Fix URL parser

Fixes #3062.

3 files changed, 25 insertions, 21 deletions

diff --git a/src/libexpr/flake/flakeref.cc b/src/libexpr/flake/flakeref.cc
index a34d766d7..c46661df8 100644
--- a/src/libexpr/flake/flakeref.cc
+++ b/src/libexpr/flake/flakeref.cc
@@ -67,8 +67,10 @@ std::pair<FlakeRef, std::string> parseFlakeRefWithFragment(
 {
     using namespace fetchers;
 
+    static std::string fnRegex = "[0-9a-zA-Z-._~!$&'\"()*+,;=]+";
+
     static std::regex pathUrlRegex(
-        "(" + pathRegex + "/?)"
+        "(/?" + fnRegex + "(?:/" + fnRegex + ")*/?)"
         + "(?:\\?(" + queryRegex + "))?"
         + "(?:#(" + queryRegex + "))?",
         std::regex::ECMAScript);
diff --git a/src/libstore/fetchers/parse.cc b/src/libstore/fetchers/parse.cc
index dc1b3efe6..4f7cb3c6b 100644
--- a/src/libstore/fetchers/parse.cc
+++ b/src/libstore/fetchers/parse.cc
@@ -11,24 +11,22 @@ std::regex flakeIdRegex(flakeIdRegexS, std::regex::ECMAScript);
 ParsedURL parseURL(const std::string & url)
 {
     static std::regex uriRegex(
-        "(((" + schemeRegex + "):"
-        + "(//(" + authorityRegex + "))?"
-        + "(" + pathRegex + "))"
+        "((" + schemeRegex + "):"
+        + "(?:(?://(" + authorityRegex + ")(" + absPathRegex + "))|(/?" + pathRegex + ")))"
         + "(?:\\?(" + queryRegex + "))?"
-        + "(?:#(" + queryRegex + "))?"
-        + ")",
+        + "(?:#(" + queryRegex + "))?",
         std::regex::ECMAScript);
 
     std::smatch match;
 
     if (std::regex_match(url, match, uriRegex)) {
-        auto & base = match[2];
-        std::string scheme = match[3];
+        auto & base = match[1];
+        std::string scheme = match[2];
         auto authority = match[4].matched
             ? std::optional<std::string>(match[5]) : std::nullopt;
-        std::string path = match[6];
-        auto & query = match[7];
-        auto & fragment = match[8];
+        std::string path = match[4].matched ? match[4] : match[5];
+        auto & query = match[6];
+        auto & fragment = match[7];
 
         auto isFile = scheme.find("file") != std::string::npos;
 
diff --git a/src/libstore/fetchers/regex.hh b/src/libstore/fetchers/regex.hh
index 504d7bf18..e0989edfc 100644
--- a/src/libstore/fetchers/regex.hh
+++ b/src/libstore/fetchers/regex.hh
@@ -5,16 +5,20 @@
 namespace nix::fetchers {
 
 // URI stuff.
-const static std::string pctEncoded = "%[0-9a-fA-F][0-9a-fA-F]";
-const static std::string schemeRegex = "[a-z+]+";
-const static std::string authorityRegex =
-    "(?:(?:[a-z])*@)?"
-    "[a-zA-Z0-9._~-]*";
-const static std::string segmentRegex = "[a-zA-Z0-9._~-]+";
-const static std::string pathRegex = "(?:/?" + segmentRegex + "(?:/" + segmentRegex + ")*|/?)";
-const static std::string pcharRegex =
-    "(?:[a-zA-Z0-9-._~!$&'\"()*+,;=:@ ]|" + pctEncoded + ")";
-const static std::string queryRegex = "(?:" + pcharRegex + "|[/?])*";
+const static std::string pctEncoded = "(?:%[0-9a-fA-F][0-9a-fA-F])";
+const static std::string schemeRegex = "(?:[a-z+]+)";
+const static std::string ipv6AddressRegex = "(?:\\[[0-9a-fA-F:]+\\])";
+const static std::string unreservedRegex = "(?:[a-zA-Z0-9-._~])";
+const static std::string subdelimsRegex = "(?:[!$&'\"()*+,;=])";
+const static std::string hostnameRegex = "(?:(?:" + unreservedRegex + "|" + pctEncoded + "|" + subdelimsRegex + ")*)";
+const static std::string hostRegex = "(?:" + ipv6AddressRegex + "|" + hostnameRegex + ")";
+const static std::string userRegex = "(?:(?:" + unreservedRegex + "|" + pctEncoded + "|" + subdelimsRegex + "|:)*)";
+const static std::string authorityRegex = "(?:" + userRegex + "@)?" + hostRegex + "(?::[0-9]+)?";
+const static std::string pcharRegex = "(?:" + unreservedRegex + "|" + pctEncoded + "|" + subdelimsRegex + "|[:@])";
+const static std::string queryRegex = "(?:" + pcharRegex + "|[/? \"])*";
+const static std::string segmentRegex = "(?:" + pcharRegex + "+)";
+const static std::string absPathRegex = "(?:(?:/" + segmentRegex + ")*/?)";
+const static std::string pathRegex = "(?:" + segmentRegex + "(?:/" + segmentRegex + ")*/?)";
 
 // A Git ref (i.e. branch or tag name).
 const static std::string refRegexS = "[a-zA-Z0-9][a-zA-Z0-9_.-]*"; // FIXME: check