Add uprop_get_gcb()

1 files changed, 83 insertions, 0 deletions

diff --git a/gen/prop/gcb b/gen/prop/gcb
new file mode 100755
index 0000000..17d3136
--- /dev/null
+++ b/gen/prop/gcb
@@ -0,0 +1,83 @@
+#!/bin/sh
+
+set -e
+cd "${0%/*}/../.."
+exec >lib/unicode/prop/uprop_get_gcb.c
+
+gawk '
+BEGIN {
+	FS = "( *#.*| *; +)"
+
+	map["Control"]            = "CN"
+	map["CR"]                 = "CR"
+	map["E_Base"]             = "EB"
+	map["E_Base_GAZ"]         = "EBG"
+	map["E_Modifier"]         = "EM"
+	map["Extend"]             = "EX"
+	map["Glue_After_Zwj"]     = "GAZ"
+	map["LF"]                 = "LF"
+	map["L"]                  = "L"
+	map["LV"]                 = "LV"
+	map["LVT"]                = "LVT"
+	map["Other"]              = "XX"
+	map["Prepend"]            = "PP"
+	map["Regional_Indicator"] = "RI"
+	map["SpacingMark"]        = "SM"
+	map["T"]                  = "T"
+	map["V"]                  = "V"
+	map["ZWJ"]                = "ZWJ"
+
+	print "/* This file is autogenerated by gen/prop/gcb; 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] = "GCB_" map[$2]
+	}
+}
+
+END {
+	print "static constexpr enum uprop_gcb lookup_lat1[] = {"
+	for (i = 0; i < 0x100; i++) {
+		if (i % 8 == 0)
+			printf "\t"
+		printf "%-7s%s", (props[i] ? props[i] : "GCB_XX") ",", \
+			i % 8 == 7 ? "\n" : " "
+	}
+	print "};"
+	print ""
+	print "static const struct {"
+	print "\trune lo, hi;"
+	print "\tenum uprop_gcb val;"
+	print "} lookup[] = {"
+
+	for (i = 0x100; i <= 0x10FFFF; i++) {
+		if (!props[i])
+			continue
+		for (lo = i; props[lo] == props[i + 1]; i++)
+			;
+		printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[i]
+	}
+
+	print "};"
+	print ""
+	print "_MLIB_DEFINE_BSEARCH(enum uprop_gcb, lookup, GCB_XX)"
+	print ""
+	print "enum uprop_gcb"
+	print "uprop_get_gcb(rune ch)"
+	print "{"
+	print "\treturn ch < lengthof(lookup_lat1) ? lookup_lat1[ch] : mlib_lookup(ch);"
+	print "}"
+}
+' data/GraphemeBreakProperty | sed 's/\s*$//'