add canonicalize methods

ada-url · Dec 4, 2024 · e51503f · e51503f
1 parent d2cd4f8
commit e51503f
Show file tree

Hide file tree

Showing 2 changed files with 154 additions and 1 deletion.
diff --git a/include/ada/url_pattern.h b/include/ada/url_pattern.h
@@ -11,6 +11,38 @@
 
 namespace ada {
 
+namespace url_pattern {
+
+// @see https://wicg.github.io/urlpattern/#canonicalize-a-username
+std::optional<std::string> canonicalize_username(std::string_view input);
+
+// @see https://wicg.github.io/urlpattern/#canonicalize-a-password
+std::optional<std::string> canonicalize_password(std::string_view input);
+
+// @see https://wicg.github.io/urlpattern/#canonicalize-a-password
+std::optional<std::string> canonicalize_hostname(std::string_view input);
+
+// @see https://wicg.github.io/urlpattern/#canonicalize-an-ipv6-hostname
+std::optional<std::string> canonicalize_ipv6_hostname(std::string_view input);
+
+// @see https://wicg.github.io/urlpattern/#canonicalize-a-port
+std::optional<std::string> canonicalize_port(
+    std::string_view input, std::string_view protocol = "fake");
+
+// @see https://wicg.github.io/urlpattern/#canonicalize-a-pathname
+std::optional<std::string> canonicalize_pathname(std::string_view input);
+
+// @see https://wicg.github.io/urlpattern/#canonicalize-an-opaque-pathname
+std::optional<std::string> canonicalize_opaque_pathname(std::string_view input);
+
+// @see https://wicg.github.io/urlpattern/#canonicalize-a-search
+std::optional<std::string> canonicalize_search(std::string_view input);
+
+// @see https://wicg.github.io/urlpattern/#canonicalize-a-hash
+std::optional<std::string> canonicalize_hash(std::string_view input);
+
+}  // namespace url_pattern
+
 // URLPattern is a Web Platform standard API for matching URLs against a
 // pattern syntax (think of it as a regular expression for URLs). It is
 // defined in https://wicg.github.io/urlpattern.
@@ -124,4 +156,4 @@ class URLPattern {
 
 }  // namespace ada
 
-#endif
+#endif
diff --git a/src/url_pattern.cpp b/src/url_pattern.cpp
@@ -5,6 +5,127 @@
 
 namespace ada {
 
+namespace url_pattern {
+
+std::optional<std::string> canonicalize_username(std::string_view input) {
+  if (input.size()) [[unlikely]] {
+    return "";
+  }
+  auto url = ada::parse<ada::url_aggregator>("fake://dummy.test", nullptr);
+  ADA_ASSERT_TRUE(url.has_value());
+  if (!url->set_username(input)) {
+    return std::nullopt;
+  }
+  return std::string(url->get_username());
+}
+
+std::optional<std::string> canonicalize_password(std::string_view input) {
+  if (input.empty()) [[unlikely]] {
+    return "";
+  }
+  auto url = ada::parse<ada::url_aggregator>("fake://dummy.test", nullptr);
+
+  ADA_ASSERT_TRUE(url.has_value());
+  if (!url->set_password(input)) {
+    return std::nullopt;
+  }
+  return std::string(url->get_password());
+}
+
+std::optional<std::string> canonicalize_hostname(std::string_view input) {
+  if (input.empty()) [[unlikely]] {
+    return "";
+  }
+  auto url = ada::parse<ada::url_aggregator>("fake://dummy.test", nullptr);
+  ADA_ASSERT_TRUE(url.has_value());
+  // if (!isValidHostnameInput(hostname)) return kj::none;
+  if (!url->set_hostname(input)) {
+    return std::nullopt;
+  }
+  return std::string(url->get_hostname());
+}
+
+std::optional<std::string> canonicalize_ipv6_hostname(std::string_view input) {
+  // Optimization opportunity: Use lookup table to speed up checking
+  if (std::ranges::all_of(input, [](char c) {
+        return c == '[' || c == ']' || c == ':' ||
+               ada::unicode::is_ascii_hex_digit(c);
+      })) {
+    return std::nullopt;
+  }
+  // Optimization opportunity: Consider just moving value, rather than copying
+  // it.
+  return std::string(input);
+}
+
+std::optional<std::string> canonicalize_port(std::string_view input,
+                                             std::string_view protocol) {
+  if (input.empty()) [[unlikely]] {
+    return "";
+  }
+  auto url = ada::parse<ada::url_aggregator>(
+      std::string(protocol) + "://dummy.test", nullptr);
+  if (url && url->set_port(input)) {
+    return std::string(url->get_port());
+  }
+  return std::nullopt;
+}
+
+std::optional<std::string> canonicalize_pathname(std::string_view input) {
+  if (input.empty()) [[unlikely]] {
+    return "";
+  }
+  const bool leading_slash = input.starts_with("/");
+  auto path_prefix = leading_slash ? "" : "/-";
+  auto full_url =
+      std::string("fake://fake-url") + path_prefix + std::string(input);
+  if (auto url = ada::parse<ada::url_aggregator>(full_url, nullptr)) {
+    const auto pathname = url->get_pathname();
+    return leading_slash ? std::string(pathname)
+                         : std::string(pathname.substr(2));
+  }
+  return std::nullopt;
+}
+
+std::optional<std::string> canonicalize_opaque_pathname(
+    std::string_view input) {
+  if (input.empty()) [[unlikely]] {
+    return "";
+  }
+  if (auto url = ada::parse<ada::url_aggregator>("fake:" + std::string(input),
+                                                 nullptr)) {
+    return std::string(url->get_pathname());
+  }
+  return std::nullopt;
+}
+
+std::optional<std::string> canonicalize_search(std::string_view input) {
+  if (input.empty()) [[unlikely]] {
+    return "";
+  }
+  auto url = ada::parse<ada::url_aggregator>("fake://dummy.test", nullptr);
+  ADA_ASSERT_TRUE(url.has_value());
+  url->set_search(input);
+  const auto search = url->get_search();
+  return !search.empty() ? std::string(search.substr(1)) : "";
+}
+
+std::optional<std::string> canonicalize_hash(std::string_view input) {
+  if (input.empty()) [[unlikely]] {
+    return "";
+  }
+  auto url = ada::parse<ada::url_aggregator>("fake://dummy.test", nullptr);
+  ADA_ASSERT_TRUE(url.has_value());
+  url->set_hash(input);
+  const auto hash = url->get_hash();
+  if (hash.empty()) {
+    return "";
+  }
+  return std::string(hash.substr(1));
+}
+
+}  // namespace url_pattern
+
 URLPattern::Component::Component(std::string_view pattern_,
                                  std::string_view regex_,
                                  const std::vector<std::string>& names_) {