hare

commit bd52e831d9fb1024c98ef5a1f501a408a3a8328c
parent 0fa2a28048a20a83813b0ec4ec6b29774d1662d1
Author: Sebastian <sebastian@sebsite.pw>
Date:   Wed, 13 Apr 2022 20:33:46 -0400

ascii: add validstr

Signed-off-by: Sebastian <sebastian@sebsite.pw>

Diffstat:
M
ascii/strcmp.ha
 | 
4
++--
M
ascii/valid.ha
 | 
15
+++++++++++++++
M
net/dns/decode.ha
 | 
2
+-
M
scripts/gen-stdlib
 | 
5
+++--
M
stdlib.mk
 | 
10
++++++----

5 files changed, 27 insertions(+), 9 deletions(-)
diff --git a/ascii/strcmp.ha b/ascii/strcmp.ha
@@ -80,13 +80,13 @@ export fn strcasecmp(a: str, b: str) (int | void) = {
 			case void =>
 				break;
 			case let r: rune =>
-				if (!isascii(r)) return;
+				if (!valid(r)) return;
 			};
 			return 1;
 		case let r: rune =>
 			yield r;
 		};
-		if (!isascii(ra) || !isascii(rb)) {
+		if (!valid(ra) || !valid(rb)) {
 			return;
 		};
 		let ra = tolower(ra), rb = tolower(rb);
diff --git a/ascii/valid.ha b/ascii/valid.ha
@@ -1,11 +1,26 @@
 // License: MPL-2.0
 // (c) 2022 Sebastian <sebastian@sebsite.pw>
 // (c) 2021 Drew Devault <sir@cmpwn.com>
+use strings;
 
 // Returns true if a rune is a valid ASCII character.
 export fn valid(c: rune) bool = c: u32 <= 0o177;
 
+// Returns true if all runes in a string are valid ASCII characters.
+export fn validstr(s: str) bool = {
+	const s = strings::toutf8(s);
+	for (let i = 0z; i < len(s); i += 1) {
+		if (s[i] >= 0200) {
+			return false;
+		};
+	};
+	return true;
+};
+
 @test fn valid() void = {
 	assert(valid('a') && valid('\0') && valid('\x7F'));
 	assert(!valid('\x80') && !valid('こ'));
+	assert(validstr("abc\0"));
+	assert(!validstr("こんにちは"));
+	assert(!validstr("ABCこんにちは"));
 };
diff --git a/net/dns/decode.ha b/net/dns/decode.ha
@@ -112,7 +112,7 @@ fn decode_name(dec: *decoder) ([]str | format) = {
 		const name = dec.cur[..z];
 		dec.cur = dec.cur[z..];
 		for (let i = 0z; i < len(name); i += 1) {
-			if (!ascii::isascii(name[i]: u32: rune)) {
+			if (!ascii::valid(name[i]: u32: rune)) {
 				return format;
 			};
 		};
diff --git a/scripts/gen-stdlib b/scripts/gen-stdlib
@@ -157,7 +157,8 @@ EOF
 ascii() {
 	gen_srcs ascii \
 		ctype.ha \
-		strcmp.ha
+		strcmp.ha \
+		valid.ha
 	gen_ssa ascii strings
 }
 
@@ -601,7 +602,7 @@ hare_types() {
 	else
 		gensrcs_hare_types
 	fi
-	gen_ssa hare::types hare::ast hare::parse hash hash::fnv endian io
+	gen_ssa hare::types hare::ast hare::parse hash hash::fnv endian io strings
 }
 
 gensrcs_hare_unit() {
diff --git a/stdlib.mk b/stdlib.mk
@@ -689,7 +689,8 @@ stdlib_uuid_freebsd=$(stdlib_uuid_any)
 # ascii (+any)
 stdlib_ascii_any_srcs= \
 	$(STDLIB)/ascii/ctype.ha \
-	$(STDLIB)/ascii/strcmp.ha
+	$(STDLIB)/ascii/strcmp.ha \
+	$(STDLIB)/ascii/valid.ha
 
 $(HARECACHE)/ascii/ascii-any.ssa: $(stdlib_ascii_any_srcs) $(stdlib_rt) $(stdlib_strings_$(PLATFORM))
 	@printf 'HAREC \t$@\n'
@@ -1182,7 +1183,7 @@ stdlib_hare_types_any_srcs= \
 	$(STDLIB)/hare/types/store.ha \
 	$(STDLIB)/hare/types/types.ha
 
-$(HARECACHE)/hare/types/hare_types-any.ssa: $(stdlib_hare_types_any_srcs) $(stdlib_rt) $(stdlib_hare_ast_$(PLATFORM)) $(stdlib_hare_parse_$(PLATFORM)) $(stdlib_hash_$(PLATFORM)) $(stdlib_hash_fnv_$(PLATFORM)) $(stdlib_endian_$(PLATFORM)) $(stdlib_io_$(PLATFORM))
+$(HARECACHE)/hare/types/hare_types-any.ssa: $(stdlib_hare_types_any_srcs) $(stdlib_rt) $(stdlib_hare_ast_$(PLATFORM)) $(stdlib_hare_parse_$(PLATFORM)) $(stdlib_hash_$(PLATFORM)) $(stdlib_hash_fnv_$(PLATFORM)) $(stdlib_endian_$(PLATFORM)) $(stdlib_io_$(PLATFORM)) $(stdlib_strings_$(PLATFORM))
 	@printf 'HAREC \t$@\n'
 	@mkdir -p $(HARECACHE)/hare/types
 	@HARECACHE=$(HARECACHE) $(HAREC) $(HAREFLAGS) -o $@ -Nhare::types \
@@ -2658,7 +2659,8 @@ testlib_uuid_freebsd=$(testlib_uuid_any)
 # ascii (+any)
 testlib_ascii_any_srcs= \
 	$(STDLIB)/ascii/ctype.ha \
-	$(STDLIB)/ascii/strcmp.ha
+	$(STDLIB)/ascii/strcmp.ha \
+	$(STDLIB)/ascii/valid.ha
 
 $(TESTCACHE)/ascii/ascii-any.ssa: $(testlib_ascii_any_srcs) $(testlib_rt) $(testlib_strings_$(PLATFORM))
 	@printf 'HAREC \t$@\n'
@@ -3180,7 +3182,7 @@ testlib_hare_types_any_srcs= \
 	$(STDLIB)/hare/types/types.ha \
 	$(STDLIB)/hare/types/+test.ha
 
-$(TESTCACHE)/hare/types/hare_types-any.ssa: $(testlib_hare_types_any_srcs) $(testlib_rt) $(testlib_hare_ast_$(PLATFORM)) $(testlib_hare_parse_$(PLATFORM)) $(testlib_hash_$(PLATFORM)) $(testlib_hash_fnv_$(PLATFORM)) $(testlib_endian_$(PLATFORM)) $(testlib_io_$(PLATFORM))
+$(TESTCACHE)/hare/types/hare_types-any.ssa: $(testlib_hare_types_any_srcs) $(testlib_rt) $(testlib_hare_ast_$(PLATFORM)) $(testlib_hare_parse_$(PLATFORM)) $(testlib_hash_$(PLATFORM)) $(testlib_hash_fnv_$(PLATFORM)) $(testlib_endian_$(PLATFORM)) $(testlib_io_$(PLATFORM)) $(testlib_strings_$(PLATFORM))
 	@printf 'HAREC \t$@\n'
 	@mkdir -p $(TESTCACHE)/hare/types
 	@HARECACHE=$(TESTCACHE) $(HAREC) $(TESTHAREFLAGS) -o $@ -Nhare::types \