[geary/wip/336-smtp-rcpt-to-quoting: 1/3] Correctly handle escaping UTF-8 characters in RFC822 mailboxes

[Michael Gratton <mjog src gnome org>]

commit c306c2c5faee6f7c58baea1e802e6bb4c9c8364c
Author: Michael Gratton <mike vee net>
Date:   Wed Apr 24 20:38:11 2019 +1000

    Correctly handle escaping UTF-8 characters in RFC822 mailboxes
    
    Use string.get_next_char() to handle iterating over an
    RFC822.MailboxAddress mailbox local part, and explicitly allow UTF-8
    2, 3, and 4 byte chars without quoting, per RFC 5322.
    
    This lets us fix the test cases that use UTF-8 chars in the local part.

 src/engine/rfc822/rfc822-mailbox-address.vala | 25 +++++++++++++++++--------
 test/engine/rfc822-mailbox-address-test.vala  | 10 ++++++++--
 2 files changed, 25 insertions(+), 10 deletions(-)
---
diff --git a/src/engine/rfc822/rfc822-mailbox-address.vala b/src/engine/rfc822/rfc822-mailbox-address.vala
index a1864baf..eda28fe1 100644
--- a/src/engine/rfc822/rfc822-mailbox-address.vala
+++ b/src/engine/rfc822/rfc822-mailbox-address.vala
@@ -21,7 +21,7 @@ public class Geary.RFC822.MailboxAddress :
     Gee.Hashable<MailboxAddress>,
     BaseObject {
 
-    private static char[] ATEXT = {
+    private static unichar[] ATEXT = {
         '!', '#', '$', '%', '&', '\'', '*', '+', '-',
         '/', '=', '?', '^', '_', '`', '{', '|', '}', '~'
     };
@@ -64,17 +64,26 @@ public class Geary.RFC822.MailboxAddress :
         if (!String.is_empty(local_part)) {
             int index = 0;
             for (;;) {
-                char ch = local_part[index++];
-                if (ch == String.EOS)
+                unichar ch;
+                if (!local_part.get_next_char(ref index, out ch)) {
                     break;
+                }
 
                 is_dot = (ch == '.');
 
-                if (!(ch >= 0x41 && ch <= 0x5A) && // A-Z
-                    !(ch >= 0x61 && ch <= 0x7A) && // a-z
-                    !(ch >= 0x30 && ch <= 0x39) && // 0-9
-                    !(ch in ATEXT) &&
-                    !(is_dot && index > 1)) { // no leading dots
+                if (!(
+                        // RFC 5322 ASCII
+                        (ch >= 0x61 && ch <= 0x7A) || // a-z
+                        (ch >= 0x41 && ch <= 0x5A) || // A-Z
+                        (ch >= 0x30 && ch <= 0x39) || // 0-9
+                        // RFC 6532 UTF8
+                        (ch >= 0x80 && ch <= 0x07FF) ||      // UTF-8 2 byte
+                        (ch >= 0x800 && ch <= 0xFFFF) ||     // UTF-8 3 byte
+                        (ch >= 0x10000 && ch <= 0x10FFFF) || // UTF-8 4 byte
+                        // RFC 5322 atext
+                        (ch in ATEXT) ||
+                        // RFC 5322 dot-atom (no leading quotes)
+                        (is_dot && index > 1))) {
                     needs_quote = true;
                     break;
                 }
diff --git a/test/engine/rfc822-mailbox-address-test.vala b/test/engine/rfc822-mailbox-address-test.vala
index 761cf796..d65a5869 100644
--- a/test/engine/rfc822-mailbox-address-test.vala
+++ b/test/engine/rfc822-mailbox-address-test.vala
@@ -228,6 +228,7 @@ class Geary.RFC822.MailboxAddressTest : TestCase {
         //       "\"test\" test\"@example.com");
         //assert(new MailboxAddress(null, "test\"test example com").to_rfc822_address() ==
         //       "\"test\"test\"@example.com");
+
         assert_string(
             "$test example com",
             new MailboxAddress(null, "$test example com").to_rfc822_address()
@@ -236,14 +237,19 @@ class Geary.RFC822.MailboxAddressTest : TestCase {
             "\"test@test\"@example.com",
             new MailboxAddress(null, "test@test example com").to_rfc822_address()
         );
+
+        // Likewise, Unicode chars should be passed through. Note that
+        // these can only be sent if a UTF8 connection is negotiated
+        // with the SMTP server
         assert_string(
-            "=?iso-8859-1?b?qQ==?=@example.com",
+            "©@example.com",
             new MailboxAddress(null, "©@example.com").to_rfc822_address()
         );
         assert_string(
-            "=?UTF-8?b?8J+YuA==?=@example.com",
+            "😸@example.com",
             new MailboxAddress(null, "😸@example.com").to_rfc822_address()
         );
+
     }
 
     public void to_rfc822_string() throws Error {