From f90d98c3c451aa5be2979ff05863b5aa8bee9336 Mon Sep 17 00:00:00 2001
From: Leo Tenenbaum <pommicket@gmail.com>
Date: Tue, 20 Apr 2021 14:18:56 -0400
Subject: cfg syntax highlighting

---
 README.md   |  2 +-
 keywords.h  |  9 +++++++
 keywords.py |  4 +++
 main.c      |  2 --
 syntax.c    | 85 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++-----
 ted.cfg     |  1 +
 ted.h       |  2 ++
 7 files changed, 96 insertions(+), 9 deletions(-)

diff --git a/README.md b/README.md
index 054ce0e..9450b98 100644
--- a/README.md
+++ b/README.md
@@ -23,7 +23,7 @@ in other editors.
 - Multiple tabs, each with a different file
 - Split screen (default: Ctrl+\\, Ctrl+Shift+\\)
 - Auto-indent
-- Syntax highlighting for C, C++, Rust, Python, LaTeX, and Markdown.
+- Syntax highlighting for C, C++, HTML, LaTeX, Markdown, Python, and Rust.
 - Find and replace (with regular expressions!)
 - Run build command (F4), go to errors
 - Run any shell command (Ctrl+!)
diff --git a/keywords.h b/keywords.h
index 5fdb0e5..c880b77 100644
--- a/keywords.h
+++ b/keywords.h
@@ -203,3 +203,12 @@ static Keyword const *const syntax_all_keywords_html[] = {
 	['a'] = syntax_keywords_html_a, ['b'] = syntax_keywords_html_b, ['c'] = syntax_keywords_html_c, ['d'] = syntax_keywords_html_d, ['e'] = syntax_keywords_html_e, ['f'] = syntax_keywords_html_f, ['h'] = syntax_keywords_html_h, ['i'] = syntax_keywords_html_i, ['k'] = syntax_keywords_html_k, ['l'] = syntax_keywords_html_l, ['m'] = syntax_keywords_html_m, ['n'] = syntax_keywords_html_n, ['o'] = syntax_keywords_html_o, ['p'] = syntax_keywords_html_p, ['r'] = syntax_keywords_html_r, ['s'] = syntax_keywords_html_s, ['t'] = syntax_keywords_html_t, ['u'] = syntax_keywords_html_u, ['v'] = syntax_keywords_html_v, ['w'] = syntax_keywords_html_w
 };
 
+static Keyword const syntax_keywords_config_f[2] = {{"false", SYNTAX_CONSTANT}};
+static Keyword const syntax_keywords_config_n[2] = {{"no", SYNTAX_CONSTANT}};
+static Keyword const syntax_keywords_config_o[3] = {{"off", SYNTAX_CONSTANT},{"on", SYNTAX_CONSTANT}};
+static Keyword const syntax_keywords_config_t[2] = {{"true", SYNTAX_CONSTANT}};
+static Keyword const syntax_keywords_config_y[2] = {{"yes", SYNTAX_CONSTANT}};
+static Keyword const *const syntax_all_keywords_config[] = {
+	['f'] = syntax_keywords_config_f, ['n'] = syntax_keywords_config_n, ['o'] = syntax_keywords_config_o, ['t'] = syntax_keywords_config_t, ['y'] = syntax_keywords_config_y
+};
+
diff --git a/keywords.py b/keywords.py
index c8f7e9c..2413dd9 100755
--- a/keywords.py
+++ b/keywords.py
@@ -243,6 +243,9 @@ attributes_html = [
 	'srclang','srcset','start','step','style','tabindex','target',
 	'title','translate','type','usemap','value','width','wrap'
 ]
+constants_config = [
+	'on', 'off', 'yes', 'no', 'true', 'false'
+]
 
 assert len(attributes_html) == len(set(attributes_html))
 
@@ -274,4 +277,5 @@ output_keywords(file, cpp_things, 'cpp')
 output_keywords(file, label(keywords_rust, SYNTAX_KEYWORD) + label(builtins_rust, SYNTAX_BUILTIN) + label(constants_rust, SYNTAX_CONSTANT), 'rust')
 output_keywords(file, label(keywords_python, SYNTAX_KEYWORD) + label(builtins_python, SYNTAX_BUILTIN), 'python')
 output_keywords(file, label(builtins_html, SYNTAX_BUILTIN), 'html')
+output_keywords(file, label(constants_config, SYNTAX_CONSTANT), 'config')
 file.close()
diff --git a/main.c b/main.c
index 044a11b..d140f29 100644
--- a/main.c
+++ b/main.c
@@ -1,5 +1,3 @@
-// Highlighting for:
-// - Config (cfg)
 // :set-language command; -1 = default language for this extension
 
 #include "base.h"
diff --git a/syntax.c b/syntax.c
index 2e361cf..bc8803c 100644
--- a/syntax.c
+++ b/syntax.c
@@ -16,12 +16,18 @@ Language language_from_str(char const *str) {
 // start of single line comment for language l -- used for comment/uncomment selection
 char const *language_comment_start(Language l) {
 	switch (l) {
-	case LANG_C:   return "/* ";
+	case LANG_C:
+		return "/* ";
 	case LANG_RUST:
-	case LANG_CPP: return "// ";
-	case LANG_PYTHON: return "# ";
-	case LANG_TEX: return "% ";
-	case LANG_HTML: return "<!-- ";
+	case LANG_CPP:
+		return "// ";
+	case LANG_CONFIG:
+	case LANG_PYTHON:
+		return "# ";
+	case LANG_TEX:
+		return "% ";
+	case LANG_HTML:
+		return "<!-- ";
 	case LANG_NONE:
 	case LANG_MARKDOWN:
 	case LANG_COUNT:
@@ -697,7 +703,7 @@ static void syntax_highlight_markdown(SyntaxState *state, char32_t const *line,
 	}
 
 	if (multiline_code) {
-		static_assert_if_possible(sizeof *char_types == 1)
+		static_assert_if_possible(sizeof *char_types == 1) // NOTE: memset is used extensively in this file this way
 		memset(char_types, SYNTAX_CODE, line_len);
 		return;
 	}
@@ -947,6 +953,10 @@ static void syntax_highlight_html(SyntaxState *state, char32_t const *line, u32
 				break;
 			default:
 				if (char_types) {
+				
+					if ((i && is32_ident(line[i - 1])) || !is32_ident(line[i]))
+						break; // can't be a keyword on its own.
+					
 					u32 keyword_len = syntax_keyword_len(LANG_HTML, line, i, line_len);
 					Keyword const *keyword = syntax_keyword_lookup(syntax_all_keywords_html, arr_count(syntax_all_keywords_html),
 						&line[i], keyword_len);
@@ -973,6 +983,66 @@ 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;
+	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);
+		return;
+	}
+	if (line[0] == '[' && line[line_len - 1] == ']') {
+		memset(char_types, SYNTAX_BUILTIN, line_len);
+		return;
+	}
+	bool string = false;
+	for (u32 i = 0; i < line_len; ++i) {
+		char_types[i] = string ? SYNTAX_STRING : SYNTAX_NORMAL;
+		switch (line[i]) {
+		case '"':
+			string = !string;
+			if (string)
+				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);
+			i = line_len;
+			break;
+		case ANY_DIGIT:
+			if (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(line, line_len, i))
+					char_types[i++] = SYNTAX_CONSTANT;
+			}
+			break;
+		default: {
+			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]) || !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);
+			if (keyword) {
+				SyntaxCharType type = keyword->type;
+				for (size_t j = 0; j < keyword_len; ++j) {
+					char_types[i++] = type;
+				}
+				--i; // we'll increment i from the for loop
+				break;
+			}
+		} break;
+		}
+	}
+}
+
 // 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.
@@ -1004,6 +1074,9 @@ void syntax_highlight(SyntaxState *state, Language lang, char32_t const *line, u
 	case LANG_HTML:
 		syntax_highlight_html(state, line, line_len, char_types);
 		break;
+	case LANG_CONFIG:
+		syntax_highlight_config(state, line, line_len, char_types);
+		break;
 	case LANG_COUNT: assert(0); break;
 	}
 }
diff --git a/ted.cfg b/ted.cfg
index aa15520..bad356e 100644
--- a/ted.cfg
+++ b/ted.cfg
@@ -210,3 +210,4 @@ Python = .py
 Tex = .tex
 Markdown = .md 
 HTML = .html, .php
+Config = .cfg
diff --git a/ted.h b/ted.h
index 6e61ea9..e08f222 100644
--- a/ted.h
+++ b/ted.h
@@ -52,6 +52,7 @@ ENUM_U16 {
 	LANG_TEX,
 	LANG_MARKDOWN,
 	LANG_HTML,
+	LANG_CONFIG, // .cfg files, e.g. ted.cfg
 	LANG_COUNT
 } ENUM_U16_END(Language);
 
@@ -69,6 +70,7 @@ static LanguageName const language_names[] = {
 	{LANG_TEX, "Tex"},
 	{LANG_MARKDOWN, "Markdown"},
 	{LANG_HTML, "HTML"},
+	{LANG_CONFIG, "Config"},
 };
 
 static_assert_if_possible(arr_count(language_names) == LANG_COUNT)
-- 
cgit v1.2.3