Merge branch 'master' of github.com:sebbacon/alaveteli

5 files changed, 94 insertions, 10 deletions

diff --git a/app/models/incoming_message.rb b/app/models/incoming_message.rb
index f437ab2c9..c808dc6a1 100644
--- a/app/models/incoming_message.rb
+++ b/app/models/incoming_message.rb
@@ -395,11 +395,12 @@ class FOIAttachment
             tempfile.print self.body
             tempfile.flush
 
-            if self.content_type == 'application/vnd.ms-word'
-                # XXX do something with PNG files this spits out so they view too :)
-                system("/usr/bin/wvHtml --charset=UTF-8 " + tempfile.path + " " + tempfile.path + ".html")
-                html = File.read(tempfile.path + ".html")
-                File.unlink(tempfile.path + ".html")
+            # Use google docs for the view for these - hanging server
+            # if self.content_type == 'application/vnd.ms-word'
+            #     # XXX do something with PNG files this spits out so they view too :)
+            #     system("/usr/bin/wvHtml --charset=UTF-8 " + tempfile.path + " " + tempfile.path + ".html")
+            #     html = File.read(tempfile.path + ".html")
+            #     File.unlink(tempfile.path + ".html")
 #            elsif self.content_type == 'application/vnd.ms-excel'
 #                # Don't colorise, e.g. otherwise this one comes out with white
 #                # text which is nasty:
@@ -408,7 +409,7 @@ class FOIAttachment
 #                    html = child.read()
 #                    wrapper_id = "wrapper_xlhtml"
 #                end
-            elsif self.content_type == 'application/pdf'
+            if self.content_type == 'application/pdf'
                 IO.popen("/usr/bin/pdftohtml -nodrm -zoom 1.0 -stdout -enc UTF-8 -noframes " + tempfile.path + "", "r") do |child|
                     html = child.read()
                 end
diff --git a/commonlib b/commonlib
-Subproject b43b5949501f763e84bed7fce30458578bf92f0
+Subproject 38e0a641eb63323f6a2f895bfbc1274b23e98a7
diff --git a/lib/tmail_extensions.rb b/lib/tmail_extensions.rb
index c0ed1033b..6a5044cdb 100644
--- a/lib/tmail_extensions.rb
+++ b/lib/tmail_extensions.rb
@@ -5,8 +5,10 @@
 # Email: francis@mysociety.org; WWW: http://www.mysociety.org/
 #
 # $Id: tmail_extensions.rb,v 1.7 2009-10-02 23:31:01 francis Exp $
-
+require 'racc/parser'
 require 'tmail'
+require 'tmail/scanner'
+require 'tmail/utils'
 require 'tmail/interface'
 
 # Monkeypatch!
@@ -32,7 +34,7 @@ module TMail
             else
                 return nil
             end
-        end 
+        end
 
         # Monkeypatch! Generalisation of To:, Cc:
         def envelope_to(default = nil)
@@ -72,7 +74,7 @@ module TMail
         # a name and an email
         def Address.address_from_name_and_email(name, email)
             if !MySociety::Validate.is_valid_email(email)
-                raise "invalid email " + email + " passed to address_from_name_and_email"    
+                raise "invalid email " + email + " passed to address_from_name_and_email"
             end
             if name.nil?
                 return TMail::Address.parse(email)
@@ -84,7 +86,7 @@ module TMail
     end
 
     module TextUtils
-        # Monkeypatch! Much more aggressive list of characters to cause quoting 
+        # Monkeypatch! Much more aggressive list of characters to cause quoting
         # than in normal TMail. e.g. Have found real cases where @ needs quoting.
         # We list characters to allow, rather than characters not to allow.
         NEW_PHRASE_UNSAFE=/[^A-Za-z0-9!#\$%&'*+\-\/=?^_`{|}~ ]/n
@@ -94,4 +96,52 @@ module TMail
     end
 end
 
+# Monkeypatch! TMail 1.2.7.1 will parse only one address out of a list of addresses with 
+# unquoted display parts https://github.com/mikel/tmail/issues#issue/9 - this monkeypatch 
+# fixes this issue.
+module TMail
+
+  class Parser < Racc::Parser
+
+module_eval <<'..end lib/tmail/parser.y modeval..id2dd1c7d21d', 'lib/tmail/parser.y', 340
+
+  def self.special_quote_address(str) #:nodoc:
+    # Takes a string which is an address and adds quotation marks to special
+    # edge case methods that the RACC parser can not handle.
+    #
+    # Right now just handles two edge cases:
+    #
+    # Full stop as the last character of the display name:
+    #   Mikel L. <mikel@me.com>
+    # Returns:
+    #   "Mikel L." <mikel@me.com>
+    #
+    # Unquoted @ symbol in the display name:
+    #   mikel@me.com <mikel@me.com>
+    # Returns:
+    #   "mikel@me.com" <mikel@me.com>
+    #
+    # Any other address not matching these patterns just gets returned as is.
+    case
+    # This handles the missing "" in an older version of Apple Mail.app
+    # around the display name when the display name contains a '@'
+    # like 'mikel@me.com <mikel@me.com>'
+    # Just quotes it to: '"mikel@me.com" <mikel@me.com>'
+    when str =~ /\A([^"][^<]+@[^>]+[^"])\s(<.*?>)\Z/
+      return "\"#{$1}\" #{$2}"
+    # This handles cases where 'Mikel A. <mikel@me.com>' which is a trailing
+    # full stop before the address section.  Just quotes it to
+    # '"Mikel A." <mikel@me.com>'
+    when str =~ /\A(.*?\.)\s(<.*?>)\s*\Z/
+      return "\"#{$1}\" #{$2}"
+    else
+      str
+    end
+  end
+
+..end lib/tmail/parser.y modeval..id2dd1c7d21d
+  end   # class Parser
+
+end   # module TMail
+
 
diff --git a/spec/fixtures/multiple-unquoted-display-names.email b/spec/fixtures/multiple-unquoted-display-names.email
new file mode 100644
index 000000000..e3a742c83
--- /dev/null
+++ b/spec/fixtures/multiple-unquoted-display-names.email
@@ -0,0 +1,27 @@
+From foi@example.com Mon Mar 21 09:58:48 2011
+Return-path: <foi@example.com>
+Envelope-to: foi@wildfire.ukcod.org.uk
+Delivery-date: Mon, 21 Mar 2011 09:58:48 +0000
+From: UAS Foi <foi@admin.ox.ac.uk>
+To: WDTK User <request-66666-caa77777@whatdotheyknow.com>, Another address
+	<foi@example.com>
+Date: Mon, 21 Mar 2011 09:58:30 +0000
+Subject: RE: Freedom of Information request - Participation in the IPCC
+ Assessment Process
+Thread-Topic: Freedom of Information request - Participation in the IPCC
+ Assessment Process
+Thread-Index: Acvlavo2wQr6iwN5SQy7Y/pEUUVYaQCQnamw
+Message-ID: <C351A7C3885B7141ACFE44C69D66A98B387D486FAB@EXMBX02.ad.oak.ox.ac.uk>
+References: <E1Q0Z7o-0007ie-FL@wildfire.ukcod.org.uk>
+In-Reply-To: <E1Q0Z7o-0007ie-FL@wildfire.ukcod.org.uk>
+Accept-Language: en-US, en-GB
+Content-Language: en-US
+X-MS-Has-Attach:
+X-MS-TNEF-Correlator:
+acceptlanguage: en-US, en-GB
+Content-Type: text/plain; charset="utf-8"
+Content-Transfer-Encoding: base64
+MIME-Version: 1.0
+
+stuff
+
diff --git a/spec/lib/tmail_extensions_spec.rb b/spec/lib/tmail_extensions_spec.rb
index d1a56f2e2..41e8c8f4e 100644
--- a/spec/lib/tmail_extensions_spec.rb
+++ b/spec/lib/tmail_extensions_spec.rb
@@ -22,5 +22,11 @@ describe "when using TMail" do
         incoming_message.get_body_for_html_display()
     end
 
+    it 'should parse multiple to addresses with unqoted display names' do
+        example_file = File.join(Spec::Runner.configuration.fixture_path, 'multiple-unquoted-display-names.email')
+        mail = TMail::Mail.parse(File.read(example_file))
+        mail.to.should == ["request-66666-caa77777@whatdotheyknow.com", "foi@example.com"]
+    end
+
 end