From 4368c8e0647804a33cd029f35158914a5736b183 Mon Sep 17 00:00:00 2001
From: Michael Stapelberg <michael@stapelberg.de>
Date: Fri, 13 Apr 2012 13:25:29 +0200
Subject: [PATCH] Support 'charset = "UTF-8"' (with spaces)

This does not conform to the RFC, but PHPMailer in older versions does that.
Therefore, we should support it :-/.
---
 lib/heliotrope/message.rb |  2 +-
 test/broken_charset.msg   | 57 +++++++++++++++++++++++++++++++++++++++
 test/test_heliotrope.rb   | 12 +++++++++
 3 files changed, 70 insertions(+), 1 deletion(-)
 create mode 100644 test/broken_charset.msg

diff --git a/lib/heliotrope/message.rb b/lib/heliotrope/message.rb
index 0c928cd..c49af66 100644
--- a/lib/heliotrope/message.rb
+++ b/lib/heliotrope/message.rb
@@ -257,7 +257,7 @@ def mime_content_for mime_part, preferred_type
 
     mt = mime_type_for(mime_part) || "text/plain" # i guess
     content_type = if mt =~ /^(.+);/ then $1.downcase else mt end
-    source_charset = if mt =~ /charset="?(.*?)"?(;|$)/i then $1 else "US-ASCII" end
+    source_charset = if mt =~ /charset\s*=\s*"?(.*?)"?(;|$)/i then $1 else "US-ASCII" end
 
     content = mime_part.decode
     converted_content, converted_charset = if(converter = CONVERSIONS[[content_type, preferred_type]])
diff --git a/test/broken_charset.msg b/test/broken_charset.msg
new file mode 100644
index 0000000..b549bd8
--- /dev/null
+++ b/test/broken_charset.msg
@@ -0,0 +1,57 @@
+Return-Path: <moodle@bar.invalid>
+Date: Fri, 13 Apr 2012 12:00:02 +0200
+Return-Path: moodle@bar.invalid
+To: "Michael Stapelberg" <foo@bar.invalid>
+From: "=?utf-8?B?THV0eiBTdHLDvG5nbWFubg==?=" <l.struengmann@bar.invalid>
+Subject: =?utf-8?B?QUZTMjAxMjogQXVmZ2FiZW46IFB1bmt0ZSDDnGJ1bmcgMQ==?=
+Message-ID: <0ba81486243611670b534d4564ee3718@localhost.localdomain>
+X-Priority: 3
+X-Mailer: PHPMailer [version Moodle 2007101590]
+MIME-Version: 1.0
+Content-Type: multipart/alternative;
+	boundary="b1_0ba81486243611670b534d4564ee3718"
+X-DSPAM-Result: Innocent
+X-DSPAM-Processed: Fri Apr 13 12:08:16 2012
+X-DSPAM-Confidence: 0.9899
+X-DSPAM-Probability: 0.0000
+X-DSPAM-Signature: 4f87fb10319744089468961
+
+
+--b1_0ba81486243611670b534d4564ee3718
+Content-Type: text/plain; charset = "UTF-8"
+Content-Transfer-Encoding: quoted-printable
+
+
+AFS2012 -> Aufgaben -> Punkte =C3=9Cbung 1
+---------------------------------------------------------------------
+Lutz Str=C3=BCngmann hat einen Kommentar zu Ihrer eingereichten Aufgabe =
+'Punkte
+=C3=9Cbung 1' verfasst.
+
+Sie finden ihn im Anhang Ihrer abgegebenen Aufgabe:
+
+http://moodle.hs-mannheim.de/mod/assignment/view.php?id=3D11205
+---------------------------------------------------------------------
+
+
+
+--b1_0ba81486243611670b534d4564ee3718
+Content-Type: text/html; charset = "UTF-8"
+Content-Transfer-Encoding: quoted-printable
+
+<p><font face=3D"sans-serif"><a =
+href=3D"http://moodle.hs-mannheim.de/course/view.php?id=3D666">AFS2012</a> =
+-><a href=3D"http://moodle.hs-mannheim.de/mod/assignment/index.php?id=3D66=
+6">Aufgaben</a> -><a href=3D"http://moodle.hs-mannheim.de/mod/assignment/v=
+iew.php?id=3D11205">Punkte =C3=9Cbung 1</a></font></p><hr /><font =
+face=3D"sans-serif"><p>Lutz Str=C3=BCngmann hat einen Kommentar zu Ihrer =
+abgegebenen Aufgabe verfasst.</i>'<i>Punkte =C3=9Cbung 1</i>'<br /><br />
+Sie finden ihn im Anhang Ihrer=20
+<a href=3D"http://moodle.hs-mannheim.de/mod/assignment/view.php?id=3D11205=
+">eingereichten Aufgabe</a>:.</p></font><hr />
+
+
+
+--b1_0ba81486243611670b534d4564ee3718--
+
+
diff --git a/test/test_heliotrope.rb b/test/test_heliotrope.rb
index e396b91..c480d29 100644
--- a/test/test_heliotrope.rb
+++ b/test/test_heliotrope.rb
@@ -1,3 +1,5 @@
+# encoding: UTF-8
+
 require 'test/unit'
 require 'fileutils'
 require 'digest/md5'
@@ -394,6 +396,16 @@ def test_labellist_pruning_removes_labels_without_corresponding_threads
     assert_does_not_include "potato", @metaindex.all_labels
   end
 
+  def test_recognize_charset
+    msg = nil
+    File.open('test/broken_charset.msg', 'r') do |f|
+      msg = Heliotrope::Message.new(f).parse!
+    end
+    assert_not_nil msg
+    mime_parts = msg.mime_parts "text/plain"
+    assert_equal mime_parts, [["text/plain; charset = \"utf-8\"", nil, nil, "\nAFS2012 -> Aufgaben -> Punkte Übung 1\n---------------------------------------------------------------------\nLutz Strüngmann hat einen Kommentar zu Ihrer eingereichten Aufgabe 'Punkte\nÜbung 1' verfasst.\n\nSie finden ihn im Anhang Ihrer abgegebenen Aufgabe:\n\nhttp://moodle.hs-mannheim.de/mod/assignment/view.php?id=11205\n---------------------------------------------------------------------\n\n\n"]]
+  end
+
 private
 
   def assert_includes v, set # standard one seems to have these things reversed