Fix Windows UNC support on URI::to_path()

src/core/uri/filesystem.cc

@@ -2,10 +2,24 @@
 
 #include "escaping.h"
 
-#include <algorithm>  // std::ranges::replace
-#include <filesystem> // std::filesystem
-#include <iterator>   // std::advance, std::next
-#include <string>     // std::string
+#include <algorithm>   // std::ranges::equal, std::ranges::replace
+#include <cctype>      // std::tolower
+#include <filesystem>  // std::filesystem
+#include <iterator>    // std::advance, std::next
+#include <string>      // std::string
+#include <string_view> // std::string_view
+
+namespace {
+
+auto is_localhost_host(const std::string_view host) -> bool {
+  constexpr std::string_view localhost{"localhost"};
+  return std::ranges::equal(
+      host, localhost, [](const char left, const char right) {
+        return std::tolower(static_cast<unsigned char>(left)) == right;
+      });
+}
+
+} // namespace
 
 namespace sourcemeta::core {
 
@@ -17,6 +31,26 @@ auto URI::to_path() const -> std::filesystem::path {
     return path;
   }
 
+  // RFC 8089: a non-empty, non-localhost host on a file URI denotes a UNC
+  // server. The "localhost" host is equivalent to no host
+  const auto host_value = this->host();
+  const auto is_unc = host_value.has_value() && !host_value->empty() &&
+                      !is_localhost_host(host_value.value());
+  if (is_unc) {
+    if (!path.empty() && path.front() == '/') {
+      path.erase(0, 1);
+    }
+    std::ranges::replace(path, '/', '\\');
+    uri_unescape_all_inplace(path);
+    std::string unc{"\\\\"};
+    unc.append(host_value.value());
+    if (!path.empty()) {
+      unc.push_back('\\');
+      unc.append(path);
+    }
+    return unc;
+  }
+
   // Check for Windows absolute path (e.g., /C:/)
   const auto is_windows_absolute =
       path.size() >= 3 && path[0] == '/' && path[2] == ':';

test/uri/uri_to_path_test.cc

@@ -92,3 +92,45 @@ TEST(URI_to_path, unicode_windows) {
   const std::filesystem::path expected{u8R"(C:\data\résumé.doc)"};
   EXPECT_EQ(uri.to_path(), expected);
 }
+
+TEST(URI_to_path, windows_unc_simple) {
+  const sourcemeta::core::URI uri{"file://server/share/file.txt"};
+  const std::filesystem::path expected{R"(\\server\share\file.txt)"};
+  EXPECT_EQ(uri.to_path(), expected);
+}
+
+TEST(URI_to_path, windows_unc_with_space) {
+  const sourcemeta::core::URI uri{"file://srv/My%20Docs/a%20b.txt"};
+  const std::filesystem::path expected{R"(\\srv\My Docs\a b.txt)"};
+  EXPECT_EQ(uri.to_path(), expected);
+}
+
+TEST(URI_to_path, windows_unc_unicode) {
+  const sourcemeta::core::URI uri{"file://server/data/%C3%A9clair.txt"};
+  const std::filesystem::path expected{u8R"(\\server\data\éclair.txt)"};
+  EXPECT_EQ(uri.to_path(), expected);
+}
+
+TEST(URI_to_path, localhost_treated_as_no_host) {
+  const sourcemeta::core::URI uri{"file://localhost/foo/bar"};
+  const std::filesystem::path expected{"/foo/bar"};
+  EXPECT_EQ(uri.to_path(), expected);
+}
+
+TEST(URI_to_path, localhost_with_windows_drive) {
+  const sourcemeta::core::URI uri{"file://localhost/C:/foo"};
+  const std::filesystem::path expected{R"(C:\foo)"};
+  EXPECT_EQ(uri.to_path(), expected);
+}
+
+TEST(URI_to_path, localhost_uppercase) {
+  const sourcemeta::core::URI uri{"file://LOCALHOST/foo/bar"};
+  const std::filesystem::path expected{"/foo/bar"};
+  EXPECT_EQ(uri.to_path(), expected);
+}
+
+TEST(URI_to_path, localhost_mixed_case) {
+  const sourcemeta::core::URI uri{"file://LocalHost/foo/bar"};
+  const std::filesystem::path expected{"/foo/bar"};
+  EXPECT_EQ(uri.to_path(), expected);
+}