Remove ICU related code from source and cmake files

upa-url · Aug 12, 2024 · ec5f342 · ec5f342
1 parent 822407b
commit ec5f342
Show file tree

Hide file tree

Showing 9 changed files with 12 additions and 361 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -1,10 +1,9 @@
-# FindICU requires 3.7
 # add_link_options() requires 3.13
 cmake_minimum_required(VERSION 3.13)
 
 set(CMAKE_SUPPRESS_REGENERATION true)
 
-# use ICU_ROOT
+# https://cmake.org/cmake/help/latest/policy/CMP0074.html
 if(POLICY CMP0074)
   cmake_policy(SET CMP0074 NEW)
 endif()
@@ -23,7 +22,6 @@ project(upa_url VERSION ${UPA_URL_VERSION} LANGUAGES CXX)
 # ${upa_lib_name}-config.cmake
 # It also must be used as the package name argument to find_package
 set(upa_lib_name upa)
-set(upa_lib_name_in upa-icu)
 # Exported name for library target files; also used to create an alias
 # target: upa::${upa_lib_export}
 set(upa_lib_export url)
@@ -55,7 +53,6 @@ option(UPA_BUILD_TOOLS "Build tools." OFF)
 option(UPA_INSTALL "Generate the install target." ON)
 # library options
 option(UPA_AMALGAMATED "Use amalgamated URL library source." OFF)
-option(UPA_USE_WINDOWS_ICU "Use ICU library bundled with Windows 10 version 1903 or later." OFF)
 # tests build options
 option(UPA_TEST_COVERAGE "Build tests with code coverage reporting" OFF)
 option(UPA_TEST_COVERAGE_CLANG "Build tests with Clang source-based code coverage" OFF)
@@ -114,14 +111,9 @@ endif()
 
 include_directories(deps)
 
-# Are Upa URL and ICU libraries needed?
+# Is the Upa URL library needed?
 if (UPA_BUILD_TESTS OR UPA_BUILD_BENCH OR UPA_BUILD_FUZZER OR UPA_BUILD_EXAMPLES OR
     UPA_BUILD_EXTRACTED OR UPA_INSTALL OR NOT UPA_BUILD_TOOLS)
-  if (NOT UPA_USE_WINDOWS_ICU)
-    # This library depends on ICU
-    find_package(ICU REQUIRED COMPONENTS i18n uc)
-  endif()
-
   if (UPA_AMALGAMATED)
     add_library(${upa_lib_target} STATIC
       single_include/upa/url.cpp)
@@ -131,7 +123,6 @@ if (UPA_BUILD_TESTS OR UPA_BUILD_BENCH OR UPA_BUILD_FUZZER OR UPA_BUILD_EXAMPLES
     add_library(${upa_lib_target} STATIC
       src/idna.cpp
       src/url.cpp
-      src/url_idna.cpp
       src/url_ip.cpp
       src/url_percent_encode.cpp
       src/url_search_params.cpp
@@ -143,13 +134,6 @@ if (UPA_BUILD_TESTS OR UPA_BUILD_BENCH OR UPA_BUILD_FUZZER OR UPA_BUILD_EXAMPLES
   add_library(upa::${upa_lib_export} ALIAS ${upa_lib_target})
   set_target_properties(${upa_lib_target} PROPERTIES
     EXPORT_NAME ${upa_lib_export})
-  if (UPA_USE_WINDOWS_ICU)
-    target_compile_definitions(${upa_lib_target} PRIVATE UPA_USE_WINDOWS_ICU=1)
-    set(upa_lib_name_in ${upa_lib_name})
-  else()
-    target_include_directories(${upa_lib_target} PRIVATE ${ICU_INCLUDE_DIR})
-    target_link_libraries(${upa_lib_target} INTERFACE ICU::i18n ICU::uc)
-  endif()
 endif()
 
 # Test targets
@@ -274,7 +258,7 @@ if (UPA_INSTALL AND NOT UPA_AMALGAMATED)
 
   # generate the config file that includes the exports
   configure_package_config_file(
-    ${CMAKE_CURRENT_SOURCE_DIR}/cmake/${upa_lib_name_in}-config.cmake.in
+    ${CMAKE_CURRENT_SOURCE_DIR}/cmake/${upa_lib_name}-config.cmake.in
     ${CMAKE_CURRENT_BINARY_DIR}/${upa_lib_name}-config.cmake
     INSTALL_DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/${upa_lib_name}
     NO_SET_AND_CHECK_MACRO

diff --git a/cmake/upa-icu-config.cmake.in b/cmake/upa-icu-config.cmake.in
diff --git a/include/upa/url_host.h b/include/upa/url_host.h
@@ -6,18 +6,10 @@
 #ifndef UPA_URL_HOST_H
 #define UPA_URL_HOST_H
 
-#ifndef UPA_USE_ICU
-#define UPA_USE_ICU 0  // NOLINT(*-macro-*)
-#endif // UPA_USE_ICU
-
 #include "buffer.h"
 #include "config.h"
+#include "idna.h"
 #include "str_arg.h"
-#if UPA_USE_ICU
-# include "url_idna.h"
-#else
-# include "idna.h"
-#endif
 #include "url_ip.h"
 #include "url_percent_encode.h"
 #include "url_result.h"
@@ -221,62 +213,6 @@ inline validation_errc host_parser::parse_host(const CharT* first, const CharT*
             return validation_errc::domain_invalid_code_point;
     }
 
-#if UPA_USE_ICU
-    // Input for domain_to_ascii
-    simple_buffer<char16_t> buff_uc;
-
-    // copy ASCII chars
-    for (auto it = first; it != ptr; ++it) {
-        const auto uch = static_cast<UCharT>(*it);
-        buff_uc.push_back(static_cast<char16_t>(uch));
-    }
-
-    // Let buff_uc be the result of running UTF-8 decode (to UTF-16) without BOM
-    // on the percent decoding of UTF-8 encode on input
-    for (auto it = ptr; it != last;) {
-        const auto uch = static_cast<UCharT>(*it++);
-        if (uch < 0x80) {
-            if (uch != '%') {
-                buff_uc.push_back(static_cast<char16_t>(uch));
-                continue;
-            }
-            // uch == '%'
-            unsigned char uc8; // NOLINT(cppcoreguidelines-init-variables)
-            if (detail::decode_hex_to_byte(it, last, uc8)) {
-                if (uc8 < 0x80) {
-                    buff_uc.push_back(static_cast<char16_t>(uc8));
-                    continue;
-                }
-                // percent encoded utf-8 sequence
-                // TODO: gal po vieną code_point, tuomet užtektų utf-8 buferio vienam simboliui
-                simple_buffer<char> buff_utf8;
-                buff_utf8.push_back(static_cast<char>(uc8));
-                while (it != last && *it == '%') {
-                    ++it; // skip '%'
-                    if (!detail::decode_hex_to_byte(it, last, uc8))
-                        uc8 = '%';
-                    buff_utf8.push_back(static_cast<char>(uc8));
-                }
-                url_utf::convert_utf8_to_utf16(buff_utf8.data(), buff_utf8.data() + buff_utf8.size(), buff_uc);
-                //buff_utf8.clear();
-                continue;
-            }
-            // detected an invalid percent-encoding sequence
-            buff_uc.push_back('%');
-        } else { // uch >= 0x80
-            --it;
-            url_utf::append_utf16(url_utf::read_utf_char(it, last).value, buff_uc);
-        }
-    }
-
-
-    // domain to ASCII
-    simple_buffer<char16_t> buff_ascii;
-
-    const auto res = domain_to_ascii(buff_uc.data(), buff_uc.size(), buff_ascii);
-    if (res != validation_errc::ok)
-        return res;
-#else
     std::string buff_ascii;
 
     const auto pes = std::find(ptr, last, '%');
@@ -337,7 +273,6 @@ inline validation_errc host_parser::parse_host(const CharT* first, const CharT*
         if (!idna::domain_to_ascii(buff_ascii, buff_uc.begin(), buff_uc.end()))
             return validation_errc::domain_to_ascii;
     }
-#endif
 
     if (detail::contains_forbidden_domain_char(buff_ascii.data(), buff_ascii.data() + buff_ascii.size())) {
         // 7. If asciiDomain contains a forbidden domain code point, domain-invalid-code-point