From d9dda44ced75c07b4e5b2897b74ca97ac7205ca7 Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Thu, 11 Apr 2024 14:56:15 +0200
Subject: Add uprop_get_lb()

---
 gen/data-files |  1 +
 gen/prop/lb    | 65 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 66 insertions(+)
 create mode 100755 gen/prop/lb

(limited to 'gen')

diff --git a/gen/data-files b/gen/data-files
index 3c87b73..550ef82 100755
--- a/gen/data-files
+++ b/gen/data-files
@@ -15,6 +15,7 @@ emoji/emoji-data.txt
 extracted/DerivedBinaryProperties.txt
 extracted/DerivedDecompositionType.txt
 extracted/DerivedEastAsianWidth.txt
+extracted/DerivedLineBreak.txt
 extracted/DerivedNumericType.txt
 extracted/DerivedNumericValues.txt
 PropList.txt
diff --git a/gen/prop/lb b/gen/prop/lb
new file mode 100755
index 0000000..25febdb
--- /dev/null
+++ b/gen/prop/lb
@@ -0,0 +1,65 @@
+#!/bin/sh
+
+set -e
+cd "${0%/*}/../.."
+exec >lib/unicode/prop/uprop_get_lb.c
+
+gawk '
+BEGIN {
+	FS = "( *#.*| *; +)"
+
+	print "/* This file is autogenerated by gen/prop/lb; DO NOT EDIT. */"
+	print ""
+	print "#include \"__bsearch.h\""
+	print "#include \"macros.h\""
+	print "#include \"rune.h\""
+	print "#include \"unicode/prop.h\""
+	print ""
+}
+
+/^[^#]/ {
+	n = split($1, a, /\.\./)
+	lo = strtonum("0X" a[1])
+	hi = strtonum("0X" a[n])
+
+	for (i = lo; i <= hi; i++) {
+		gsub(/^; /, "", $2)
+		props[i] = "LB_" $2
+	}
+}
+
+END {
+	print "static constexpr enum uprop_lb lookup_lat1[] = {"
+	for (i = 0; i < 0x100; i++) {
+		if (i % 8 == 0)
+			printf "\t"
+		printf "%-6s%s", (props[i] ? props[i] : "LB_XX") ",", \
+			i % 8 == 7 ? "\n" : " "
+	}
+	print "};"
+	print ""
+	print "static const struct {"
+	print "\trune lo, hi;"
+	print "\tenum uprop_lb val;"
+	print "} lookup[] = {"
+
+	for (i = 0x100; i <= 0x10FFFF; i++) {
+		if (!props[i])
+			continue
+		lo = i
+		while (props[lo] == props[i + 1])
+			i++
+		printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[i]
+	}
+
+	print "};"
+	print ""
+	print "__MLIB_DEFINE_BSEARCH(enum uprop_lb, lookup, LB_XX)"
+	print ""
+	print "enum uprop_lb"
+	print "uprop_get_lb(rune ch)"
+	print "{"
+	print "\treturn ch < lengthof(lookup_lat1) ? lookup_lat1[ch] : mlib_lookup(ch);"
+	print "}"
+}
+' data/DerivedLineBreak.txt | sed 's/\s*$//'
-- 
cgit v1.2.3