Add custom semicolon insertion to odin/tokenizer and odin/parser

3 files changed, 157 insertions, 26 deletions

diff --git a/core/odin/parser/parser.odin b/core/odin/parser/parser.odin
index 4d94ed479..583a4cc03 100644
--- a/core/odin/parser/parser.odin
+++ b/core/odin/parser/parser.odin
@@ -190,6 +190,50 @@ peek_token_kind :: proc(p: ^Parser, kind: tokenizer.Token_Kind, lookahead := 0)
 	return;
 }
 
+peek_token :: proc(p: ^Parser, lookahead := 0) -> (tok: tokenizer.Token) {
+	prev_parser := p^;
+	defer p^ = prev_parser;
+
+	p.tok.err = nil;
+	for i := 0; i <= lookahead; i += 1 {
+		advance_token(p);
+	}
+	tok = p.curr_tok;
+	return;
+}
+skip_possible_newline :: proc(p: ^Parser) -> bool {
+	if .Insert_Semicolon not_in p.tok.flags {
+		return false;
+	}
+
+	prev := p.curr_tok;
+	if tokenizer.is_newline(prev) {
+		advance_token(p);
+		return true;
+	}
+	return false;
+}
+
+skip_possible_newline_for_literal :: proc(p: ^Parser) -> bool {
+	if .Insert_Semicolon not_in p.tok.flags {
+		return false;
+	}
+
+	curr_pos := p.curr_tok.pos;
+	if tokenizer.is_newline(p.curr_tok) {
+		next := peek_token(p);
+		if curr_pos.line+1 >= next.pos.line {
+			#partial switch next.kind {
+			case .Open_Brace, .Else, .Where:
+				advance_token(p);
+				return true;
+			}
+		}
+	}
+
+	return false;
+}
+
 
 next_token0 :: proc(p: ^Parser) -> bool {
 	p.curr_tok = tokenizer.scan(&p.tok);
@@ -280,7 +324,7 @@ expect_token :: proc(p: ^Parser, kind: tokenizer.Token_Kind) -> tokenizer.Token
 	prev := p.curr_tok;
 	if prev.kind != kind {
 		e := tokenizer.to_string(kind);
-		g := tokenizer.to_string(prev.kind);
+		g := tokenizer.token_to_string(prev);
 		error(p, prev.pos, "expected '%s', got '%s'", e, g);
 	}
 	advance_token(p);
@@ -291,7 +335,7 @@ expect_token_after :: proc(p: ^Parser, kind: tokenizer.Token_Kind, msg: string)
 	prev := p.curr_tok;
 	if prev.kind != kind {
 		e := tokenizer.to_string(kind);
-		g := tokenizer.to_string(prev.kind);
+		g := tokenizer.token_to_string(prev);
 		error(p, prev.pos, "expected '%s' after %s, got '%s'", e, msg, g);
 	}
 	advance_token(p);
@@ -303,7 +347,7 @@ expect_operator :: proc(p: ^Parser) -> tokenizer.Token {
 	if prev.kind == .If || prev.kind == .When {
 		// okay
 	} else if !tokenizer.is_operator(prev.kind) {
-		g := tokenizer.to_string(prev.kind);
+		g := tokenizer.token_to_string(prev);
 		error(p, prev.pos, "expected an operator, got '%s'", g);
 	}
 	advance_token(p);
@@ -400,7 +444,16 @@ expect_semicolon :: proc(p: ^Parser, node: ^ast.Node) -> bool {
 	}
 
 	if node != nil {
-		if prev.pos.line != p.curr_tok.pos.line {
+		if .Insert_Semicolon in p.tok.flags  {
+			#partial switch p.curr_tok.kind {
+			case .Close_Brace, .Close_Paren, .Else, .EOF:
+				return true;
+			}
+
+			if is_semicolon_optional_for_node(p, node) {
+				return true;
+			}
+		} else if prev.pos.line != p.curr_tok.pos.line {
 			if is_semicolon_optional_for_node(p, node) {
 				return true;
 			}
@@ -420,7 +473,7 @@ expect_semicolon :: proc(p: ^Parser, node: ^ast.Node) -> bool {
 		}
 	}
 
-	error(p, prev.pos, "expected ';', got %s", tokenizer.to_string(p.curr_tok.kind));
+	error(p, prev.pos, "expected ';', got %s", tokenizer.token_to_string(p.curr_tok));
 	return false;
 }
 
@@ -493,6 +546,7 @@ parse_when_stmt :: proc(p: ^Parser) -> ^ast.When_Stmt {
 		body = convert_stmt_to_body(p, parse_stmt(p));
 	} else {
 		body = parse_block_stmt(p, true);
+		skip_possible_newline_for_literal(p);
 	}
 
 	if allow_token(p, .Else) {
@@ -568,6 +622,7 @@ parse_if_stmt :: proc(p: ^Parser) -> ^ast.If_Stmt {
 		body = convert_stmt_to_body(p, parse_stmt(p));
 	} else {
 		body = parse_block_stmt(p, false);
+		skip_possible_newline_for_literal(p);
 	}
 
 	if allow_token(p, .Else) {
@@ -629,6 +684,7 @@ parse_for_stmt :: proc(p: ^Parser) -> ^ast.Stmt {
 				body = convert_stmt_to_body(p, parse_stmt(p));
 			} else {
 				body = parse_body(p);
+				skip_possible_newline_for_literal(p);
 			}
 
 			range_stmt := ast.new(ast.Range_Stmt, tok.pos, body.end);
@@ -663,6 +719,7 @@ parse_for_stmt :: proc(p: ^Parser) -> ^ast.Stmt {
 		body = convert_stmt_to_body(p, parse_stmt(p));
 	} else {
 		body = parse_body(p);
+		skip_possible_newline_for_literal(p);
 	}
 
 
@@ -840,6 +897,8 @@ parse_attribute :: proc(p: ^Parser, tok: tokenizer.Token, open_kind, close_kind:
 	attribute.elems = elems[:];
 	attribute.close = close.pos;
 
+	skip_possible_newline(p);
+
 	decl := parse_stmt(p);
 	switch d in &decl.derived {
 	case ast.Value_Decl:
@@ -1028,10 +1087,11 @@ parse_stmt :: proc(p: ^Parser) -> ^ast.Stmt {
 	    		body = convert_stmt_to_body(p, parse_stmt(p));
 	    	} else {
 	    		body = parse_block_stmt(p, false);
+			skip_possible_newline_for_literal(p);
 	    	}
 
 	    	if bad_stmt {
-				return ast.new(ast.Bad_Stmt, inline_tok.pos, end_pos(p.prev_tok));
+			return ast.new(ast.Bad_Stmt, inline_tok.pos, end_pos(p.prev_tok));
 	    	}
 
 	    	range_stmt := ast.new(ast.Inline_Range_Stmt, inline_tok.pos, body.end);
@@ -1206,7 +1266,7 @@ parse_stmt :: proc(p: ^Parser) -> ^ast.Stmt {
 	}
 
 	tok := advance_token(p);
-	error(p, tok.pos, "expected a statement, got %s", tokenizer.to_string(tok.kind));
+	error(p, tok.pos, "expected a statement, got %s", tokenizer.token_to_string(tok));
 	s := ast.new(ast.Bad_Stmt, tok.pos, end_pos(tok));
 	return s;
 }
@@ -2158,7 +2218,10 @@ parse_operand :: proc(p: ^Parser, lhs: bool) -> ^ast.Expr {
 
 		where_token: tokenizer.Token;
 		where_clauses: []^ast.Expr;
-		if (p.curr_tok.kind == .Where) {
+
+		skip_possible_newline_for_literal(p);
+
+		if p.curr_tok.kind == .Where {
 			where_token = expect_token(p, .Where);
 			prev_level := p.expr_level;
 			p.expr_level = -1;
@@ -2334,7 +2397,10 @@ parse_operand :: proc(p: ^Parser, lhs: bool) -> ^ast.Expr {
 
 		where_token: tokenizer.Token;
 		where_clauses: []^ast.Expr;
-		if (p.curr_tok.kind == .Where) {
+
+		skip_possible_newline_for_literal(p);
+
+		if p.curr_tok.kind == .Where {
 			where_token = expect_token(p, .Where);
 			where_prev_level := p.expr_level;
 			p.expr_level = -1;
@@ -2397,7 +2463,10 @@ parse_operand :: proc(p: ^Parser, lhs: bool) -> ^ast.Expr {
 
 		where_token: tokenizer.Token;
 		where_clauses: []^ast.Expr;
-		if (p.curr_tok.kind == .Where) {
+
+		skip_possible_newline_for_literal(p);
+
+		if p.curr_tok.kind == .Where {
 			where_token = expect_token(p, .Where);
 			where_prev_level := p.expr_level;
 			p.expr_level = -1;
@@ -2730,7 +2799,7 @@ parse_atom_expr :: proc(p: ^Parser, value: ^ast.Expr, lhs: bool) -> (operand: ^a
 			case .Colon:
 				interval = advance_token(p);
 				is_slice_op = true;
-				if (p.curr_tok.kind != .Close_Bracket && p.curr_tok.kind != .EOF) {
+				if p.curr_tok.kind != .Close_Bracket && p.curr_tok.kind != .EOF {
 					indicies[1] = parse_expr(p, false);
 				}
 			}
diff --git a/core/odin/tokenizer/token.odin b/core/odin/tokenizer/token.odin
index 54110cf02..997b4967d 100644
--- a/core/odin/tokenizer/token.odin
+++ b/core/odin/tokenizer/token.odin
@@ -283,6 +283,19 @@ tokens := [Token_Kind.COUNT]string {
 
 custom_keyword_tokens: []string;
 
+
+is_newline :: proc(tok: Token) -> bool {
+	return tok.kind == .Semicolon && tok.text == "\n";
+}
+
+
+token_to_string :: proc(tok: Token) -> string {
+	if is_newline(tok) {
+		return "newline";
+	}
+	return to_string(tok.kind);
+}
+
 to_string :: proc(kind: Token_Kind) -> string {
 	if Token_Kind.Invalid <= kind && kind < Token_Kind.COUNT {
 		return tokens[kind];
diff --git a/core/odin/tokenizer/tokenizer.odin b/core/odin/tokenizer/tokenizer.odin
index 132b63572..3df65e49b 100644
--- a/core/odin/tokenizer/tokenizer.odin
+++ b/core/odin/tokenizer/tokenizer.odin
@@ -1,22 +1,31 @@
 package odin_tokenizer
 
 import "core:fmt"
+import "core:unicode"
 import "core:unicode/utf8"
 
 Error_Handler :: #type proc(pos: Pos, fmt: string, args: ..any);
 
+Flag :: enum {
+	Insert_Semicolon,
+}
+Flags :: distinct bit_set[Flag; u32];
+
 Tokenizer :: struct {
 	// Immutable data
 	path: string,
 	src:  []byte,
 	err:  Error_Handler,
 
+	flags: Flags,
+
 	// Tokenizing state
 	ch:          rune,
 	offset:      int,
 	read_offset: int,
 	line_offset: int,
 	line_count:  int,
+	insert_semicolon: bool,
 
 	// Mutable data
 	error_count: int,
@@ -105,11 +114,18 @@ peek_byte :: proc(t: ^Tokenizer, offset := 0) -> byte {
 }
 
 skip_whitespace :: proc(t: ^Tokenizer) {
-	for t.ch == ' '  ||
-	    t.ch == '\t' ||
-	    t.ch == '\n' ||
-	    t.ch == '\r' {
-		advance_rune(t);
+	for {
+		switch t.ch {
+		case ' ', '\t', '\r':
+			advance_rune(t);
+		case '\n':
+			if t.insert_semicolon {
+				return;
+			}
+			advance_rune(t);
+		case:
+			return;
+		}
 	}
 }
 
@@ -122,12 +138,13 @@ is_letter :: proc(r: rune) -> bool {
 			return true;
 		}
 	}
-	// TODO(bill): Add unicode lookup tables
-	return false;
+	return unicode.is_letter(r);
 }
 is_digit :: proc(r: rune) -> bool {
-	// TODO(bill): Add unicode lookup tables
-	return '0' <= r && r <= '9';
+	if '0' <= r && r <= '9' {
+		return true;
+	}
+	return unicode.is_digit(r);
 }
 
 
@@ -491,6 +508,8 @@ scan :: proc(t: ^Tokenizer) -> Token {
 	lit:  string;
 	pos := offset_to_pos(t, offset);
 
+	insert_semicolon := false;
+
 	switch ch := t.ch; true {
 	case is_letter(ch):
 		lit = scan_identifier(t);
@@ -509,24 +528,39 @@ scan :: proc(t: ^Tokenizer) -> Token {
 					break check_keyword;
 				}
 			}
-			if kind == .Ident && lit == "notin" {
-				kind = .Not_In;
+
+			#partial switch kind {
+			case .Ident, .Context, .Typeid, .Break, .Continue, .Fallthrough, .Return:
+				insert_semicolon = true;
 			}
 		}
 	case '0' <= ch && ch <= '9':
+		insert_semicolon = true;
 		kind, lit = scan_number(t, false);
 	case:
 		advance_rune(t);
 		switch ch {
 		case -1:
 			kind = .EOF;
+			if t.insert_semicolon {
+				t.insert_semicolon = false;
+				kind = .Semicolon;
+				lit = "\n";
+			}
+		case '\n':
+			t.insert_semicolon = false;
+			kind = .Semicolon;
+			lit = "\n";
 		case '"':
+			insert_semicolon = true;
 			kind = .String;
 			lit = scan_string(t);
 		case '\'':
+			insert_semicolon = true;
 			kind = .Rune;
 			lit = scan_rune(t);
 		case '`':
+			insert_semicolon = true;
 			kind = .String;
 			lit = scan_raw_string(t);
 		case '=':
@@ -540,10 +574,13 @@ scan :: proc(t: ^Tokenizer) -> Token {
 		case '#':
 			kind = .Hash;
 			if t.ch == '!' {
+				insert_semicolon = t.insert_semicolon;
 				kind = .Comment;
 				lit = scan_comment(t);
 			}
-		case '?': kind = .Question;
+		case '?':
+			insert_semicolon = true;
+			kind = .Question;
 		case '@': kind = .At;
 		case '$': kind = .Dollar;
 		case '^': kind = .Pointer;
@@ -562,6 +599,7 @@ scan :: proc(t: ^Tokenizer) -> Token {
 		case '*': kind = switch2(t, .Mul, .Mul_Eq);
 		case '/':
 			if t.ch == '/' || t.ch == '*' {
+				insert_semicolon = t.insert_semicolon;
 				kind = .Comment;
 				lit = scan_comment(t);
 			} else {
@@ -604,11 +642,17 @@ scan :: proc(t: ^Tokenizer) -> Token {
 		case ',': kind = .Comma;
 		case ';': kind = .Semicolon;
 		case '(': kind = .Open_Paren;
-		case ')': kind = .Close_Paren;
+		case ')':
+			insert_semicolon = true;
+			kind = .Close_Paren;
 		case '[': kind = .Open_Bracket;
-		case ']': kind = .Close_Bracket;
+		case ']':
+			insert_semicolon = true;
+			kind = .Close_Bracket;
 		case '{': kind = .Open_Brace;
-		case '}': kind = .Close_Brace;
+		case '}':
+			insert_semicolon = true;
+			kind = .Close_Brace;
 
 		case '\\': kind = .Back_Slash;
 
@@ -616,10 +660,15 @@ scan :: proc(t: ^Tokenizer) -> Token {
 			if ch != utf8.RUNE_BOM {
 				error(t, t.offset, "illegal character '%r': %d", ch, ch);
 			}
+			insert_semicolon = t.insert_semicolon; // preserve insert_semicolon info
 			kind = .Invalid;
 		}
 	}
 
+	if .Insert_Semicolon in t.flags {
+		t.insert_semicolon = insert_semicolon;
+	}
+
 	if lit == "" {
 		lit = string(t.src[offset : t.offset]);
 	}