diff options
Diffstat (limited to 'lib/mail_handler')
| -rw-r--r-- | lib/mail_handler/backends/mail_backend.rb | 76 | ||||
| -rw-r--r-- | lib/mail_handler/backends/mail_extensions.rb | 48 | ||||
| -rw-r--r-- | lib/mail_handler/mail_handler.rb | 12 |
3 files changed, 119 insertions, 17 deletions
diff --git a/lib/mail_handler/backends/mail_backend.rb b/lib/mail_handler/backends/mail_backend.rb index f7893a60d..28c486e1b 100644 --- a/lib/mail_handler/backends/mail_backend.rb +++ b/lib/mail_handler/backends/mail_backend.rb @@ -1,4 +1,35 @@ require 'mail' +require 'mapi/msg' +require 'mapi/convert' + +module Mail + class Message + + # The behaviour of the 'to' and 'cc' methods have changed + # between TMail and Mail; this monkey-patching restores the + # TMail behaviour. The key difference is that when there's an + # invalid address, e.g. '<foo@example.org', Mail returns the + # string as an ActiveSupport::Multibyte::Chars, whereas + # previously TMail would return nil. + + alias_method :old_to, :to + alias_method :old_cc, :cc + + def clean_addresses(old_method, val) + old_result = self.send(old_method, val) + old_result.class == Mail::AddressContainer ? old_result : nil + end + + def to(val = nil) + self.clean_addresses :old_to, val + end + + def cc(val = nil) + self.clean_addresses :old_cc, val + end + + end +end module MailHandler module Backends @@ -38,7 +69,11 @@ module MailHandler # Get the body of a mail part def get_part_body(part) - part.body.decoded + decoded = part.body.decoded + if part.content_type =~ /^text\// + decoded = convert_string_to_utf8_or_binary decoded, part.charset + end + decoded end # Return the first from field if any @@ -77,7 +112,7 @@ module MailHandler if first_from.is_a?(ActiveSupport::Multibyte::Chars) return nil else - return first_from.display_name ? eval(%Q{"#{first_from.display_name}"}) : nil + return (first_from.display_name || nil) end else return nil @@ -141,9 +176,14 @@ module MailHandler end elsif get_content_type(part) == 'application/ms-tnef' # A set of attachments in a TNEF file - part.rfc822_attachment = mail_from_tnef(part.body.decoded) - if part.rfc822_attachment.nil? - # Attached mail didn't parse, so treat as binary + begin + part.rfc822_attachment = mail_from_tnef(part.body.decoded) + if part.rfc822_attachment.nil? + # Attached mail didn't parse, so treat as binary + part.content_type = 'application/octet-stream' + end + rescue TNEFParsingError + part.rfc822_attachment = nil part.content_type = 'application/octet-stream' end end @@ -160,8 +200,11 @@ module MailHandler part.parts.each{ |sub_part| expand_and_normalize_parts(sub_part, parent_mail) } else part_filename = get_part_file_name(part) - charset = part.charset # save this, because overwriting content_type also resets charset - + if part.has_charset? + original_charset = part.charset # save this, because overwriting content_type also resets charset + else + original_charset = nil + end # Don't allow nil content_types if get_content_type(part).nil? part.content_type = 'application/octet-stream' @@ -180,7 +223,9 @@ module MailHandler # Use standard content types for Word documents etc. part.content_type = normalise_content_type(get_content_type(part)) decode_attached_part(part, parent_mail) - part.charset = charset + if original_charset + part.charset = original_charset + end end end @@ -228,8 +273,15 @@ module MailHandler def _get_attachment_leaves_recursive(part, within_rfc822_attachment, parent_mail) leaves_found = [] if part.multipart? - raise "no parts on multipart mail" if part.parts.size == 0 - if part.sub_type == 'alternative' + if part.parts.size == 0 + # This is typically caused by a missing final + # MIME boundary, in which case the text of the + # message (including the opening MIME + # boundary) is in part.body, so just add this + # part as a leaf and treat it as text/plain: + part.content_type = "text/plain" + leaves_found += [part] + elsif part.sub_type == 'alternative' best_part = choose_best_alternative(part) leaves_found += _get_attachment_leaves_recursive(best_part, within_rfc822_attachment, @@ -315,7 +367,9 @@ module MailHandler end def address_from_string(string) - Mail::Address.new(string).address + mail = Mail.new + mail.from = string + mail.from[0] end end end diff --git a/lib/mail_handler/backends/mail_extensions.rb b/lib/mail_handler/backends/mail_extensions.rb index d25012e39..a5b6bc97e 100644 --- a/lib/mail_handler/backends/mail_extensions.rb +++ b/lib/mail_handler/backends/mail_extensions.rb @@ -73,7 +73,15 @@ module Mail if match encoding = match[1] str = Ruby18.decode_base64(match[2]) - str = Iconv.conv('UTF-8//IGNORE', fix_encoding(encoding), str) + # Adding and removing trailing spaces is a workaround + # for Iconv.conv throwing an exception if it finds an + # invalid character at the end of the string, even + # with UTF-8//IGNORE: + # http://po-ru.com/diary/fixing-invalid-utf-8-in-ruby-revisited/ + begin + str = Iconv.conv('UTF-8//IGNORE', fix_encoding(encoding), str + " ")[0...-4] + rescue Iconv::InvalidEncoding + end end str end @@ -86,7 +94,12 @@ module Mail # Remove trailing = if it exists in a Q encoding string = string.sub(/\=$/, '') str = Encodings::QuotedPrintable.decode(string) - str = Iconv.conv('UTF-8//IGNORE', fix_encoding(encoding), str) + # Adding and removing trailing spaces is a workaround + # for Iconv.conv throwing an exception if it finds an + # invalid character at the end of the string, even + # with UTF-8//IGNORE: + # http://po-ru.com/diary/fixing-invalid-utf-8-in-ruby-revisited/ + str = Iconv.conv('UTF-8//IGNORE', fix_encoding(encoding), str + " ")[0...-4] end str end @@ -102,4 +115,35 @@ module Mail end end end + class Ruby19 + + def Ruby19.b_value_decode(str) + match = str.match(/\=\?(.+)?\?[Bb]\?(.+)?\?\=/m) + if match + encoding = match[1] + str = Ruby19.decode_base64(match[2]) + # Rescue an ArgumentError arising from an unknown encoding. + begin + str.force_encoding(fix_encoding(encoding)) + rescue ArgumentError + end + end + decoded = str.encode("utf-8", :invalid => :replace, :replace => "") + decoded.valid_encoding? ? decoded : decoded.encode("utf-16le", :invalid => :replace, :replace => "").encode("utf-8") + end + + def Ruby19.q_value_decode(str) + match = str.match(/\=\?(.+)?\?[Qq]\?(.+)?\?\=/m) + if match + encoding = match[1] + str = Encodings::QuotedPrintable.decode(match[2].gsub(/_/, '=20')) + # Backport line from mail 2.5 to strip a trailing = character + # Remove trailing = if it exists in a Q encoding + str = str.sub(/\=$/, '') + str.force_encoding(fix_encoding(encoding)) + end + decoded = str.encode("utf-8", :invalid => :replace, :replace => "") + decoded.valid_encoding? ? decoded : decoded.encode("utf-16le", :invalid => :replace, :replace => "").encode("utf-8") + end + end end diff --git a/lib/mail_handler/mail_handler.rb b/lib/mail_handler/mail_handler.rb index d9ebee854..9c955cccd 100644 --- a/lib/mail_handler/mail_handler.rb +++ b/lib/mail_handler/mail_handler.rb @@ -8,20 +8,23 @@ module MailHandler require 'backends/mail_backend' include Backends::MailBackend + class TNEFParsingError < StandardError + end + # Returns a set of attachments from the given TNEF contents # The TNEF contents also contains the message body, but in general this is the # same as the message body in the message proper. def tnef_attachments(content) attachments = [] Dir.mktmpdir do |dir| - IO.popen("#{`which tnef`.chomp} -K -C #{dir}", "wb") do |f| + IO.popen("tnef -K -C #{dir} 2> /dev/null", "wb") do |f| f.write(content) f.close if $?.signaled? raise IOError, "tnef exited with signal #{$?.termsig}" end if $?.exited? && $?.exitstatus != 0 - raise IOError, "tnef exited with status #{$?.exitstatus}" + raise TNEFParsingError, "tnef exited with status #{$?.exitstatus}" end end found = 0 @@ -34,7 +37,7 @@ module MailHandler end end if found == 0 - raise IOError, "tnef produced no attachments" + raise TNEFParsingError, "tnef produced no attachments" end end attachments @@ -77,7 +80,8 @@ module MailHandler tempfile.flush default_params = { :append_to => text, :binary_output => false } if content_type == 'application/vnd.ms-word' - AlaveteliExternalCommand.run("wvText", tempfile.path, tempfile.path + ".txt") + AlaveteliExternalCommand.run("wvText", tempfile.path, tempfile.path + ".txt", + { :memory_limit => 536870912 } ) # Try catdoc if we get into trouble (e.g. for InfoRequestEvent 2701) if not File.exists?(tempfile.path + ".txt") AlaveteliExternalCommand.run("catdoc", tempfile.path, default_params) |