From 8c2f9268ab976015edcff16e26a13398bba8396c Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Mon, 1 Apr 2024 16:03:29 +0200
Subject: Add unicode/prop.h

---
 gen/prop/gc | 87 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 87 insertions(+)
 create mode 100755 gen/prop/gc

(limited to 'gen/prop/gc')

diff --git a/gen/prop/gc b/gen/prop/gc
new file mode 100755
index 0000000..48c6659
--- /dev/null
+++ b/gen/prop/gc
@@ -0,0 +1,87 @@
+#!/bin/sh
+
+set -e
+cd "${0%/*}/../.."
+exec >include/unicode/__gc.h
+
+cat <<C
+/* This file is autogenerated by gen/prop/gc; DO NOT EDIT. */
+
+#ifndef MLIB_UNICODE_PROP___GC_H
+#define MLIB_UNICODE_PROP___GC_H
+
+#include "unicode/prop.h"
+
+static const enum uprop_gc __mlib_gc_lat1_lt[] = {
+C
+
+gawk '
+BEGIN {
+	FS = ";"
+}
+
+{
+	s = "GC_" toupper($3)
+	lo = strtonum("0X" $1)
+
+	if ($2 ~ /First/) {
+		getline
+		hi = strtonum("0X" $1)
+	} else
+		hi = lo
+
+	for (i = lo; i <= hi; i++)
+		props[i] = s
+}
+
+END {
+	for (i = 0; i <= 0xFF; i++)
+		print props[i] ","
+}
+' data/UnicodeData.txt | paste -d' ' - - - - - - - - | sed 's/^/\t/'
+
+cat <<C
+};
+
+static const struct {
+	rune lo, hi;
+	enum uprop_gc val;
+} __mlib_gc_lt[] = {
+C
+
+gawk '
+BEGIN {
+	FS = ";"
+}
+
+{
+	s = "GC_" toupper($3)
+	lo = strtonum("0X" $1)
+
+	if ($2 ~ /First/) {
+		getline
+		hi = strtonum("0X" $1)
+	} else
+		hi = lo
+
+	for (i = lo; i <= hi; i++)
+		props[i] = s
+}
+
+END {
+	for (i = 0x100; i <= 0x10FFFF; i++) {
+		if (!props[i])
+			continue
+		lo = i
+		while (props[lo] == props[i + 1])
+			i++
+		printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[lo]
+	}
+}
+' data/UnicodeData.txt | sort
+
+cat <<C
+};
+
+#endif /* !MLIB_UNICODE_PROP___GC_H */
+C
-- 
cgit v1.2.3