Add uprop_get_nf{,k}{c,d}_qc()

1 files changed, 77 insertions, 0 deletions

diff --git a/gen/prop/nfXX_qc b/gen/prop/nfXX_qc
new file mode 100755
index 0000000..fed603e
--- /dev/null
+++ b/gen/prop/nfXX_qc
@@ -0,0 +1,77 @@
+#!/bin/sh
+
+set -e
+cd "${0%/*}/../.."
+
+for x in c d kc kd
+do
+	gawk -v s=$x '
+	BEGIN {
+		FS = "[ ;]+"
+
+		_default = "NF" toupper(s) "_QC_Y"
+
+		print "/* This file is autogenerated by gen/prop/nfXX_qc; DO NOT EDIT. */"
+		print ""
+		print "/* The macros.h include may be unused */"
+		print ""
+		print "#include \"_bsearch.h\""
+		print "#include \"macros.h\""
+		print "#include \"rune.h\""
+		print "#include \"unicode/prop.h\""
+		print ""
+	}
+
+	/^[A-F0-9]/ && $2 == "NF" toupper(s) "_QC" {
+		n = split($1, a, /\.\./)
+		lo = strtonum("0X" a[1])
+		hi = strtonum("0X" a[n])
+
+		for (i = lo; i <= hi; i++)
+			props[i] = "NF" toupper(s) "_QC_" $3
+		if (lo < 0x100)
+			want_lat1_tbl = 1
+	}
+
+	END {
+		if (want_lat1_tbl) {
+			print "static constexpr enum uprop_nf" s "_qc lookup_lat1[] = {"
+			for (i = 0; i < 0x100; i++) {
+				if (i % 8 == 0)
+					printf "\t"
+				printf "%s,%s", props[i] ? props[i] : _default, \
+					i % 8 == 7 ? "\n" : " "
+			}
+			print "};"
+			print ""
+		}
+
+		print "static const struct {"
+		print "\trune lo, hi;"
+		print "\tenum uprop_nf" s "_qc val;"
+		print "} lookup[] = {"
+
+		for (i = want_lat1_tbl ? 0x100 : 0; i <= 0x10FFFF; i++) {
+			if (!props[i])
+				continue
+			for (lo = i; props[lo] == props[i + 1]; i++)
+				;
+			printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[i]
+		}
+
+		print "};"
+		print ""
+		print "_MLIB_DEFINE_BSEARCH(enum uprop_nf" s "_qc, lookup, " _default ")"
+		print ""
+		print "enum uprop_nf" s "_qc"
+		print "uprop_get_nf" s "_qc(rune ch)"
+		print "{"
+		if (want_lat1_tbl)
+			print "\treturn ch < lengthof(lookup_lat1) ? lookup_lat1[ch] : mlib_lookup(ch);"
+		else
+			print "\treturn ch < lookup[0].lo ? " _default " : mlib_lookup(ch);"
+		print "}"
+	}
+	' data/DerivedNormalizationProps \
+	| sed 's/\s*$//' >lib/unicode/prop/uprop_get_nf${x}_qc.c
+done