moving functions around

include/ada.h

@@ -21,6 +21,7 @@
 #include "ada/url_base-inl.h"
 #include "ada/url-inl.h"
 #include "ada/url_components.h"
+#include "ada/url_components-inl.h"
 #include "ada/url_aggregator.h"
 #include "ada/url_aggregator-inl.h"
 #include "ada/url_search_params.h"

include/ada/url-inl.h

@@ -43,6 +43,10 @@ inline std::ostream &operator<<(std::ostream &out, const ada::url &u) {
   return path.size();
 }
 
+[[nodiscard]] constexpr std::string_view url::get_pathname() const noexcept {
+  return path;
+}
+
 [[nodiscard]] ada_really_inline ada::url_components url::get_components()
     const noexcept {
   url_components out{};

include/ada/url_aggregator-inl.h

@@ -732,7 +732,8 @@ url_aggregator::get_components() const noexcept {
   return components;
 }
 
-[[nodiscard]] constexpr bool ada::url_aggregator::has_authority() const noexcept {
+[[nodiscard]] constexpr bool ada::url_aggregator::has_authority()
+    const noexcept {
   ada_log("url_aggregator::has_authority");
   // Performance: instead of doing this potentially expensive check, we could
   // have a boolean in the struct.
@@ -844,8 +845,8 @@ constexpr bool url_aggregator::has_port() const noexcept {
          buffer[components.host_end + 1] == '.';
 }
 
-[[nodiscard]] constexpr std::string_view url_aggregator::get_href() const noexcept
-    ada_lifetime_bound {
+[[nodiscard]] constexpr std::string_view url_aggregator::get_href()
+    const noexcept ada_lifetime_bound {
   ada_log("url_aggregator::get_href");
   return buffer;
 }
@@ -919,6 +920,191 @@ constexpr void url_aggregator::set_protocol_as_file() {
   ADA_ASSERT_TRUE(validate());
 }
 
+[[nodiscard]] constexpr bool url_aggregator::validate() const noexcept {
+  if (!is_valid) {
+    return true;
+  }
+  if (!components.check_offset_consistency()) {
+    ada_log("url_aggregator::validate inconsistent components \n",
+            to_diagram());
+    return false;
+  }
+  // We have a credible components struct, but let us investivate more
+  // carefully:
+  /**
+   * https://user:pass@example.com:1234/foo/bar?baz#quux
+   *       |     |    |          | ^^^^|       |   |
+   *       |     |    |          | |   |       |   `----- hash_start
+   *       |     |    |          | |   |       `--------- search_start
+   *       |     |    |          | |   `----------------- pathname_start
+   *       |     |    |          | `--------------------- port
+   *       |     |    |          `----------------------- host_end
+   *       |     |    `---------------------------------- host_start
+   *       |     `--------------------------------------- username_end
+   *       `--------------------------------------------- protocol_end
+   */
+  if (components.protocol_end == url_components::omitted) {
+    ada_log("url_aggregator::validate omitted protocol_end \n", to_diagram());
+    return false;
+  }
+  if (components.username_end == url_components::omitted) {
+    ada_log("url_aggregator::validate omitted username_end \n", to_diagram());
+    return false;
+  }
+  if (components.host_start == url_components::omitted) {
+    ada_log("url_aggregator::validate omitted host_start \n", to_diagram());
+    return false;
+  }
+  if (components.host_end == url_components::omitted) {
+    ada_log("url_aggregator::validate omitted host_end \n", to_diagram());
+    return false;
+  }
+  if (components.pathname_start == url_components::omitted) {
+    ada_log("url_aggregator::validate omitted pathname_start \n", to_diagram());
+    return false;
+  }
+
+  if (components.protocol_end > buffer.size()) {
+    ada_log("url_aggregator::validate protocol_end overflow \n", to_diagram());
+    return false;
+  }
+  if (components.username_end > buffer.size()) {
+    ada_log("url_aggregator::validate username_end overflow \n", to_diagram());
+    return false;
+  }
+  if (components.host_start > buffer.size()) {
+    ada_log("url_aggregator::validate host_start overflow \n", to_diagram());
+    return false;
+  }
+  if (components.host_end > buffer.size()) {
+    ada_log("url_aggregator::validate host_end overflow \n", to_diagram());
+    return false;
+  }
+  if (components.pathname_start > buffer.size()) {
+    ada_log("url_aggregator::validate pathname_start overflow \n",
+            to_diagram());
+    return false;
+  }
+
+  if (components.protocol_end > 0) {
+    if (buffer[components.protocol_end - 1] != ':') {
+      ada_log(
+          "url_aggregator::validate missing : at the end of the protocol \n",
+          to_diagram());
+      return false;
+    }
+  }
+
+  if (components.username_end != buffer.size() &&
+      components.username_end > components.protocol_end + 2) {
+    if (buffer[components.username_end] != ':' &&
+        buffer[components.username_end] != '@') {
+      ada_log(
+          "url_aggregator::validate missing : or @ at the end of the username "
+          "\n",
+          to_diagram());
+      return false;
+    }
+  }
+
+  if (components.host_start != buffer.size()) {
+    if (components.host_start > components.username_end) {
+      if (buffer[components.host_start] != '@') {
+        ada_log(
+            "url_aggregator::validate missing @ at the end of the password \n",
+            to_diagram());
+        return false;
+      }
+    } else if (components.host_start == components.username_end &&
+               components.host_end > components.host_start) {
+      if (components.host_start == components.protocol_end + 2) {
+        if (buffer[components.protocol_end] != '/' ||
+            buffer[components.protocol_end + 1] != '/') {
+          ada_log(
+              "url_aggregator::validate missing // between protocol and host "
+              "\n",
+              to_diagram());
+          return false;
+        }
+      } else {
+        if (components.host_start > components.protocol_end &&
+            buffer[components.host_start] != '@') {
+          ada_log(
+              "url_aggregator::validate missing @ at the end of the username "
+              "\n",
+              to_diagram());
+          return false;
+        }
+      }
+    } else {
+      if (components.host_end != components.host_start) {
+        ada_log("url_aggregator::validate expected omitted host \n",
+                to_diagram());
+        return false;
+      }
+    }
+  }
+  if (components.host_end != buffer.size() &&
+      components.pathname_start > components.host_end) {
+    if (components.pathname_start == components.host_end + 2 &&
+        buffer[components.host_end] == '/' &&
+        buffer[components.host_end + 1] == '.') {
+      if (components.pathname_start + 1 >= buffer.size() ||
+          buffer[components.pathname_start] != '/' ||
+          buffer[components.pathname_start + 1] != '/') {
+        ada_log(
+            "url_aggregator::validate expected the path to begin with // \n",
+            to_diagram());
+        return false;
+      }
+    } else if (buffer[components.host_end] != ':') {
+      ada_log("url_aggregator::validate missing : at the port \n",
+              to_diagram());
+      return false;
+    }
+  }
+  if (components.pathname_start != buffer.size() &&
+      components.pathname_start < components.search_start &&
+      components.pathname_start < components.hash_start && !has_opaque_path) {
+    if (buffer[components.pathname_start] != '/') {
+      ada_log("url_aggregator::validate missing / at the path \n",
+              to_diagram());
+      return false;
+    }
+  }
+  if (components.search_start != url_components::omitted) {
+    if (buffer[components.search_start] != '?') {
+      ada_log("url_aggregator::validate missing ? at the search \n",
+              to_diagram());
+      return false;
+    }
+  }
+  if (components.hash_start != url_components::omitted) {
+    if (buffer[components.hash_start] != '#') {
+      ada_log("url_aggregator::validate missing # at the hash \n",
+              to_diagram());
+      return false;
+    }
+  }
+
+  return true;
+}
+
+[[nodiscard]] constexpr std::string_view url_aggregator::get_pathname()
+    const noexcept ada_lifetime_bound {
+  ada_log("url_aggregator::get_pathname pathname_start = ",
+          components.pathname_start, " buffer.size() = ", buffer.size(),
+          " components.search_start = ", components.search_start,
+          " components.hash_start = ", components.hash_start);
+  auto ending_index = uint32_t(buffer.size());
+  if (components.search_start != url_components::omitted) {
+    ending_index = components.search_start;
+  } else if (components.hash_start != url_components::omitted) {
+    ending_index = components.hash_start;
+  }
+  return helpers::substring(buffer, components.pathname_start, ending_index);
+}
+
 inline std::ostream &operator<<(std::ostream &out,
                                 const ada::url_aggregator &u) {
   return out << u.to_string();

include/ada/url_aggregator.h

@@ -144,7 +144,8 @@ struct url_aggregator : url_base {
    * A URL includes credentials if its username or password is not the empty
    * string.
    */
-  [[nodiscard]] ada_really_inline constexpr bool has_credentials() const noexcept;
+  [[nodiscard]] ada_really_inline constexpr bool has_credentials()
+      const noexcept;
 
   /**
    * Useful for implementing efficient serialization for the URL.

include/ada/url_base-inl.h

@@ -21,7 +21,8 @@
 
 namespace ada {
 
-[[nodiscard]] ada_really_inline constexpr bool url_base::is_special() const noexcept {
+[[nodiscard]] ada_really_inline constexpr bool url_base::is_special()
+    const noexcept {
   return type != ada::scheme::NOT_SPECIAL;
 }
 

include/ada/url_components-inl.h

@@ -0,0 +1,88 @@
+/**
+ * @file url_components.h
+ * @brief Declaration for the URL Components
+ */
+#ifndef ADA_URL_COMPONENTS_INL_H
+#define ADA_URL_COMPONENTS_INL_H
+
+#include "ada/url_components.h"
+
+namespace ada {
+
+[[nodiscard]] constexpr bool url_components::check_offset_consistency() const noexcept {
+  /**
+   * https://user:pass@example.com:1234/foo/bar?baz#quux
+   *       |     |    |          | ^^^^|       |   |
+   *       |     |    |          | |   |       |   `----- hash_start
+   *       |     |    |          | |   |       `--------- search_start
+   *       |     |    |          | |   `----------------- pathname_start
+   *       |     |    |          | `--------------------- port
+   *       |     |    |          `----------------------- host_end
+   *       |     |    `---------------------------------- host_start
+   *       |     `--------------------------------------- username_end
+   *       `--------------------------------------------- protocol_end
+   */
+  // These conditions can be made more strict.
+  uint32_t index = 0;
+
+  if (protocol_end == url_components::omitted) {
+    return false;
+  }
+  if (protocol_end < index) {
+    return false;
+  }
+  index = protocol_end;
+
+  if (username_end == url_components::omitted) {
+    return false;
+  }
+  if (username_end < index) {
+    return false;
+  }
+  index = username_end;
+
+  if (host_start == url_components::omitted) {
+    return false;
+  }
+  if (host_start < index) {
+    return false;
+  }
+  index = host_start;
+
+  if (port != url_components::omitted) {
+    if (port > 0xffff) {
+      return false;
+    }
+    uint32_t port_length = helpers::fast_digit_count(port) + 1;
+    if (index + port_length < index) {
+      return false;
+    }
+    index += port_length;
+  }
+
+  if (pathname_start == url_components::omitted) {
+    return false;
+  }
+  if (pathname_start < index) {
+    return false;
+  }
+  index = pathname_start;
+
+  if (search_start != url_components::omitted) {
+    if (search_start < index) {
+      return false;
+    }
+    index = search_start;
+  }
+
+  if (hash_start != url_components::omitted) {
+    if (hash_start < index) {
+      return false;
+    }
+  }
+
+  return true;
+}
+
+}  // namespace ada
+#endif

include/ada/url_components.h

@@ -75,6 +75,5 @@ struct url_components {
   [[nodiscard]] std::string to_string() const;
 
 };  // struct url_components
-
 }  // namespace ada
 #endif

src/url-getters.cpp

@@ -62,10 +62,6 @@ namespace ada {
   return host.value_or("");
 }
 
-[[nodiscard]] constexpr std::string_view url::get_pathname() const noexcept {
-  return path;
-}
-
 [[nodiscard]] std::string url::get_search() const noexcept {
   // If this's URL's query is either null or the empty string, then return the
   // empty string. Return U+003F (?), followed by this's URL's query.