different syntax highlighting depending on the file extension

6 files changed, 87 insertions, 6 deletions

diff --git a/buffer.c b/buffer.c
index 2b9957c..fa16c20 100644
--- a/buffer.c
+++ b/buffer.c
@@ -177,6 +177,33 @@ static inline Settings const *buffer_settings(TextBuffer *buffer) {
 	return &buffer->ted->settings;
 }
 
+// what programming language is this?
+Language buffer_language(TextBuffer *buffer) {
+	Settings const *settings = buffer_settings(buffer);
+	char const *filename = buffer->filename;
+	if (!filename)
+		return LANG_NONE;
+	size_t filename_len = strlen(filename);
+
+	for (u16 l = 0; l < LANG_COUNT; ++l) {
+		char const *extensions = settings->language_extensions[l];
+		if (extensions) {
+			// extensions is a string with commas separating each extension.
+			size_t len = 0;
+			for (char const *p = extensions; *p; p += len) {
+				if (*p == ',') ++p; // move past comma
+				len = strcspn(p, ",");
+				if (filename_len >= len && strncmp(&filename[filename_len - len], p, len) == 0) {
+					// found a match!
+					return (Language)l;
+				}
+			}
+		}
+	}
+	// no extensions matched
+	return LANG_NONE;
+}
+
 // NOTE: this string will be invalidated when the line is edited!!!
 // only use it briefly!!
 static String32 buffer_get_line(TextBuffer *buffer, u32 line_number) {
@@ -1976,11 +2003,12 @@ void buffer_render(TextBuffer *buffer, Rect r) {
 	text_state.y -= (float)(buffer->scroll_y - start_line) * char_height;
 
 	SyntaxState syntax_state = {0};
+	Language language = buffer_language(buffer);
 	// dynamic array of character types, to be filled by syntax_highlight
 	SyntaxCharType *char_types = NULL;
 	for (u32 line_idx = 0; line_idx < start_line; ++line_idx) {
 		Line *line = &lines[line_idx];
-		syntax_highlight(&syntax_state, buffer->language, line->str, line->len, NULL);
+		syntax_highlight(&syntax_state, language, line->str, line->len, NULL);
 	}
 
 	for (u32 line_idx = start_line; line_idx < nlines; ++line_idx) {
@@ -1988,7 +2016,7 @@ void buffer_render(TextBuffer *buffer, Rect r) {
 		if (arr_len(char_types) < line->len) {
 			arr_set_len(char_types, line->len);
 		}
-		syntax_highlight(&syntax_state, buffer->language, line->str, line->len, char_types);
+		syntax_highlight(&syntax_state, language, 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];
diff --git a/config.c b/config.c
index 2ebc094..bf3b284 100644
--- a/config.c
+++ b/config.c
@@ -11,7 +11,8 @@ typedef enum {
 	SECTION_NONE,
 	SECTION_CORE,
 	SECTION_KEYBOARD,
-	SECTION_COLORS
+	SECTION_COLORS,
+	SECTION_EXTENSIONS
 } Section;
 
 // all worth it for the -Wformat warnings
@@ -227,6 +228,8 @@ void config_read(Ted *ted, char const *filename) {
 								section = SECTION_COLORS;
 							} else if (streq(section_name, "core")) {
 								section = SECTION_CORE;
+							} else if (streq(section_name, "extensions")) {
+								section = SECTION_EXTENSIONS;
 							} else {
 								config_err(cfg, "Unrecognized section: [%s].", section_name);
 							}
@@ -294,6 +297,25 @@ void config_read(Ted *ted, char const *filename) {
 										config_err(cfg, "Expected ':' for key action. This line should look something like: %s = :command.", key);
 									}
 								} break;
+								case SECTION_EXTENSIONS: {
+									Language lang = language_from_str(key);
+									if (lang == LANG_NONE) {
+										config_err(cfg, "Invalid programming language: %s.", key);
+									} else {
+										char *new_str = malloc(strlen(value) + 1);
+										if (!new_str) {
+											config_err(cfg, "Out of memory.");
+										} else {
+											char *dst = new_str;
+											// get rid of whitespace in extension list
+											for (char const *src = value; *src; ++src)
+												if (!isspace(*src))
+													*dst++ = *src;
+											*dst = 0;
+											settings->language_extensions[lang] = new_str;
+										}
+									}
+								} break;
 								case SECTION_CORE: {
 									char const *endptr;
 									long long const integer = strtoll(value, (char **)&endptr, 10);
@@ -380,3 +402,10 @@ void config_read(Ted *ted, char const *filename) {
 		ted_seterr(ted, "Couldn't open file %s.", filename);
 	}
 }
+
+static void settings_free(Settings *settings) {
+	for (u16 i = 0; i < LANG_COUNT; ++i) {
+		free(settings->language_extensions[i]);
+		settings->language_extensions[i] = NULL;
+	}
+}
diff --git a/main.c b/main.c
index 0932609..bcad077 100644
--- a/main.c
+++ b/main.c
@@ -591,6 +591,7 @@ int main(int argc, char **argv) {
 	buffer_free(&ted->line_buffer);
 	text_font_free(ted->font);
 	text_font_free(ted->font_bold);
+	settings_free(&ted->settings);
 	free(ted);
 #if _WIN32
 	for (int i = 0; i < argc; ++i)
diff --git a/syntax.c b/syntax.c
index eff776c..a4378b1 100644
--- a/syntax.c
+++ b/syntax.c
@@ -1,3 +1,12 @@
+// returns the language this string is referring to, or LANG_NONE if it's invalid.
+Language language_from_str(char const *str) {
+	for (int i = 0; i < LANG_COUNT; ++i) {
+		if (streq(language_names[i].name, str))
+			return language_names[i].lang;
+	}
+	return LANG_NONE;
+}
+
 // NOTE: returns the color setting, not the color
 ColorSetting syntax_char_type_to_color(SyntaxCharType t) {
 	switch (t) {
@@ -210,7 +219,7 @@ static void syntax_highlight_c(SyntaxStateC *state, char32_t *line, u32 line_len
 // 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_TEXT:
+	case LANG_NONE:
 		memset(char_types, 0, line_len * sizeof *char_types);
 		break;
 	case LANG_C:
diff --git a/ted.cfg b/ted.cfg
index 7e2877f..f76c6bd 100644
--- a/ted.cfg
+++ b/ted.cfg
@@ -137,3 +137,6 @@ string = #f77
 character = #a7f
 comment = #999
 constant = #8ff
+
+[extensions]
+C = .c, .h
diff --git a/ted.h b/ted.h
index 2fcc541..59247fb 100644
--- a/ted.h
+++ b/ted.h
@@ -16,11 +16,21 @@ typedef union {
 } SyntaxState;
 
 ENUM_U16 {
-	LANG_TEXT,
+	LANG_NONE,
 	LANG_C,
 	LANG_COUNT
 } ENUM_U16_END(Language);
 
+typedef struct {
+	Language lang;
+	char const *name;
+} LanguageName;
+
+static LanguageName const language_names[] = {
+	{LANG_NONE, "None"},
+	{LANG_C, "C"},
+};
+
 ENUM_U8 {
 	SYNTAX_NORMAL,
 	SYNTAX_KEYWORD,
@@ -45,6 +55,8 @@ typedef struct {
 	u8 border_thickness;
 	u8 padding;
 	u8 scrolloff;
+	// [i] = comma-separated string of file extensions for language i, or NULL for none
+	char *language_extensions[LANG_COUNT];
 } Settings;
 
 #define SCANCODE_COUNT 0x120 // SDL scancodes should be less than this value.
@@ -91,7 +103,6 @@ typedef struct {
 	double scroll_x, scroll_y; // number of characters scrolled in the x/y direction
 	BufferPos cursor_pos;
 	BufferPos selection_pos; // if selection is true, the text between selection_pos and cursor_pos is selected.
-	Language language;
 	bool is_line_buffer; // "line buffers" are buffers which can only have one line of text (used for inputs)
 	bool selection;
 	bool store_undo_events; // set to false to disable undo events