highlight TODOs in comments

11 files changed, 76 insertions, 17 deletions

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 994f286..e808041 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -28,7 +28,8 @@ if(MSVC)
 	target_link_libraries(ted ${SDL2_LIB_DIR}/SDL2.lib)
 	target_link_libraries(ted ${CMAKE_SOURCE_DIR}/pcre2-32-static.lib)
 else()
-	set(CMAKE_C_FLAGS "-Wall -Wextra -Wshadow -Wconversion -Wpedantic -pedantic -std=gnu11 -Wno-unused-function -Wno-fixed-enum-extension -Wimplicit-fallthrough -Wno-format-truncation -Wno-unknown-warning-option")
+	# NOTE: -gdwarf-4 is needed for valgrind to work
+	set(CMAKE_C_FLAGS "-Wall -Wextra -Wshadow -Wconversion -Wpedantic -pedantic -std=gnu11 -gdwarf-4 -Wno-unused-function -Wno-fixed-enum-extension -Wimplicit-fallthrough -Wno-format-truncation -Wno-unknown-warning-option")
 	target_link_libraries(ted m SDL2)
 	target_link_libraries(ted ${CMAKE_SOURCE_DIR}/libpcre2-32.a)
 endif()
diff --git a/colors.c b/colors.c
index 1a3f649..7157dae 100644
--- a/colors.c
+++ b/colors.c
@@ -41,6 +41,7 @@ static ColorName color_names[] = {
 	{COLOR_STRING, "string"},
 	{COLOR_CHARACTER, "character"},
 	{COLOR_CONSTANT, "constant"},
+	{COLOR_TODO, "todo"},
 	{COLOR_AUTOCOMPLETE_BG, "autocomplete-bg"},
 	{COLOR_AUTOCOMPLETE_HL, "autocomplete-hl"},
 	{COLOR_AUTOCOMPLETE_BORDER, "autocomplete-border"},
diff --git a/colors.h b/colors.h
index eda97af..d026619 100644
--- a/colors.h
+++ b/colors.h
@@ -55,6 +55,7 @@ typedef enum {
 
 	COLOR_KEYWORD,
 	COLOR_BUILTIN,
+	COLOR_TODO,
 	COLOR_COMMENT,
 	COLOR_PREPROCESSOR,
 	COLOR_STRING,
diff --git a/keywords.h b/keywords.h
index ea05647..fae4a16 100644
--- a/keywords.h
+++ b/keywords.h
@@ -403,3 +403,12 @@ static const KeywordList syntax_all_keywords_css[128] = {
 	['!'] = {syntax_keywords_css_x21, arr_count(syntax_keywords_css_x21)}, ['-'] = {syntax_keywords_css_x2d, arr_count(syntax_keywords_css_x2d)}, ['@'] = {syntax_keywords_css_x40, arr_count(syntax_keywords_css_x40)}, ['a'] = {syntax_keywords_css_a, arr_count(syntax_keywords_css_a)}, ['b'] = {syntax_keywords_css_b, arr_count(syntax_keywords_css_b)}, ['c'] = {syntax_keywords_css_c, arr_count(syntax_keywords_css_c)}, ['d'] = {syntax_keywords_css_d, arr_count(syntax_keywords_css_d)}, ['e'] = {syntax_keywords_css_e, arr_count(syntax_keywords_css_e)}, ['f'] = {syntax_keywords_css_f, arr_count(syntax_keywords_css_f)}, ['g'] = {syntax_keywords_css_g, arr_count(syntax_keywords_css_g)}, ['h'] = {syntax_keywords_css_h, arr_count(syntax_keywords_css_h)}, ['i'] = {syntax_keywords_css_i, arr_count(syntax_keywords_css_i)}, ['j'] = {syntax_keywords_css_j, arr_count(syntax_keywords_css_j)}, ['k'] = {syntax_keywords_css_k, arr_count(syntax_keywords_css_k)}, ['l'] = {syntax_keywords_css_l, arr_count(syntax_keywords_css_l)}, ['m'] = {syntax_keywords_css_m, arr_count(syntax_keywords_css_m)}, ['n'] = {syntax_keywords_css_n, arr_count(syntax_keywords_css_n)}, ['o'] = {syntax_keywords_css_o, arr_count(syntax_keywords_css_o)}, ['p'] = {syntax_keywords_css_p, arr_count(syntax_keywords_css_p)}, ['q'] = {syntax_keywords_css_q, arr_count(syntax_keywords_css_q)}, ['r'] = {syntax_keywords_css_r, arr_count(syntax_keywords_css_r)}, ['s'] = {syntax_keywords_css_s, arr_count(syntax_keywords_css_s)}, ['t'] = {syntax_keywords_css_t, arr_count(syntax_keywords_css_t)}, ['u'] = {syntax_keywords_css_u, arr_count(syntax_keywords_css_u)}, ['v'] = {syntax_keywords_css_v, arr_count(syntax_keywords_css_v)}, ['w'] = {syntax_keywords_css_w, arr_count(syntax_keywords_css_w)}, ['x'] = {syntax_keywords_css_x, arr_count(syntax_keywords_css_x)}, ['y'] = {syntax_keywords_css_y, arr_count(syntax_keywords_css_y)}, ['z'] = {syntax_keywords_css_z, arr_count(syntax_keywords_css_z)}
 };
 
+static const Keyword syntax_keywords_comment_B[1] = {{"BUG", SYNTAX_TODO}};
+static const Keyword syntax_keywords_comment_F[1] = {{"FIXME", SYNTAX_TODO}};
+static const Keyword syntax_keywords_comment_O[1] = {{"OPTIMIZE", SYNTAX_TODO}};
+static const Keyword syntax_keywords_comment_T[2] = {{"TEMP", SYNTAX_TODO},{"TODO", SYNTAX_TODO}};
+static const Keyword syntax_keywords_comment_X[1] = {{"XXX", SYNTAX_TODO}};
+static const KeywordList syntax_all_keywords_comment[128] = {
+	['B'] = {syntax_keywords_comment_B, arr_count(syntax_keywords_comment_B)}, ['F'] = {syntax_keywords_comment_F, arr_count(syntax_keywords_comment_F)}, ['O'] = {syntax_keywords_comment_O, arr_count(syntax_keywords_comment_O)}, ['T'] = {syntax_keywords_comment_T, arr_count(syntax_keywords_comment_T)}, ['X'] = {syntax_keywords_comment_X, arr_count(syntax_keywords_comment_X)}
+};
+
diff --git a/keywords.py b/keywords.py
index 9a74b85..75d02dc 100755
--- a/keywords.py
+++ b/keywords.py
@@ -8,7 +8,8 @@
 import ast
 
 types = [
-	'SYNTAX_KEYWORD', 'SYNTAX_CONSTANT', 'SYNTAX_BUILTIN'
+	'SYNTAX_KEYWORD', 'SYNTAX_CONSTANT', 'SYNTAX_BUILTIN',
+	'SYNTAX_TODO'
 ]
 exec('\n'.join(['{} = {}'.format(type, i) for (i, type) in enumerate(types)]))
 
@@ -599,6 +600,9 @@ keywords_css =  [
 	'xmp', 'svg'
 ]
 
+keywords_comment = [
+	'TODO', 'FIXME', 'XXX', 'BUG', 'TEMP', 'OPTIMIZE',
+]
 
 
 file = open('keywords.h', 'w')
@@ -638,4 +642,5 @@ output_keywords(file, label(constants_config, SYNTAX_CONSTANT), 'config')
 output_keywords(file, label(keywords_glsl, SYNTAX_KEYWORD) + label(constants_glsl, SYNTAX_CONSTANT)
 	+ label(builtins_glsl, SYNTAX_BUILTIN), 'glsl')
 output_keywords(file, label(builtins_css, SYNTAX_BUILTIN) + label(constants_css, SYNTAX_CONSTANT) + label(keywords_css, SYNTAX_KEYWORD), 'css')
+output_keywords(file, label(keywords_comment, SYNTAX_TODO), 'comment')
 file.close()
diff --git a/main.c b/main.c
index 49d9663..eb883a1 100644
--- a/main.c
+++ b/main.c
@@ -1,6 +1,7 @@
 /*
+TODO:
+- select sole completion if all completions' textEdits are identical (e.g. SyntaxCharT___ in ted)
 FUTURE FEATURES:
-- highlight TODO, FIXME, XXX, others(?) in comments
 - autodetect indentation (tabs vs spaces)
 - font setting & support for multiple fonts to cover more characters
 - support for variable-width fonts
@@ -23,12 +24,6 @@ FUTURE FEATURES:
 - with macros we can really test performance of buffer_insert_text_at_pos, etc. (which should ideally be fast)
 - manual.md
 - LSP request timeout
-BUG REPORTS IM TO LAZY TO FILE (RIGHT NOW)
-- rust-analyzer:
-    - bad json can give "Unexpected error: client exited without proper shutdown sequence"
-    - containerName not always given in workspace/symbols
-- texlab:
-    - textDocument/definition gives LocationLink regardless of client capabilities
 */
 
 #include "ted.h"
diff --git a/syntax.c b/syntax.c
index 258ff20..e7ee69d 100644
--- a/syntax.c
+++ b/syntax.c
@@ -109,6 +109,7 @@ ColorSetting syntax_char_type_to_color_setting(SyntaxCharType t) {
 	case SYNTAX_CHARACTER: return COLOR_CHARACTER;
 	case SYNTAX_CONSTANT: return COLOR_CONSTANT;
 	case SYNTAX_BUILTIN: return COLOR_BUILTIN;
+	case SYNTAX_TODO: return COLOR_TODO;
 	}
 	return COLOR_TEXT;
 }
@@ -168,7 +169,7 @@ bool syntax_is_opening_bracket(Language lang, char32_t c) {
 }
 
 // lookup the given string in the keywords table
-static Keyword const *syntax_keyword_lookup(const KeywordList *all_keywords, const char32_t *str, size_t len) {
+static const Keyword *syntax_keyword_lookup(const KeywordList *all_keywords, const char32_t *str, size_t len) {
 	if (!len) return NULL;
 
 	const KeywordList *list = &all_keywords[str[0] % 128];
@@ -184,6 +185,48 @@ static Keyword const *syntax_keyword_lookup(const KeywordList *all_keywords, con
 	return NULL;
 }
 
+// fast function to figure out if something can be in a comment keyword (like _TODO_)
+static bool can_be_comment_keyword(char32_t c) {
+	return c < 128 && c >= 'A' && c <= 'Z';//fuck you ebcdic
+}
+
+// this is used to highlight comments across languages, for stuff like TODOs
+static SyntaxCharType syntax_highlight_comment(const char32_t *line, u32 i, u32 line_len) {
+	assert(i < line_len);
+	if (!can_be_comment_keyword(line[i]))
+		return SYNTAX_COMMENT; // cannot be a keyword
+	
+	u32 max_len = 10; // maximum length of any comment keyword
+	
+	// find end of keyword
+	u32 end = i + 1;
+	while (end < line_len && i + max_len > end && is32_alpha(line[end])) {
+		if (!can_be_comment_keyword(line[end]))
+			return SYNTAX_COMMENT;
+		++end;
+	}
+	if (end < line_len && (line[end] == '_' || is32_digit(line[end])))
+		return SYNTAX_COMMENT;
+	
+	// find start of keyword
+	u32 start = i;
+	while (start > 0 && start + max_len > i && is32_alpha(line[start])) {
+		if (!can_be_comment_keyword(line[start]))
+			return SYNTAX_COMMENT;
+		--start;
+	}
+	if (!is32_alpha(line[start]))
+		++start;
+	if (start > 0 && (line[start - 1] == '_' || is32_digit(line[start - 1])))
+		return SYNTAX_COMMENT;
+	
+	// look it up
+	const Keyword *kwd = syntax_keyword_lookup(syntax_all_keywords_comment, &line[start], end - start);
+	if (kwd)
+		return kwd->type;
+	return SYNTAX_COMMENT;
+}
+
 // does i continue the number literal from i-1
 static bool syntax_number_continues(Language lang, const char32_t *line, u32 line_len, u32 i) {
 	if (line[i] == '.') {
@@ -423,7 +466,7 @@ static void syntax_highlight_c_cpp(SyntaxState *state_ptr, const char32_t *line,
 		if (char_types && !dealt_with) {
 			SyntaxCharType type = SYNTAX_NORMAL;
 			if (in_single_line_comment || in_multi_line_comment)
-				type = SYNTAX_COMMENT;
+				type = syntax_highlight_comment(line, i, line_len);
 			else if (in_string)
 				type = SYNTAX_STRING;
 			else if (in_char)
@@ -648,7 +691,7 @@ static void syntax_highlight_rust(SyntaxState *state, const char32_t *line, u32
 		if (char_types && !dealt_with) {
 			SyntaxCharType type = SYNTAX_NORMAL;
 			if (comment_depth) {
-				type = SYNTAX_COMMENT;
+				type = syntax_highlight_comment(line, i, line_len);
 			} else if (in_string) {
 				type = SYNTAX_STRING;
 			} else if (in_number) {
@@ -1451,7 +1494,7 @@ static void syntax_highlight_javascript_like(
 		if (char_types && !dealt_with) {
 			SyntaxCharType type = SYNTAX_NORMAL;
 			if (in_multiline_comment)
-				type = SYNTAX_COMMENT;
+				type = syntax_highlight_comment(line, i, line_len);
 			else if (in_string)
 				type = SYNTAX_STRING;
 			else if (in_number)
@@ -1574,7 +1617,7 @@ static void syntax_highlight_java(SyntaxState *state_ptr, const char32_t *line,
 		if (char_types && !dealt_with) {
 			SyntaxCharType type = SYNTAX_NORMAL;
 			if (in_multiline_comment)
-				type = SYNTAX_COMMENT;
+				type = syntax_highlight_comment(line, i, line_len);
 			else if (in_string)
 				type = SYNTAX_STRING;
 			else if (in_char)
@@ -1714,7 +1757,7 @@ static void syntax_highlight_go(SyntaxState *state_ptr, const char32_t *line, u3
 		if (char_types && !dealt_with) {
 			SyntaxCharType type = SYNTAX_NORMAL;
 			if (in_multiline_comment)
-				type = SYNTAX_COMMENT;
+				type = syntax_highlight_comment(line, i, line_len);
 			else if (in_string)
 				type = SYNTAX_STRING;
 			else if (in_char)
@@ -1935,7 +1978,7 @@ static void syntax_highlight_css(SyntaxState *state_ptr, const char32_t *line, u
 		if (char_types && !dealt_with && i < line_len) {
 			SyntaxCharType type = SYNTAX_NORMAL;
 			if (in_comment)
-				type = SYNTAX_COMMENT;
+				type = syntax_highlight_comment(line, i, line_len);
 			char_types[i] = type;
 		}
 	}
diff --git a/ted.h b/ted.h
index 5b79599..abfe991 100644
--- a/ted.h
+++ b/ted.h
@@ -101,7 +101,7 @@ typedef u32 SyntaxState;
 
 /// types of syntax highlighting
 enum SyntaxCharType {
-	// do not change these numbers as it will break backwards compatibility
+	// do not change these numbers as it will break backwards compatibility with plugins
 	SYNTAX_NORMAL = 0,
 	SYNTAX_KEYWORD = 1,
 	SYNTAX_BUILTIN = 2,
@@ -110,6 +110,7 @@ enum SyntaxCharType {
 	SYNTAX_STRING = 5,
 	SYNTAX_CHARACTER = 6,
 	SYNTAX_CONSTANT = 7,
+	SYNTAX_TODO = 8,
 };
 /// Type of syntax highlighting.
 typedef u8 SyntaxCharType;
diff --git a/themes/classic-light.ted.cfg b/themes/classic-light.ted.cfg
index c154020..a7458ab 100644
--- a/themes/classic-light.ted.cfg
+++ b/themes/classic-light.ted.cfg
@@ -42,6 +42,7 @@ keyword = #062
 preprocessor = #11a
 string = #b33
 character = #960
+todo = #a00
 builtin = #70a
 comment = #555
 constant = #077
diff --git a/themes/classic.ted.cfg b/themes/classic.ted.cfg
index b7a9025..0202a81 100644
--- a/themes/classic.ted.cfg
+++ b/themes/classic.ted.cfg
@@ -67,6 +67,7 @@ character = #fa7
 builtin = #a7f
 comment = #999
 constant = #8ff
+todo = #f44
 
 line-numbers = #779
 cursor-line-number = #ddf
diff --git a/themes/extradark.ted.cfg b/themes/extradark.ted.cfg
index 0b3603e..808d975 100644
--- a/themes/extradark.ted.cfg
+++ b/themes/extradark.ted.cfg
@@ -43,6 +43,7 @@ string = #f40
 character = #fa0
 builtin = #c0f
 comment = #999
+todo = #f00
 constant = #0ff
 line-numbers = #8a8
 cursor-line-number = #dfd