Refactored punycode and domain functions (#58)

Author: glynos

include/skyr/domain/domain.hpp

@@ -0,0 +1,40 @@
+// Copyright 2018-20 Glyn Matthews.
+// Distributed under the Boost Software License, Version 1.0.
+// (See accompanying file LICENSE_1_0.txt or copy at
+// http://www.boost.org/LICENSE_1_0.txt)
+
+#ifndef SKYR_DOMAIN_DOMAIN_HPP
+#define SKYR_DOMAIN_DOMAIN_HPP
+
+/// \file domain.hpp
+
+#include <string>
+#include <string_view>
+#include <system_error>
+#include <tl/expected.hpp>
+
+namespace skyr {
+inline namespace v1 {
+/// Converts a UTF-8 encoded domain to ASCII using
+/// [IDNA processing](https://www.domain.org/reports/tr46/#Processing)
+///
+/// \param domain A domain
+/// \param be_strict Tells the processor to be strict
+/// \returns An ASCII domain, or an error
+auto domain_to_ascii(
+    std::string_view domain,
+    bool be_strict = false) -> tl::expected<std::string, std::error_code>;
+
+/// Converts a UTF-32 encoded domain to ASCII using
+/// [IDNA processing](https://www.domain.org/reports/tr46/#Processing)
+///
+/// \param domain A domain
+/// \param be_strict Tells the processor to be strict
+/// \returns An ASCII domain, or an error
+auto domain_to_ascii(
+    std::u32string_view domain,
+    bool be_strict = false) -> tl::expected<std::string, std::error_code>;
+}  // namespace v1
+}  // namespace skyr
+
+#endif // SKYR_DOMAIN_DOMAIN_HPP

include/skyr/domain/errors.hpp

@@ -0,0 +1,38 @@
+// Copyright 2020 Glyn Matthews.
+// Distributed under the Boost Software License, Version 1.0.
+// (See accompanying file LICENSE_1_0.txt or copy at
+// http://www.boost.org/LICENSE_1_0.txt)
+
+#ifndef SKYR_DOMAIN_ERRORS_HPP
+#define SKYR_DOMAIN_ERRORS_HPP
+
+#include <system_error>
+
+namespace skyr {
+inline namespace v1 {
+/// \enum domain_errc
+/// Enumerates domain processing errors
+enum class domain_errc {
+  /// The domain code point is disallowed
+  disallowed_code_point = 1,
+  /// The encoder or decoder received bad input
+  bad_input,
+  /// Overflow
+  overflow,
+  /// Unicode encoding error
+  encoding_error,
+};
+
+/// Creates a `std::error_code` given a `skyr::domain_errc` value
+/// \param error A domain error
+/// \returns A `std::error_code` object
+auto make_error_code(domain_errc error) noexcept -> std::error_code;
+}  // namespace v1
+}  // namespace skyr
+
+namespace std {
+template <>
+struct is_error_code_enum<skyr::v1::domain_errc> : true_type {};
+}  // namespace std
+
+#endif //SKYR_DOMAIN_ERRORS_HPP

include/skyr/unicode/idna.hpp include/skyr/domain/idna.hppinclude/skyr/unicode/idna.hpp renamed to include/skyr/domain/idna.hpp

@@ -3,16 +3,16 @@
 // (See accompanying file LICENSE_1_0.txt or copy at
 // http://www.boost.org/LICENSE_1_0.txt)
 
-#ifndef SKYR_UNICODE_IDNA_HPP
-#define SKYR_UNICODE_IDNA_HPP
+#ifndef SKYR_DOMAIN_IDNA_HPP
+#define SKYR_DOMAIN_IDNA_HPP
 
 namespace skyr {
 inline namespace v1 {
-namespace unicode {
+namespace domain {
 /// \enum idna_status
-/// The status values come from the IDNA mapping table in unicode TR46:
+/// The status values come from the IDNA mapping table in domain TR46:
 ///
-/// https://unicode.org/reports/tr46/#IDNA_Mapping_Table
+/// https://domain.org/reports/tr46/#IDNA_Mapping_Table
 ///
 enum class idna_status {
   /// The code point is disallowed
@@ -45,8 +45,8 @@ auto map_idna_status(char32_t code_point) -> idna_status;
 /// \return The code point or mapped value, depending on the status of the code
 /// point
 auto map_idna_code_point(char32_t code_point) -> char32_t;
-}  // namespace unicode
+}  // namespace domain
 }  // namespace v1
 }  // namespace skyr
 
-#endif  // SKYR_UNICODE_IDNA_HPP
+#endif  // SKYR_DOMAIN_IDNA_HPP

include/skyr/domain/punycode.hpp

@@ -0,0 +1,42 @@
+// Copyright 2020 Glyn Matthews.
+// Distributed under the Boost Software License, Version 1.0.
+// (See accompanying file LICENSE_1_0.txt or copy at
+// http://www.boost.org/LICENSE_1_0.txt)
+
+#ifndef SKYR_DOMAIN_PUNYCODE_HPP
+#define SKYR_DOMAIN_PUNYCODE_HPP
+
+#include <string>
+#include <string_view>
+#include <system_error>
+#include <tl/expected.hpp>
+
+namespace skyr {
+inline namespace v1 {
+/// Performs Punycode encoding based on a reference implementation
+/// defined in [RFC 3492](https://tools.ietf.org/html/rfc3492)
+///
+/// \param input A UTF-8 encoded domain to be encoded
+/// \returns The encoded ASCII domain, or an error
+auto punycode_encode(
+    std::string_view input) -> tl::expected<std::string, std::error_code>;
+
+/// Performs Punycode encoding based on a reference implementation
+/// defined in [RFC 3492](https://tools.ietf.org/html/rfc3492)
+///
+/// \param input A UTF-32 encoded domain to be encoded
+/// \returns The encoded ASCII domain, or an error
+auto punycode_encode(
+    std::u32string_view input) -> tl::expected<std::string, std::error_code>;
+
+/// Performs Punycode decoding based on a reference implementation
+/// defined in [RFC 3492](https://tools.ietf.org/html/rfc3492)
+///
+/// \param input An ASCII encoded domain to be decoded
+/// \returns The decoded UTF-8 domain, or an error
+auto punycode_decode(
+    std::string_view input) -> tl::expected<std::string, std::error_code>;
+}  // namespace v1
+}  // namespace skyr
+
+#endif // SKYR_DOMAIN_PUNYCODE_HPP

include/skyr/unicode/domain.hpp

@@ -10,8 +10,10 @@ configure_file(
 
 set(skyr_SRCS
         unicode/errors.cpp
-        unicode/idna.cpp
-        unicode/domain.cpp
+        domain/errors.cpp
+        domain/domain.cpp
+        domain/punycode.cpp
+        domain/idna.cpp
         core/url_parser_context.hpp
         core/url_parser_context.cpp
         core/url_record.cpp
@@ -47,8 +49,9 @@ set(skyr_SRCS
         ${PROJECT_SOURCE_DIR}/include/skyr/unicode/ranges/transforms/byte_transform.hpp
         ${PROJECT_SOURCE_DIR}/include/skyr/unicode/ranges/transforms/u16_transform.hpp
         ${PROJECT_SOURCE_DIR}/include/skyr/unicode/ranges/transforms/u32_transform.hpp
-        ${PROJECT_SOURCE_DIR}/include/skyr/unicode/idna.hpp
-        ${PROJECT_SOURCE_DIR}/include/skyr/unicode/domain.hpp
+        ${PROJECT_SOURCE_DIR}/include/skyr/domain/errors.hpp
+        ${PROJECT_SOURCE_DIR}/include/skyr/domain/idna.hpp
+        ${PROJECT_SOURCE_DIR}/include/skyr/domain/domain.hpp
         ${PROJECT_SOURCE_DIR}/include/skyr/percent_encoding/errors.hpp
         ${PROJECT_SOURCE_DIR}/include/skyr/percent_encoding/percent_decode_range.hpp
         ${PROJECT_SOURCE_DIR}/include/skyr/percent_encoding/percent_encode_range.hpp

src/CMakeLists.txt

@@ -11,7 +11,7 @@
 #include <locale>
 #include <skyr/network/ipv4_address.hpp>
 #include <skyr/network/ipv6_address.hpp>
-#include <skyr/unicode/domain.hpp>
+#include <skyr/domain/domain.hpp>
 #include <skyr/percent_encoding/percent_decode_range.hpp>
 #include "url_parser_context.hpp"
 #include "url_schemes.hpp"
@@ -137,7 +137,7 @@ tl::expected<std::string, url_parse_errc> parse_host(
     return tl::make_unexpected(url_parse_errc::cannot_decode_host_point);
   }
 
-  auto ascii_domain = unicode::domain_to_ascii(domain.value());
+  auto ascii_domain = domain_to_ascii(domain.value());
   if (!ascii_domain) {
     return tl::make_unexpected(url_parse_errc::domain_error);
   }