Add single codepoint unicode character support

include/terml.h

@@ -14,14 +14,21 @@ typedef void (*terml_resize_callback)(
 	unsigned int previous_height
 );
 
+struct tcell
+{
+	int codepoint;
+	int foreground;
+	int background;
+};
+
 int terml_init();
 int terml_deinit();
 
 unsigned int terml_get_width();
 unsigned int terml_get_height();
 
-char terml_get(unsigned int x, unsigned int y, int* foreground_color, int* background_color);
+int  terml_get(unsigned int x, unsigned int y, const tcell** cell);
-void terml_set(unsigned int x, unsigned int y, char c, int foreground_color, int background_color);
+int  terml_set(unsigned int x, unsigned int y, tcell  cell);
 void terml_flush();
 
 void terml_set_main_callback(terml_main_callback);

source/terml.cpp

@@ -9,108 +9,102 @@
 #include <cstdio>
 #include <cstring>
 
+namespace
+{
+	constexpr int BYTE_ONE_MASK   = 0b0000'0000'0000'0000'0011'1111;
+	constexpr int BYTE_TWO_MASK   = 0b0000'0000'0000'1111'1100'0000;
+	constexpr int BYTE_THREE_MASK = 0b0000'0011'1111'0000'0000'0000;
+	constexpr int BYTE_FOUR_MASK  = 0b0001'1100'0000'0000'0000'0000;
+
+	constexpr int BYTE_ONE_MASK_OFFSET   = 0;
+	constexpr int BYTE_TWO_MASK_OFFSET   = BYTE_ONE_MASK_OFFSET   + 2;
+	constexpr int BYTE_THREE_MASK_OFFSET = BYTE_TWO_MASK_OFFSET   + 2;
+	constexpr int BYTE_FOUR_MASK_OFFSET  = BYTE_THREE_MASK_OFFSET + 2;
+
+	constexpr int ONE_BYTE_FILL   =                            0b10000000;
+	constexpr int TWO_BYTE_FILL   =                   0b11000000'10000000;
+	constexpr int THREE_BYTE_FILL =          0b11100000'10000000'10000000;
+	constexpr int FOUR_BYTE_FILL  = 0b11110000'10000000'10000000'10000000;
+
+	void print_cell(tcell cell, const tcell* last)
+	{
+		if (!last || cell.foreground != last->foreground)
+		{
+			printf(FGP, (cell.foreground & 0xFF0000) >> 16, (cell.foreground & 0xFF00) >> 8, cell.foreground & 0xFF);
+		}
+		if (!last || cell.background != last->background)
+		{
+			printf(BGP, (cell.background & 0xFF0000) >> 16, (cell.background & 0xFF00) >> 8, cell.background & 0xFF);
+		}
+
+		// one-byte codepoints
+		if (cell.codepoint < 0x80)
+		{
+			const char str[] = { cell.codepoint, '\0' };
+			printf("%s", str);
+		}
+		// two-byte codepoints
+		else if (cell.codepoint < 0x800)
+		{
+			const int byte_one = (cell.codepoint & BYTE_ONE_MASK) << BYTE_ONE_MASK_OFFSET;
+			const int byte_two = (cell.codepoint & BYTE_TWO_MASK) << BYTE_TWO_MASK_OFFSET;
+			const int utf8 = TWO_BYTE_FILL | byte_one | byte_two;
+			
+			const char str[] = { (utf8 & 0xFF00) >> 8, utf8 & 0xFF, '\0' };
+			printf("%s", str);
+		}
+		// three-byte codepoints
+		else if (cell.codepoint < 0x10000)
+		{
+			const int byte_one   = (cell.codepoint & BYTE_ONE_MASK)   << BYTE_ONE_MASK_OFFSET;
+			const int byte_two   = (cell.codepoint & BYTE_TWO_MASK)   << BYTE_TWO_MASK_OFFSET;
+			const int byte_three = (cell.codepoint & BYTE_THREE_MASK) << BYTE_THREE_MASK_OFFSET;
+			const int utf8 = THREE_BYTE_FILL | byte_one | byte_two | byte_three;
+
+			const char str[] = { (utf8 & 0xFF0000) >> 16, (utf8 & 0xFF00) >> 8, utf8 & 0xFF, '\0' };
+			printf("%s", str);
+		}
+		// four-byte codepoints
+		else if (cell.codepoint < 0x110000)
+		{
+			const int byte_one   = (cell.codepoint & BYTE_ONE_MASK)   << BYTE_ONE_MASK_OFFSET;
+			const int byte_two   = (cell.codepoint & BYTE_TWO_MASK)   << BYTE_TWO_MASK_OFFSET;
+			const int byte_three = (cell.codepoint & BYTE_THREE_MASK) << BYTE_THREE_MASK_OFFSET;
+			const int byte_four  = (cell.codepoint & BYTE_FOUR_MASK)  << BYTE_FOUR_MASK_OFFSET;
+			const int utf8 = FOUR_BYTE_FILL | byte_one | byte_two | byte_three | byte_four;
+
+			const char str[] = { (utf8 & 0xFF000000) >> 24, (utf8 & 0xFF0000) >> 16, (utf8 & 0xFF00) >> 8, utf8 & 0xFF, '\0' };
+			printf("%s", str);
+		}
+	}
+}
+
 terml::terml() :
+	cells(nullptr),
+	width(0),
+	height(0),
 	main(nullptr),
 	quit(nullptr),
 	key(nullptr),
 	resize(nullptr),
-	buffer(nullptr),
+	should_quit(false),
-	width(0),
+	really_should_quit(false)
-	height(0)
 {
-	printf(ALT_BUF() HIDE_CURSOR());
-	fflush(stdout);
 }
 
 terml::~terml()
 {
-	printf(REG_BUF() SHOW_CURSOR());
+	if (cells) { delete[] cells; }
-	fflush(stdout);
-
-	delete[] buffer;
 }
 
-char terml::get(unsigned int x, unsigned int y, int* fg, int* bg) const
+const tcell& terml::get(unsigned int x, unsigned int y) const
 {
-	const unsigned int offset = x + y * width;
+	return cells[x + y * width];
-
-	if (fg)
-	{
-		const unsigned char r =
-			(buffer[offset * CELL_SIZE + 9] - '0') +
-			(buffer[offset * CELL_SIZE + 8] - '0') * 10 +
-			(buffer[offset * CELL_SIZE + 7] - '0') * 100;
-
-		const unsigned char g =
-			(buffer[offset * CELL_SIZE + 13] - '0') +
-			(buffer[offset * CELL_SIZE + 12] - '0') * 10 +
-			(buffer[offset * CELL_SIZE + 11] - '0') * 100;
-
-		const unsigned char b =
-			(buffer[offset * CELL_SIZE + 17] - '0') +
-			(buffer[offset * CELL_SIZE + 16] - '0') * 10 +
-			(buffer[offset * CELL_SIZE + 15] - '0') * 100;
-
-		*fg = (r << 16) | (g << 8) | b;
-	}
-
-	if (bg)
-	{
-		const unsigned char r =
-			(buffer[offset * CELL_SIZE + 28] - '0') +
-			(buffer[offset * CELL_SIZE + 27] - '0') * 10 +
-			(buffer[offset * CELL_SIZE + 26] - '0') * 100;
-
-		const unsigned char g =
-			(buffer[offset * CELL_SIZE + 32] - '0') +
-			(buffer[offset * CELL_SIZE + 31] - '0') * 10 +
-			(buffer[offset * CELL_SIZE + 30] - '0') * 100;
-
-		const unsigned char b =
-			(buffer[offset * CELL_SIZE + 36] - '0') +
-			(buffer[offset * CELL_SIZE + 35] - '0') * 10 +
-			(buffer[offset * CELL_SIZE + 34] - '0') * 100;
-
-		*bg = (r << 16) | (g << 8) | b;
-	}
-
-	return buffer[offset * CELL_SIZE + CELL_SIZE - 1];
 }
 
-void terml::set(unsigned int x, unsigned int y, char c, int fg, int bg)
+void terml::set(unsigned int x, unsigned int y, tcell cell)
 {
-	const unsigned int offset = x + y * width;
+	cells[x + y * width] = cell;
-
-	unsigned char fgr = (fg & 0xFF0000) >> 16;
-	unsigned char fgg = (fg & 0x00FF00) >> 8;
-	unsigned char fgb = (fg & 0x0000FF);
-
-	unsigned char bgr = (bg & 0xFF0000) >> 16;
-	unsigned char bgg = (bg & 0x00FF00) >> 8;
-	unsigned char bgb = (bg & 0x0000FF);
-
-	for (int i = 0; i < 3; i++)
-	{
-		buffer[offset * CELL_SIZE + (9 - i)] = (fgr % 10) + '0';
-		fgr /= 10;
-
-		buffer[offset * CELL_SIZE + (13 - i)] = (fgg % 10) + '0';
-		fgg /= 10;
-
-		buffer[offset * CELL_SIZE + (17 - i)] = (fgb % 10) + '0';
-		fgb /= 10;
-
-		buffer[offset * CELL_SIZE + (28 - i)] = (bgr % 10) + '0';
-		bgr /= 10;
-
-		buffer[offset * CELL_SIZE + (32 - i)] = (bgg % 10) + '0';
-		bgg /= 10;
-
-		buffer[offset * CELL_SIZE + (36 - i)] = (bgb % 10) + '0';
-		bgb /= 10;
-	}
-
-	buffer[offset * CELL_SIZE + CELL_SIZE - 1] = c;
 }
 
 void terml::flush() const
@@ -118,7 +112,11 @@ void terml::flush() const
 	printf(CUP(1, 1));
 	fflush(stdout);
 
-	printf("%s", buffer);
+	print_cell(cells[0], nullptr);
+	for (int i = 1; i < width * height; i++)
+	{
+		print_cell(cells[i], &cells[i - 1]);
+	}
 	fflush(stdout);
 }
 
@@ -155,7 +153,7 @@ void terml::mainloop()
 	should_quit = false;
 	really_should_quit = false;
 
-	const unsigned long long wait_time = timer_frequency() / 1000;
+	const unsigned long long wait_time = timer_frequency() / 60;
 	unsigned long long last_time = timer();
 
 	while (!really_should_quit)
@@ -230,17 +228,9 @@ void terml::setup_buffer()
 		width = new_width;
 		height = new_height;
 
-		if (buffer) { delete[] buffer; }
+		if (cells) { delete[] cells; }
-		buffer = new char[CELL_SIZE * width * height + 1];
+		cells = new tcell[width * height];
-
+		memset(cells, 0, sizeof(tcell) * width * height);
-		constexpr char BLANK_CELL[] = FG(255, 255, 255) CSI "48;2;000;000;000m" " ";
-
-		for (int i = 0; i < width * height; i++)
-		{
-			memcpy(buffer + i * CELL_SIZE, BLANK_CELL, CELL_SIZE);
-		}
-
-		buffer[CELL_SIZE * width * height] = '\0';
 
 		if (resize)
 		{
@@ -249,6 +239,23 @@ void terml::setup_buffer()
 	}
 }
 
+void terml::set_console_settings()
+{
+	setvbuf(stdout, nullptr, _IOFBF, BUFSIZ * BUFSIZ);
+	printf(ALT_BUF() HIDE_CURSOR());
+	fflush(stdout);
+	set_console_settings_impl();
+}
+
+void terml::reset_console_settings()
+{
+	setvbuf(stdout, nullptr, _IOLBF, BUFSIZ);
+	printf(REG_BUF() SHOW_CURSOR());
+	fflush(stdout);
+
+	reset_console_settings_impl();
+}
+
 static terml* TERML_G;
 static const char* LAST_ERROR = nullptr;
 
@@ -266,16 +273,20 @@ extern "C"
 				TERML_G = new terml_linux;
 #endif
 			}
+			else
+			{
+				throw "terml already initialized.";
+			}
 
 			TERML_G->set_console_settings();
 			TERML_G->setup_buffer();
 
-			return 1;
+			return 0;
 		}
 		catch (const char* c)
 		{
 			LAST_ERROR = c;
-			return 0;
+			return 1;
 		}
 	}
 
@@ -289,12 +300,12 @@ extern "C"
 				delete TERML_G;
 			}
 
-			return 1;
+			return 0;
 		}
 		catch (const char* c)
 		{
 			LAST_ERROR = c;
-			return 0;
+			return 1;
 		}
 	}
 
@@ -308,14 +319,37 @@ extern "C"
 		return TERML_G->get_height();
 	}
 
-	char terml_get(unsigned int x, unsigned int y, int* fg, int* bg)
+	int terml_get(unsigned int x, unsigned int y, const tcell** cell)
 	{
-		return TERML_G->get(x, y, fg, bg);
+		if (x >= TERML_G->get_width() || y >= TERML_G->get_height())
+		{
+			LAST_ERROR = "Coordinates out of bounds.";
+			return 1;
+		}
+		else if (!cell)
+		{
+			LAST_ERROR = "Null output pointer.";
+			return 1;
+		}
+		else
+		{
+			*cell = &TERML_G->get(x, y);
+			return 0;
+		}
 	}
 
-	void terml_set(unsigned int x, unsigned int y, char c, int fg, int bg)
+	int terml_set(unsigned int x, unsigned int y, tcell cell)
 	{
-		TERML_G->set(x, y, c, fg, bg);
+		if (x >= TERML_G->get_width() || y >= TERML_G->get_height())
+		{
+			LAST_ERROR = "Coordinates out of bounds.";
+			return 1;
+		}
+		else
+		{
+			TERML_G->set(x, y, cell);
+			return 0;
+		}
 	}
 
 	void terml_flush()

source/terml_linux.cpp

@@ -7,7 +7,7 @@
 #include <cstdio>
 #include <poll.h>
 
-void terml_linux::set_console_settings()
+void terml_linux::set_console_settings_impl()
 {
 	tcgetattr(STDIN_FILENO, &old_input_settings);
 	tcgetattr(STDOUT_FILENO, &old_output_settings);
@@ -18,7 +18,7 @@ void terml_linux::set_console_settings()
 	tcsetattr(STDIN_FILENO, TCSANOW, &t);
 }
 
-void terml_linux::reset_console_settings()
+void terml_linux::reset_console_settings_impl()
 {
 	tcsetattr(STDIN_FILENO, TCSANOW, &old_input_settings);
 	tcsetattr(STDOUT_FILENO, TCSANOW, &old_output_settings);

source/terml_linux.h

@@ -11,8 +11,8 @@
 class terml_linux : public terml
 {
 protected:
-	virtual void set_console_settings() override;
+	virtual void set_console_settings_impl() override;
-	virtual void reset_console_settings() override;
+	virtual void reset_console_settings_impl() override;
 	virtual void read_stdin(char* buffer, unsigned int buffer_size) override;
 	virtual unsigned long long timer() override;
 	virtual unsigned long long timer_frequency() override;

source/terml_private.h

@@ -25,6 +25,9 @@
 #define FG(r, g, b) CSI "38;2;" STRINGIFY(r) ";" STRINGIFY(g) ";" STRINGIFY(b) "m"
 #define BG(r, g, b) CSI "48;2;" STRINGIFY(r) ";" STRINGIFY(g) ";" STRINGIFY(b) "m"
 
+#define FGP CSI "38;2;%d;%d;%dm"
+#define BGP CSI "48;2;%d;%d;%dm"
+
 class terml
 {
 public:
@@ -37,8 +40,8 @@ public:
 	terml& operator=(const terml&) = delete;
 	terml& operator=(terml&&) = delete;
 
-	char get(unsigned int x, unsigned int y, int* fg, int* bg) const;
+	const tcell& get(unsigned int x, unsigned int y) const;
-	void set(unsigned int x, unsigned int y, char c, int fg, int bg);
+	void set(unsigned int x, unsigned int y, tcell cell);
 	void flush() const;
 
 	void set_main_callback(terml_main_callback);
@@ -54,10 +57,12 @@ public:
 
 	void setup_buffer();
 
-	virtual void set_console_settings() = 0;
+	void set_console_settings();
-	virtual void reset_console_settings() = 0;
+	void reset_console_settings();
 
 protected:
+	virtual void set_console_settings_impl() = 0;
+	virtual void reset_console_settings_impl() = 0;
 	virtual void read_stdin(char* buffer, unsigned int buffer_size) = 0;
 	virtual unsigned long long timer() = 0;
 	virtual unsigned long long timer_frequency() = 0;
@@ -66,7 +71,7 @@ protected:
 	void key_event(char code) const;
 
 private:
-	char* buffer;
+	tcell* cells;
 	unsigned int width;
 	unsigned int height;
 	terml_main_callback main;
@@ -75,8 +80,6 @@ private:
 	terml_resize_callback resize;
 	bool should_quit;
 	bool really_should_quit;
-
-	static const unsigned int CELL_SIZE = sizeof(FG(255, 255, 255) BG(255, 255, 255) " ") - 1;
 };
 
 #endif//TERML_TERML_PRIVATE_H

source/terml_windows.cpp

@@ -2,7 +2,7 @@
 
 #ifdef _WIN32
 
-void terml_windows::set_console_settings()
+void terml_windows::set_console_settings_impl()
 {
 	handle_stdin = GetStdHandle(STD_INPUT_HANDLE);
 	GetConsoleMode(handle_stdin, &previous_input_mode);
@@ -13,7 +13,7 @@ void terml_windows::set_console_settings()
 	new_mode |= ENABLE_VIRTUAL_TERMINAL_INPUT;
 	if (!SetConsoleMode(handle_stdin, new_mode))
 	{
-		throw "Failed to set stdin mode";
+		throw "Failed to set stdin mode.";
 	}
 
 	HANDLE handle = GetStdHandle(STD_OUTPUT_HANDLE);
@@ -23,19 +23,27 @@ void terml_windows::set_console_settings()
 	new_mode |= ENABLE_VIRTUAL_TERMINAL_PROCESSING;
 	if (!SetConsoleMode(handle, new_mode))
 	{
-		throw "Failed to set stdout mode";
+		throw "Failed to set stdout mode.";
 	}
+
+	previous_codepage = GetConsoleOutputCP();
+	SetConsoleOutputCP(CP_UTF8);
 }
 
-void terml_windows::reset_console_settings()
+void terml_windows::reset_console_settings_impl()
 {
 	if (!SetConsoleMode(handle_stdin, previous_input_mode))
 	{
-		throw "Failed to reset stdin mode";
+		throw "Failed to reset stdin mode.";
 	}
 	if (!SetConsoleMode(GetStdHandle(STD_OUTPUT_HANDLE), previous_output_mode))
 	{
-		throw "Failed to reset stdout mode";
+		throw "Failed to reset stdout mode.";
+	}
+
+	if (!SetConsoleOutputCP(previous_codepage))
+	{
+		throw "Failed to reset codepage.";
 	}
 }
 

source/terml_windows.h

@@ -11,8 +11,8 @@
 class terml_windows : public terml
 {
 protected:
-	virtual void set_console_settings() override;
+	virtual void set_console_settings_impl() override;
-	virtual void reset_console_settings() override;
+	virtual void reset_console_settings_impl() override;
 	virtual void read_stdin(char* buffer, unsigned int buffer_size) override;
 	virtual unsigned long long timer() override;
 	virtual unsigned long long timer_frequency() override;
@@ -21,6 +21,7 @@ protected:
 private:
 	DWORD previous_input_mode;
 	DWORD previous_output_mode;
+	UINT previous_codepage;
 	HANDLE handle_stdin;
 };