Improved filename detection of incoming postmaster emails (use subject.eml and no file-1).

2017-11-06 04:26:00 +01:00 · 2017-11-06 04:26:00 +01:00 · 99af4f187a
commit 99af4f187a
parent 967f814ff3
2 changed files with 111 additions and 16 deletions
--- a/app/models/channel/email_parser.rb
+++ b/app/models/channel/email_parser.rb
@ -158,7 +158,7 @@ class Channel::EmailParser
      end

      # text attachment/body exists
-      if data[:body].empty? && mail.text_part
+      if data[:body].blank? && mail.text_part
        data[:body] = mail.text_part.body.decoded
        data[:body] = Encode.conv(mail.text_part.charset, data[:body])
        data[:body] = data[:body].to_s.force_encoding('utf-8')
@ -170,7 +170,7 @@ class Channel::EmailParser
      end

      # any other attachments
-      if data[:body].empty?
+      if data[:body].blank?
        data[:body] = 'no visible content'
        data[:content_type] = 'text/plain'
      end
@ -296,13 +296,13 @@ class Channel::EmailParser
  def _get_attachment(file, attachments, mail)

    # check if sub parts are available
-    if !file.parts.empty?
-      a = []
+    if file.parts.present?
+      list = []
      file.parts.each do |p|
        attachment = _get_attachment(p, attachments, mail)
-        a.concat(attachment)
+        list.concat(attachment)
      end
-      return a
+      return list
    end

    # ignore text/plain attachments - already shown in view
@ -327,6 +327,12 @@ class Channel::EmailParser
      end
    end

+    # cleanup content id, <> will be added automatically later
+    if headers_store['Content-ID']
+      headers_store['Content-ID'].gsub!(/^</, '')
+      headers_store['Content-ID'].gsub!(/>$/, '')
+    end
+
    # get filename from content-disposition
    filename = nil

@ -361,7 +367,38 @@ class Channel::EmailParser
    # for some broken sm mail clients (X-MimeOLE: Produced By Microsoft Exchange V6.5)
    filename ||= file.header[:content_location].to_s

-    # generate file name
+    # generate file name based on content-id
+    if filename.blank? && headers_store['Content-ID'].present?
+      if headers_store['Content-ID'] =~ /(.+?)@.+?/i
+        filename = $1
+      end
+    end
+
+    # generate file name based on content type
+    if filename.blank? && headers_store['Content-Type'].present?
+      if headers_store['Content-Type'] =~ %r{^message/delivery-status$}i
+        filename = if headers_store['Content-Description'].present?
+                     "#{headers_store['Content-Description']}.txt"
+                   else
+                     'delivery-status.txt'
+                   end
+      elsif headers_store['Content-Type'] =~ %r{^message/rfc822$}i
+        begin
+          parser = Channel::EmailParser.new
+          mail_local = parser.parse(file.body.to_s)
+          filename = if mail_local[:subject].present?
+                       "#{mail_local[:subject]}.eml"
+                     elsif headers_store['Content-Description'].present?
+                       "#{headers_store['Content-Description']}.eml"
+                     else
+                       'Mail.eml'
+                     end
+        rescue
+          filename = 'Mail.eml'
+        end
+      end
+    end
+
    if filename.blank?
      attachment_count = 0
      (1..1000).each do |count|
@ -390,12 +427,6 @@ class Channel::EmailParser
    headers_store.delete('Content-Transfer-Encoding')
    headers_store.delete('Content-Disposition')

-    # cleanup content id, <> will be added automatically later
-    if headers_store['Content-ID']
-      headers_store['Content-ID'].gsub!(/^</, '')
-      headers_store['Content-ID'].gsub!(/>$/, '')
-    end
-
    # workaround for mail gem
    # https://github.com/zammad/zammad/issues/928
    filename = Mail::Encodings.value_decode(filename)
@ -667,7 +698,7 @@ returns
      end
    end

-    if data.empty? || data[:from_email].blank?
+    if data.blank? || data[:from_email].blank?
      from.strip!
      if from =~ /^(.+?)<(.+?)@(.+?)>$/
        data[:from_email]        = "#{$2}@#{$3}"
--- a/test/unit/email_parser_test.rb
+++ b/test/unit/email_parser_test.rb
@ -421,11 +421,11 @@ Managing Director: Martin Edenhofer
        attachments: [
          {
            md5: '5536be23f647953dc39c1673205d6f5b',
-            filename: 'file-1',
+            filename: 'Neue Anfrage erstellt - 33284.eml',
          },
          {
            md5: '4eeeae078b920f9d0708353ba0f6aa63',
-            filename: 'file-2',
+            filename: 'Call: HW-Anforderung; Best-nr.47524152.eml',
          },
        ],
        params: {
@ -1191,6 +1191,70 @@ Old programmers never die. They just branch to a new address."
 Old programmers never die. They just branch to a new address."
        },
      },
+      {
+        data: IO.binread('test/fixtures/mail62.box'),
+        body_md5: '10e7158e65a12b5850163d4d4b8ca2f8',
+        attachments: [
+          {
+            md5: '2b615b93ed76877dddbb7d0e3855916b',
+            filename: 'message.html',
+          },
+          {
+            md5: 'a618d671348735744d4c9a4005b56799',
+            filename: 'image001.jpg',
+            cid: 'image001.jpg@01CDB132.D8A510F0',
+          },
+          {
+            md5: '2cca28177acb4190cbf79cf3624cddae',
+            filename: 'image000.jpg',
+            cid: 'image000.jpg@01CDB132.D8A510F0',
+          },
+        ],
+        params: {
+          from: 'Smith Sepp <smith@example.com>',
+          from_email: 'smith@example.com',
+          from_display_name: 'Smith Sepp',
+          subject: 'Gruß aus Oberalteich',
+          content_type: 'text/html',
+          body: "<div>
+<p>Herzliche Grüße aus Oberalteich sendet Herrn Smith</p><p>&nbsp;</p><p>Sepp Smith - Dipl.Ing. agr. (FH)</p><p>Geschäftsführer der example Straubing-Bogen</p><p>Klosterhof 1 | 94327 Bogen-Oberalteich</p><p>Tel: 09422-505601 | Fax: 09422-505620</p><p>Internet: <a href=\"http://example-straubing-bogen.de/\" rel=\"nofollow noreferrer noopener\" target=\"_blank\">http://example-straubing-bogen.de</a></p><p>Facebook: <a href=\"http://facebook.de/examplesrbog\" rel=\"nofollow noreferrer noopener\" target=\"_blank\">http://facebook.de/examplesrbog</a></p><p><b><img border=\"0\" src=\"cid:image001.jpg@01CDB132.D8A510F0\" alt=\"Beschreibung: Beschreibung: efqmLogo\" style=\"width:60px;height:19px;\"><img border=\"0\" src=\"cid:image000.jpg@01CDB132.D8A510F0\" alt=\"Beschreibung: Beschreibung: efqmLogo\" style=\"width:60px;height:19px;\"></b><b> - European Foundation für Quality Management</b></p><p>&nbsp;</p></div>",
+        },
+      },
+      {
+        data: IO.binread('test/fixtures/mail63.box'),
+        body_md5: 'dbed0b09656d17bf4e832b2c18381c24',
+        attachments: [
+          {
+            md5: 'b8be1b421733bf2fab2eb24e11d53139',
+            filename: 'Delivery report.txt',
+          },
+          {
+            md5: '47744d00358baaeff355ca91ede93859',
+            filename: 'hello 123 äöüß.eml',
+          },
+        ],
+        params: {
+          from: 'MAILER-DAEMON@mx1.example.com (Mail Delivery System)',
+          from_email: 'MAILER-DAEMON@mx1.example.com',
+          from_display_name: 'Mail Delivery System',
+          subject: 'Undelivered Mail Returned to Sender',
+          content_type: 'text/plain',
+          body: "This is the mail system at host mx1.example.com.
+
+I'm sorry to have to inform you that your message could not
+be delivered to one or more recipients. It's attached below.
+
+For further assistance, please send mail to postmaster.
+
+If you do so, please include this problem report. You can
+delete your own text from the attached returned message.
+
+                   The mail system
+
+<notextisting@example.com>: user unknown
+",
+        },
+      },
    ]

    count = 0