Support RFC 2231 language tags in RFC 2047 header strings.

diff --git a/sbr/fmt_rfc2047.c b/sbr/fmt_rfc2047.c
index 25c98c3ded2ae2b392eb652a7ff1d41a49f12318..7f73efd20009c36000d3d7cca89ab32c8f851f55 100644 (file)
--- a/sbr/fmt_rfc2047.c
+++ b/sbr/fmt_rfc2047.c
@@ -61,7 +61,7 @@ int
 decode_rfc2047 (char *str, char *dst, size_t dstlen)
 {
     char *p, *q, *pp;
-    char *startofmime, *endofmime;
+    char *startofmime, *endofmime, *endofcharset;
     int c, quoted_printable;
     int encoding_found = 0;    /* did we decode anything?                */
     int between_encodings = 0; /* are we between two encodings?          */
@@ -127,11 +127,25 @@ decode_rfc2047 (char *str, char *dst, size_t dstlen)
            if (!*pp)
                continue;
 
+           /*
+            * RFC 2231 specifies that language information can appear
+            * in a charset specification like so:
+            *
+            * =?us-ascii*en?Q?Foo?=
+            *
+            * Right now we don't use language information, so ignore it.
+            */
+
+           for (endofcharset = startofmime;
+                       *endofcharset != '*' && endofcharset < pp;
+                                                       endofcharset++)
+               ;
+
            /* Check if character set can be handled natively */
-           if (!check_charset(startofmime, pp - startofmime)) {
+           if (!check_charset(startofmime, endofcharset - startofmime)) {
 #ifdef HAVE_ICONV
                /* .. it can't. We'll use iconv then. */
-               *pp = '\0';
+               *endofcharset = '\0';
                cd = iconv_open(get_charset(), startofmime);
                fromutf8 = !mh_strcasecmp(startofmime, "UTF-8");
                *pp = '?';

diff --git a/test/scan/test-scan-multibyte b/test/scan/test-scan-multibyte
index 98dd9a2608248e98f6a1bd43406ac707f9a1a1ae..255adacedb53315afdd7faf7398fac6deb4ac31b 100755 (executable)
--- a/test/scan/test-scan-multibyte
+++ b/test/scan/test-scan-multibyte
@@ -37,10 +37,17 @@ LC_ALL=en_US.UTF-8; export LC_ALL
 # We use a test program to determine what the output width of U+0308 is
 # and adjust our test output appropriately.
 #
+# True Spın̈al Tap fans will note that David st Hubbins was born in Squatney,
+# London, England, and thus having his name language-tagged with "cy" is almost
+# certainly incorrect.  But in his own words: "Here lies David st Hubbins,
+# and why not?".
+#
+# The second "* in the To line is just to exercise the parser a bit.
+#
 
 cat > "${MH_TEST_DIR}/Mail/inbox/11" <<EOF
-From: David =?utf-8?q?=EF=AC=86?= Hubbins <hubbins@example.com>
-To: Sir Denis =?utf-8?q?Eton=E2=80=93Hogg? <sirdenis@example.com>
+From: David =?utf-8*cy?q?=EF=AC=86?= Hubbins <hubbins@example.com>
+To: Sir Denis =?utf-8*?q?Eton=E2=80=93Hogg? <sirdenis@example.com>
 Date: Friday, 2 Mar 1984 00:00:00
 Subject: =?utf-8?q?Sp=C4=B1n=CC=88al_Tap_=E2=86=92_Tap_into_America!?=