Add wchar_t and envp support, as well as utf8 conversion, to app_parameters

src/common/app_parameters.cpp

@@ -17,24 +17,89 @@
  */
 
 #include "app_parameters.hpp"
+#include <cuchar>
+#include "jessilib/unicode.hpp"
+#include "jessilib/split.hpp"
 
 namespace jessilib {
 
-app_parameters::app_parameters(int in_argc, char** in_argv)
-	: app_parameters{ in_argc, const_cast<const char**>(in_argv) } {
+// Convert null-terminated multi-byte string to UTF-8
+std::u8string ntmbs_to_u8string(const char* in_ntmbs) {
+	std::u8string result;
+
+	std::string_view ntmbs_view = in_ntmbs;
+	std::mbstate_t mbstate{};
+	char32_t codepoint{};
+	while (!ntmbs_view.empty()) {
+		size_t bytes_read = std::mbrtoc32(&codepoint, ntmbs_view.data(), ntmbs_view.size(), &mbstate);
+		if (bytes_read > ntmbs_view.size()) {
+			// Some sort of error; just break
+			break;
+		}
+		// bytes_read will never be 0 except for null characters, which are excluded from our view; here for future reuse
+		bytes_read = std::max(size_t{1}, bytes_read);
+		ntmbs_view.remove_prefix(bytes_read);
+		encode_codepoint(result, codepoint);
+	}
+
+	return result;
+}
+
+std::vector<std::u8string> vectorize(const char** in_argv) {
+	std::vector<std::u8string> result;
+	if (in_argv == nullptr) {
+		return result;
+	}
+
+	for (auto argv = in_argv; *argv != nullptr; ++argv) {
+		result.emplace_back(ntmbs_to_u8string(*argv));
+	}
+	return result;
+}
+
+std::vector<std::u8string> vectorize(const wchar_t** in_argv) {
+	std::vector<std::u8string> result;
+	if (in_argv == nullptr) {
+		return result;
+	}
+
+	for (auto argv = in_argv; *argv != nullptr; ++argv) {
+		result.emplace_back(jessilib::string_cast<char8_t>(std::wstring_view{ *argv }));
+	}
+	return result;
+}
+
+app_parameters::app_parameters(int in_argc, char** in_argv, char** in_envp)
+	: app_parameters{ in_argc, const_cast<const char**>(in_argv), const_cast<const char**>(in_envp) } {
 	// Empty ctor body
 }
 
-app_parameters::app_parameters(int in_argc, const char** in_argv) {
-	// TODO: discard argc/argv and use GetCommandLineW on Windows
-	// TODO: not assume argv is utf-8; it often will not be
+app_parameters::app_parameters(int, const char** in_argv, const char** in_envp)
+	: app_parameters{ vectorize(in_argv), vectorize(in_envp) } {
+}
+
+app_parameters::app_parameters(int in_argc, wchar_t** in_argv, wchar_t** in_envp)
+	: app_parameters{ in_argc, const_cast<const wchar_t**>(in_argv), const_cast<const wchar_t**>(in_envp) } {
+}
+
+app_parameters::app_parameters(int, const wchar_t** in_argv, const wchar_t** in_envp)
+	: app_parameters{ vectorize(in_argv), vectorize(in_envp) } {
+}
+
+app_parameters::app_parameters(std::vector<std::u8string> in_args, std::vector<std::u8string> in_env) {
+	// Parse in environment variables first to ensure they're parsed before the early-out
+	for (const auto& env : in_env) {
+		auto split = jessilib::split_once(env, u8'=');
+		m_env_values[split.first] = split.second;
+	}
+
 	// Sanity safety check; should never happen
-	if (in_argc <= 0 || in_argv == nullptr) {
+	if (in_args.empty()) {
 		return;
 	}
 
 	// Populate path
-	m_path = reinterpret_cast<const char8_t*>(in_argv[0]);
+	m_path = in_args[0];
 
 	// Process args
 	std::u8string_view key;
@@ -46,23 +111,24 @@ app_parameters::app_parameters(int in_argc, const char** in_argv) {
 				m_switches.emplace_back(key);
 			}
 			else {
-				m_values.emplace(key, std::move(value));
+				m_arg_values.emplace(key, std::move(value));
 				value.clear();
 			}
 		}
 	};
 
-	for (int index = 1; index < in_argc; ++index) {
-		const char8_t* arg = reinterpret_cast<const char8_t*>(in_argv[index]);
-		if (arg != nullptr && *arg != '\0') {
+	for (size_t index = 1; index < in_args.size(); ++index) {
+		const std::u8string& arg = in_args[index];
+		if (!arg.empty()) {
 			// Check if this is a key or value
-			if (*arg == '-') {
+			if (arg.front() == '-') {
 				// Flush pending value (if any)
 				flush_value();
 
 				// Strip any leading '-' or '--' and set key
-				key = arg + 1;
-				if (key[0] == '-') {
+				key = arg;
+				key.remove_prefix(1);
+				if (key.front() == '-') {
 					key.remove_prefix(1);
 				}
 
@@ -93,27 +159,27 @@ app_parameters::app_parameters(int in_argc, const char** in_argv) {
 	flush_value();
 
 	// Populate m_switches_set from m_switches
-	m_switches_set = std::unordered_set<std::u8string_view>{ m_switches.begin(), m_switches.end() };
+	m_switches_set = std::unordered_set<std::u8string>{ m_switches.begin(), m_switches.end() };
 }
 
 std::u8string_view app_parameters::path() const {
 	return m_path;
 }
 
-const std::vector<std::u8string_view>& app_parameters::arguments() const {
+const std::vector<std::u8string>& app_parameters::arguments() const {
 	return m_args;
 }
 
-const std::vector<std::u8string_view>& app_parameters::switches() const {
+const std::vector<std::u8string>& app_parameters::switches() const {
 	return m_switches;
 }
 
-const std::unordered_set<std::u8string_view>& app_parameters::switches_set() const {
+const std::unordered_set<std::u8string>& app_parameters::switches_set() const {
 	return m_switches_set;
 }
 
-const std::unordered_map<std::u8string_view, std::u8string>& app_parameters::values() const {
-	return m_values;
+const std::unordered_map<std::u8string, std::u8string>& app_parameters::values() const {
+	return m_arg_values;
 }
 
 object app_parameters::as_object() const {
@@ -130,23 +196,51 @@ object app_parameters::as_object() const {
 		{ u8"Path"s, m_path },
 		{ u8"Args"s, m_args },
 		{ u8"Switches"s, m_switches },
-		{ u8"Values"s, m_values }
+		{ u8"Values"s, m_arg_values }
 	};
 }
 
+#ifdef __cpp_lib_generic_unordered_lookup
+#define WRAP_MAP_KEY(in_key) in_key
+#else // We can't use std::string_view for InKeyType until GCC 11 & clang 12, and I still want to support GCC 9
+#define WRAP_MAP_KEY(in_key) static_cast<std::u8string>(in_key)
+#endif // __cpp_lib_generic_unordered_lookup
+
 bool app_parameters::has_switch(std::u8string_view in_switch) const {
-	return m_switches_set.find(in_switch) != m_switches_set.end();
+	return m_switches_set.find(WRAP_MAP_KEY(in_switch)) != m_switches_set.end();
 }
 
-std::u8string_view app_parameters::get_value(std::u8string_view in_key, std::u8string_view in_default) const {
-	auto result = m_values.find(in_key);
+std::u8string_view app_parameters::get_arg_value(std::u8string_view in_key, std::u8string_view in_default) const {
+	auto result = m_arg_values.find(WRAP_MAP_KEY(in_key));
 
 	// Safety check
-	if (result == m_values.end()) {
+	if (result == m_arg_values.end()) {
 		return in_default;
 	}
 
 	return result->second;
 }
 
+std::u8string_view app_parameters::get_env_value(std::u8string_view in_key, std::u8string_view in_default) const {
+	auto result = m_env_values.find(WRAP_MAP_KEY(in_key));
+
+	// Safety check
+	if (result == m_env_values.end()) {
+		return in_default;
+	}
+
+	return result->second;
+}
+
+std::u8string_view app_parameters::get_value(std::u8string_view in_key, std::u8string_view in_default) const {
+	// Explicit args take priority
+	auto result = m_arg_values.find(WRAP_MAP_KEY(in_key));
+	if (result != m_arg_values.end()) {
+		return result->second;
+	}
+
+	// Fallback to env
+	return get_env_value(in_key, in_default);
+}
+
 } // namespace jessilib

src/include/jessilib/app_parameters.hpp

@@ -22,27 +22,34 @@ namespace jessilib {
 
 class app_parameters {
 public:
-	app_parameters(int in_argc, char** in_argv);
-	app_parameters(int in_argc, const char** in_argv);
+	app_parameters(int in_argc, char** in_argv, char** in_envp = nullptr);
+	app_parameters(int in_argc, const char** in_argv, const char** in_envp = nullptr);
+	app_parameters(int in_argc, wchar_t** in_argv, wchar_t** in_envp = nullptr);
+	app_parameters(int in_argc, const wchar_t** in_argv, const wchar_t** in_envp = nullptr);
+	app_parameters(std::vector<std::u8string> in_args, std::vector<std::u8string> in_env = {});
 
 	std::u8string_view path() const;
-	const std::vector<std::u8string_view>& arguments() const;
-	const std::vector<std::u8string_view>& switches() const;
-	const std::unordered_set<std::u8string_view>& switches_set() const;
-	const std::unordered_map<std::u8string_view, std::u8string>& values() const;
+	const std::vector<std::u8string>& arguments() const;
+
+	const std::vector<std::u8string>& switches() const;
+	const std::unordered_set<std::u8string>& switches_set() const;
+	const std::unordered_map<std::u8string, std::u8string>& values() const;
 	jessilib::object as_object() const;
 
 	bool has_switch(std::u8string_view in_switch) const;
+	std::u8string_view get_arg_value(std::u8string_view in_key, std::u8string_view in_default = {}) const;
+	std::u8string_view get_env_value(std::u8string_view in_key, std::u8string_view in_default = {}) const;
 	std::u8string_view get_value(std::u8string_view in_key, std::u8string_view in_default = {}) const;
 
 	operator jessilib::object() const { return as_object(); }
 
 private:
-	std::u8string_view m_path;
-	std::vector<std::u8string_view> m_args;
-	std::vector<std::u8string_view> m_switches;
-	std::unordered_set<std::u8string_view> m_switches_set;
-	std::unordered_map<std::u8string_view, std::u8string> m_values;
+	std::u8string m_path;
+	std::vector<std::u8string> m_args;
+	std::vector<std::u8string> m_switches;
+	std::unordered_set<std::u8string> m_switches_set;
+	std::unordered_map<std::u8string, std::u8string> m_arg_values;
+	std::unordered_map<std::u8string, std::u8string> m_env_values;
 };
 
 } // namespace jessilib

src/include/jessilib/unicode_base.hpp

@@ -454,7 +454,7 @@ constexpr decode_result decode_codepoint_utf32(std::basic_string_view<CharT> in_
 		return { 0, 0 };
 	}
 
-	return { in_string.front(), 1 };
+	return { static_cast<char32_t>(in_string.front()), 1 };
 }
 
 template<typename CharT>

src/test/app_parameters.cpp

@@ -22,24 +22,26 @@
 using namespace jessilib;
 using namespace std::literals;
 
+template<typename CharT = char>
 class ArgWrapper {
 public:
 	template<typename... Args>
 	ArgWrapper(Args... in_args)
-		: ArgWrapper{ std::vector<std::string>{ in_args... } } {
+		: ArgWrapper{ std::vector<std::basic_string<CharT>>{ in_args... } } {
 		// Empty ctor body
 	}
 
-	ArgWrapper(std::vector<std::string> in_args)
+	ArgWrapper(std::vector<std::basic_string<CharT>> in_args)
 		: m_args{ in_args },
-		m_argv{ new const char*[in_args.size()] } {
+		m_argv{ new const CharT*[in_args.size() + 1] } {
 		// Populate m_argv
 		for (size_t index = 0; index != m_args.size(); ++index) {
 			m_argv[index] = m_args[index].c_str();
 		}
+		m_argv[in_args.size()] = nullptr; // last arg is always nullptr
 	}
 
-	const char** argv() const {
+	const CharT** argv() const {
 		return m_argv.get();
 	}
 
@@ -48,8 +50,8 @@ public:
 	}
 
 private:
-	std::vector<std::string> m_args;
-	std::unique_ptr<const char*[]> m_argv;
+	std::vector<std::basic_string<CharT>> m_args;
+	std::unique_ptr<const CharT*[]> m_argv;
 };
 
 TEST(AppParametersTest, null) {
@@ -86,6 +88,21 @@ TEST(AppParametersTest, path_only) {
 	EXPECT_EQ(obj[u8"Path"], u8"/path/to/exe");
 }
 
+TEST(AppParametersTest, path_only_w) {
+	ArgWrapper<wchar_t> args{ L"/path/to/exe" };
+	app_parameters parameters{ args.argc(), args.argv() };
+
+	EXPECT_EQ(parameters.path(), u8"/path/to/exe");
+	EXPECT_TRUE(parameters.arguments().empty());
+	EXPECT_TRUE(parameters.switches().empty());
+	EXPECT_TRUE(parameters.switches_set().empty());
+	EXPECT_TRUE(parameters.values().empty());
+
+	auto obj = parameters.as_object();
+	EXPECT_FALSE(obj.null());
+	EXPECT_EQ(obj[u8"Path"], u8"/path/to/exe");
+}
+
 TEST(AppParametersTest, single_switch) {
 	ArgWrapper args{ "/path/to/exe", "-switch" };
 	app_parameters parameters{ args.argc(), args.argv() };
@@ -279,3 +296,28 @@ TEST(AppParametersTest, switch_and_value) {
 	EXPECT_FALSE(parameters.has_switch(u8"switch2"));
 	EXPECT_EQ(parameters.get_value(u8"key"), u8"value");
 }
+
+TEST(AppParametersTest, switch_and_value_w) {
+	ArgWrapper<wchar_t> args{ L"/path/to/exe", L"--switch", L"-key", L"value" };
+	app_parameters parameters{ args.argc(), args.argv() };
+
+	EXPECT_FALSE(parameters.path().empty());
+	EXPECT_EQ(parameters.arguments().size(), 3U);
+	EXPECT_EQ(parameters.switches().size(), 1U);
+	EXPECT_EQ(parameters.switches_set().size(), 1U);
+	EXPECT_EQ(parameters.values().size(), 1U);
+
+	auto obj = parameters.as_object();
+	std::vector<std::u8string> expected_args{ u8"--switch", u8"-key", u8"value" };
+	std::vector<std::u8string> expected_switches{ u8"switch" };
+	std::map<std::u8string, object> expected_values{ { u8"key", u8"value" } };
+	EXPECT_FALSE(obj.null());
+	EXPECT_EQ(obj[u8"Path"], u8"/path/to/exe");
+	EXPECT_EQ(obj[u8"Args"], expected_args);
+	EXPECT_EQ(obj[u8"Switches"], expected_switches);
+	EXPECT_EQ(obj[u8"Values"], expected_values);
+
+	EXPECT_TRUE(parameters.has_switch(u8"switch"));
+	EXPECT_FALSE(parameters.has_switch(u8"switch2"));
+	EXPECT_EQ(parameters.get_value(u8"key"), u8"value");
+}