started syntax higlighting (still need to add state to each line)

7 files changed, 179 insertions, 3 deletions

diff --git a/base.h b/base.h
index afd320a..293d95e 100644
--- a/base.h
+++ b/base.h
@@ -65,6 +65,14 @@ typedef long long llong;
 typedef unsigned long long ullong;
 
 #if __clang__
+#define ENUM_U8 typedef enum : u8
+#define ENUM_U8_END(name) name
+#else
+#define ENUM_U8 enum
+#define ENUM_U8_END(name) ; typedef u8 name
+#endif
+
+#if __clang__
 #define ENUM_U16 typedef enum : u16
 #define ENUM_U16_END(name) name
 #else
diff --git a/buffer.c b/buffer.c
index 41c5d09..6f1b607 100644
--- a/buffer.c
+++ b/buffer.c
@@ -132,6 +132,8 @@ static bool buffer_pos_valid(TextBuffer *buffer, BufferPos p) {
 
 // are there any unsaved changes?
 bool buffer_unsaved_changes(TextBuffer *buffer) {
+	if (streq(buffer->filename, TED_UNTITLED) && buffer_empty(buffer))
+		return false; // don't worry about empty untitled buffers
 	return buffer->modified;
 }
 
@@ -1969,13 +1971,21 @@ void buffer_render(TextBuffer *buffer, Rect r) {
 	// scroll <= sel - scrolloff
 	text_state.y -= (float)(buffer->scroll_y - start_line) * char_height;
 
-	gl_color_rgba(colors[COLOR_TEXT]);
+	SyntaxState syntax_state = {0};
+	// dynamic array of character types, to be filled by syntax_highlight
+	SyntaxCharType *char_types = NULL;
 
 	for (u32 line_idx = start_line; line_idx < nlines; ++line_idx) {
 		Line *line = &lines[line_idx];
-		for (char32_t *p = line->str, *end = p + line->len; p != end; ++p) {
-			char32_t c = *p;
+		if (arr_len(char_types) < line->len) {
+			arr_set_len(char_types, line->len);
+		}
+		syntax_highlight(&syntax_state, LANG_C, line->str, line->len, char_types);
+		for (u32 i = 0; i < line->len; ++i) {
+			char32_t c = line->str[i];
+			SyntaxCharType type = char_types[i];
 
+			gl_color_rgba(colors[syntax_char_type_to_color(type)]);
 			switch (c) {
 			case '\n': assert(0); break;
 			case '\r': break; // for CRLF line endings
@@ -2003,6 +2013,9 @@ void buffer_render(TextBuffer *buffer, Rect r) {
 		column = 0;
 	}
 	
+	arr_free(char_types);
+
+	
 
 	text_chars_end(font);
 
diff --git a/colors.h b/colors.h
index 7945beb..64d7ca9 100644
--- a/colors.h
+++ b/colors.h
@@ -23,6 +23,13 @@ ENUM_U16 {
 	COLOR_NO,
 	COLOR_CANCEL,
 
+	COLOR_KEYWORD,
+	COLOR_COMMENT,
+	COLOR_PREPROCESSOR,
+	COLOR_STRING,
+	COLOR_CHARACTER,
+	COLOR_NUMBER,
+
 	COLOR_COUNT
 } ENUM_U16_END(ColorSetting);
 
@@ -50,6 +57,12 @@ static ColorName const color_names[COLOR_COUNT] = {
 	{COLOR_ERROR_BG, "error-bg"},
 	{COLOR_ERROR_BORDER, "error-border"},
 	{COLOR_ACTIVE_TAB_HL, "active-tab-hl"},
+	{COLOR_KEYWORD, "keyword"},
+	{COLOR_COMMENT, "comment"},
+	{COLOR_PREPROCESSOR, "preprocessor"},
+	{COLOR_STRING, "string"},
+	{COLOR_CHARACTER, "character"},
+	{COLOR_NUMBER, "number"},
 	{COLOR_YES, "yes"},
 	{COLOR_NO, "no"},
 	{COLOR_CANCEL, "cancel"}
@@ -106,3 +119,4 @@ static Status color_from_str(char const *str, u32 *color) {
 		*color = (u32)r << 24 | (u32)g << 16 | (u32)b << 8 | (u32)a;
 	return true;
 }
+
diff --git a/main.c b/main.c
index 8b68662..0932609 100644
--- a/main.c
+++ b/main.c
@@ -39,6 +39,7 @@ no_warn_end
 #include "time.c"
 #include "string32.c"
 #include "arr.c"
+#include "syntax.c"
 #include "buffer.c"
 #include "ted.c"
 #include "ui.c"
diff --git a/string32.c b/string32.c
index fc6117b..381bf15 100644
--- a/string32.c
+++ b/string32.c
@@ -110,3 +110,16 @@ size_t str32_remove_all_instances_of_char(String32 *s, char32_t c) {
 	s->len = out;
 	return ndeleted;
 }
+
+bool is32_space(char32_t c) {
+	return c <= WINT_MAX && iswspace((wint_t)c);
+}
+
+bool is32_alpha(char32_t c) {
+	return c <= WINT_MAX && iswalpha((wint_t)c);
+}
+
+bool is32_alnum(char32_t c) {
+	return c <= WINT_MAX && iswalnum((wint_t)c);
+}
+
diff --git a/syntax.c b/syntax.c
new file mode 100644
index 0000000..50e87d9
--- /dev/null
+++ b/syntax.c
@@ -0,0 +1,121 @@
+typedef struct {
+	bool multi_line_comment:1; // are we in a multi-line comment? (delineated by /* */)
+	bool continued_single_line_comment:1; // if you add a \ to the end of a single-line comment, it is continued to the next line.
+	bool continued_preprocessor:1; // similar to above
+	bool continued_string:1;
+} SyntaxStateC;
+
+typedef union {
+	SyntaxStateC c;
+} SyntaxState;
+
+ENUM_U16 {
+	LANG_C
+} ENUM_U16_END(Language);
+
+ENUM_U8 {
+	SYNTAX_NORMAL,
+	SYNTAX_KEYWORD,
+	SYNTAX_COMMENT,
+	SYNTAX_PREPROCESSOR,
+	SYNTAX_STRING,
+	SYNTAX_CHARACTER,
+	SYNTAX_NUMBER
+} ENUM_U8_END(SyntaxCharType);
+
+// NOTE: returns the color setting, not the color
+ColorSetting syntax_char_type_to_color(SyntaxCharType t) {
+	switch (t) {
+	case SYNTAX_NORMAL: return COLOR_TEXT;
+	case SYNTAX_KEYWORD: return COLOR_KEYWORD;
+	case SYNTAX_COMMENT: return COLOR_COMMENT;
+	case SYNTAX_PREPROCESSOR: return COLOR_PREPROCESSOR;
+	case SYNTAX_STRING: return COLOR_STRING;
+	case SYNTAX_CHARACTER: return COLOR_CHARACTER;
+	case SYNTAX_NUMBER: return COLOR_NUMBER;
+	}
+}
+
+static void syntax_highlight_c(SyntaxStateC *state, char32_t *line, u32 line_len, SyntaxCharType *char_types) {
+	(void)state;
+	bool in_preprocessor = state->continued_preprocessor;
+	bool in_string = state->continued_string;
+	bool in_single_line_comment = state->continued_single_line_comment; // this kind of comment :)
+	bool in_multi_line_comment = state->multi_line_comment;
+	int backslashes = 0;
+	for (u32 i = 0; i < line_len; ++i) {
+		SyntaxCharType type = SYNTAX_NORMAL;
+		// necessary for the final " of a string to be highlighted
+		bool in_string_now = in_string;
+		bool in_multi_line_comment_now = in_multi_line_comment;
+
+		// are there 1/2 characters left in the line?
+		bool has_1_char = i + 1 < line_len;
+
+		switch (line[i]) {
+		case '#':
+			if (!in_single_line_comment && !in_multi_line_comment)
+				in_preprocessor = true;
+			break;
+		case '\\':
+			++backslashes;
+			break;
+		case '/':
+			if (!in_multi_line_comment && !in_single_line_comment && !in_string && has_1_char) {
+				if (line[i + 1] == '/')
+					in_single_line_comment = true; // //
+				else if (line[i + 1] == '*')
+					in_multi_line_comment = in_multi_line_comment_now = true; // /*
+			} else if (in_multi_line_comment) {
+				if (i && line[i - 1] == '*') {
+					// */
+					in_multi_line_comment = false;
+				}
+			}
+			break;
+		case '"':
+			if (in_string && backslashes % 2 == 0)
+				in_string = false;
+			else if (!in_multi_line_comment && !in_single_line_comment)
+				in_string = in_string_now = true;
+			break;
+		case '<': // preprocessor string, e.g. <stdio.h>
+			if (in_preprocessor)
+				in_string = in_string_now = true;
+			break;
+		case '>':
+			if (in_preprocessor && in_string)
+				in_string = false;
+			break;
+		}
+		if (line[i] != '\\') backslashes = 0;
+
+		if (in_single_line_comment || in_multi_line_comment_now)
+			type = SYNTAX_COMMENT;
+		else if (in_string_now)
+			type = SYNTAX_STRING;
+		else if (in_preprocessor)
+			type = SYNTAX_PREPROCESSOR;
+
+		if (char_types) {
+			char_types[i] = type;
+		}
+	}
+	state->continued_single_line_comment = backslashes && in_single_line_comment;
+	state->continued_preprocessor = backslashes && in_preprocessor;
+	state->continued_string = backslashes && in_string;
+
+	state->multi_line_comment = in_multi_line_comment;
+}
+
+// This is the main syntax highlighting function. It will determine which colors to use for each character.
+// Rather than returning colors, it returns a character type (e.g. comment) which can be converted to a color.
+// To highlight multiple lines, start out with a zeroed SyntaxState, and pass a pointer to it each time.
+// You can set char_types to NULL if you just want to advance the state, and don't care about the character types.
+void syntax_highlight(SyntaxState *state, Language lang, char32_t *line, u32 line_len, SyntaxCharType *char_types) {
+	switch (lang) {
+	case LANG_C:
+		syntax_highlight_c(&state->c, line, line_len, char_types);
+		break;
+	}
+}
diff --git a/ted.cfg b/ted.cfg
index a639917..79c9e16 100644
--- a/ted.cfg
+++ b/ted.cfg
@@ -125,3 +125,9 @@ error-text = #fdd
 yes = #afa
 no = #faa
 cancel = #ffa
+
+preprocessor = #77f
+string = #f77
+character = #f7f
+comment = #777
+number = #aff