go syntax highlighting

also :goto-line now clamps numbers < 1 and > nlines instead of rejecting them

8 files changed, 222 insertions, 4 deletions

diff --git a/README.md b/README.md
index da61dd5..bee145e 100644
--- a/README.md
+++ b/README.md
@@ -30,7 +30,7 @@ in other editors.
 - Multiple tabs, each with a different file
 - Split screen
 - Auto-indent
-- Syntax highlighting for C, C++, HTML, Java, JavaScript, LaTeX, Markdown, Python, and Rust.
+- Syntax highlighting for C, C++, Go, HTML, Java, JavaScript, LaTeX, Markdown, Python, and Rust.
 - Find and replace (with regular expressions!)
 - Run build command, go to errors
 - Run any shell command
@@ -146,7 +146,7 @@ Then, open windows\_installer\\ted\\ted.sln, and build.
 <tr><td>1.0r1</td> <td>Windows-specific bugfixes, update to new version of PCRE2</td> <td>2022 Jan 1</td></tr>
 <tr><td>1.0r2</td> <td>Various bugfixes involving closing tabs and windows</td> <td>2022 Mar 26</td></tr>
 <tr><td>1.0r3</td> <td>Better TeX syntax highlighting, move to cursor on backspace/delete</td> <td>2022 Jul 7</td></tr>
-<tr><td>1.1</td> <td>Minor fixes, syntax highlighting for JavaScript and Java</td> <td>2022 Jul 22</td></tr>
+<tr><td>1.1</td> <td>Minor fixes, syntax highlighting for JavaScript, Java, and Go</td> <td>2022 Jul 22</td></tr>
 </table>
 
 ## License
diff --git a/keywords.h b/keywords.h
index 723f8f6..6cf14f8 100644
--- a/keywords.h
+++ b/keywords.h
@@ -170,6 +170,27 @@ static Keyword const *const syntax_all_keywords_javascript[] = {
 	['A'] = syntax_keywords_javascript_A, ['B'] = syntax_keywords_javascript_B, ['D'] = syntax_keywords_javascript_D, ['E'] = syntax_keywords_javascript_E, ['F'] = syntax_keywords_javascript_F, ['G'] = syntax_keywords_javascript_G, ['I'] = syntax_keywords_javascript_I, ['J'] = syntax_keywords_javascript_J, ['M'] = syntax_keywords_javascript_M, ['N'] = syntax_keywords_javascript_N, ['O'] = syntax_keywords_javascript_O, ['P'] = syntax_keywords_javascript_P, ['R'] = syntax_keywords_javascript_R, ['S'] = syntax_keywords_javascript_S, ['T'] = syntax_keywords_javascript_T, ['U'] = syntax_keywords_javascript_U, ['W'] = syntax_keywords_javascript_W, ['a'] = syntax_keywords_javascript_a, ['b'] = syntax_keywords_javascript_b, ['c'] = syntax_keywords_javascript_c, ['d'] = syntax_keywords_javascript_d, ['e'] = syntax_keywords_javascript_e, ['f'] = syntax_keywords_javascript_f, ['g'] = syntax_keywords_javascript_g, ['i'] = syntax_keywords_javascript_i, ['l'] = syntax_keywords_javascript_l, ['n'] = syntax_keywords_javascript_n, ['p'] = syntax_keywords_javascript_p, ['r'] = syntax_keywords_javascript_r, ['s'] = syntax_keywords_javascript_s, ['t'] = syntax_keywords_javascript_t, ['u'] = syntax_keywords_javascript_u, ['v'] = syntax_keywords_javascript_v, ['w'] = syntax_keywords_javascript_w, ['y'] = syntax_keywords_javascript_y
 };
 
+static Keyword const syntax_keywords_go_a[2] = {{"append", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_b[4] = {{"break", SYNTAX_KEYWORD},{"bool", SYNTAX_BUILTIN},{"byte", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_c[12] = {{"case", SYNTAX_KEYWORD},{"chan", SYNTAX_KEYWORD},{"const", SYNTAX_KEYWORD},{"continue", SYNTAX_KEYWORD},{"cap", SYNTAX_BUILTIN},{"close", SYNTAX_BUILTIN},{"comparable", SYNTAX_BUILTIN},{"complex", SYNTAX_BUILTIN},{"complex128", SYNTAX_BUILTIN},{"complex64", SYNTAX_BUILTIN},{"copy", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_d[4] = {{"default", SYNTAX_KEYWORD},{"defer", SYNTAX_KEYWORD},{"delete", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_e[3] = {{"else", SYNTAX_KEYWORD},{"error", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_f[7] = {{"fallthrough", SYNTAX_KEYWORD},{"for", SYNTAX_KEYWORD},{"func", SYNTAX_KEYWORD},{"false", SYNTAX_CONSTANT},{"float32", SYNTAX_BUILTIN},{"float64", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_g[3] = {{"go", SYNTAX_KEYWORD},{"goto", SYNTAX_KEYWORD}};
+static Keyword const syntax_keywords_go_i[11] = {{"if", SYNTAX_KEYWORD},{"import", SYNTAX_KEYWORD},{"interface", SYNTAX_KEYWORD},{"iota", SYNTAX_CONSTANT},{"imag", SYNTAX_BUILTIN},{"int", SYNTAX_BUILTIN},{"int16", SYNTAX_BUILTIN},{"int32", SYNTAX_BUILTIN},{"int64", SYNTAX_BUILTIN},{"int8", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_l[2] = {{"len", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_m[3] = {{"map", SYNTAX_KEYWORD},{"make", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_n[3] = {{"nil", SYNTAX_CONSTANT},{"new", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_p[5] = {{"package", SYNTAX_KEYWORD},{"panic", SYNTAX_BUILTIN},{"print", SYNTAX_BUILTIN},{"println", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_r[6] = {{"range", SYNTAX_KEYWORD},{"return", SYNTAX_KEYWORD},{"real", SYNTAX_BUILTIN},{"recover", SYNTAX_BUILTIN},{"rune", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_s[5] = {{"select", SYNTAX_KEYWORD},{"struct", SYNTAX_KEYWORD},{"switch", SYNTAX_KEYWORD},{"string", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_t[3] = {{"type", SYNTAX_KEYWORD},{"true", SYNTAX_CONSTANT}};
+static Keyword const syntax_keywords_go_u[7] = {{"uint", SYNTAX_BUILTIN},{"uint16", SYNTAX_BUILTIN},{"uint32", SYNTAX_BUILTIN},{"uint64", SYNTAX_BUILTIN},{"uint8", SYNTAX_BUILTIN},{"uintptr", SYNTAX_BUILTIN}};
+static Keyword const syntax_keywords_go_v[2] = {{"var", SYNTAX_KEYWORD}};
+static Keyword const *const syntax_all_keywords_go[] = {
+	['a'] = syntax_keywords_go_a, ['b'] = syntax_keywords_go_b, ['c'] = syntax_keywords_go_c, ['d'] = syntax_keywords_go_d, ['e'] = syntax_keywords_go_e, ['f'] = syntax_keywords_go_f, ['g'] = syntax_keywords_go_g, ['i'] = syntax_keywords_go_i, ['l'] = syntax_keywords_go_l, ['m'] = syntax_keywords_go_m, ['n'] = syntax_keywords_go_n, ['p'] = syntax_keywords_go_p, ['r'] = syntax_keywords_go_r, ['s'] = syntax_keywords_go_s, ['t'] = syntax_keywords_go_t, ['u'] = syntax_keywords_go_u, ['v'] = syntax_keywords_go_v
+};
+
 static Keyword const syntax_keywords_java_a[3] = {{"abstract", SYNTAX_KEYWORD},{"assert", SYNTAX_KEYWORD}};
 static Keyword const syntax_keywords_java_b[4] = {{"boolean", SYNTAX_KEYWORD},{"break", SYNTAX_KEYWORD},{"byte", SYNTAX_KEYWORD}};
 static Keyword const syntax_keywords_java_c[7] = {{"case", SYNTAX_KEYWORD},{"catch", SYNTAX_KEYWORD},{"char", SYNTAX_KEYWORD},{"class", SYNTAX_KEYWORD},{"const", SYNTAX_KEYWORD},{"continue", SYNTAX_KEYWORD}};
diff --git a/keywords.py b/keywords.py
index b7f7432..1e345d9 100755
--- a/keywords.py
+++ b/keywords.py
@@ -300,6 +300,28 @@ constants_java = [
 	'true', 'false', 'null'
 ]
 
+keywords_go = [
+	'break', 'default', 'func', 'interface', 'select',
+	'case', 'defer', 'go', 'map', 'struct',
+	'chan', 'else', 'goto', 'package', 'switch',
+	'const', 'fallthrough', 'if', 'range', 'type',
+	'continue', 'for', 'import', 'return', 'var'
+]
+
+constants_go = [
+	'true', 'false', 'iota', 'nil'
+]
+
+builtins_go = [
+	'new', 'make', 'cap', 'len', 'close', 'append', 'copy', 'delete',
+	'complex', 'real', 'imag', 'panic', 'recover', 'print', 'println',
+	'bool', 'uint8', 'uint16', 'uint32', 'uint64',
+	'int8', 'int16', 'int32', 'int64',
+	'float32', 'float64', 'complex64', 'complex128',
+	'byte', 'rune', 'uint', 'int', 'uintptr', 'string', 'error',
+	'comparable'
+]
+
 file = open('keywords.h', 'w')
 file.write('''// keywords for all languages ted supports
 // This file was auto-generated by keywords.py
@@ -325,6 +347,8 @@ 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_javascript, SYNTAX_KEYWORD) + label(builtins_javascript, SYNTAX_BUILTIN) +
 	label(constants_javascript, SYNTAX_CONSTANT), 'javascript')
+output_keywords(file, label(keywords_go, SYNTAX_KEYWORD) + label(builtins_go, SYNTAX_BUILTIN) +
+	label(constants_go, SYNTAX_CONSTANT), 'go')
 output_keywords(file, label(keywords_java, SYNTAX_KEYWORD) + label(constants_java, SYNTAX_CONSTANT), 'java')
 output_keywords(file, label(keywords_python, SYNTAX_KEYWORD) + label(builtins_python, SYNTAX_BUILTIN), 'python')
 output_keywords(file, label(builtins_html, SYNTAX_BUILTIN), 'html')
diff --git a/menu.c b/menu.c
index 6977552..dfb3db9 100644
--- a/menu.c
+++ b/menu.c
@@ -244,7 +244,9 @@ static void menu_update(Ted *ted) {
 		char *end;
 		long line_number = strtol(contents, &end, 0);
 		TextBuffer *buffer = ted->prev_active_buffer;
-		if (line_number > 0 && *end == '\0' && line_number <= (long)buffer->nlines) {
+		if (*contents != '\0' && *end == '\0') {
+			if (line_number < 1) line_number = 1;
+			if (line_number > buffer->nlines) line_number = (long)buffer->nlines; 
 			BufferPos pos = {(u32)line_number - 1, 0};
 			
 			if (line_buffer->line_buffer_submitted) {
diff --git a/syntax.c b/syntax.c
index 99aae45..e26560d 100644
--- a/syntax.c
+++ b/syntax.c
@@ -1169,7 +1169,7 @@ static void syntax_highlight_javascript(SyntaxState *state, char32_t const *line
 static void syntax_highlight_java(SyntaxState *state_ptr, char32_t const *line, u32 line_len, SyntaxCharType *char_types) {
 	SyntaxState state = *state_ptr;
 	bool in_string = false;
-	bool in_multiline_comment = (state & SYNTAX_STATE_CPP_MULTI_LINE_COMMENT) != 0;
+	bool in_multiline_comment = (state & SYNTAX_STATE_JAVA_MULTILINE_COMMENT) != 0;
 	bool in_char = false;
 	bool in_number = false;
 	
@@ -1291,6 +1291,148 @@ static void syntax_highlight_java(SyntaxState *state_ptr, char32_t const *line,
 	);
 }
 
+
+static void syntax_highlight_go(SyntaxState *state_ptr, char32_t const *line, u32 line_len, SyntaxCharType *char_types) {
+	SyntaxState state = *state_ptr;
+	bool string_is_raw = (state & SYNTAX_STATE_GO_RAW_STRING) != 0;
+	bool in_string = string_is_raw;
+	bool in_multiline_comment = (state & SYNTAX_STATE_GO_MULTILINE_COMMENT) != 0;
+	bool in_char = false;
+	bool in_number = false;
+	
+	int backslashes = 0;
+	for (u32 i = 0; i < line_len; ++i) {
+
+		// are there 1/2 characters left in the line?
+		bool has_1_char =  i + 1 < line_len;
+		
+		bool dealt_with = false;
+		
+		char32_t c = line[i];
+		
+		switch (c) {
+		case '\\':
+			++backslashes;
+			break;
+		case '/':
+			if (!in_multiline_comment && !in_string && !in_char && has_1_char) {
+				if (line[i + 1] == '/') {
+						if (char_types) memset(&char_types[i], SYNTAX_COMMENT, line_len - i);
+						i = line_len - 1;
+						dealt_with = true;
+				} else if (line[i + 1] == '*') {
+					in_multiline_comment = true; // /*
+				}
+			} else if (in_multiline_comment) {
+				if (i > 0 && line[i - 1] == '*') {
+					// */
+					in_multiline_comment = false;
+					if (char_types) {
+						dealt_with = true;
+						char_types[i] = SYNTAX_COMMENT;
+					}
+				}
+			}
+			break;
+		case '"':
+			if (in_string && !string_is_raw && backslashes % 2 == 0) {
+				in_string = false;
+				if (char_types) {
+					dealt_with = true;
+					char_types[i] = SYNTAX_STRING;
+				}
+			} else if (!in_multiline_comment && !in_char) {
+				in_string = true;
+				string_is_raw = false;
+			}
+			break;
+		case '`':
+			if (in_string && string_is_raw) {
+				// end of raw string
+				in_string = false;
+				string_is_raw = false;
+				if (char_types) char_types[i] = SYNTAX_STRING;
+				dealt_with = true;
+			} else if (!in_string && !in_multiline_comment && !in_char) {
+				// start of raw string
+				in_string = true;
+				string_is_raw = true;
+			}
+			break;
+		case '\'':
+			if (in_char && backslashes % 2 == 0) {
+				in_char = false;
+				if (char_types) {
+					dealt_with = true;
+					char_types[i] = SYNTAX_CHARACTER;
+				}
+			} else if (!in_multiline_comment && !in_string) {
+				in_char = true;
+			}
+			break;
+		case ANY_DIGIT:
+			// a number!
+			if (char_types && !in_multiline_comment && !in_string && !in_number && !in_char) {
+				in_number = true;
+				if (i) {
+					if (line[i - 1] == '.') {
+						// support .6, for example
+						char_types[i - 1] = SYNTAX_CONSTANT;
+					} else if (is32_ident(line[i - 1])) {
+						// actually, this isn't a number. it's something like a*6* or u3*2*.
+						in_number = false;
+					}
+				}
+			}
+			break;
+		default: {
+			if ((i && is32_ident(line[i - 1])) || !is32_ident(c))
+				break; // can't be a keyword on its own.
+			
+			// keywords don't matter for advancing the state
+			if (char_types && !in_multiline_comment && !in_number && !in_string && !in_char) {
+				u32 keyword_len = syntax_keyword_len(LANG_GO, line, i, line_len);
+				Keyword const *keyword = syntax_keyword_lookup(syntax_all_keywords_go, arr_count(syntax_all_keywords_go),
+						&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
+					dealt_with = true;
+					break;
+				}
+			}
+		} break;
+		}
+		if (c != '\\') backslashes = 0;
+		if (in_number && !syntax_number_continues(line, line_len, i)) {
+			in_number = false;
+		}
+
+		if (char_types && !dealt_with) {
+			SyntaxCharType type = SYNTAX_NORMAL;
+			if (in_multiline_comment)
+				type = SYNTAX_COMMENT;
+			else if (in_string)
+				type = SYNTAX_STRING;
+			else if (in_char)
+				type = SYNTAX_CHARACTER;
+			else if (in_number)
+				type = SYNTAX_CONSTANT;
+
+			char_types[i] = type;
+		}
+	}
+	*state_ptr = (SyntaxState)(
+		  (in_multiline_comment * SYNTAX_STATE_GO_MULTILINE_COMMENT)
+		| ((in_string && string_is_raw) * SYNTAX_STATE_GO_RAW_STRING)
+	);
+}
+
 // 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.
@@ -1331,6 +1473,9 @@ void syntax_highlight(SyntaxState *state, Language lang, char32_t const *line, u
 	case LANG_JAVA:
 		syntax_highlight_java(state, line, line_len, char_types);
 		break;
+	case LANG_GO:
+		syntax_highlight_go(state, line, line_len, char_types);
+		break;
 	case LANG_COUNT: assert(0); break;
 	}
 }
diff --git a/ted.cfg b/ted.cfg
index 27b33e5..fa37904 100644
--- a/ted.cfg
+++ b/ted.cfg
@@ -229,3 +229,4 @@ HTML = .html, .php, .xml, .xhtml
 Config = .cfg
 Javascript = .js
 Java = .java
+Go = .go
diff --git a/ted.h b/ted.h
index 2400aa0..01ad115 100644
--- a/ted.h
+++ b/ted.h
@@ -50,6 +50,11 @@ enum {
 	SYNTAX_STATE_JAVA_MULTILINE_COMMENT = 0x01u
 };
 
+enum {
+	SYNTAX_STATE_GO_RAW_STRING = 0x01u, // backtick-enclosed string
+	SYNTAX_STATE_GO_MULTILINE_COMMENT = 0x02u
+};
+
 typedef u8 SyntaxState;
 
 // If you are adding new languages, DO NOT change the constant values
@@ -66,6 +71,7 @@ ENUM_U16 {
 	LANG_CONFIG = 8, // .cfg files, e.g. ted.cfg
 	LANG_JAVASCRIPT = 9,
 	LANG_JAVA = 10,
+	LANG_GO = 11,
 	LANG_COUNT
 } ENUM_U16_END(Language);
 
@@ -86,6 +92,7 @@ static LanguageName const language_names[] = {
 	{LANG_CONFIG, "Config"},
 	{LANG_JAVASCRIPT, "Javascript"},
 	{LANG_JAVA, "Java"},
+	{LANG_GO, "Go"},
 };
 
 static_assert_if_possible(arr_count(language_names) == LANG_COUNT)
diff --git a/test.go b/test.go
new file mode 100644
index 0000000..98eea24
--- /dev/null
+++ b/test.go
@@ -0,0 +1,18 @@
+package main
+
+import "fmt"
+
+/*
+what
+a
+wonderful
+day
+*///yes
+
+func main() {
+	var x []int = make([]int, 10)
+	fmt.Println(x != nil)
+	println(`hello
+	world\`)
+	println("yes\"\\")
+}