keysymtab: Add (re)generation script

It does not produce exactly the same content yet, as the
left/rightanglebracket unicode equivalent was not defined in xterm as
what Xorg currently says.

2 files changed, 43 insertions, 2 deletions

diff --git a/registryd/ucs2keysym.c b/registryd/ucs2keysym.c
index a077ce03..09f5fd3c 100644
--- a/registryd/ucs2keysym.c
+++ b/registryd/ucs2keysym.c
@@ -34,6 +34,9 @@
 #include <X11/X.h>
 #include "deviceeventcontroller.h"	/* for prototype */
 
+/* DO NOT UPATE BY HAND!
+ * This table can be regenerated from Xorg's keysymdef.h with the ucs2keysym.sh
+ * script.  */
 struct codepair {
   unsigned short keysym;
   unsigned short ucs;
@@ -603,8 +606,8 @@ struct codepair {
   { 0x0afa, 0x2315 }, /*           telephonerecorder ⌕ TELEPHONE RECORDER */
   { 0x08a4, 0x2320 }, /*                 topintegral ⌠ TOP HALF INTEGRAL */
   { 0x08a5, 0x2321 }, /*                 botintegral ⌡ BOTTOM HALF INTEGRAL */
-  { 0x0abc, 0x2329 }, /*            leftanglebracket 〈 LEFT-POINTING ANGLE BRACKET */
-  { 0x0abe, 0x232a }, /*           rightanglebracket 〉 RIGHT-POINTING ANGLE BRACKET */
+  { 0x0abc, 0x2329 }, /*            leftanglebracket 〈 LEFT-POINTING ANGLE BRACKET, not U+27E8 per xterm source */
+  { 0x0abe, 0x232a }, /*           rightanglebracket 〉 RIGHT-POINTING ANGLE BRACKET, not U+27E9 per xterm source */
   { 0x0bcc, 0x2395 }, /*                        quad ⎕ APL FUNCTIONAL SYMBOL QUAD */
   { 0x08ab, 0x239b }, /*               topleftparens ⎛ LEFT PARENTHESIS UPPER HOOK */
   { 0x08ac, 0x239d }, /*               botleftparens ⎝ LEFT PARENTHESIS LOWER HOOK */
diff --git a/registryd/ucs2keysym.sh b/registryd/ucs2keysym.sh
new file mode 100755
index 00000000..ff37f53b
--- /dev/null
+++ b/registryd/ucs2keysym.sh
@@ -0,0 +1,38 @@
+#!/bin/bash
+
+if [ "$#" = 0 ]
+then
+	echo "Usage: $0 /path/to/keysymdef.h"
+	exit 1
+fi
+
+# We are only interested in
+# - keysyms which have well-defined unicode equivalent
+# - and are not just trivial unicode keysyms
+# - non-latin1 keysyms
+# - not the lamda aliases
+# and we tinker with the alias parentheses to make sorting easier
+
+grep '^#define' "$1" | \
+	grep -i "U+" | \
+	grep -vi "0x100[0-9a-f][0-9a-f][0-9a-f][0-9a-f]" | \
+	grep -vi " 0x00[0-9a-f][0-9a-f] " | \
+	grep -vi "_lamda " | \
+	sed -e 's/\/\*(/& /' | \
+	sed -e 's/)\*\// &/' | \
+	sort -k 5 | \
+	perl -CS -e '
+my $last = 0;
+while (<>) {
+	chomp;
+	if ( /^\#define XK_([a-zA-Z_0-9]+)(\s*)   0x([0-9a-f]+)\s*\/\*(\(?) U\+([0-9A-F]{4,6}) (.*) \)?\*\/\s*$/ ) {
+		my ( $xk, $space, $keysym, $paren, $unicode, $unistr ) = ( $1, $2, $3, $4, $5, $6);
+		$unicode = hex("0x".$unicode);
+
+		print "\n" if (int($unicode / 256) != int($last / 256));
+		$last = $unicode;
+
+		printf "  { 0x$keysym, 0x%04x }, /* $space$xk %lc $unistr */\n", $unicode, $unicode;
+	}
+}
+	'