Test suite: Start testing our output for valid UTF-8.

Feed the output of `tshark -G <glossary>` to `iconv -f UTF-8`. Adjust a
couple of the Bluetooth dissectors and X11 keysyms accordingly.

Change-Id: I5b04dc3fa4734c8f0a795daf44bd398fe5ebc1bd
Reviewed-on: https://code.wireshark.org/review/6146
Petri-Dish: Gerald Combs <gerald@wireshark.org>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Gerald Combs <gerald@wireshark.org>

1 files changed, 100 insertions, 96 deletions

diff --git a/epan/dissectors/x11-keysym.h b/epan/dissectors/x11-keysym.h
index fd4ea9d9e8..6009002d32 100644
--- a/epan/dissectors/x11-keysym.h
+++ b/epan/dissectors/x11-keysym.h
@@ -23,7 +23,9 @@
  */
 
 /* Note; Entries should be kept in ascending order so that val_to_str_ext()
- *   and etc can do the lookup using a binary search.
+ * and etc can do the lookup using a binary search.
+ *
+ * Strings must be valid UTF-8 sequences.
  */
 
 static const value_string x11_keysym_vals_source[] = {
@@ -124,101 +126,103 @@ static const value_string x11_keysym_vals_source[] = {
       { 0x07d, ", HFILL }" },
       { 0x07e, "~" },
       { 0x0a0, "nobreakspace" },
-      { 0x0a1, "\x0a1" },
-      { 0x0a2, "\x0a2" },
-      { 0x0a3, "\x0a3" },
-      { 0x0a4, "\x0a4" },
-      { 0x0a5, "\x0a5" },
-      { 0x0a6, "\x0a6" },
-      { 0x0a7, "\x0a7" },
-      { 0x0a8, "\x0a8" },
-      { 0x0a9, "\x0a9" },
-      { 0x0aa, "\x0aa" },
-      { 0x0ab, "\x0ab" },
-      { 0x0ac, "\x0ac" },
-      { 0x0ad, "\x0ad" },
-      { 0x0ae, "\x0ae" },
-      { 0x0af, "\x0af" },
-      { 0x0b0, "\x0b0" },
-      { 0x0b1, "\x0b1" },
-      { 0x0b2, "\x0b2" },
-      { 0x0b3, "\x0b3" },
-      { 0x0b4, "\x0b4" },
-      { 0x0b5, "\x0b5" },
-      { 0x0b6, "\x0b6" },
-      { 0x0b7, "\x0b7" },
-      { 0x0b8, "\x0b8" },
-      { 0x0b9, "\x0b9" },
-      { 0x0ba, "\x0ba" },
-      { 0x0bb, "\x0bb" },
-      { 0x0bc, "\x0bc" },
-      { 0x0bd, "\x0bd" },
-      { 0x0be, "\x0be" },
-      { 0x0bf, "\x0bf" },
-      { 0x0c0, "\x0c0" },
-      { 0x0c1, "\x0c1" },
-      { 0x0c2, "\x0c2" },
-      { 0x0c3, "\x0c3" },
-      { 0x0c4, "\x0c4" },
-      { 0x0c5, "\x0c5" },
-      { 0x0c6, "\x0c6" },
-      { 0x0c7, "\x0c7" },
-      { 0x0c8, "\x0c8" },
-      { 0x0c9, "\x0c9" },
-      { 0x0ca, "\x0ca" },
-      { 0x0cb, "\x0cb" },
-      { 0x0cc, "\x0cc" },
-      { 0x0cd, "\x0cd" },
-      { 0x0ce, "\x0ce" },
-      { 0x0cf, "\x0cf" },
-      { 0x0d0, "\x0d0" },
-      { 0x0d1, "\x0d1" },
-      { 0x0d2, "\x0d2" },
-      { 0x0d3, "\x0d3" },
-      { 0x0d4, "\x0d4" },
-      { 0x0d5, "\x0d5" },
-      { 0x0d6, "\x0d6" },
-      { 0x0d7, "\x0d7" },
-      { 0x0d8, "\x0d8" },
-      { 0x0d9, "\x0d9" },
-      { 0x0da, "\x0da" },
-      { 0x0db, "\x0db" },
-      { 0x0dc, "\x0dc" },
-      { 0x0dd, "\x0dd" },
-      { 0x0de, "\x0de" },
-      { 0x0df, "\x0df" },
-      { 0x0e0, "\x0e0" },
-      { 0x0e1, "\x0e1" },
-      { 0x0e2, "\x0e2" },
-      { 0x0e3, "\x0e3" },
-      { 0x0e4, "\x0e4" },
-      { 0x0e5, "\x0e5" },
-      { 0x0e6, "\x0e6" },
-      { 0x0e7, "\x0e7" },
-      { 0x0e8, "\x0e8" },
-      { 0x0e9, "\x0e9" },
-      { 0x0ea, "\x0ea" },
-      { 0x0eb, "\x0eb" },
-      { 0x0ec, "\x0ec" },
-      { 0x0ed, "\x0ed" },
-      { 0x0ee, "\x0ee" },
-      { 0x0ef, "\x0ef" },
-      { 0x0f0, "\x0f0" },
-      { 0x0f1, "\x0f1" },
-      { 0x0f2, "\x0f2" },
-      { 0x0f3, "\x0f3" },
-      { 0x0f4, "\x0f4" },
-      { 0x0f5, "\x0f5" },
-      { 0x0f6, "\x0f6" },
-      { 0x0f7, "\x0f7" },
-      { 0x0f8, "\x0f8" },
-      { 0x0f9, "\x0f9" },
-      { 0x0fa, "\x0fa" },
-      { 0x0fb, "\x0fb" },
-      { 0x0fc, "\x0fc" },
-      { 0x0fd, "\x0fd" },
-      { 0x0fe, "\x0fe" },
-      { 0x0ff, "\x0ff" },
+
+      /* Map raw Latin 1 / 8859-1 to Latin-1 Supplement UTF-8 sequences */
+      { 0x0a1, "\xc2\xa1" },
+      { 0x0a2, "\xc2\xa2" },
+      { 0x0a3, "\xc2\xa3" },
+      { 0x0a4, "\xc2\xa4" },
+      { 0x0a5, "\xc2\xa5" },
+      { 0x0a6, "\xc2\xa6" },
+      { 0x0a7, "\xc2\xa7" },
+      { 0x0a8, "\xc2\xa8" },
+      { 0x0a9, "\xc2\xa9" },
+      { 0x0aa, "\xc2\xaa" },
+      { 0x0ab, "\xc2\xab" },
+      { 0x0ac, "\xc2\xac" },
+      { 0x0ad, "\xc2\xad" },
+      { 0x0ae, "\xc2\xae" },
+      { 0x0af, "\xc2\xaf" },
+      { 0x0b0, "\xc2\xb0" },
+      { 0x0b1, "\xc2\xb1" },
+      { 0x0b2, "\xc2\xb2" },
+      { 0x0b3, "\xc2\xb3" },
+      { 0x0b4, "\xc2\xb4" },
+      { 0x0b5, "\xc2\xb5" },
+      { 0x0b6, "\xc2\xb6" },
+      { 0x0b7, "\xc2\xb7" },
+      { 0x0b8, "\xc2\xb8" },
+      { 0x0b9, "\xc2\xb9" },
+      { 0x0ba, "\xc2\xba" },
+      { 0x0bb, "\xc2\xbb" },
+      { 0x0bc, "\xc2\xbc" },
+      { 0x0bd, "\xc2\xbd" },
+      { 0x0be, "\xc2\xbe" },
+      { 0x0bf, "\xc2\xbf" },
+      { 0x0c0, "\xc3\x80" },
+      { 0x0c1, "\xc3\x81" },
+      { 0x0c2, "\xc3\x82" },
+      { 0x0c3, "\xc3\x83" },
+      { 0x0c4, "\xc3\x84" },
+      { 0x0c5, "\xc3\x85" },
+      { 0x0c6, "\xc3\x86" },
+      { 0x0c7, "\xc3\x87" },
+      { 0x0c8, "\xc3\x88" },
+      { 0x0c9, "\xc3\x89" },
+      { 0x0ca, "\xc3\x8a" },
+      { 0x0cb, "\xc3\x8b" },
+      { 0x0cc, "\xc3\x8c" },
+      { 0x0cd, "\xc3\x8d" },
+      { 0x0ce, "\xc3\x8e" },
+      { 0x0cf, "\xc3\x8f" },
+      { 0x0d0, "\xc3\x90" },
+      { 0x0d1, "\xc3\x91" },
+      { 0x0d2, "\xc3\x92" },
+      { 0x0d3, "\xc3\x93" },
+      { 0x0d4, "\xc3\x94" },
+      { 0x0d5, "\xc3\x95" },
+      { 0x0d6, "\xc3\x96" },
+      { 0x0d7, "\xc3\x97" },
+      { 0x0d8, "\xc3\x98" },
+      { 0x0d9, "\xc3\x99" },
+      { 0x0da, "\xc3\x9a" },
+      { 0x0db, "\xc3\x9b" },
+      { 0x0dc, "\xc3\x9c" },
+      { 0x0dd, "\xc3\x9d" },
+      { 0x0de, "\xc3\x9e" },
+      { 0x0df, "\xc3\x9f" },
+      { 0x0e0, "\xc3\xa0" },
+      { 0x0e1, "\xc3\xa1" },
+      { 0x0e2, "\xc3\xa2" },
+      { 0x0e3, "\xc3\xa3" },
+      { 0x0e4, "\xc3\xa4" },
+      { 0x0e5, "\xc3\xa5" },
+      { 0x0e6, "\xc3\xa6" },
+      { 0x0e7, "\xc3\xa7" },
+      { 0x0e8, "\xc3\xa8" },
+      { 0x0e9, "\xc3\xa9" },
+      { 0x0ea, "\xc3\xaa" },
+      { 0x0eb, "\xc3\xab" },
+      { 0x0ec, "\xc3\xac" },
+      { 0x0ed, "\xc3\xad" },
+      { 0x0ee, "\xc3\xae" },
+      { 0x0ef, "\xc3\xaf" },
+      { 0x0f0, "\xc3\xb0" },
+      { 0x0f1, "\xc3\xb1" },
+      { 0x0f2, "\xc3\xb2" },
+      { 0x0f3, "\xc3\xb3" },
+      { 0x0f4, "\xc3\xb4" },
+      { 0x0f5, "\xc3\xb5" },
+      { 0x0f6, "\xc3\xb6" },
+      { 0x0f7, "\xc3\xb7" },
+      { 0x0f8, "\xc3\xb8" },
+      { 0x0f9, "\xc3\xb9" },
+      { 0x0fa, "\xc3\xba" },
+      { 0x0fb, "\xc3\xbb" },
+      { 0x0fc, "\xc3\xbc" },
+      { 0x0fd, "\xc3\xbd" },
+      { 0x0fe, "\xc3\xbe" },
+      { 0x0ff, "\xc3\xbf" },
 
       { 0x1a1, "Aogonek" },
       { 0x1a2, "breve" },