Fixed issue #2227 - Unprocessable email ERROR: #<ArgumentError: invalid byte sequence in UTF-8> and improved email parser tests to also verify attachments.

app/models/channel/email_parser.rb

@@ -599,15 +599,29 @@ process unprocessable_mails (tmp/unprocessable_mail/*.eml) again
     # get filename from content-disposition
 
     # workaround for: NoMethodError: undefined method `filename' for #<Mail::UnstructuredField:0x007ff109e80678>
+    begin
       filename = file.header[:content_disposition].try(:filename)
+    rescue
+      begin
+        if file.header[:content_disposition].to_s =~ /(filename|name)(\*{0,1})="(.+?)"/i
+          filename = $3
+        elsif file.header[:content_disposition].to_s =~ /(filename|name)(\*{0,1})='(.+?)'/i
+          filename = $3
+        elsif file.header[:content_disposition].to_s =~ /(filename|name)(\*{0,1})=(.+?);/i
+          filename = $3
+        end
+      rescue
+        Rails.logger.debug { 'Unable to get filename' }
+      end
+    end
 
     begin
-      if file.header[:content_disposition].to_s =~ /filename="(.+?)"/i
+      if file.header[:content_disposition].to_s =~ /(filename|name)(\*{0,1})="(.+?)"/i
-        filename = $1
+        filename = $3
-      elsif file.header[:content_disposition].to_s =~ /filename='(.+?)'/i
+      elsif file.header[:content_disposition].to_s =~ /(filename|name)(\*{0,1})='(.+?)'/i
-        filename = $1
+        filename = $3
-      elsif file.header[:content_disposition].to_s =~ /filename=(.+?);/i
+      elsif file.header[:content_disposition].to_s =~ /(filename|name)(\*{0,1})=(.+?);/i
-        filename = $1
+        filename = $3
       end
     rescue
       Rails.logger.debug { 'Unable to get filename' }
@@ -615,12 +629,12 @@ process unprocessable_mails (tmp/unprocessable_mail/*.eml) again
 
     # as fallback, use raw values
     if filename.blank?
-      if headers_store['Content-Disposition'].to_s =~ /filename="(.+?)"/i
+      if headers_store['Content-Disposition'].to_s =~ /(filename|name)(\*{0,1})="(.+?)"/i
-        filename = $1
+        filename = $3
-      elsif headers_store['Content-Disposition'].to_s =~ /filename='(.+?)'/i
+      elsif headers_store['Content-Disposition'].to_s =~ /(filename|name)(\*{0,1})='(.+?)'/i
-        filename = $1
+        filename = $3
-      elsif headers_store['Content-Disposition'].to_s =~ /filename=(.+?);/i
+      elsif headers_store['Content-Disposition'].to_s =~ /(filename|name)(\*{0,1})=(.+?);/i
-        filename = $1
+        filename = $3
       end
     end
 
@@ -654,9 +668,9 @@ process unprocessable_mails (tmp/unprocessable_mail/*.eml) again
 
       # e. g. Content-Type: video/quicktime; name="Video.MOV";
       if filename.blank?
-        ['name="(.+?)"(;|$)', "name='(.+?)'(;|$)", 'name=(.+?)(;|$)'].each do |regexp|
+        ['(filename|name)(\*{0,1})="(.+?)"(;|$)', '(filename|name)(\*{0,1})=\'(.+?)\'(;|$)', '(filename|name)(\*{0,1})=(.+?)(;|$)'].each do |regexp|
           if headers_store['Content-Type'] =~ /#{regexp}/i
-            filename = $1
+            filename = $3
             break
           end
         end

test/data/mail/mail014.yml

@@ -2,9 +2,8 @@
 from: '"Müller, Bernd" <Bernd.Mueller@example.com>'
 from_email: Bernd.Mueller@example.com
 from_display_name: Müller, Bernd
-subject: 'AW: OTRS [Ticket#118192]'
 to: "'Martin Edenhofer via Znuny Sales' <sales@znuny.com>"
-content_type: text/plain
+subject: 'AW: OTRS [Ticket#118192]'
 body: |
   äöüß ad asd
 
@@ -12,3 +11,62 @@ body: |
 
   --
   Old programmers never die. They just branch to a new address.
+content_type: text/plain
+attachments:
+- !ruby/hash:ActiveSupport::HashWithIndifferentAccess
+  data: "Received: from NTEXCHANGE3.example.dom ([172.17.253.64]) by ntexchange.example.dom\n
+    with Microsoft SMTPSVC(6.0.3790.4675);\t Wed, 5 Dec 2012 21:15:07 +0100\nReceived:
+    from NTEXCHANGE2.example.dom (172.17.253.213) by NTEXCHANGE3.example.dom\n (172.17.253.64)
+    with Microsoft SMTP Server (TLS) id 14.2.318.4; Wed, 5 Dec\n 2012 21:15:07 +0100\nReceived:
+    from ntserver-1.example.com (172.17.253.224) by NTEXCHANGE2.example.dom\n (172.17.253.213)
+    with Microsoft SMTP Server id 14.2.318.4; Wed, 5 Dec 2012\n 21:15:06 +0100\nReceived:
+    from ntserver-1.example.com ([192.168.100.117]) by ntserver-1.example.com with\n
+    Microsoft SMTPSVC(7.5.7601.17514);\t Wed, 5 Dec 2012 21:15:05 +0100\nReceived:
+    from mailout02.ispamone.de ( [212.211.146.205]) by ntserver-1.example.com\n\t(Reddoxx
+    engine) with SMTP id 56767845A33; Wed, 5 Dec 2012 21:15:05 +0100\nFrom: \"helpdesk@example.com\"
+    <helpdesk@example.com>\nTo: HelpDesk <HelpDesk@example.com>\nSubject: Neue Anfrage
+    erstellt - 33284\nThread-Topic: Neue Anfrage erstellt - 33284\nThread-Index: Ac3TJBhe4TPAltHrS6m3GwtF5lDkwA==\nContent-Class:
+    urn:content-classes:message\nDate: Wed, 5 Dec 2012 20:07:04 +0000\nMessage-ID:
+    <174905662.1354738495468.JavaMail.babeldoc@atlas>\nContent-Language: de-DE\nX-MS-Has-Attach:\nX-MS-Exchange-Organization-SCL:
+    0\nX-MS-TNEF-Correlator:\nContent-Type: multipart/alternative;\n\tboundary=\"_000_1749056621354738495468JavaMailbabeldocatlas_\"\nMIME-Version:
+    1.0\n\n--_000_1749056621354738495468JavaMailbabeldocatlas_\nContent-Type: text/plain;
+    charset=\"utf-8\"\nContent-Transfer-Encoding: base64\n\ndGhpcyBpcyBhIHRlc3Q=\n\n--_000_1749056621354738495468JavaMailbabeldocatlas_\nContent-Type:
+    text/html; charset=\"utf-8\"\nContent-Transfer-Encoding: base64\n\nPGgxPnRoaXMgaXMgYSB0ZXN0PC9oMT4=\n\n--_000_1749056621354738495468JavaMailbabeldocatlas_--\n"
+  filename: Neue Anfrage erstellt - 33284.eml
+  preferences: !ruby/hash:ActiveSupport::HashWithIndifferentAccess
+    Content-Type: message/rfc822
+    Mime-Type: message/rfc822
+    Charset: UTF-8
+- !ruby/hash:ActiveSupport::HashWithIndifferentAccess
+  data: "Received: from NTEXCHANGE3.example.dom ([172.17.253.64]) by ntexchange.example.dom\n
+    with Microsoft SMTPSVC(6.0.3790.4675);\t Wed, 5 Dec 2012 21:06:21 +0100\nReceived:
+    from NTEXCHANGE2.example.dom (172.17.253.213) by NTEXCHANGE3.example.dom\n (172.17.253.64)
+    with Microsoft SMTP Server (TLS) id 14.2.318.4; Wed, 5 Dec\n 2012 21:06:21 +0100\nReceived:
+    from ntserver-1.example.com (172.17.253.224) by NTEXCHANGE2.example.dom\n (172.17.253.213)
+    with Microsoft SMTP Server id 14.2.318.4; Wed, 5 Dec 2012\n 21:06:21 +0100\nReceived:
+    from ntserver-1.example.com ([192.168.100.117]) by ntserver-1.example.com with\n
+    Microsoft SMTPSVC(7.5.7601.17514);\t Wed, 5 Dec 2012 21:06:21 +0100\nReceived:
+    from mailout01.ispamone.de ( [212.211.146.204]) by ntserver-1.example.com\n\t(Reddoxx
+    engine) with SMTP id 40D026EFCF5; Wed, 5 Dec 2012 21:06:19 +0100\nFrom: \"bestellung.example@example.com\"
+    <bestellung.example@example.com>\nTo: HelpDesk <HelpDesk@example.com>\nSubject:
+    Call: HW-Anforderung; Best-nr.47524152\nThread-Topic: Call: HW-Anforderung; Best-nr.47524152\nThread-Index:
+    Ac3TI/8Xh/P5FTJFSYuBz+vjlXSWbg==\nDate: Wed, 5 Dec 2012 20:06:21 +0000\nMessage-ID:
+    <13553079.1354737948277.JavaMail.jboss@fpep>\nContent-Language: de-DE\nX-MS-Has-Attach:\nX-MS-Exchange-Organization-SCL:
+    0\nX-MS-TNEF-Correlator:\nContent-Type: text/plain; charset=\"iso-8859-1\"\nContent-Transfer-Encoding:
+    quoted-printable\nMIME-Version: 1.0\n\n\n\n\nWarenempf=E4nger:\nFirma:            Example
+    GmbH\nName:             Rehm\nVorname:          Joachim\nStra=DFe            K=E4ssstr.
+    19\nOrt:              00000 Ulm\nE-Mail:           helpdesk@example.com\nTelefon:
+    \         0000 166-2399\nFax:              0000 166-2309\nAbteilung:        TN
+    23\nGeb=E4ude:          EG, Zimmer E60\n\n\nBestellpositionen:\nPos. Menge  Art.-Nr.
+    \ Bezeichnung                        Ansprechpartner   =\n       Aufstellort                        Bemerkungen
+    des Bestellers\n10   1.0    588164    Dell UltraSharp U2410 24           Andreas
+    Ring, GE-P=\nEW     K=E4ssstr. 19, 3. OG, Zi 306\n20   1.0    71705     Crossoverkabel
+    S/FTP, PiMF, Cat... Andreas Ring, GE-P=\nEW     K=E4ssstr. 19, 3. OG, Zi 306\n30
+    \  1.0    777577_638Dell Latitude E6430 mit UMTS / ... Andreas Ring, GE-P=\nEW
+    \    K=E4ssstr. 19, 3. OG, Zi 306\n40   1.0    920-003052Logitech K270 cordless
+    Keyboard... Andreas Ring, GE-P=\nEW     K=E4ssstr. 19, 3. OG, Zi 306\n"
+  filename: 'Call: HW-Anforderung; Best-nr.47524152.eml'
+  preferences: !ruby/hash:ActiveSupport::HashWithIndifferentAccess
+    Content-Type: message/rfc822
+    Mime-Type: message/rfc822
+    Charset: UTF-8

test/data/mail/mail024.yml

@@ -2,6 +2,17 @@
 from: oracle@IG0-1-DB01.example.com
 from_email: oracle@IG0-1-DB01.example.com
 from_display_name: ''
-subject: 'Regelsets im Test-Status gefunden: 1'
 to: support@example.com
+subject: 'Regelsets im Test-Status gefunden: 1'
 body: no visible content
+content_type: text/plain
+attachments:
+- !ruby/hash:ActiveSupport::HashWithIndifferentAccess
+  data: |
+    RULESET_ID;NAME;ACTIV;RUN_MODE;AUDIT_MODIFY_DATE
+    387;DP DHL JOIN - EN : Einladung eAC;T;SM;1.09.14
+  filename: rulesets-report.csv
+  preferences: !ruby/hash:ActiveSupport::HashWithIndifferentAccess
+    Content-Type: text/csv
+    Mime-Type: text/csv
+    Charset: UTF-8

test/unit/email_parser_test.rb

@@ -8,23 +8,25 @@ class EmailParserTest < ActiveSupport::TestCase
 
 to write new .yml files for emails you can use the following code:
 
-File.write('test/data/mail/mailXXX.yml', Channel::EmailParser.new.parse(File.read('test/data/mail/mailXXX.box')).slice(:from, :from_email, :from_display_name, :to, :cc, :subject, :body, :content_type, :'reply-to').to_yaml)
+File.write('test/data/mail/mailXXX.yml', Channel::EmailParser.new.parse(File.read('test/data/mail/mailXXX.box')).slice(:from, :from_email, :from_display_name, :to, :cc, :subject, :body, :content_type, :'reply-to', :attachments).to_yaml)
 
 =end
 
   test 'parse' do
     msg_files = Dir.glob(Rails.root.join('test', 'data', 'mail', 'mail*.box')).sort
-
+    messages = []
-    messages = msg_files.select { |f| File.exists?(f.ext('yml')) }
+    msg_files.each do |f|
-                        .map do |f|
+      next if !File.exists?(f.ext('yml'))
-                          {
+      item = {
         source:  File.basename(f),
         content: YAML.load(File.read(f.ext('yml'))),
         parsed:  Channel::EmailParser.new.parse(File.read(f)),
       }
+      messages.push item
     end
 
     messages.each do |m|
+
       # assert: raw content hash is a subset of parsed message hash
       expected_msg = m[:content].except(:attachments)
       parsed_msg = m[:parsed].slice(*expected_msg.keys)
@@ -35,23 +37,24 @@ File.write('test/data/mail/mailXXX.yml', Channel::EmailParser.new.parse(File.rea
 
       # assert: attachments in parsed message hash match metadata in raw hash
       next if m[:content][:attachments].blank?
-
+      attachments_found = []
-      # the formats of m[:content][:attachments] and m[:parsed][:attachments] don't match,
+      m[:content][:attachments].each do |expected_attachment|
-      # so we have to convert one to the other
+        expected_attachment_md5 = Digest::MD5.hexdigest(expected_attachment[:data])
-      parsed_attachment_metadata = m[:parsed][:attachments].map do |a|
+        m[:parsed][:attachments].each do |parsed_attachment|
-                                     {
+          parsed_attachment_md5 = Digest::MD5.hexdigest(parsed_attachment[:data])
-                                       md5:      Digest::MD5.hexdigest(a[:data]),
+          next if attachments_found.include?(parsed_attachment_md5)
-                                       cid:      a[:preferences]['Content-ID'],
+          next if expected_attachment_md5 != parsed_attachment_md5
-                                       filename: a[:filename],
+          attachments_found.push parsed_attachment_md5
-                                     }.with_indifferent_access
+          expected_attachment.each do |key, value|
+            assert_equal(value, parsed_attachment[key], "#{key} is different")
           end
-
+          next
-      m[:content][:attachments].sort_by { |a| a[:md5] }
+        end
-       .zip(parsed_attachment_metadata.sort_by { |a| a[:md5] })
+      end
-       .each do |content, parsed|
+      next if attachments_found.count == m[:content][:attachments].count
-        assert_operator(content, :<=, parsed,
+      m[:content][:attachments].each do |expected_attachment|
-                        "parsed attachment data from #{m[:source]} does not match " \
+        next if attachments_found.include?(Digest::MD5.hexdigest(expected_attachment[:data]))
-                        "attachment metadata from #{m[:source].ext('yml')}")
+        assert(false, "Attachment not found test/data/mail/#{m[:source]}: #{expected_attachment.inspect}")
       end
     end
   end