00
else
data[:body] = 'no visible content'
end
2012-05-07 22:37:07 +00:00
# add body as attachment
headers_store = {
'content-alternative' => true,
}
if mail.mime_type
headers_store['Mime-Type'] = mail.mime_type
end
if mail.charset
headers_store['Charset'] = mail.charset
2012-05-07 22:37:07 +00:00
end
attachment = {
data: mail.body.decoded,
filename: mail.filename || filename,
preferences: headers_store
}
data[:attachments].push attachment
2012-05-04 19:30:22 +00:00
end
2012-07-02 18:52:27 +00:00
# strip not wanted chars
2015-12-14 09:23:14 +00:00
data[:body].gsub!(/\n\r/, "\n")
data[:body].gsub!(/\r\n/, "\n")
data[:body].tr!("\r", "\n")
2012-07-02 18:52:27 +00:00
# remember original mail instance
data[:mail_instance] = mail
data
end
2015-12-14 09:23:14 +00:00
def _get_attachment(file, attachments, mail)
# check if sub parts are available
if !file.parts.empty?
a = []
file.parts.each {|p|
2015-12-14 09:23:14 +00:00
attachment = _get_attachment(p, attachments, mail)
a.concat(attachment)
}
return a
end
# ignore text/plain attachments - already shown in view
return [] if mail.text_part && mail.text_part.body.to_s == file.body.to_s
# ignore text/html - html part, already shown in view
return [] if mail.html_part && mail.html_part.body.to_s == file.body.to_s
# get file preferences
headers_store = {}
file.header.fields.each { |field|
headers_store[field.name.to_s] = field.value.to_s
}
# get filename from content-disposition
filename = nil
# workaround for: NoMethodError: undefined method `filename' for #<Mail::UnstructuredField:0x007ff109e80678>
begin
filename = file.header[:content_disposition].filename
rescue
begin
result = file.header[:content_disposition].to_s.scan( /filename=("|)(.+?)("|);/i )
if result && result[0] && result[0][1]
filename = result[0][1]
end
2016-01-28 19:08:13 +00:00
rescue
Rails.logger.debug 'Unable to get filename'
end
end
# for some broken sm mail clients (X-MimeOLE: Produced By Microsoft Exchange V6.5)
if !filename
filename = file.header[:content_location].to_s
end
# generate file name
if !filename || filename.empty?
attachment_count = 0
(1..1000).each {|count|
filename_exists = false
filename = 'file-' + count.to_s
attachments.each {|attachment|
if attachment[:filename] == filename
filename_exists = true
end
}
break if filename_exists == false
}
end
# get mime type
if file.header[:content_type] && file.header[:content_type].string
headers_store['Mime-Type'] = file.header[:content_type].string
end
# get charset
if file.header && file.header.charset
headers_store['Charset'] = file.header.charset
end
# remove not needed header
headers_store.delete('Content-Transfer-Encoding')
headers_store.delete('Content-Disposition')
attach = {
data: file.body.to_s,
filename: filename,
preferences: headers_store,
}
[attach]
end
=begin
parser = Channel::EmailParser.new
ticket, article, user = parser.process(channel, email_raw_string)
retrns
[ticket, article, user]
=end
def process(channel, msg)
2015-12-14 09:23:14 +00:00
mail = parse(msg)
2012-10-05 06:42:12 +00:00
# run postmaster pre filter
filters = {
'0010' => Channel::Filter::Trusted,
'0020' => Channel::Filter::AutoResponseCheck,
'0030' => Channel::Filter::OutOfOfficeCheck,
'0100' => Channel::Filter::FollowUpCheck,
'0900' => Channel::Filter::BounceCheck,
2012-10-05 06:42:12 +00:00
'1000' => Channel::Filter::Database,
}
2012-10-04 07:09:27 +00:00
2015-12-14 09:23:14 +00:00
# filter(channel, mail)
filters.each {|_prio, backend|
2012-10-05 06:42:12 +00:00
begin
backend.run(channel, mail)
rescue => e
2015-05-04 18:58:28 +00:00
Rails.logger.error "can't run postmaster pre filter #{backend}"
Rails.logger.error e.inspect
2015-08-31 09:27:12 +00:00
raise e
2012-10-05 06:42:12 +00:00
end
}
2012-10-04 07:09:27 +00:00
2012-05-06 20:48:23 +00:00
# check ignore header
return true if mail[ 'x-zammad-ignore'.to_sym ] == 'true' || mail[ 'x-zammad-ignore'.to_sym ] == true
ticket = nil
article = nil
user = nil
# use transaction
ActiveRecord::Base.transaction do
2013-04-20 10:14:27 +00:00
# reset current_user
UserInfo.current_user_id = 1
# create sender
2012-05-06 20:48:23 +00:00
if mail[ 'x-zammad-customer-login'.to_sym ]
2015-12-14 09:23:14 +00:00
user = User.find_by(login: mail[ 'x-zammad-customer-login'.to_sym ])
2012-05-06 20:48:23 +00:00
end
if !user
2015-12-14 09:23:14 +00:00
user = User.find_by(email: mail[ 'x-zammad-customer-email'.to_sym ] || mail[:from_email])
2012-05-06 20:48:23 +00:00
end
if !user
user = user_create(
login: mail[ 'x-zammad-customer-login'.to_sym ] || mail[ 'x-zammad-customer-email'.to_sym ] || mail[:from_email],
firstname: mail[ 'x-zammad-customer-firstname'.to_sym ] || mail[:from_display_name],
lastname: mail[ 'x-zammad-customer-lastname'.to_sym ],
email: mail[ 'x-zammad-customer-email'.to_sym ] || mail[:from_email],
)
end
2012-10-04 07:09:27 +00:00
# create to and cc user
2015-01-12 10:55:13 +00:00
['raw-to', 'raw-cc'].each { |item|
next if !mail[item.to_sym]
next if !mail[item.to_sym].addrs
items = mail[item.to_sym].addrs
items.each {|address_data|
user_create(
firstname: address_data.display_name,
lastname: '',
email: address_data.address,
)
}
}
# set current user
UserInfo.current_user_id = user.id
2012-10-04 07:09:27 +00:00
# get ticket# based on email headers
if mail[ 'x-zammad-ticket-id'.to_sym ]
ticket = Ticket.find_by( id: mail[ 'x-zammad-ticket-id'.to_sym ] )
end
if mail[ 'x-zammad-ticket-number'.to_sym ]
ticket = Ticket.find_by( number: mail[ 'x-zammad-ticket-number'.to_sym ] )
end
# set ticket state to open if not new
if ticket
2015-12-14 09:23:14 +00:00
state = Ticket::State.find(ticket.state_id)
state_type = Ticket::StateType.find(state.state_type_id)
2012-08-28 05:21:45 +00:00
# if tickte is merged, find linked ticket
if state_type.name == 'merged'
2012-08-28 05:21:45 +00:00
end
# set ticket to open again
if state_type.name != 'new' && !mail[ 'x-zammad-out-of-office'.to_sym ]
ticket.state = Ticket::State.find_by(name: 'open')
ticket.save
end
end
# create new ticket
2012-05-06 20:48:23 +00:00
if !ticket
2015-12-14 09:23:14 +00:00
preferences = {}
if channel[:id]
preferences = {
channel_id: channel[:id]
}
end
ticket = Ticket.new(
group_id: channel[:group_id] || 1,
customer_id: user.id,
title: mail[:subject] || '',
2015-12-14 09:23:14 +00:00
state_id: Ticket::State.find_by(name: 'new').id,
priority_id: Ticket::Priority.find_by(name: '2 normal').id,
preferences: preferences,
)
2012-05-06 20:48:23 +00:00
2015-12-14 09:23:14 +00:00
set_attributes_by_x_headers(ticket, 'ticket', mail)
2012-05-06 20:48:23 +00:00
# create ticket
ticket.save
end
# import mail
2012-05-06 20:48:23 +00:00
# set attributes
article = Ticket::Article.new(
ticket_id: ticket.id,
2015-12-14 09:23:14 +00:00
type_id: Ticket::Article::Type.find_by(name: 'email').id,
sender_id: Ticket::Article::Sender.find_by(name: 'Customer').id,
body: mail[:body],
from: mail[:from],
to: mail[:to],
cc: mail[:cc],
subject: mail[:subject],
message_id: mail[:message_id],
internal: false,
)
2012-05-06 20:48:23 +00:00
# x-headers lookup
2015-12-14 09:23:14 +00:00
set_attributes_by_x_headers(article, 'article', mail)
2012-05-06 20:48:23 +00:00
# create article
article.save
# store mail plain
Store.add(
object: 'Ticket::Article::Mail',
o_id: article.id,
data: msg,
filename: "ticket-#{ticket.number}-#{article.id}.eml",
preferences: {}
)
# store attachments
if mail[:attachments]
mail[:attachments].each do |attachment|
Store.add(
object: 'Ticket::Article',
o_id: article.id,
data: attachment[:data],
filename: attachment[:filename],
preferences: attachment[:preferences]
)
end
end
end
# execute ticket events
2012-11-18 10:56:48 +00:00
Observer::Ticket::Notification.transaction
2012-05-06 20:48:23 +00:00
2012-10-05 06:42:12 +00:00
# run postmaster post filter
filters = {
2015-12-14 09:23:14 +00:00
# '0010' => Channel::Filter::Trusted,
2012-10-05 06:42:12 +00:00
}
# filter( channel, mail )
filters.each {|_prio, backend|
2012-10-05 06:42:12 +00:00
begin
2015-12-14 09:23:14 +00:00
backend.run(channel, mail, ticket, article, user)
rescue => e
2015-05-04 18:58:28 +00:00
Rails.logger.error "can't run postmaster post filter #{backend}"
Rails.logger.error e.inspect
2012-10-05 06:42:12 +00:00
end
}
2012-05-06 20:48:23 +00:00
# return new objects
[ticket, article, user, mail]
end
def user_create(data)
# return existing
2015-12-14 09:23:14 +00:00
user = User.find_by(login: data[:email].downcase)
return user if user
# create new user
2015-12-14 09:23:14 +00:00
roles = Role.where(name: 'Customer')
# fillup
%w(firstname lastname).each { |item|
if data[item.to_sym].nil?
data[item.to_sym] = ''
end
}
data[:password] = ''
data[:active] = true
data[:roles] = roles
data[:updated_by_id] = 1
data[:created_by_id] = 1
user = User.create(data)
user.update_attributes(
updated_by_id: user.id,
created_by_id: user.id,
)
user
end
2015-12-14 09:23:14 +00:00
def set_attributes_by_x_headers(item_object, header_name, mail)
# loop all x-zammad-hedaer-* headers
item_object.attributes.each {|key, _value|
# ignore read only attributes
next if key == 'updated_at'
next if key == 'created_at'
next if key == 'updated_by_id'
next if key == 'created_by_id'
# check if id exists
key_short = key[ key.length - 3, key.length ]
if key_short == '_id'
key_short = key[ 0, key.length - 3 ]
header = "x-zammad-#{header_name}-#{key_short}"
if mail[ header.to_sym ]
2015-05-04 18:58:28 +00:00
Rails.logger.info "header #{header} found #{mail[ header.to_sym ]}"
item_object.class.reflect_on_all_associations.map { |assoc|
next if assoc.name.to_s != key_short
Rails.logger.info "ASSOC found #{assoc.class_name} lookup #{mail[ header.to_sym ]}"
item = assoc.class_name.constantize
if item.respond_to?(:name)
2015-12-14 09:23:14 +00:00
if item.lookup(name: mail[ header.to_sym ])
item_object[key] = item.lookup(name: mail[ header.to_sym ]).id
end
elsif item.respond_to?(:login)
2015-12-14 09:23:14 +00:00
if item.lookup(login: mail[ header.to_sym ])
item_object[key] = item.lookup(login: mail[ header.to_sym ]).id
end
end
}
2012-10-05 06:42:12 +00:00
end
end
# check if attribute exists
header = "x-zammad-#{header_name}-#{key}"
if mail[ header.to_sym ]
2015-05-04 18:58:28 +00:00
Rails.logger.info "header #{header} found #{mail[ header.to_sym ]}"
item_object[key] = mail[ header.to_sym ]
end
2012-10-05 06:42:12 +00:00
}
2012-10-05 06:42:12 +00:00
end
end
# workaround to parse subjects with 2 different encodings correctly (e. g. quoted-printable see test/fixtures/mail9.box)
module Mail
module Encodings
def self.value_decode(str)
# Optimization: If there's no encoded-words in the string, just return it
return str unless str.index('=?')
str = str.gsub(/\?=(\s*)=\?/, '?==?') # Remove whitespaces between 'encoded-word's
# Split on white-space boundaries with capture, so we capture the white-space as well
str.split(/([ \t])/).map do |text|
if text.index('=?') .nil?
text
else
# Join QP encoded-words that are adjacent to avoid decoding partial chars
# text.gsub!(/\?\=\=\?.+?\?[Qq]\?/m, '') if text =~ /\?==\?/
# Search for occurences of quoted strings or plain strings
text.scan(/( # Group around entire regex to include it in matches
2013-06-13 07:01:06 +00:00
\=\?[^?]+\?([QB])\?[^?]+?\?\= # Quoted String with subgroup for encoding method
| # or
.+?(?=\=\?|$) # Plain String
)/xmi).map do |matches|
string, method = *matches
if method == 'b' || method == 'B'
b_value_decode(string)
elsif method == 'q' || method == 'Q'
q_value_decode(string)
else
string
end
end
end
end.join('')
end
end
end