From bf212e80fb0f1d83902de520e9cee098187383f7 Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Mon, 8 Apr 2024 22:34:51 +0200
Subject: Tweaks to uprop_get_*()

---
 gen/prop/gc | 85 ++++++++++++++++++++++++-------------------------------------
 gen/prop/nv |  2 +-
 2 files changed, 34 insertions(+), 53 deletions(-)

(limited to 'gen')

diff --git a/gen/prop/gc b/gen/prop/gc
index 48c6659..219df8a 100755
--- a/gen/prop/gc
+++ b/gen/prop/gc
@@ -2,56 +2,19 @@
 
 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
+exec >lib/unicode/prop/uprop_get_gc.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 = ";"
+	print "/* This file is autogenerated by gen/prop/gc; DO NOT EDIT. */"
+	print ""
+	print "#include \"__bsearch.h\""
+	print "#include \"macros.h\""
+	print "#include \"rune.h\""
+	print "#include \"unicode/prop.h\""
+	print ""
 }
 
 {
@@ -69,6 +32,20 @@ BEGIN {
 }
 
 END {
+	print "static constexpr enum uprop_gc lookup_lat1[] = {"
+	for (i = 0; i < 0x100; i++) {
+		if (i % 8 == 0)
+			printf "\t"
+		printf "%5s,%s", props[i], i % 8 == 7 ? "\n" : " "
+	}
+	print "};"
+	print ""
+	print "static const struct {"
+	print "\trune lo, hi;"
+	print "\tenum uprop_gc val;"
+	print "} lookup[] = {"
+	print ""
+
 	for (i = 0x100; i <= 0x10FFFF; i++) {
 		if (!props[i])
 			continue
@@ -77,11 +54,15 @@ END {
 			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
+	print "};"
+	print ""
+	print "__MLIB_DEFINE_BSEARCH(enum uprop_gc, lookup, GC_CN)"
+	print ""
+	print "enum uprop_gc"
+	print "uprop_get_gc(rune ch)"
+	print "{"
+	print "\treturn ch < lengthof(lookup_lat1) ? lookup_lat1[ch] : mlib_lookup(ch);"
+	print "}"
+}
+' data/UnicodeData.txt
diff --git a/gen/prop/nv b/gen/prop/nv
index 7510704..4d81fa9 100755
--- a/gen/prop/nv
+++ b/gen/prop/nv
@@ -26,7 +26,7 @@ BEGIN {
 
 	for (i = lo; i <= hi; i++) {
 		gsub(/^; /, "", $3)
-		if ($3 ~ /\//)
+		if ($3 ~ /[^.]\//)
 			sub(/\//, "./", $3)
 		props[i] = $3
 	}
-- 
cgit v1.2.3