Add uprop_get_age()

2 files changed, 56 insertions, 0 deletions

diff --git a/gen/data-files b/gen/data-files
index 1f154ac..801f591 100755
--- a/gen/data-files
+++ b/gen/data-files
@@ -9,6 +9,7 @@ readonly BASE=https://www.unicode.org/Public/UCD/latest/ucd
 readonly PATHS='
 auxiliary/GraphemeBreakProperty
 BidiBrackets
+DerivedAge
 DerivedCoreProperties
 DerivedNormalizationProps
 emoji/emoji-data
diff --git a/gen/prop/age b/gen/prop/age
new file mode 100755
index 0000000..d0f742f
--- /dev/null
+++ b/gen/prop/age
@@ -0,0 +1,55 @@
+#!/bin/sh
+
+set -e
+cd "${0%/*}/../.."
+exec >lib/unicode/prop/uprop_get_age.c
+
+gawk '
+BEGIN {
+	FS = " *(; *|#.*)"
+
+	print "/* This file is autogenerated by gen/prop/age; DO NOT EDIT. */"
+	print ""
+	print "#include \"__bsearch.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] = "AGE_V" int($2) "_" ($2 % 1 * 10)
+	}
+}
+
+END {
+	print "static const struct {"
+	print "\trune lo, hi;"
+	print "\tenum uprop_age val;"
+	print "} lookup[] = {"
+
+	for (i = 0x1F6; i <= 0x10FFFF; i++) {
+		if (!props[i])
+			continue
+		lo = i
+		while (props[lo] == props[i + 1])
+			i++
+		printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[i]
+	}
+
+	print "};"
+	print ""
+	print "__MLIB_DEFINE_BSEARCH(enum uprop_age, lookup, AGE_NA)"
+	print ""
+	print "enum uprop_age"
+	print "uprop_get_age(rune ch)"
+	print "{"
+	print "\treturn ch <= RUNE_C(0x01F5) ? AGE_V1_1 : mlib_lookup(ch);"
+	print "}"
+}
+' data/DerivedAge | sed 's/\s*$//'