From 903080d6b1bddde8d9097359eed21a9b9ee74fd0 Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Tue, 16 Apr 2024 23:05:29 +0200
Subject: Add uprop_get_ccc()

---
 gen/prop/ccc | 116 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 116 insertions(+)
 create mode 100755 gen/prop/ccc

(limited to 'gen')

diff --git a/gen/prop/ccc b/gen/prop/ccc
new file mode 100755
index 0000000..4f370e7
--- /dev/null
+++ b/gen/prop/ccc
@@ -0,0 +1,116 @@
+#!/bin/sh
+
+set -e
+cd "${0%/*}/../.."
+exec >lib/unicode/prop/uprop_get_ccc.c
+
+gawk '
+BEGIN {
+	FS = ";"
+
+	map[1]   = "OV"
+	map[6]   = "HANR"
+	map[7]   = "NK"
+	map[8]   = "KV"
+	map[9]   = "VR"
+	map[10]  = "CCC10"
+	map[11]  = "CCC11"
+	map[12]  = "CCC12"
+	map[13]  = "CCC13"
+	map[14]  = "CCC14"
+	map[15]  = "CCC15"
+	map[16]  = "CCC16"
+	map[17]  = "CCC17"
+	map[18]  = "CCC18"
+	map[19]  = "CCC19"
+	map[20]  = "CCC20"
+	map[21]  = "CCC21"
+	map[22]  = "CCC22"
+	map[23]  = "CCC23"
+	map[24]  = "CCC24"
+	map[25]  = "CCC25"
+	map[26]  = "CCC26"
+	map[27]  = "CCC27"
+	map[28]  = "CCC28"
+	map[29]  = "CCC29"
+	map[30]  = "CCC30"
+	map[31]  = "CCC31"
+	map[32]  = "CCC32"
+	map[33]  = "CCC33"
+	map[34]  = "CCC34"
+	map[35]  = "CCC35"
+	map[36]  = "CCC36"
+	map[84]  = "CCC84"
+	map[91]  = "CCC91"
+	map[103] = "CCC103"
+	map[107] = "CCC107"
+	map[118] = "CCC118"
+	map[122] = "CCC122"
+	map[129] = "CCC129"
+	map[130] = "CCC130"
+	map[132] = "CCC132"
+	map[133] = "CCC133"
+	map[200] = "ATBL"
+	map[202] = "ATB"
+	map[214] = "ATA"
+	map[216] = "ATAR"
+	map[218] = "BL"
+	map[220] = "B"
+	map[222] = "BR"
+	map[224] = "L"
+	map[226] = "R"
+	map[228] = "AL"
+	map[230] = "A"
+	map[232] = "AR"
+	map[233] = "DB"
+	map[234] = "DA"
+	map[240] = "IS"
+
+	print "/* This file is autogenerated by gen/prop/ccc; DO NOT EDIT. */"
+	print ""
+	print "#include \"_bsearch.h\""
+	print "#include \"macros.h\""
+	print "#include \"rune.h\""
+	print "#include \"unicode/prop.h\""
+	print ""
+}
+
+{
+	s = "CCC_" (map[$4] ? map[$4] : "NR")
+	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 {
+	print "static const struct {"
+	print "\trune lo, hi;"
+	print "\tenum uprop_ccc val;"
+	print "} lookup[] = {"
+
+	for (i = 0; i <= 0x10FFFF; i++) {
+		if (!props[i] || props[i] == "CCC_NR")
+			continue
+		for (lo = i; props[lo] == props[i + 1]; i++)
+			;
+		printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[lo]
+	}
+
+	print "};"
+	print ""
+	print "_MLIB_DEFINE_BSEARCH(enum uprop_ccc, lookup, CCC_NR)"
+	print ""
+	print "enum uprop_ccc"
+	print "uprop_get_ccc(rune ch)"
+	print "{"
+	print "\treturn ch < lookup[0].lo ? CCC_NR : mlib_lookup(ch);"
+	print "}"
+}
+' data/UnicodeData
-- 
cgit v1.2.3