more syntax highlighting

4 files changed, 73 insertions, 7 deletions

diff --git a/buffer.c b/buffer.c
index 6f1b607..41dc216 100644
--- a/buffer.c
+++ b/buffer.c
@@ -1974,6 +1974,10 @@ void buffer_render(TextBuffer *buffer, Rect r) {
 	SyntaxState syntax_state = {0};
 	// 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, LANG_C, line->str, line->len, NULL);
+	}
 
 	for (u32 line_idx = start_line; line_idx < nlines; ++line_idx) {
 		Line *line = &lines[line_idx];
diff --git a/string32.c b/string32.c
index 381bf15..03f96c4 100644
--- a/string32.c
+++ b/string32.c
@@ -123,3 +123,7 @@ bool is32_alnum(char32_t c) {
 	return c <= WINT_MAX && iswalnum((wint_t)c);
 }
 
+// could this character appear in a C-style identifier?
+bool is32_ident(char32_t c) {
+	return c <= WINT_MAX && (iswalnum((wint_t)c) || c == '_');
+}
diff --git a/syntax.c b/syntax.c
index 50e87d9..9a8ba75 100644
--- a/syntax.c
+++ b/syntax.c
@@ -52,6 +52,8 @@ static void syntax_highlight_c(SyntaxStateC *state, char32_t *line, u32 line_len
 		// are there 1/2 characters left in the line?
 		bool has_1_char = i + 1 < line_len;
 
+		bool dealt_with = false;
+
 		switch (line[i]) {
 		case '#':
 			if (!in_single_line_comment && !in_multi_line_comment)
@@ -87,17 +89,72 @@ static void syntax_highlight_c(SyntaxStateC *state, char32_t *line, u32 line_len
 			if (in_preprocessor && in_string)
 				in_string = false;
 			break;
+		default: {
+			// split keywords by starting letter to speed this up
+			static char const *const all_keywords[][10] = {
+				['a'] = {"auto"},
+				['b'] = {"break", "bool"},
+				['c'] = {"case", "char", "const", "continue", "char8_t", "char16_t", "char32_t"},
+				['d'] = {"default", "do", "double"},
+				['e'] = {"else", "enum", "extern"},
+				['f'] = {"float", "for"},
+				['g'] = {"goto"},
+				['i'] = {"if", "inline", "int", "int8_t", "int16_t", "int32_t", "int64_t"},
+				['l'] = {"long"},
+				['r'] = {"register", "restrict", "return"},
+				['s'] = {"short", "signed", "sizeof", "static", "struct", "switch"},
+				['t'] = {"typedef"},
+				['u'] = {"union", "unsigned", "uint8_t", "uint16_t", "uint32_t", "uint64_t"},
+				['v'] = {"void", "volatile"},
+				['w'] = {"while", "wchar_t", "wint_t"},
+				['_'] = {"_Alignas", "_Alignof", "_Atomic", "_Bool", "_Complex", "_Generic",
+					"_Imaginary", "_Noreturn", "_Static_assert", "_Thread_local"},
+			};
+
+			char const *const *keywords = line[i] < arr_count(all_keywords) ? all_keywords[line[i]] : NULL;
+			if (char_types && keywords && !in_single_line_comment && !in_multi_line_comment && !in_string && !in_preprocessor) {
+				// keywords don't matter for advancing the state
+				for (size_t k = 0; keywords[k]; ++k) {
+					bool matches = true;
+					char const *keyword = keywords[k];
+					size_t keyword_len = strlen(keyword);
+					if (i + keyword_len <= line_len) {
+						// make sure we don't catch "print" as containing the keyword "int"
+						bool separated = (i == 0 || !is32_ident(line[i-1])) && (i + keyword_len == line_len || !is32_ident(line[i + keyword_len]));
+						if (separated) {
+							char32_t *p = &line[i];
+							// check if `p` starts with `keyword`
+							for (char const *q = keyword; *q; ++p, ++q) {
+								if (*p != (char32_t)*q) {
+									matches = false;
+									break;
+								}
+							}
+							if (matches) {
+								for (size_t c = 0; keyword[c]; ++c) {
+									char_types[i++] = SYNTAX_KEYWORD;
+								}
+								--i; // we'll increment i from the for loop
+								dealt_with = true;
+								break;
+							}
+						}
+					}
+				}
+
+			}
+		} 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 && !dealt_with) {
+			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;
 		}
 	}
diff --git a/ted.cfg b/ted.cfg
index 79c9e16..6fe6a8c 100644
--- a/ted.cfg
+++ b/ted.cfg
@@ -126,6 +126,7 @@ yes = #afa
 no = #faa
 cancel = #ffa
 
+keyword = #0c0
 preprocessor = #77f
 string = #f77
 character = #f7f