Add uprop_get_sb()

1 files changed, 78 insertions, 0 deletions

diff --git a/gen/prop/sb b/gen/prop/sb
new file mode 100755
index 0000000..aff06fd
--- /dev/null
+++ b/gen/prop/sb
@@ -0,0 +1,78 @@
+#!/bin/sh
+
+set -e
+cd "${0%/*}/../.."
+exec >lib/unicode/prop/uprop_get_sb.c
+
+gawk '
+BEGIN {
+	FS = " *(; *|#.*)"
+
+	map["ATerm"]     = "AT"
+	map["Close"]     = "CL"
+	map["CR"]        = "CR"
+	map["Extend"]    = "EX"
+	map["Format"]    = "FO"
+	map["LF"]        = "LF"
+	map["Lower"]     = "LO"
+	map["Numeric"]   = "NU"
+	map["OLetter"]   = "LE"
+	map["Other"]     = "XX"
+	map["SContinue"] = "SC"
+	map["Sep"]       = "SE"
+	map["Sp"]        = "SP"
+	map["STerm"]     = "ST"
+	map["Upper"]     = "UP"
+
+	print "/* This file is autogenerated by gen/prop/sb; DO NOT EDIT. */"
+	print ""
+	print "#include \"_bsearch.h\""
+	print "#include \"macros.h\""
+	print "#include \"rune.h\""
+	print "#include \"unicode/prop.h\""
+	print ""
+}
+
+/^[A-F0-9]/ {
+	n = split($1, a, /\.\./)
+	lo = strtonum("0X" a[1])
+	hi = strtonum("0X" a[n])
+
+	for (i = lo; i <= hi; i++)
+		props[i] = "SB_" map[$2]
+}
+
+END {
+	print "static constexpr enum uprop_sb lookup_lat1[] = {"
+	for (i = 0; i < 0x100; i++) {
+		if (i % 8 == 0)
+			printf "\t"
+		printf "%s%s", (props[i] ? props[i] : "SB_XX") ",", \
+			i % 8 == 7 ? "\n" : " "
+	}
+	print "};"
+	print ""
+	print "static const struct {"
+	print "\trune lo, hi;"
+	print "\tenum uprop_sb val;"
+	print "} lookup[] = {"
+
+	for (i = 0x100; i <= 0x10FFFF; i++) {
+		if (!props[i])
+			continue
+		for (lo = i; props[lo] == props[i + 1]; i++)
+			;
+		printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[i]
+	}
+
+	print "};"
+	print ""
+	print "_MLIB_DEFINE_BSEARCH(enum uprop_sb, lookup, SB_XX)"
+	print ""
+	print "enum uprop_sb"
+	print "uprop_get_sb(rune ch)"
+	print "{"
+	print "\treturn ch < lengthof(lookup_lat1) ? lookup_lat1[ch] : mlib_lookup(ch);"
+	print "}"
+}
+' data/SentenceBreakProperty | sed 's/\s*$//'