Remove Iconv

We do not use this anywhere since years.
Is a maintenance burden and nobody tests this.

Fix #465

Author: Ghabry

CMakeLists.txt

@@ -8,7 +8,7 @@ project(liblcf VERSION 0.8 LANGUAGES CXX)
 
 # Compilation options
 option(BUILD_SHARED_LIBS "Build shared library, disable for building the static library (default: ON)" ON)
-option(LIBLCF_WITH_ICU "ICU encoding detection (when OFF fallback to iconv, not recommended, default: ON)" ON)
+option(LIBLCF_WITH_ICU "ICU encoding handling (disable only for testing purposes, default: ON)" ON)
 option(LIBLCF_WITH_XML "XML reading support (expat, default: ON)" ON)
 option(LIBLCF_UPDATE_MIMEDB "Whether to run update-mime-database after install (default: ON)" ON)
 option(LIBLCF_ENABLE_TOOLS "Whether to build the tools (default: ON)" ON)
@@ -343,16 +343,13 @@ set_property(TARGET lcf PROPERTY WINDOWS_EXPORT_ALL_SYMBOLS ON)
 # Name of the exported library
 set_property(TARGET lcf PROPERTY EXPORT_NAME liblcf)
 
-# icu or fallback to iconv
+# icu
 set(LCF_SUPPORT_ICU 0)
 if(LIBLCF_WITH_ICU)
 	find_package(ICU COMPONENTS i18n uc data REQUIRED)
 	target_link_libraries(lcf ICU::i18n ICU::uc ICU::data)
 	list(APPEND LIBLCF_DEPS "icu-i18n")
 	set(LCF_SUPPORT_ICU 1)
-else()
-	find_package(Iconv REQUIRED)
-	target_link_libraries(lcf Iconv::Iconv)
 endif()
 
 # expat

builds/cmake/Modules/FindIconv.cmake

@@ -4,8 +4,6 @@ include(CMakeFindDependencyMacro)
 
 if(@LCF_SUPPORT_ICU@)
 	find_dependency(ICU COMPONENTS i18n uc data REQUIRED)
-else()
-	find_dependency(Iconv REQUIRED)
 endif()
 
 if(@LCF_SUPPORT_XML@)

builds/cmake/liblcf-config.cmake.in

@@ -22,7 +22,7 @@ AM_CONDITIONAL(CROSS_COMPILING,[test "x$cross_compiling" = "xyes"])
 
 # Checks for libraries.
 AC_SUBST([LCF_SUPPORT_ICU],[0])
-AC_ARG_ENABLE([icu],[AS_HELP_STRING([--disable-icu],[Disable ICU encoding detection (fallback to iconv) [default=no]])])
+AC_ARG_ENABLE([icu],[AS_HELP_STRING([--disable-icu],[Disable ICU encoding detection (only for testing purposes) [default=no]])])
 AS_IF([test "x$enable_icu" != "xno"],[
 	AX_PKG_CHECK_MODULES([ICU],[],[icu-i18n],[LCF_SUPPORT_ICU=1])
 ])

configure.ac

@@ -13,7 +13,6 @@
 #include "lcf/scope_guard.h"
 #include <cstdio>
 #include <cstdlib>
-#include <exception>
 
 #if LCF_SUPPORT_ICU
 #   include <unicode/ucsdet.h>
@@ -27,9 +26,6 @@
 #ifdef _WIN32
 #   include <windows.h>
 #else
-#   if !LCF_SUPPORT_ICU
-#       include <iconv.h>
-#   endif
 #   include <locale>
 #endif
 
@@ -77,12 +73,12 @@ void Encoder::Decode(std::string& str) {
 	Convert(str, _conv_storage, _conv_runtime);
 }
 
+#if LCF_SUPPORT_ICU
 void Encoder::Init() {
 	if (_encoding.empty()) {
 		return;
 	}
 
-#if LCF_SUPPORT_ICU
 	auto code_page = atoi(_encoding.c_str());
 	const auto& storage_encoding = code_page > 0
 		? ReaderUtil::CodepageToEncoding(code_page)
@@ -110,27 +106,22 @@ void Encoder::Init() {
 
 	_conv_runtime = conv_runtime;
 	_conv_storage = conv_storage;
-#else
-	_conv_runtime = const_cast<char*>("UTF-8");
-	_conv_storage = const_cast<char*>(_encoding.c_str());
-#endif
 }
 
 void Encoder::Reset() {
-#if LCF_SUPPORT_ICU
-	auto* conv = reinterpret_cast<UConverter*>(_conv_runtime);
-	if (conv) ucnv_close(conv);
-	conv = reinterpret_cast<UConverter*>(_conv_storage);
-	if (conv) ucnv_close(conv);
-#endif
-}
+	if (_conv_runtime) {
+		ucnv_close(_conv_runtime);
+		_conv_runtime = nullptr;
+	}
 
+	if (_conv_storage) {
+		ucnv_close(_conv_storage);
+		_conv_storage = nullptr;
+	}
+}
 
-void Encoder::Convert(std::string& str, void* conv_dst_void, void* conv_src_void) {
-#if LCF_SUPPORT_ICU
+void Encoder::Convert(std::string& str, UConverter* conv_dst, UConverter* conv_src) {
 	const auto& src = str;
-	auto* conv_dst = reinterpret_cast<UConverter*>(conv_dst_void);
-	auto* conv_src = reinterpret_cast<UConverter*>(conv_src_void);
 
 	auto status = U_ZERO_ERROR;
 	_buffer.resize(src.size() * 4);
@@ -151,36 +142,8 @@ void Encoder::Convert(std::string& str, void* conv_dst_void, void* conv_src_void
 	}
 
 	str.assign(_buffer.data(), dst_p);
-	return;
-#else
-	auto* conv_dst = reinterpret_cast<const char*>(conv_dst_void);
-	auto* conv_src = reinterpret_cast<const char*>(conv_src_void);
-	iconv_t cd = iconv_open(conv_dst, conv_src);
-	if (cd == (iconv_t)-1)
-		return;
-	char *src = &str.front();
-	size_t src_left = str.size();
-	size_t dst_size = str.size() * 5 + 10;
-	_buffer.resize(dst_size);
-	char *dst = _buffer.data();
-	size_t dst_left = dst_size;
-#    ifdef ICONV_CONST
-	char ICONV_CONST *p = src;
-#    else
-	char *p = src;
-#    endif
-	char *q = dst;
-	size_t status = iconv(cd, &p, &src_left, &q, &dst_left);
-	iconv_close(cd);
-	if (status == (size_t) -1 || src_left > 0) {
-		str.clear();
-		return;
-	}
-	*q++ = '\0';
-	str.assign(dst, dst_size - dst_left);
-	return;
-#endif
 }
+#endif
 
 } //namespace lcf
 

src/encoder.cpp

@@ -12,6 +12,10 @@
 #include <vector>
 #include <string>
 
+#if LCF_SUPPORT_ICU
+class UConverter;
+#endif
+
 namespace lcf {
 
 class Encoder {
@@ -23,20 +27,39 @@ class Encoder {
 
 		~Encoder();
 
+		/**
+		 * Converts from the specified encoding to UTF-8
+		 *
+		 * @param str String to encode to UTF-8
+		 */
 		void Encode(std::string& str);
+
+		/**
+		 * Converts from UTF-8 to the specified encoding
+		 *
+		 * @param str String to decode from UTF-8
+		 */
 		void Decode(std::string& str);
 
 		bool IsOk() const;
 
 		const std::string& GetEncoding() const;
 	private:
+#if LCF_SUPPORT_ICU
 		void Init();
 		void Reset();
-		void Convert(std::string& str, void* conv_dst, void* conv_src);
-	private:
+		void Convert(std::string& str, UConverter* conv_dst, UConverter* conv_src);
+
+		UConverter* _conv_storage = nullptr;
+		UConverter* _conv_runtime = nullptr;
+		std::vector<char> _buffer;
+#else
+		void Init() {}
+		void Reset() {}
+		void Convert(std::string&, void*, void*) {}
 		void* _conv_storage = nullptr;
 		void* _conv_runtime = nullptr;
-		std::vector<char> _buffer;
+#endif
 		std::string _encoding;
 };
 

src/lcf/encoder.h

@@ -104,27 +104,14 @@ namespace ReaderUtil {
 	 */
 	std::string Recode(StringView str_to_encode, StringView source_encoding);
 
-	/**
-	 * Converts a string between encodings.
-	 *
-	 * @param str_to_encode the string to convert.
-	 * @param src_enc the source encoding.
-	 * @param dst_enc the destination encoding.
-	 * @return the recoded string.
-	 */
-	std::string Recode(StringView str_to_encode,
-					   StringView src_enc,
-					   StringView dst_enc);
-
 	/**
 	 * Converts a UTF-8 string to lowercase and then decomposes it.
-	 * 
+	 *
 	 * @param str the string to normalize.
 	 * @return the normalized string.
 	 */
 	std::string Normalize(StringView str);
 
-
 	/**
 	 * Helper function that returns an element from a vector using a 1-based
 	 * index as usually used by LCF data structures.