Add experimental -insert-semicolon functionality to tokenizer and parser

core/math/math.odin

@@ -10,7 +10,7 @@ Float_Class :: enum {
 	Neg_Zero,  // the negative zero
 	NaN,       // Not-A-Number (NaN)
 	Inf,       // positive infinity
-	Neg_Inf    // negative infinity
+	Neg_Inf,   // negative infinity
 };
 
 TAU          :: 6.28318530717958647692528676655900576;

core/mem/allocators.odin

@@ -935,7 +935,7 @@ Small_Allocator :: struct(BUFFER_SIZE: int)
 	start:      uintptr,
 	curr:       uintptr,
 	end:        uintptr,
-	chunk_size: int
+	chunk_size: int,
 }
 
 small_allocator :: proc(s: ^$S/Small_Allocator, backing := context.allocator) -> (a: Allocator) {

core/os/stat_windows.odin

@@ -114,10 +114,10 @@ cleanpath_strip_prefix :: proc(buf: []u16) -> []u16 {
 	buf = buf[:N];
 
 	if len(buf) >= 4 {
-		if buf[0] == '\\'
+		if buf[0] == '\\' &&
-		&& buf[1] == '\\'
+		   buf[1] == '\\' &&
-		&& buf[2] == '?'
+		   buf[2] == '?'  &&
-		&& buf[3] == '\\' {
+		   buf[3] == '\\' {
 			buf = buf[4:];
 		}
 	}

core/runtime/core.odin

@@ -64,7 +64,7 @@ Type_Info_Boolean    :: struct {};
 Type_Info_Any        :: struct {};
 Type_Info_Type_Id    :: struct {};
 Type_Info_Pointer :: struct {
-	elem: ^Type_Info // nil -> rawptr
+	elem: ^Type_Info, // nil -> rawptr
 };
 Type_Info_Procedure :: struct {
 	params:     ^Type_Info, // Type_Info_Tuple
@@ -296,7 +296,7 @@ Logger_Option :: enum {
 	Line,
 	Procedure,
 	Terminal_Color,
-	Thread_Id
+	Thread_Id,
 }
 
 Logger_Options :: bit_set[Logger_Option];

core/sys/windows/kernel32.odin

@@ -183,8 +183,7 @@ foreign kernel32 {
 		lpOverlapped: LPOVERLAPPED,
 	) -> BOOL ---
 	CloseHandle :: proc(hObject: HANDLE) -> BOOL ---
-	MoveFileExW :: proc(lpExistingFileName: LPCWSTR, lpNewFileName: LPCWSTR, dwFlags: DWORD)
+	MoveFileExW :: proc(lpExistingFileName: LPCWSTR, lpNewFileName: LPCWSTR, dwFlags: DWORD) -> BOOL ---
-	-> BOOL ---
 	SetFilePointerEx :: proc(
 		hFile: HANDLE,
 		liDistanceToMove: LARGE_INTEGER,

core/sys/windows/types.odin

@@ -118,12 +118,12 @@ SYNCHRONIZE: DWORD : 0x00100000;
 GENERIC_READ: DWORD : 0x80000000;
 GENERIC_WRITE: DWORD : 0x40000000;
 STANDARD_RIGHTS_WRITE: DWORD : READ_CONTROL;
-FILE_GENERIC_WRITE: DWORD : STANDARD_RIGHTS_WRITE
+FILE_GENERIC_WRITE: DWORD : STANDARD_RIGHTS_WRITE |
-	| FILE_WRITE_DATA
+	FILE_WRITE_DATA |
-	| FILE_WRITE_ATTRIBUTES
+	FILE_WRITE_ATTRIBUTES |
-	| FILE_WRITE_EA
+	FILE_WRITE_EA |
-	| FILE_APPEND_DATA
+	FILE_APPEND_DATA |
-	| SYNCHRONIZE;
+	SYNCHRONIZE;
 
 FILE_FLAG_OPEN_REPARSE_POINT: DWORD : 0x00200000;
 FILE_FLAG_BACKUP_SEMANTICS: DWORD : 0x02000000;
@@ -722,7 +722,7 @@ SYSTEM_INFO :: struct {
 
 // https://docs.microsoft.com/en-us/windows-hardware/drivers/ddi/wdm/ns-wdm-_osversioninfoexw
 OSVERSIONINFOEXW :: struct {
-    dwOSVersionInfoSize: ULONG, 
+    dwOSVersionInfoSize: ULONG,
     dwMajorVersion:      ULONG,
     dwMinorVersion:      ULONG,
     dwBuildNumber:       ULONG,
@@ -733,4 +733,4 @@ OSVERSIONINFOEXW :: struct {
     wSuiteMask:          USHORT,
     wProductType:        UCHAR,
     wReserved:           UCHAR,
-};
+};

src/build_settings.cpp

@@ -157,6 +157,7 @@ struct BuildContext {
 	bool   different_os;
 	bool   keep_object_files;
 	bool   disallow_do;
+	bool   insert_semicolon;
 
 	bool   use_llvm_api;
 

src/main.cpp

@@ -587,6 +587,7 @@ enum BuildFlagKind {
 
 	BuildFlag_DisallowDo,
 	BuildFlag_DefaultToNilAllocator,
+	BuildFlag_InsertSemicolon,
 
 	BuildFlag_Compact,
 	BuildFlag_GlobalDefinitions,
@@ -687,6 +688,7 @@ bool parse_build_flags(Array<String> args) {
 
 	add_flag(&build_flags, BuildFlag_DisallowDo,            str_lit("disallow-do"),              BuildFlagParam_None);
 	add_flag(&build_flags, BuildFlag_DefaultToNilAllocator, str_lit("default-to-nil-allocator"), BuildFlagParam_None);
+	add_flag(&build_flags, BuildFlag_InsertSemicolon,       str_lit("insert-semicolon"),         BuildFlagParam_None);
 
 	add_flag(&build_flags, BuildFlag_Compact, str_lit("compact"), BuildFlagParam_None);
 	add_flag(&build_flags, BuildFlag_GlobalDefinitions, str_lit("global-definitions"), BuildFlagParam_None);
@@ -1127,6 +1129,10 @@ bool parse_build_flags(Array<String> args) {
 							build_context.ODIN_DEFAULT_TO_NIL_ALLOCATOR = true;
 							break;
 
+						case BuildFlag_InsertSemicolon:
+							build_context.insert_semicolon = true;
+							break;
+
 						case BuildFlag_Compact:
 							if (!build_context.query_data_set_settings.ok) {
 								gb_printf_err("Invalid use of -compact flag, only allowed with 'odin query'\n");

src/parser.cpp

@@ -1247,11 +1247,69 @@ bool peek_token_kind(AstFile *f, TokenKind kind) {
 	return false;
 }
 
+Token peek_token(AstFile *f) {
+	for (isize i = f->curr_token_index+1; i < f->tokens.count; i++) {
+		Token tok = f->tokens[i];
+		if (tok.kind == Token_Comment) {
+			continue;
+		}
+		return tok;
+	}
+	return {};
+}
+
+
+bool token_is_newline(Token const &tok) {
+	return tok.kind == Token_Semicolon && tok.string == "\n";
+}
+
+bool skip_possible_newline(AstFile *f) {
+	if ((f->tokenizer.flags & TokenizerFlag_InsertSemicolon) == 0) {
+		return false;
+	}
+	Token *prev = &f->curr_token;
+	if (prev->kind == Token_Semicolon && prev->string == "\n") {
+		advance_token(f);
+		return true;
+	}
+	return false;
+}
+
+bool skip_possible_newline_for_literal(AstFile *f) {
+	if ((f->tokenizer.flags & TokenizerFlag_InsertSemicolon) == 0) {
+		return false;
+	}
+	TokenPos curr_pos = f->curr_token.pos;
+	if (token_is_newline(f->curr_token)) {
+		Token next = peek_token(f);
+		if (curr_pos.line+1 >= next.pos.line) {
+			switch (next.kind) {
+			case Token_OpenBrace:
+			case Token_else:
+			case Token_where:
+				advance_token(f);
+				return true;
+			}
+		}
+	}
+
+	return false;
+}
+
+String token_to_string(Token const &tok) {
+	String p = token_strings[tok.kind];
+	if (token_is_newline(tok)) {
+		p = str_lit("newline");
+	}
+	return p;
+}
+
+
 Token expect_token(AstFile *f, TokenKind kind) {
 	Token prev = f->curr_token;
 	if (prev.kind != kind) {
 		String c = token_strings[kind];
-		String p = token_strings[prev.kind];
+		String p = token_to_string(prev);
 		syntax_error(f->curr_token, "Expected '%.*s', got '%.*s'", LIT(c), LIT(p));
 		if (prev.kind == Token_EOF) {
 			gb_exit(1);
@@ -1265,7 +1323,7 @@ Token expect_token(AstFile *f, TokenKind kind) {
 Token expect_token_after(AstFile *f, TokenKind kind, char const *msg) {
 	Token prev = f->curr_token;
 	if (prev.kind != kind) {
-		String p = token_strings[prev.kind];
+		String p = token_to_string(prev);
 		syntax_error(f->curr_token, "Expected '%.*s' after %s, got '%.*s'",
 		             LIT(token_strings[kind]),
 		             msg,
@@ -1296,11 +1354,13 @@ Token expect_operator(AstFile *f) {
 	} else if (prev.kind == Token_if || prev.kind == Token_when) {
 		// okay
 	} else if (!gb_is_between(prev.kind, Token__OperatorBegin+1, Token__OperatorEnd-1)) {
+		String p = token_to_string(prev);
 		syntax_error(f->curr_token, "Expected an operator, got '%.*s'",
-		             LIT(token_strings[prev.kind]));
+		             LIT(p));
 	} else if (!f->allow_range && is_token_range(prev)) {
+		String p = token_to_string(prev);
 		syntax_error(f->curr_token, "Expected an non-range operator, got '%.*s'",
-		             LIT(token_strings[prev.kind]));
+		             LIT(p));
 	}
 	advance_token(f);
 	return prev;
@@ -1309,8 +1369,9 @@ Token expect_operator(AstFile *f) {
 Token expect_keyword(AstFile *f) {
 	Token prev = f->curr_token;
 	if (!gb_is_between(prev.kind, Token__KeywordBegin+1, Token__KeywordEnd-1)) {
+		String p = token_to_string(prev);
 		syntax_error(f->curr_token, "Expected a keyword, got '%.*s'",
-		             LIT(token_strings[prev.kind]));
+		             LIT(p));
 	}
 	advance_token(f);
 	return prev;
@@ -1470,7 +1531,22 @@ void expect_semicolon(AstFile *f, Ast *s) {
 
 
 	if (s != nullptr) {
-		if (prev_token.pos.line != f->curr_token.pos.line) {
+		bool insert_semi = (f->tokenizer.flags & TokenizerFlag_InsertSemicolon) != 0;
+		if (insert_semi) {
+			switch (f->curr_token.kind) {
+			case Token_CloseBrace:
+			case Token_CloseParen:
+			case Token_else:
+			case Token_EOF:
+				return;
+
+			default:
+				if (is_semicolon_optional_for_node(f, s)) {
+					return;
+				}
+				break;
+			}
+		} else if (prev_token.pos.line != f->curr_token.pos.line) {
 			if (is_semicolon_optional_for_node(f, s)) {
 				return;
 			}
@@ -1488,14 +1564,16 @@ void expect_semicolon(AstFile *f, Ast *s) {
 			}
 		}
 		String node_string = ast_strings[s->kind];
+		String p = token_to_string(f->curr_token);
 		syntax_error(prev_token, "Expected ';' after %.*s, got %.*s",
-		             LIT(node_string), LIT(token_strings[f->curr_token.kind]));
+		             LIT(node_string), LIT(p));
 	} else {
 		switch (f->curr_token.kind) {
 		case Token_EOF:
 			return;
 		}
-		syntax_error(prev_token, "Expected ';'");
+		String p = token_to_string(f->curr_token);
+		syntax_error(prev_token, "Expected ';', got %.*s", LIT(p));
 	}
 	fix_advance_to_next_stmt(f);
 }
@@ -1990,6 +2068,7 @@ Ast *parse_operand(AstFile *f, bool lhs) {
 		Token where_token = {};
 		Array<Ast *> where_clauses = {};
 		u64 tags = 0;
+		skip_possible_newline_for_literal(f);
 
 		if (f->curr_token.kind == Token_where) {
 			where_token = expect_token(f, Token_where);
@@ -2166,6 +2245,8 @@ Ast *parse_operand(AstFile *f, bool lhs) {
 		Token where_token = {};
 		Array<Ast *> where_clauses = {};
 
+		skip_possible_newline_for_literal(f);
+
 		if (f->curr_token.kind == Token_where) {
 			where_token = expect_token(f, Token_where);
 			isize prev_level = f->expr_level;
@@ -2237,6 +2318,7 @@ Ast *parse_operand(AstFile *f, bool lhs) {
 			syntax_error(f->curr_token, "#maybe and #no_nil cannot be applied together");
 		}
 
+		skip_possible_newline_for_literal(f);
 
 		Token where_token = {};
 		Array<Ast *> where_clauses = {};
@@ -3361,7 +3443,8 @@ bool parse_expect_field_separator(AstFile *f, Ast *param) {
 		return true;
 	}
 	if (token.kind == Token_Semicolon) {
-		syntax_error(f->curr_token, "Expected a comma, got a semicolon");
+		String p = token_to_string(token);
+		syntax_error(f->curr_token, "Expected a comma, got a %.*s", LIT(p));
 		advance_token(f);
 		return true;
 	}
@@ -3684,6 +3767,7 @@ Ast *parse_if_stmt(AstFile *f) {
 		}
 	} else {
 		body = parse_block_stmt(f, false);
+		skip_possible_newline_for_literal(f);
 	}
 
 	if (allow_token(f, Token_else)) {
@@ -3739,6 +3823,7 @@ Ast *parse_when_stmt(AstFile *f) {
 		}
 	} else {
 		body = parse_block_stmt(f, true);
+		skip_possible_newline_for_literal(f);
 	}
 
 	if (allow_token(f, Token_else)) {
@@ -3844,6 +3929,7 @@ Ast *parse_for_stmt(AstFile *f) {
 				}
 			} else {
 				body = parse_block_stmt(f, false);
+				skip_possible_newline_for_literal(f);
 			}
 			return ast_range_stmt(f, token, nullptr, nullptr, in_token, rhs, body);
 		}
@@ -3879,6 +3965,7 @@ Ast *parse_for_stmt(AstFile *f) {
 		}
 	} else {
 		body = parse_block_stmt(f, false);
+		skip_possible_newline_for_literal(f);
 	}
 
 	if (is_range) {
@@ -4160,6 +4247,8 @@ Ast *parse_attribute(AstFile *f, Token token, TokenKind open_kind, TokenKind clo
 	}
 	Ast *attribute = ast_attribute(f, token, open, close, elems);
 
+	skip_possible_newline(f);
+
 	Ast *decl = parse_stmt(f);
 	if (decl->kind == Ast_ValueDecl) {
 		array_add(&decl->ValueDecl.attributes, attribute);
@@ -4228,6 +4317,7 @@ Ast *parse_stmt(AstFile *f) {
 				}
 			} else {
 				body = parse_block_stmt(f, false);
+				skip_possible_newline_for_literal(f);
 			}
 			if (bad_stmt) {
 				return ast_bad_stmt(f, inline_token, f->curr_token);
@@ -4432,7 +4522,11 @@ ParseFileError init_ast_file(AstFile *f, String fullpath, TokenPos *err_pos) {
 	if (!string_ends_with(f->fullpath, str_lit(".odin"))) {
 		return ParseFile_WrongExtension;
 	}
-	TokenizerInitError err = init_tokenizer(&f->tokenizer, f->fullpath);
+	TokenizerFlags tokenizer_flags = TokenizerFlag_None;
+	if (build_context.insert_semicolon) {
+		tokenizer_flags = TokenizerFlag_InsertSemicolon;
+	}
+	TokenizerInitError err = init_tokenizer(&f->tokenizer, f->fullpath, tokenizer_flags);
 	if (err != TokenizerInit_None) {
 		switch (err) {
 		case TokenizerInit_Empty:

src/tokenizer.cpp

@@ -527,6 +527,12 @@ struct TokenizerState {
 	u8 *  read_curr;   // pos from start
 	u8 *  line;        // current line pos
 	isize line_count;
+	bool  insert_semicolon;
+};
+
+enum TokenizerFlags {
+	TokenizerFlag_None = 0,
+	TokenizerFlag_InsertSemicolon = 1<<0,
 };
 
 struct Tokenizer {
@@ -542,6 +548,9 @@ struct Tokenizer {
 
 	isize error_count;
 	Array<String> allocated_strings;
+
+	TokenizerFlags flags;
+	bool insert_semicolon;
 };
 
 
@@ -552,15 +561,17 @@ TokenizerState save_tokenizer_state(Tokenizer *t) {
 	state.read_curr  = t->read_curr;
 	state.line       = t->line;
 	state.line_count = t->line_count;
+	state.insert_semicolon = t->insert_semicolon;
 	return state;
 }
 
 void restore_tokenizer_state(Tokenizer *t, TokenizerState *state) {
-	 t->curr_rune  = state->curr_rune;
+	t->curr_rune  = state->curr_rune;
-	 t->curr       = state->curr;
+	t->curr       = state->curr;
-	 t->read_curr  = state->read_curr;
+	t->read_curr  = state->read_curr;
-	 t->line       = state->line;
+	t->line       = state->line;
-	 t->line_count = state->line_count;
+	t->line_count = state->line_count;
+	t->insert_semicolon = state->insert_semicolon;
 }
 
 
@@ -615,7 +626,7 @@ void advance_to_next_rune(Tokenizer *t) {
 	}
 }
 
-TokenizerInitError init_tokenizer(Tokenizer *t, String fullpath) {
+TokenizerInitError init_tokenizer(Tokenizer *t, String fullpath, TokenizerFlags flags = TokenizerFlag_None) {
 	TokenizerInitError err = TokenizerInit_None;
 
 	char *c_str = alloc_cstring(heap_allocator(), fullpath);
@@ -625,6 +636,7 @@ TokenizerInitError init_tokenizer(Tokenizer *t, String fullpath) {
 	gbFileContents fc = gb_file_read_contents(heap_allocator(), true, c_str);
 	gb_zero_item(t);
 
+	t->flags = flags;
 	t->fullpath = fullpath;
 	t->line_count = 1;
 
@@ -888,9 +900,13 @@ void tokenizer_get_token(Tokenizer *t, Token *token) {
 	// Skip whitespace
 	for (;;) {
 		switch (t->curr_rune) {
+		case '\n':
+			if (t->insert_semicolon) {
+				break;
+			}
+			/*fallthrough*/
 		case ' ':
 		case '\t':
-		case '\n':
 		case '\r':
 			advance_to_next_rune(t);
 			continue;
@@ -907,6 +923,8 @@ void tokenizer_get_token(Tokenizer *t, Token *token) {
 	token->pos.offset = t->curr - t->start;
 	token->pos.column = t->curr - t->line + 1;
 
+	bool insert_semicolon = false;
+
 	Rune curr_rune = t->curr_rune;
 	if (rune_is_letter(curr_rune)) {
 		token->kind = Token_Ident;
@@ -930,19 +948,51 @@ void tokenizer_get_token(Tokenizer *t, Token *token) {
 				}
 			}
 		}
+
+		switch (token->kind) {
+		case Token_Ident:
+		case Token_context:
+		case Token_typeid: // Dunno?
+		case Token_break:
+		case Token_continue:
+		case Token_fallthrough:
+		case Token_return:
+			insert_semicolon = true;
+			break;
+		}
+
+
+		if (t->flags & TokenizerFlag_InsertSemicolon) {
+			t->insert_semicolon = insert_semicolon;
+		}
 		return;
 
 	} else if (gb_is_between(curr_rune, '0', '9')) {
+		insert_semicolon = true;
 		scan_number_to_token(t, token, false);
 	} else {
 		advance_to_next_rune(t);
 		switch (curr_rune) {
 		case GB_RUNE_EOF:
 			token->kind = Token_EOF;
+			if (t->insert_semicolon) {
+				t->insert_semicolon = false; // EOF consumed
+				token->string = str_lit("\n");
+				token->kind = Token_Semicolon;
+				return;
+			}
 			break;
 
+		case '\n':
+			t->insert_semicolon = false;
+			token->string = str_lit("\n");
+			token->kind = Token_Semicolon;
+			return;
+
 		case '\'': // Rune Literal
 		{
+			insert_semicolon = true;
+
 			token->kind = Token_Rune;
 			Rune quote = curr_rune;
 			bool valid = true;
@@ -978,12 +1028,19 @@ void tokenizer_get_token(Tokenizer *t, Token *token) {
 			} else {
 				tokenizer_err(t, "Invalid rune literal");
 			}
+
+			if (t->flags & TokenizerFlag_InsertSemicolon) {
+				t->insert_semicolon = insert_semicolon;
+			}
+
 			return;
 		} break;
 
 		case '`': // Raw String Literal
 		case '"': // String Literal
 		{
+			insert_semicolon = true;
+
 			bool has_carriage_return = false;
 			i32 success;
 			Rune quote = curr_rune;
@@ -1028,6 +1085,11 @@ void tokenizer_get_token(Tokenizer *t, Token *token) {
 			} else {
 				tokenizer_err(t, "Invalid string literal");
 			}
+
+			if (t->flags & TokenizerFlag_InsertSemicolon) {
+				t->insert_semicolon = insert_semicolon;
+			}
+
 			return;
 		} break;
 
@@ -1048,17 +1110,32 @@ void tokenizer_get_token(Tokenizer *t, Token *token) {
 
 		case '@':  token->kind = Token_At;           break;
 		case '$':  token->kind = Token_Dollar;       break;
-		case '?':  token->kind = Token_Question;     break;
+		case '?':
-		case '^':  token->kind = Token_Pointer;      break;
+			insert_semicolon = true;
+			token->kind = Token_Question;
+			break;
+		case '^':
+			insert_semicolon = true;
+			token->kind = Token_Pointer;
+			break;
 		case ';':  token->kind = Token_Semicolon;    break;
 		case ',':  token->kind = Token_Comma;        break;
 		case ':':  token->kind = Token_Colon;        break;
 		case '(':  token->kind = Token_OpenParen;    break;
-		case ')':  token->kind = Token_CloseParen;   break;
+		case ')':
-		case '[':  token->kind = Token_OpenBracket;  break;
+			insert_semicolon = true;
-		case ']':  token->kind = Token_CloseBracket; break;
+			token->kind = Token_CloseParen;
+			break;
+		case '[': token->kind = Token_OpenBracket;  break;
+		case ']':
+			insert_semicolon = true;
+			token->kind = Token_CloseBracket;
+			break;
 		case '{':  token->kind = Token_OpenBrace;    break;
-		case '}':  token->kind = Token_CloseBrace;   break;
+		case '}':
+			insert_semicolon = true;
+			token->kind = Token_CloseBrace;
+			break;
 		case '\\': token->kind = Token_BackSlash;    break;
 
 		case '%':
@@ -1131,10 +1208,12 @@ void tokenizer_get_token(Tokenizer *t, Token *token) {
 
 		case '#':
 			if (t->curr_rune == '!') {
+				insert_semicolon = t->insert_semicolon;
+				token->kind = Token_Comment;
+
 				while (t->curr_rune != '\n' && t->curr_rune != GB_RUNE_EOF) {
 					advance_to_next_rune(t);
 				}
-				token->kind = Token_Comment;
 			} else {
 				token->kind = Token_Hash;
 			}
@@ -1144,6 +1223,7 @@ void tokenizer_get_token(Tokenizer *t, Token *token) {
 		case '/': {
 			token->kind = Token_Quo;
 			if (t->curr_rune == '/') {
+				insert_semicolon = t->insert_semicolon;
 				token->kind = Token_Comment;
 
 				while (t->curr_rune != '\n' && t->curr_rune != GB_RUNE_EOF) {
@@ -1255,11 +1335,18 @@ void tokenizer_get_token(Tokenizer *t, Token *token) {
 				int len = cast(int)gb_utf8_encode_rune(str, curr_rune);
 				tokenizer_err(t, "Illegal character: %.*s (%d) ", len, str, curr_rune);
 			}
+			insert_semicolon = t->insert_semicolon; // Preserve insert_semicolon info
 			token->kind = Token_Invalid;
 			break;
 		}
 	}
 
+	if (t->flags & TokenizerFlag_InsertSemicolon) {
+		t->insert_semicolon = insert_semicolon;
+	}
+
 	token->string.len = t->curr - token->string.text;
+
+
 	return;
 }