multiline string highlighting for ted.cfg

4 files changed, 64 insertions, 24 deletions

diff --git a/buffer.c b/buffer.c
index 4a54ff8..6e7ba7e 100644
--- a/buffer.c
+++ b/buffer.c
@@ -253,8 +253,12 @@ Language buffer_language(TextBuffer *buffer) {
 		return LANG_NONE;
 	size_t filename_len = strlen(filename);
 
+	int match_score = 0;
+	Language match = LANG_NONE;
+	
 	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;
@@ -262,14 +266,18 @@ Language buffer_language(TextBuffer *buffer) {
 				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;
+					int score = (int)len;
+					if (score > match_score) {
+						// found a better match!
+						match_score = score;
+						match = l;
+					}
 				}
 			}
 		}
 	}
-	// no extensions matched
-	return LANG_NONE;
+	
+	return match;
 }
 
 // score is higher if context is closer match.
diff --git a/syntax.c b/syntax.c
index c84a77d..6912671 100644
--- a/syntax.c
+++ b/syntax.c
@@ -25,6 +25,7 @@ char const *language_comment_start(Language l) {
 	case LANG_GO:
 		return "// ";
 	case LANG_CONFIG:
+	case LANG_TED_CFG:
 	case LANG_PYTHON:
 		return "# ";
 	case LANG_TEX:
@@ -1079,50 +1080,60 @@ static void syntax_highlight_html(SyntaxState *state, char32_t const *line, u32
 	*state = (comment * SYNTAX_STATE_HTML_COMMENT);
 }
 
-static void syntax_highlight_config(SyntaxState *state, char32_t const *line, u32 line_len, SyntaxCharType *char_types) {
-	(void)state;
+static void syntax_highlight_config(SyntaxState *state, char32_t const *line, u32 line_len, SyntaxCharType *char_types, bool is_ted_cfg) {
+	bool string = (*state & SYNTAX_STATE_TED_CFG_STRING) != 0;
+	
 	if (line_len == 0) return;
-	if (!char_types) return; // there's no state for config files.
 	
-	if (line[0] == '#') {
-		memset(char_types, SYNTAX_COMMENT, line_len);
+	if (!string && line[0] == '#') {
+		if (char_types) memset(char_types, SYNTAX_COMMENT, line_len);
 		return;
 	}
-	if (line[0] == '[' && line[line_len - 1] == ']') {
-		memset(char_types, SYNTAX_BUILTIN, line_len);
+	if (!string && line[0] == '[' && line[line_len - 1] == ']') {
+		if (char_types) memset(char_types, SYNTAX_BUILTIN, line_len);
 		return;
 	}
-	bool string = false;
+	
+	int backslashes = 0;
+	
 	for (u32 i = 0; i < line_len; ++i) {
-		char_types[i] = string ? SYNTAX_STRING : SYNTAX_NORMAL;
+		if (char_types)
+			char_types[i] = string ? SYNTAX_STRING : SYNTAX_NORMAL;
 		switch (line[i]) {
 		case '"':
-			string = !string;
-			if (string)
+			if (string && backslashes % 2 == 0) {
+				string = false;
+			} else {
+				string = true;
+			}
+			if (char_types)
 				char_types[i] = SYNTAX_STRING;
 			break;
 		case '#':
 			// don't try highlighting the rest of the line.
 			// for ted.cfg, this could be a color, but for other cfg files,
 			// it might be a comment
-			memset(&char_types[i], 0, line_len - i);
+			if (char_types)
+				memset(&char_types[i], 0, line_len - i);
 			i = line_len;
 			break;
 		case ANY_DIGIT:
-			if (i > 0 && !string) {
+			if (char_types && i > 0 && !string) {
 				if (is32_ident(line[i-1]) // something like e5
 					|| line[i-1] == '+') // key combinations, e.g. Alt+0
 					break;
-				while (i < line_len && syntax_number_continues(LANG_CONFIG, line, line_len, i))
+				while (i < line_len && syntax_number_continues(LANG_CONFIG, line, line_len, i)) {
 					char_types[i++] = SYNTAX_CONSTANT;
+				}
 			}
 			break;
 		default: {
+			if (!char_types)
+				break; // don't care
 			if (i == 0) // none of the keywords in syntax_all_keywords_config should appear at the start of the line
 				break;
 			if (is32_ident(line[i-1]) || line[i-1] == '-' || !is32_ident(line[i]))
 				break; // can't be a keyword on its own.
-			
 			u32 keyword_len = syntax_keyword_len(LANG_CONFIG, line, i, line_len);
 			Keyword const *keyword = syntax_keyword_lookup(syntax_all_keywords_config, arr_count(syntax_all_keywords_config),
 				&line[i], keyword_len);
@@ -1136,6 +1147,16 @@ static void syntax_highlight_config(SyntaxState *state, char32_t const *line, u3
 			}
 		} break;
 		}
+		if (i < line_len) {
+			if (line[i] == '\\')
+				++backslashes;
+			else
+				backslashes = 0;
+		}
+	}
+	
+	if (is_ted_cfg) {
+		*state = SYNTAX_STATE_TED_CFG_STRING * string;
 	}
 }
 
@@ -1587,7 +1608,10 @@ void syntax_highlight(SyntaxState *state, Language lang, char32_t const *line, u
 		syntax_highlight_html(state, line, line_len, char_types);
 		break;
 	case LANG_CONFIG:
-		syntax_highlight_config(state, line, line_len, char_types);
+		syntax_highlight_config(state, line, line_len, char_types, false);
+		break;
+	case LANG_TED_CFG:
+		syntax_highlight_config(state, line, line_len, char_types, true);
 		break;
 	case LANG_JAVASCRIPT:
 		syntax_highlight_javascript(state, line, line_len, char_types);
diff --git a/ted.cfg b/ted.cfg
index eef8098..c7a08a7 100644
--- a/ted.cfg
+++ b/ted.cfg
@@ -235,6 +235,8 @@ Tex = .tex
 Markdown = .md 
 HTML = .html, .php, .xml, .xhtml, .iml
 Config = .cfg, .toml
+# ted.cfg has its own syntax highlighting for multiline strings.
+TedCfg = ted.cfg
 Javascript = .js
 Java = .java
 Go = .go
diff --git a/ted.h b/ted.h
index 8681ab6..5eebced 100644
--- a/ted.h
+++ b/ted.h
@@ -55,11 +55,15 @@ enum {
 	SYNTAX_STATE_GO_MULTILINE_COMMENT = 0x02u
 };
 
+enum {
+	SYNTAX_STATE_TED_CFG_STRING = 0x01u,
+};
+
 typedef u8 SyntaxState;
 
 // If you are adding new languages, DO NOT change the constant values
-// of the previous languages. It will mess up config files!
-ENUM_U16 {
+// of the previous languages. It will mess up config files which use :set-language!
+typedef enum {
 	LANG_NONE = 0,
 	LANG_C = 1,
 	LANG_CPP = 2,
@@ -68,12 +72,13 @@ ENUM_U16 {
 	LANG_TEX = 5,
 	LANG_MARKDOWN = 6,
 	LANG_HTML = 7,
-	LANG_CONFIG = 8, // .cfg files, e.g. ted.cfg
+	LANG_CONFIG = 8, // .cfg files
 	LANG_JAVASCRIPT = 9,
 	LANG_JAVA = 10,
 	LANG_GO = 11,
+	LANG_TED_CFG = 12, // like LANG_CONFIG, but with multiline strings.
 	LANG_COUNT
-} ENUM_U16_END(Language);
+} Language;
 
 typedef struct {
 	Language lang;
@@ -93,6 +98,7 @@ static LanguageName const language_names[] = {
 	{LANG_JAVASCRIPT, "Javascript"},
 	{LANG_JAVA, "Java"},
 	{LANG_GO, "Go"},
+	{LANG_TED_CFG, "TedCfg"},
 };
 
 static_assert_if_possible(arr_count(language_names) == LANG_COUNT)