hare

commit 26fa7f8a86940e49eca619ec0b65102337e5bb00
parent abaf8ec95392523fc7e1155fd2fdbdbff717cdb8
Author: Haelwenn (lanodan) Monnier <contact+sr.ht@hacktivis.me>
Date:   Wed,  7 Dec 2022 17:17:57 +0100

net/uri: Acknowledge the different allowed characters

- path wasn't percent-encoded
- query and fragment had allowed characters being percent-encoded

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

Diffstat:
M
net/uri/+test.ha
 | 
14
++++++++++++--
M
net/uri/fmt.ha
 | 
35
+++++++++++++++++++++++++++--------
M
net/uri/query.ha
 | 
13
+++++++++----

3 files changed, 48 insertions(+), 14 deletions(-)
diff --git a/net/uri/+test.ha b/net/uri/+test.ha
@@ -68,7 +68,17 @@ use net::ip;
 			...
 		},
 	)!;
-
+	test_uri_roundtrip(
+		"https://sr.ht/projects?search=%23risc-v&sort=longest-active#foo",
+		uri {
+			scheme = "https",
+			host = "sr.ht",
+			path = "/projects",
+			query = "search=%23risc-v&sort=longest-active",
+			fragment = "foo",
+			...
+		},
+	)!;
 };
 
 @test fn invalid() void = {
@@ -101,7 +111,7 @@ use net::ip;
 			query = "objectClass?one",
 			...
 		},
-		"ldap://[2001:db8::7]/c=GB?objectClass%3Fone",
+		"ldap://[2001:db8::7]/c=GB?objectClass?one",
 	)!;
 
 	// https://bugs.chromium.org/p/chromium/issues/detail?id=841105
diff --git a/net/uri/fmt.ha b/net/uri/fmt.ha
@@ -7,6 +7,21 @@ use strconv;
 use strings;
 use strio;
 
+
+// Extract from RFC3986 ABNF
+// URI        = scheme ":" hier-part [ "?" query ] [ "#" fragment ]
+// unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~"
+// sub-delims = "!" / "$" / "&" / "'" / "(" / ")" / "*" / "+" / "," / ";" / "="
+// reg-name   = *( unreserved / pct-encoded / sub-delims )
+// host       = IP-literal / IPv4address / reg-name
+// pchar      = unreserved / pct-encoded / sub-delims / ":" / "@"
+// query      = *( pchar / "/" / "?" )
+// fragment   = *( pchar / "/" / "?" )
+
+def unres_host: str = "-._~!$&'()*+,;=";
+def unres_query_frag: str = "-._~!$&'()*+,;=:@/?";
+def unres_path: str = "-._~!$&'()*+,;=:@/";
+
 // Writes a formatted [[uri]] to an [[io::handle]]. Returns the number of bytes
 // written.
 export fn fmt(out: io::handle, u: *const uri) (size | io::error) = {
@@ -25,7 +40,12 @@ export fn fmt(out: io::handle, u: *const uri) (size | io::error) = {
 			if (!slashes_w) {
 				n += fmt::fprint(out, "//")?;
 			};
-			n += percent_encode(out, host)?;
+			let unres = if(u.scheme == "file") {
+				yield unres_path;
+			} else {
+				yield unres_host;
+			};
+			n += percent_encode(out, host, unres)?;
 		};
 	case let addr: ip::addr =>
 		if (!slashes_w) {
@@ -36,14 +56,14 @@ export fn fmt(out: io::handle, u: *const uri) (size | io::error) = {
 	if (u.port != 0) {
 		n += fmt::fprintf(out, ":{}", u.port)?;
 	};
-	n += fmt::fprint(out, u.path)?;
+	n += percent_encode(out, u.path, unres_path)?;
 	if (len(u.query) > 0) {
-		n += fmt::fprint(out, "?")?;
-		n += percent_encode(out, u.query)?;
+		// Always percent-encoded, see parse and encodequery/decodequery
+		n += fmt::fprintf(out, "?{}", u.query)?;
 	};
 	if (len(u.fragment) > 0) {
 		n += fmt::fprint(out, "#")?;
-		n += percent_encode(out, u.fragment)?;
+		n += percent_encode(out, u.fragment, unres_query_frag)?;
 	};
 
 	return n;
@@ -62,7 +82,7 @@ fn fmtaddr(out: io::handle, addr: ip::addr) (size | io::error) = {
 	return n;
 };
 
-fn percent_encode(out: io::handle, src: str) (size | io::error) = {
+fn percent_encode(out: io::handle, src: str, allowed: str) (size | io::error) = {
 	let iter = strings::iter(src);
 	let n = 0z;
 	for (true) {
@@ -72,8 +92,7 @@ fn percent_encode(out: io::handle, src: str) (size | io::error) = {
 		case =>
 			break;
 		};
-		// unreserved
-		if (ascii::isalnum(r) || strings::contains("-._~", r)) {
+		if (ascii::isalnum(r) || strings::contains(allowed, r)) {
 			n += fmt::fprint(out, r)?;
 		} else {
 			const en = utf8::encoderune(r);
diff --git a/net/uri/query.ha b/net/uri/query.ha
@@ -50,10 +50,10 @@ export fn encodequery(pairs: [](str, str)) str = {
 		if (i > 0) strio::appendrune(&buf, '&')!;
 
 		assert(len(pair.0) > 0);
-		percent_encode(&buf, pair.0)!;
+		percent_encode(&buf, pair.0, unres_query_frag)!;
 		if (len(pair.1) > 0) {
 			strio::appendrune(&buf, '=')!;
-			percent_encode(&buf, pair.1)!;
+			percent_encode(&buf, pair.1, unres_query_frag)!;
 		};
 	};
 
@@ -61,7 +61,7 @@ export fn encodequery(pairs: [](str, str)) str = {
 };
 
 @test fn decodequery() void = {
-	const u = parse("https://sr.ht/projects?search=%23risc-v&sort=longest-active")!;
+	const u = parse("https://sr.ht/projects?search=%23risc-v&sort=longest-active&quantity=100%25")!;
 	defer finish(&u);
 
 	const query = decodequery(u.query);
@@ -73,15 +73,20 @@ export fn encodequery(pairs: [](str, str)) str = {
 	const pair = query_next(&query)! as (str, str);
 	assert(pair.0 == "sort");
 	assert(pair.1 == "longest-active");
+
+	const pair = query_next(&query)! as (str, str);
+	assert(pair.0 == "quantity");
+	assert(pair.1 == "100%");
 };
 
 @test fn encodequery() void = {
 	const pairs = [
 		("search", "#risc-v"),
 		("sort", "longest-active"),
+		("quantity", "100%")
 	];
 	const encoded = encodequery(pairs);
 	defer free(encoded);
 
-	assert(encoded == "search=%23risc-v&sort=longest-active");
+	assert(encoded == "search=%23risc-v&sort=longest-active&quantity=100%25");
 };