TNEF (winmail.dat) parsing support via libytnef

.gitlab-ci.yml

@@ -27,7 +27,7 @@ variables:
                gtk3-devel iso-codes-devel json-glib-devel itstool
                libcanberra-devel libgee-devel libhandy-devel
                libnotify-devel libsecret-devel libunwind-devel
-               libxml2-devel sqlite-devel webkitgtk4-devel
+               libxml2-devel libytnef-devel sqlite-devel webkitgtk4-devel
   FEDORA_TEST_DEPS: Xvfb tar xz
 
   # Ubuntu packages
@@ -39,7 +39,7 @@ variables:
                libhandy-0.0-dev libjson-glib-dev libmessaging-menu-dev
                libnotify-dev libsecret-1-dev libsqlite3-dev
                libunity-dev libunwind-dev libwebkit2gtk-4.0-dev
-               libxml2-dev
+               libxml2-dev libytnef0-dev
   UBUNTU_TEST_DEPS: xauth xvfb
 
 fedora:

INSTALL

@@ -45,7 +45,8 @@ Install them by running this command:
         glib2-devel gmime-devel gnome-online-accounts-devel gtk3-devel \
         iso-codes-devel json-glib-devel libcanberra-devel \
         libgee-devel libhandy-devel libnotify-devel libsecret-devel \
-        libunwind-devel libxml2-devel sqlite-devel webkitgtk4-devel
+        libunwind-devel libxml2-devel libytnef-devel sqlite-devel \
+        webkitgtk4-devel
 
 Installing dependencies on Ubuntu/Debian
 ----------------------------------------
@@ -58,7 +59,7 @@ Install them by running this command:
         libglib2.0-dev libgmime-2.6-dev libgoa-1.0-dev libgtk-3-dev \
         libjson-glib-dev libhandy-dev libnotify-dev libsecret-1-dev \
         libsqlite3-dev libunwind-dev libwebkit2gtk-4.0-dev \
-        libxml2-dev
+        libxml2-dev libytnef0-dev
 
 And for Ubuntu Unity integration:
 

bindings/vapi/libytnef.vapi

@@ -0,0 +1,56 @@
+/*
+ * Copyright (c) 2018 Oliver Giles <ohw.giles@gmail.com>
+ *
+ * This software is licensed under the GNU Lesser General Public License
+ * (version 2.1 or later).  See the COPYING file in this distribution.
+ */
+
+[CCode (cheader_filename = "ytnef.h")]
+namespace Ytnef {
+	[CCode (cname ="variableLength", has_type_id = false)]
+	public struct VariableLength {
+		[CCode (array_length_cname = "size")]
+		uint8[] data;
+	}
+
+	[CCode (cname = "MAPI_UNDEFINED")]
+	public VariableLength* MAPI_UNDEFINED;
+
+	[CCode (cname = "int", cprefix = "PT_", has_type_id = false)]
+	public enum PropType {
+		STRING8
+	}
+
+	[CCode (cname = "int", cprefix = "PR_", has_type_id = false)]
+	public enum PropID {
+		DISPLAY_NAME,
+		ATTACH_LONG_FILENAME
+	}
+
+	[CCode (cname = "PROP_TAG")]
+	public static int PROP_TAG(PropType type, PropID id);
+
+	[CCode (cname = "MAPIProps", has_type_id = false)]
+	public struct MAPIProps {
+	}
+
+	[CCode (cname = "Attachment", has_type_id = false)]
+	public struct Attachment {
+		VariableLength Title;
+		VariableLength FileData;
+		MAPIProps MAPI;
+		Attachment? next;
+	}
+
+	[CCode (cname = "TNEFStruct", destroy_function="TNEFFree", has_type_id = false)]
+	public struct TNEFStruct {
+		Attachment starting_attach;
+	}
+
+	[CCode (cname = "TNEFParseMemory", has_type_id = false)]
+	public static int ParseMemory(uint8[] data, out TNEFStruct tnef);
+
+	[CCode (cname = "MAPIFindProperty")]
+	public static unowned VariableLength* MAPIFindProperty(MAPIProps MAPI, uint tag);
+}
+

meson.build

@@ -78,6 +78,7 @@ libunwind_generic_dep = dependency(
   'libunwind-generic', version: '>= 1.1', required: not get_option('libunwind_optional')
 )
 libxml = dependency('libxml-2.0', version: '>= 2.7.8')
+libytnef = dependency('libytnef', version: '>= 1.9.3', required: get_option('tnef-support'))
 posix = valac.find_library('posix')
 webkit2gtk_web_extension = dependency('webkit2gtk-web-extension-4.0', version: '>=' + target_webkit)
 

meson_options.txt

@@ -5,3 +5,4 @@ option('ref_tracking', type: 'boolean', value: false, description: 'Whether to u
 option('iso_639_xml', type: 'string', value: '', description: 'Full path to the ISO 639 XML file.')
 option('iso_3166_xml', type: 'string', value: '', description: 'Full path to the ISO 3166 XML file.')
 option('libunwind_optional', type: 'boolean', value: false, description: 'Determines if libunwind is required.')
+option('tnef-support', type: 'boolean', value: true, description: 'Whether to support TNEF attachments (requires libytnef).')

org.gnome.Geary.json

@@ -205,6 +205,16 @@
                 }
             ]
         },
+        {
+            "name": "libytnef",
+            "sources": [
+                {
+                    "type": "git",
+                    "url": "https://github.com/Yeraze/ytnef.git",
+                    "branch": "master"
+                }
+            ]
+        },
         {
             "name": "geary",
             "buildsystem": "meson",

src/engine/meson.build

@@ -333,6 +333,7 @@ geary_engine_dependencies = [
   gmime,
   javascriptcoregtk,
   libxml,
+  libytnef,
   posix,
   sqlite
 ]
@@ -358,6 +359,13 @@ if libunwind_dep.found()
   ]
 endif
 
+if get_option('tnef-support')
+  geary_engine_dependencies += libytnef
+  geary_engine_vala_options += [
+    '-D', 'WITH_TNEF_SUPPORT'
+  ]
+endif
+
 geary_engine_lib = static_library('geary-engine',
   geary_engine_sources,
   dependencies: geary_engine_dependencies,

src/engine/rfc822/rfc822-message.vala

@@ -880,17 +880,50 @@ public class Geary.RFC822.Message : BaseObject, EmailHeaderSet {
                 Mime.ContentType content_type =
                     part.get_effective_content_type();
 
-                // Skip text/plain and text/html parts that are INLINE
+#if WITH_TNEF_SUPPORT
-                // or UNSPECIFIED, as they will be included in the body
+                if (content_type.is_type("application", "vnd.ms-tnef")) {
+                    GMime.StreamMem stream = new GMime.StreamMem();
+                    ((GMime.Part) root).get_content_object().write_to_stream(stream);
+                    ByteArray tnef_data = stream.get_byte_array();
+                    Ytnef.TNEFStruct tn;
+                    if (Ytnef.ParseMemory(tnef_data.data, out tn) == 0) {
+                        for (unowned Ytnef.Attachment? a = tn.starting_attach.next; a != null; a = a.next) {
+                            attachments.add(new Part(tnef_attachment_to_gmime_part(a)));
+                        }
+                    }
+                } else
+#endif // WITH_TNEF_SUPPORT
                 if (actual_disposition == Mime.DispositionType.ATTACHMENT ||
                     (!content_type.is_type("text", "plain") &&
                      !content_type.is_type("text", "html"))) {
+                    // Skip text/plain and text/html parts that are INLINE
+                    // or UNSPECIFIED, as they will be included in the body
                     attachments.add(part);
                 }
             }
         }
     }
 
+#if WITH_TNEF_SUPPORT
+    private GMime.Part tnef_attachment_to_gmime_part(Ytnef.Attachment a) {
+        Ytnef.VariableLength* filenameProp = Ytnef.MAPIFindProperty(a.MAPI, Ytnef.PROP_TAG(Ytnef.PropType.STRING8, Ytnef.PropID.ATTACH_LONG_FILENAME));
+        if (filenameProp == Ytnef.MAPI_UNDEFINED) {
+            filenameProp = Ytnef.MAPIFindProperty(a.MAPI, Ytnef.PROP_TAG(Ytnef.PropType.STRING8, Ytnef.PropID.DISPLAY_NAME));
+            if (filenameProp == Ytnef.MAPI_UNDEFINED) {
+                filenameProp = &a.Title;
+            }
+        }
+        string filename = (string) filenameProp.data;
+        uint8[] data = Bytes.unref_to_data(new Bytes(a.FileData.data));
+
+        GMime.Part part = new GMime.Part();
+        part.set_filename(filename);
+        part.set_content_type(new GMime.ContentType.from_string(GLib.ContentType.guess(filename, data, null)));
+        part.set_content_object(new GMime.DataWrapper.with_stream(new GMime.StreamMem.with_buffer(data), GMime.ContentEncoding.BINARY));
+        return part;
+    }
+#endif
+
     public Gee.List<Geary.RFC822.Message> get_sub_messages() {
         Gee.List<Geary.RFC822.Message> messages = new Gee.ArrayList<Geary.RFC822.Message>();
         find_sub_messages(messages, message.get_mime_part());

test/data/org.gnome.GearyTest.gresource.xml

@@ -4,6 +4,7 @@
     <file>basic-text-plain.eml</file>
     <file>basic-text-html.eml</file>
     <file>basic-multipart-alternative.eml</file>
+    <file>basic-multipart-tnef.eml</file>
     <file>geary-0.6-db.tar.xz</file>
   </gresource>
 </gresources>

test/engine/rfc822-message-test.vala

@@ -12,6 +12,7 @@ class Geary.RFC822.MessageTest : TestCase {
     private const string BASIC_TEXT_HTML = "basic-text-html.eml";
     private const string BASIC_MULTIPART_ALTERNATIVE =
         "basic-multipart-alternative.eml";
+    private const string BASIC_MULTIPART_TNEF = "basic-multipart-tnef.eml";
 
     private const string HTML_CONVERSION_TEMPLATE =
         "<div class=\"plaintext\" style=\"white-space: pre-wrap;\">%s</div>";
@@ -38,6 +39,7 @@ This is the second line.
         add_test("text_plain_as_html", text_plain_as_html);
         add_test("text_html_as_html", text_html_as_html);
         add_test("text_html_as_plain", text_html_as_plain);
+        add_test("tnef_extract_attachments", tnef_extract_attachments);
         add_test("multipart_alternative_as_plain",
                  multipart_alternative_as_plain);
         add_test("multipart_alternative_as_converted_html",
@@ -124,6 +126,14 @@ This is the second line.
         assert_string(BASIC_HTML_BODY, test.get_html_body(null));
     }
 
+    public void tnef_extract_attachments() throws Error {
+        Message test = resource_to_message(BASIC_MULTIPART_TNEF);
+        Gee.List<Part> attachments = test.get_attachments();
+        assert_true(attachments.size == 2);
+        assert_true(attachments[0].get_clean_filename() == "zappa_av1.jpg");
+        assert_true(attachments[1].get_clean_filename() == "bookmark.htm");
+    }
+
     public void multipart_alternative_as_plain() throws Error {
         Message test = resource_to_message(BASIC_MULTIPART_ALTERNATIVE);