4 files changed, 187 insertions, 14 deletions

diff --git a/vendor/librune/gen/gbrk b/vendor/librune/gen/gbrk
index 72ee2f7..1146327 100755
--- a/vendor/librune/gen/gbrk
+++ b/vendor/librune/gen/gbrk
@@ -26,8 +26,6 @@ wait
 cat <<C
 /* This file is autogenerated by gen/gbrk; DO NOT EDIT. */
 
-/* TODO: Change tables to constexpr from const when Clangd gets better */
-
 #ifndef RUNE_INTERNAL_GBRK_LOOKUP_H
 #define RUNE_INTERNAL_GBRK_LOOKUP_H
 
diff --git a/vendor/librune/gen/rtype b/vendor/librune/gen/rtype-cat
index 21916d3..e35fb77 100755
--- a/vendor/librune/gen/rtype
+++ b/vendor/librune/gen/rtype-cat
@@ -12,26 +12,24 @@ cache()
 
 set -e
 cd "${0%/*}/.."
-exec >include/internal/rtype_lookup.h
+exec >include/internal/rtype/cat.h
 
 readonly URL='https://www.unicode.org/Public/UCD/latest/ucd/UnicodeData.txt'
 cache "$URL"
 
 cat <<C
-/* This file is autogenerated by gen/gbrk; DO NOT EDIT. */
+/* This file is autogenerated by gen/rtype-cat; DO NOT EDIT. */
 
-/* TODO: Change tables to constexpr from const when Clangd gets better */
-
-#ifndef RUNE_INTERNAL_RTYPE_LOOKUP_H
-#define RUNE_INTERNAL_RTYPE_LOOKUP_H
+#ifndef RUNE_INTERNAL_RTYPE_CAT_H
+#define RUNE_INTERNAL_RTYPE_CAT_H
 
 /* IWYU pragma: private */
 /* clang-format off */
 
-#include "types.h"
-#include "../rtype.h"
+#include "../types.h"
+#include "../../rtype.h"
 
-static const unicat rtype_lat1_tbl[] = {
+static const enum unicat rtype_cat_lat1_tbl[] = {
 C
 
 gawk '
@@ -64,7 +62,7 @@ cat <<C
 
 static const struct {
 	rune lo, hi;
-	unicat cat;
+	enum unicat val;
 } rtype_cat_tbl[] = {
 C
 
@@ -88,7 +86,7 @@ BEGIN {
 }
 
 END {
-	for (i = 0; i <= 0x10FFFF; i++) {
+	for (i = 0x100; i <= 0x10FFFF; i++) {
 		if (!props[i])
 			continue
 		lo = i
@@ -102,5 +100,5 @@ END {
 cat <<C
 };
 
-#endif /* !RUNE_INTERNAL_RTYPE_LOOKUP_H */
+#endif /* !RUNE_INTERNAL_RTYPE_CAT_H */
 C
diff --git a/vendor/librune/gen/rtype-prop b/vendor/librune/gen/rtype-prop
new file mode 100755
index 0000000..4c62884
--- /dev/null
+++ b/vendor/librune/gen/rtype-prop
@@ -0,0 +1,116 @@
+#!/bin/sh
+
+cache()
+{
+	name="/tmp/librune/rtype/$(basename "$1")"
+	if test ! -f "$name"
+	then
+		mkdir -p /tmp/librune/rtype
+		wget -q "$1" -O "$name"
+	fi
+}
+
+cd "${0%/*}/.."
+
+readonly URL1='https://www.unicode.org/Public/UCD/latest/ucd/PropList.txt'
+readonly URL2='https://www.unicode.org/Public/UCD/latest/ucd/DerivedCoreProperties.txt'
+
+cache "$URL1" &
+cache "$URL2" &
+wait
+
+props1='
+ASCII_Hex_Digit
+Bidi_Control
+Dash
+Deprecated
+Diacritic
+Extender
+Hex_Digit
+ID_Compat_Math_Continue
+ID_Compat_Math_Start
+Ideographic
+IDS_Binary_Operator
+IDS_Trinary_Operator
+IDS_Unary_Operator
+Join_Control
+Logical_Order_Exception
+Noncharacter_Code_Point
+Pattern_Syntax
+Pattern_White_Space
+Prepended_Concatenation_Mark
+Quotation_Mark
+Radical
+Regional_Indicator
+Sentence_Terminal
+Soft_Dotted
+Terminal_Punctuation
+Unified_Ideograph
+Variation_Selector
+White_Space
+'
+
+props2='
+Alphabetic
+Cased
+Case_Ignorable
+Changes_When_Casefolded
+Changes_When_Casemapped
+Changes_When_Lowercased
+Changes_When_Titlecased
+Changes_When_Uppercased
+Default_Ignorable_Code_Point
+Grapheme_Base
+Grapheme_Extend
+ID_Continue
+ID_Start
+Indic_Conjunct_Break
+Lowercase
+Math
+Uppercase
+XID_Continue
+XID_Start
+'
+# InCB
+
+{
+	for prop in $props1
+	do
+		lprop=$(echo $prop | tr A-Z a-z)
+
+		printf 'rune_has_prop_%s…' $lprop >&2
+
+		gawk -M -v prop=$prop -f gen/rtype-prop.awk \
+			/tmp/librune/rtype/PropList.txt \
+			>lib/rtype/rune_has_prop_$lprop.c
+
+		echo "[[unsequenced]] bool rune_has_prop_$lprop(rune);"
+		echo ' DONE' >&2
+	done
+
+	for prop in $props2
+	do
+		lprop=$(echo $prop | tr A-Z a-z)
+
+		printf 'rune_has_prop_%s…' $lprop >&2
+
+		gawk -M -v prop=$prop -f gen/rtype-prop.awk \
+			/tmp/librune/rtype/DerivedCoreProperties.txt \
+			>lib/rtype/rune_has_prop_$lprop.c
+
+		echo "[[unsequenced]] bool rune_has_prop_$lprop(rune);"
+		echo ' DONE' >&2
+	done
+} | gawk '
+	/PROP PREDICATES END/ { no = 0 }
+	FILENAME != "-" && !no { print }
+	FILENAME == "-" { funcs[++i] = $0 }
+
+	/PROP PREDICATES START/ {
+		no = 1
+		asort(funcs)
+		for (i = 1; i <= length(funcs); i++)
+			print funcs[i]
+	}
+' - include/rtype.h \
+| sponge include/rtype.h
diff --git a/vendor/librune/gen/rtype-prop.awk b/vendor/librune/gen/rtype-prop.awk
new file mode 100644
index 0000000..59b4a99
--- /dev/null
+++ b/vendor/librune/gen/rtype-prop.awk
@@ -0,0 +1,61 @@
+BEGIN {
+	FS = "( *#.*| +; +)"
+
+	print "/* This file is autogenerated by gen/rtype-prop; DO NOT EDIT. */"
+	print ""
+	print "#include \"rtype.h\""
+	print ""
+	print "#include \"internal/common.h\""
+	print ""
+}
+
+$2 == prop || (prop == "Indic_Conjunct_Break" && $2 ~ /InCB;/) {
+	n = split($1, a, /\.\./)
+	lo = strtonum("0x" a[1])
+	hi = strtonum("0x" a[n])
+
+	for (i = lo; i <= hi; i++)
+		xs[i] = 1
+}
+
+END {
+	for (i = 0; i <= 0xFF; i++) {
+		if (xs[i])
+			mask = or(mask, lshift(1, i))
+	}
+	print  "#if BIT_LOOKUP"
+	printf "static const unsigned _BitInt(LATIN1_MAX + 1) mask = 0x%Xuwb;\n", \
+		mask
+	print  "#endif"
+	print  ""
+	print  "static const struct {"
+	print  "\trune lo, hi;"
+	print  "} lookup_tbl[] = {"
+
+	for (i = 0; i <= 0x10FFFF; i++) {
+		if (!xs[i])
+			continue
+		lo = i
+		while (xs[i + 1])
+			i++
+		printf "\t{0x%06X, 0x%06X},\n", lo, i
+	}
+
+	print  "};"
+	print  ""
+	print  "#define TYPE      bool"
+	print  "#define TABLE     lookup_tbl"
+	print  "#define DEFAULT   false"
+	print  "#define HAS_VALUE 0"
+	print  "#include \"internal/rtype/lookup-func.h\""
+	print  ""
+	print  "bool"
+	printf "rune_has_prop_%s(rune ch)\n", tolower(prop)
+	print  "{"
+	print  "\treturn"
+	print  "#if BIT_LOOKUP"
+	print  "\t\tch <= LATIN1_MAX ? (mask & ch) :"
+	print  "#endif"
+	print  "\t\tlookup(ch);"
+	print  "}"
+}