Add uprop_get_dt()

2 files changed, 85 insertions, 0 deletions

diff --git a/gen/data-files b/gen/data-files
index 0946f7e..3c87b73 100755
--- a/gen/data-files
+++ b/gen/data-files
@@ -13,6 +13,7 @@ DerivedCoreProperties.txt
 DerivedNormalizationProps.txt
 emoji/emoji-data.txt
 extracted/DerivedBinaryProperties.txt
+extracted/DerivedDecompositionType.txt
 extracted/DerivedEastAsianWidth.txt
 extracted/DerivedNumericType.txt
 extracted/DerivedNumericValues.txt
diff --git a/gen/prop/dt b/gen/prop/dt
new file mode 100755
index 0000000..ba0582a
--- /dev/null
+++ b/gen/prop/dt
@@ -0,0 +1,84 @@
+#!/bin/sh
+
+set -e
+cd "${0%/*}/../.."
+exec >lib/unicode/prop/uprop_get_dt.c
+
+gawk '
+BEGIN {
+	FS = "( *#.*| +; +)"
+
+	map["Canonical"] = "CAN"
+	map["Compat"]    = "COM"
+	map["Circle"]    = "ENC"
+	map["Final"]     = "FIN"
+	map["Font"]      = "FONT"
+	map["Fraction"]  = "FRA"
+	map["Initial"]   = "INIT"
+	map["Isolated"]  = "ISO"
+	map["Medial"]    = "MED"
+	map["Narrow"]    = "NAR"
+	map["Nobreak"]   = "NB"
+	map["Small"]     = "SML"
+	map["Square"]    = "SQR"
+	map["Sub"]       = "SUB"
+	map["Super"]     = "SUP"
+	map["Vertical"]  = "VERT"
+	map["Wide"]      = "WIDE"
+
+
+	print "/* This file is autogenerated by gen/prop/dt; 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] = "DT_" map[$2]
+	}
+}
+
+END {
+	print "static constexpr enum uprop_dt lookup_lat1[] = {"
+	for (i = 0; i < 0x100; i++) {
+		if (i % 8 == 0)
+			printf "\t"
+		printf "%-8s%s", (props[i] ? props[i] : "DT_NONE") ",", \
+			i % 8 == 7 ? "\n" : " "
+	}
+	print "};"
+	print ""
+	print "static const struct {"
+	print "\trune lo, hi;"
+	print "\tenum uprop_dt 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_dt, lookup, DT_NONE)"
+	print ""
+	print "enum uprop_dt"
+	print "uprop_get_dt(rune ch)"
+	print "{"
+	print "\treturn ch < lengthof(lookup_lat1) ? lookup_lat1[ch] : mlib_lookup(ch);"
+	print "}"
+}
+' data/DerivedDecompositionType.txt