hare

commit a3749ad1ff35cfcd356b9aae2682001121853390
parent 26fa7f8a86940e49eca619ec0b65102337e5bb00
Author: Haelwenn (lanodan) Monnier <contact+sr.ht@hacktivis.me>
Date:   Wed,  7 Dec 2022 17:17:58 +0100

net/uri: Fix decoding multi-byte percent-data

Technically it can be non-UTF-8 but hare heavily assumes UTF-8.

Signed-off-by: Haelwenn (lanodan) Monnier <contact+sr.ht@hacktivis.me>

Diffstat:
M
net/uri/+test.ha
 | 
10
++++++++++
M
net/uri/parse.ha
 | 
26
+++++++++++++++++++++++++-
M
scripts/gen-stdlib
 | 
2
+-
M
stdlib.mk
 | 
4
++--

4 files changed, 38 insertions(+), 4 deletions(-)
diff --git a/net/uri/+test.ha b/net/uri/+test.ha
@@ -79,6 +79,16 @@ use net::ip;
 			...
 		},
 	)!;
+	test_uri_roundtrip(
+		"https://en.wiktionary.org/wiki/%E3%81%8A%E3%81%AF%E3%82%88%E3%81%86#Japanese",
+		uri {
+			scheme = "https",
+			host = "en.wiktionary.org",
+			path = "/wiki/おはよう",
+			fragment = "Japanese",
+			...
+		}
+	)!;
 };
 
 @test fn invalid() void = {
diff --git a/net/uri/parse.ha b/net/uri/parse.ha
@@ -2,6 +2,7 @@
 // (c) 2022 Alexey Yerin <yyp@disroot.org>
 // (c) 2022 Umar Getagazov <umar@handlerug.me>
 use ascii;
+use encoding::utf8;
 use io;
 use net::ip;
 use strconv;
@@ -329,6 +330,7 @@ fn percent_decode_static(out: io::handle, s: str) (void | invalid) = {
 	let iter = strings::iter(s);
 	let tmp = strio::dynamic();
 	defer io::close(&tmp)!;
+	let percent_data: []u8 = [];
 	for (true) {
 		match (strings::next(&iter)) {
 		case let r: rune =>
@@ -342,14 +344,36 @@ fn percent_decode_static(out: io::handle, s: str) (void | invalid) = {
 				match (strconv::stou8b(strio::string(&tmp),
 					strconv::base::HEX)) {
 				case let ord: u8 =>
-					strio::appendrune(out, ord: u32: rune)!;
+					append(percent_data, ord);
 				case =>
 					return invalid;
 				};
 			} else {
+				if(len(percent_data) > 0) {
+					match(strings::fromutf8(percent_data)) {
+					case let stro: str =>
+						strio::concat(out, stro)!;
+					case utf8::invalid =>
+						return invalid;
+					};
+
+					percent_data = [];
+				};
+
 				strio::appendrune(out, r)!;
 			};
 		case void =>
+			if(len(percent_data) > 0) {
+				match(strings::fromutf8(percent_data)) {
+				case let stro: str =>
+					strio::concat(out, stro)!;
+				case utf8::invalid =>
+					return invalid;
+				};
+
+				percent_data = [];
+			};
+
 			break;
 		};
 	};
diff --git a/scripts/gen-stdlib b/scripts/gen-stdlib
@@ -1100,7 +1100,7 @@ net_uri() {
 			+test.ha
 	fi
 	gen_ssa net::uri \
-		ascii ip net::ip strconv strings strio
+		ascii encoding::utf8 ip net::ip strconv strings strio
 }
 
 gensrcs_math_complex() {
diff --git a/stdlib.mk b/stdlib.mk
@@ -1755,7 +1755,7 @@ stdlib_net_uri_any_srcs = \
 	$(STDLIB)/net/uri/query.ha \
 	$(STDLIB)/net/uri/uri.ha
 
-$(HARECACHE)/net/uri/net_uri-any.ssa: $(stdlib_net_uri_any_srcs) $(stdlib_rt) $(stdlib_ascii_$(PLATFORM)) $(stdlib_ip_$(PLATFORM)) $(stdlib_net_ip_$(PLATFORM)) $(stdlib_strconv_$(PLATFORM)) $(stdlib_strings_$(PLATFORM)) $(stdlib_strio_$(PLATFORM))
+$(HARECACHE)/net/uri/net_uri-any.ssa: $(stdlib_net_uri_any_srcs) $(stdlib_rt) $(stdlib_ascii_$(PLATFORM)) $(stdlib_encoding_utf8_$(PLATFORM)) $(stdlib_ip_$(PLATFORM)) $(stdlib_net_ip_$(PLATFORM)) $(stdlib_strconv_$(PLATFORM)) $(stdlib_strings_$(PLATFORM)) $(stdlib_strio_$(PLATFORM))
 	@printf 'HAREC \t$@\n'
 	@mkdir -p $(HARECACHE)/net/uri
 	@HARECACHE=$(HARECACHE) $(HAREC) $(HAREFLAGS) -o $@ -Nnet::uri \
@@ -4010,7 +4010,7 @@ testlib_net_uri_any_srcs = \
 	$(STDLIB)/net/uri/uri.ha \
 	$(STDLIB)/net/uri/+test.ha
 
-$(TESTCACHE)/net/uri/net_uri-any.ssa: $(testlib_net_uri_any_srcs) $(testlib_rt) $(testlib_ascii_$(PLATFORM)) $(testlib_ip_$(PLATFORM)) $(testlib_net_ip_$(PLATFORM)) $(testlib_strconv_$(PLATFORM)) $(testlib_strings_$(PLATFORM)) $(testlib_strio_$(PLATFORM))
+$(TESTCACHE)/net/uri/net_uri-any.ssa: $(testlib_net_uri_any_srcs) $(testlib_rt) $(testlib_ascii_$(PLATFORM)) $(testlib_encoding_utf8_$(PLATFORM)) $(testlib_ip_$(PLATFORM)) $(testlib_net_ip_$(PLATFORM)) $(testlib_strconv_$(PLATFORM)) $(testlib_strings_$(PLATFORM)) $(testlib_strio_$(PLATFORM))
 	@printf 'HAREC \t$@\n'
 	@mkdir -p $(TESTCACHE)/net/uri
 	@HARECACHE=$(TESTCACHE) $(HAREC) $(TESTHAREFLAGS) -o $@ -Nnet::uri \