Merge pull request #14434 from NixOS/improve-ipv6-zoneid-backcompat

libstore: Improve store-reference back-compat with IPv6 ZoneId literals

Author: Ericson2314

src/libstore-tests/data/store-reference/ssh_unbracketed_ipv6_4.txt

@@ -0,0 +1 @@
+ssh://userinfo@[fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%eth0]?a=b&c=d

src/libstore-tests/data/store-reference/ssh_unbracketed_ipv6_5.txt

@@ -0,0 +1 @@
+ssh://userinfo@[fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%25eth0]?a=b&c=d

src/libstore-tests/data/store-reference/ssh_unbracketed_ipv6_6.txt

@@ -0,0 +1 @@
+ssh://userinfo@fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%25?a=b&c=d

src/libstore-tests/data/store-reference/ssh_unbracketed_ipv6_7.txt

@@ -0,0 +1 @@
+ssh://userinfo@fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%eth0?a=b&c=d

src/libstore-tests/data/store-reference/ssh_unbracketed_ipv6_8.txt

@@ -0,0 +1 @@
+ssh://fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%eth0?a=b&c=d

src/libstore-tests/data/store-reference/ssh_unbracketed_ipv6_9.txt

@@ -0,0 +1 @@
+ssh://fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%eth0

src/libstore-tests/store-reference.cc

@@ -186,4 +186,64 @@ static StoreReference sshIPv6AuthorityWithUserinfoAndParams{
 
 URI_TEST_READ(ssh_unbracketed_ipv6_3, sshIPv6AuthorityWithUserinfoAndParams)
 
+static const StoreReference sshIPv6AuthorityWithUserinfoAndParamsAndZoneId{
+    .variant =
+        StoreReference::Specified{
+            .scheme = "ssh",
+            .authority = "userinfo@[fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%25eth0]",
+        },
+    .params =
+        {
+            {"a", "b"},
+            {"c", "d"},
+        },
+};
+
+URI_TEST_READ(ssh_unbracketed_ipv6_4, sshIPv6AuthorityWithUserinfoAndParamsAndZoneId)
+URI_TEST_READ(ssh_unbracketed_ipv6_5, sshIPv6AuthorityWithUserinfoAndParamsAndZoneId)
+
+static const StoreReference sshIPv6AuthorityWithUserinfoAndParamsAndZoneIdTricky{
+    .variant =
+        StoreReference::Specified{
+            .scheme = "ssh",
+            .authority = "userinfo@[fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%2525]",
+        },
+    .params =
+        {
+            {"a", "b"},
+            {"c", "d"},
+        },
+};
+
+// Non-standard syntax where the IPv6 literal appears without brackets. In
+// this case don't considering %25 to be a pct-encoded % and just take it as a
+// literal value. 25 is a perfectly legal ZoneId value in theory.
+URI_TEST_READ(ssh_unbracketed_ipv6_6, sshIPv6AuthorityWithUserinfoAndParamsAndZoneIdTricky)
+URI_TEST_READ(ssh_unbracketed_ipv6_7, sshIPv6AuthorityWithUserinfoAndParamsAndZoneId)
+
+static const StoreReference sshIPv6AuthorityWithParamsAndZoneId{
+    .variant =
+        StoreReference::Specified{
+            .scheme = "ssh",
+            .authority = "[fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%25eth0]",
+        },
+    .params =
+        {
+            {"a", "b"},
+            {"c", "d"},
+        },
+};
+
+URI_TEST_READ(ssh_unbracketed_ipv6_8, sshIPv6AuthorityWithParamsAndZoneId)
+
+static const StoreReference sshIPv6AuthorityWithZoneId{
+    .variant =
+        StoreReference::Specified{
+            .scheme = "ssh",
+            .authority = "[fea5:23e1:3916:fc24:cb52:2837:2ecb:ea8e%25eth0]",
+        },
+};
+
+URI_TEST_READ(ssh_unbracketed_ipv6_9, sshIPv6AuthorityWithZoneId)
+
 } // namespace nix

src/libstore/store-reference.cc

@@ -121,15 +121,39 @@ StoreReference StoreReference::parse(const std::string & uri, const StoreReferen
              * greedily assumed to be the part of the host address. */
             auto authorityString = schemeAndAuthority->authority;
             auto userinfo = splitPrefixTo(authorityString, '@');
-            auto maybeIpv6 = boost::urls::parse_ipv6_address(authorityString);
+            /* Back-compat shim for ZoneId specifiers. Technically this isn't
+             * standard, but the expectation is this works with the old syntax
+             * for ZoneID specifiers. For the full story behind the fiasco that
+             * is ZoneID in URLs look at [^].
+             * [^]: https://datatracker.ietf.org/doc/html/draft-schinazi-httpbis-link-local-uri-bcp-03
+             */
+
+            /* Fish out the internals from inside square brackets. It might be that the pct-sign is unencoded and that's
+             * why we failed to parse it previously. */
+            if (authorityString.starts_with('[') && authorityString.ends_with(']')) {
+                authorityString.remove_prefix(1);
+                authorityString.remove_suffix(1);
+            }
+
+            auto maybeBeforePct = splitPrefixTo(authorityString, '%');
+            bool hasZoneId = maybeBeforePct.has_value();
+            auto maybeZoneId = hasZoneId ? std::optional{authorityString} : std::nullopt;
+
+            std::string_view maybeIpv6S = maybeBeforePct.value_or(authorityString);
+            auto maybeIpv6 = boost::urls::parse_ipv6_address(maybeIpv6S);
+
             if (maybeIpv6) {
                 std::string fixedAuthority;
                 if (userinfo) {
                     fixedAuthority += *userinfo;
                     fixedAuthority += '@';
                 }
                 fixedAuthority += '[';
-                fixedAuthority += authorityString;
+                fixedAuthority += maybeIpv6S;
+                if (maybeZoneId) {
+                    fixedAuthority += "%25"; // pct-encoded percent character
+                    fixedAuthority += *maybeZoneId;
+                }
                 fixedAuthority += ']';
                 return {
                     .variant =