Add boolean properties to unicode/prop.h

1 files changed, 79 insertions, 0 deletions

diff --git a/gen/prop/bool-props.awk b/gen/prop/bool-props.awk
new file mode 100644
index 0000000..d9c6299
--- /dev/null
+++ b/gen/prop/bool-props.awk
@@ -0,0 +1,79 @@
+BEGIN {
+	FS = "( *#.*| +; +)"
+
+	print "/* This file is autogenerated by gen/prop/bool-props; DO NOT EDIT. */"
+	print ""
+	print "#include \"__bsearch.h\""
+	print "#include \"bitset.h\""
+	print "#include \"rune.h\""
+	print "#include \"unicode/prop.h\""
+	print ""
+	print "/* clang-format off */"
+	print ""
+}
+
+$2 == prop || (prop == "Indic_Conjunct_Break" && $2 ~ /InCB;/) {
+	n = split($1, a, /\.\./)
+	lo = strtonum("0x" a[1])
+	hi = strtonum("0x" a[n])
+
+	for (i = lo; i <= hi; i++)
+		xs[i] = 1
+}
+
+END {
+	for (i = 0; i <= 0xFF; i++) {
+		if (xs[i])
+			mask = or(mask, lshift(1, i))
+	}
+
+	if (mask > 0) {
+		print "static constexpr bitset(bs, LATIN1_MAX) = {"
+		for (i = 0; i < 32; i++) {
+			if (i % 8 == 0)
+				printf "\t"
+			printf "0x%02X,", and(rshift(mask, 8 * i), 0xFF)
+			printf((i % 8 == 7) ? "\n" : " ")
+		}
+		print "};"
+		print  ""
+	}
+
+	for (i = 0x100; i <= 0x10FFFF; i++) {
+		if (xs[i]) {
+			need_big_lookup = 1
+			break
+		}
+	}
+
+	if (need_big_lookup) {
+		print "static const struct {"
+		print "\trune lo, hi;"
+		print "} lookup[] = {"
+
+		for (i = 0x100; i <= 0x10FFFF; i++) {
+			if (!xs[i])
+				continue
+			lo = i
+			while (xs[i + 1])
+				i++
+			printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X)},\n", lo, i
+		}
+
+		print  "};"
+		print  ""
+		print  "__MLIB_DEFINE_BSEARCH_CONTAINS(lookup)"
+		print  ""
+	}
+
+	print  "bool"
+	printf "uprop_is_%s(rune ch)\n", short
+	print  "{"
+	if (mask > 0 && need_big_lookup)
+		print "\treturn ch <= LATIN1_MAX ? TESTBIT(bs, ch) : mlib_lookup_contains(ch);"
+	else if (need_big_lookup)
+		print "\treturn mlib_lookup_contains(ch);"
+	else
+		print "\treturn ch <= LATIN1_MAX && TESTBIT(bs, ch);"
+	print  "}"
+}