const std = @import("std"); const logger = @import("logger.zig"); const tables = @import("js_lexer_tables.zig"); const alloc = @import("alloc.zig"); const build_options = @import("build_options"); const js_ast = @import("js_ast.zig"); usingnamespace @import("ast/base.zig"); usingnamespace @import("global.zig"); const unicode = std.unicode; const Source = logger.Source; pub const T = tables.T; pub const Keywords = tables.Keywords; pub const tokenToString = tables.tokenToString; pub const StrictModeReservedWords = tables.StrictModeReservedWords; pub const PropertyModifierKeyword = tables.PropertyModifierKeyword; pub const TypescriptStmtKeyword = tables.TypescriptStmtKeyword; pub const TypeScriptAccessibilityModifier = tables.TypeScriptAccessibilityModifier; pub const ChildlessJSXTags = tables.ChildlessJSXTags; fn notimpl() noreturn { Global.panic("not implemented yet!", .{}); } pub var emptyJavaScriptString = ([_]u16{0}); pub const JSONOptions = struct { allow_comments: bool = false, allow_trailing_commas: bool = false, starts_with_string: bool = false, }; pub const Lexer = struct { const LexerType = @This(); pub const Error = error{ UTF8Fail, OutOfMemory, SyntaxError, UnexpectedSyntax, JSONStringsMustUseDoubleQuotes, ParserError, }; // pub const Error = error{ // UnexpectedToken, // EndOfFile, // }; // err: ?LexerType.Error, log: *logger.Log, json_options: ?JSONOptions = null, for_global_name: bool = false, source: *const logger.Source, current: usize = 0, start: usize = 0, end: usize = 0, did_panic: bool = false, approximate_newline_count: usize = 0, previous_backslash_quote_in_jsx: logger.Range = logger.Range.None, token: T = T.t_end_of_file, has_newline_before: bool = false, has_pure_comment_before: bool = false, preserve_all_comments_before: bool = false, is_legacy_octal_literal: bool = false, is_log_disabled: bool = false, comments_to_preserve_before: std.ArrayList(js_ast.G.Comment), all_original_comments: ?[]js_ast.G.Comment = null, code_point: CodePoint = -1, identifier: []const u8 = "", jsx_factory_pragma_comment: ?js_ast.Span = null, jsx_fragment_pragma_comment: ?js_ast.Span = null, source_mapping_url: ?js_ast.Span = null, number: f64 = 0.0, rescan_close_brace_as_template_token: bool = false, prev_error_loc: logger.Loc = logger.Loc.Empty, allocator: *std.mem.Allocator, /// In JavaScript, strings are stored as UTF-16, but nearly every string is ascii. /// This means, usually, we can skip UTF8 -> UTF16 conversions. string_literal_buffer: std.ArrayList(u16), string_literal_slice: string = "", string_literal: JavascriptString, string_literal_is_ascii: bool = false, is_ascii_only: bool = true, pub fn clone(self: *const LexerType) LexerType { return LexerType{ .log = self.log, .json_options = self.json_options, .for_global_name = self.for_global_name, .source = self.source, .current = self.current, .start = self.start, .end = self.end, .did_panic = self.did_panic, .approximate_newline_count = self.approximate_newline_count, .previous_backslash_quote_in_jsx = self.previous_backslash_quote_in_jsx, .token = self.token, .has_newline_before = self.has_newline_before, .has_pure_comment_before = self.has_pure_comment_before, .preserve_all_comments_before = self.preserve_all_comments_before, .is_legacy_octal_literal = self.is_legacy_octal_literal, .is_log_disabled = self.is_log_disabled, .comments_to_preserve_before = self.comments_to_preserve_before, .all_original_comments = self.all_original_comments, .code_point = self.code_point, .identifier = self.identifier, .jsx_factory_pragma_comment = self.jsx_factory_pragma_comment, .jsx_fragment_pragma_comment = self.jsx_fragment_pragma_comment, .source_mapping_url = self.source_mapping_url, .number = self.number, .rescan_close_brace_as_template_token = self.rescan_close_brace_as_template_token, .prev_error_loc = self.prev_error_loc, .allocator = self.allocator, .string_literal_buffer = self.string_literal_buffer, .string_literal_slice = self.string_literal_slice, .string_literal = self.string_literal, .string_literal_is_ascii = self.string_literal_is_ascii, .is_ascii_only = self.is_ascii_only, }; } pub fn loc(self: *LexerType) logger.Loc { return logger.usize2Loc(self.start); } pub fn syntaxError(self: *LexerType) !void { @setCold(true); self.addError(self.start, "Syntax Error!!", .{}, true); return Error.SyntaxError; } pub fn addDefaultError(self: *LexerType, msg: []const u8) !void { @setCold(true); self.addError(self.start, "{s}", .{msg}, true); return Error.SyntaxError; } pub fn addSyntaxError(self: *LexerType, _loc: usize, comptime fmt: []const u8, args: anytype) !void { @setCold(true); self.addError(_loc, fmt, args, false); return Error.SyntaxError; } pub fn addError(self: *LexerType, _loc: usize, comptime format: []const u8, args: anytype, panic: bool) void { @setCold(true); if (self.is_log_disabled) return; var __loc = logger.usize2Loc(_loc); if (__loc.eql(self.prev_error_loc)) { return; } self.log.addErrorFmt(self.source, __loc, self.allocator, format, args) catch unreachable; self.prev_error_loc = __loc; } pub fn addRangeError(self: *LexerType, r: logger.Range, comptime format: []const u8, args: anytype, panic: bool) !void { @setCold(true); if (self.is_log_disabled) return; if (self.prev_error_loc.eql(r.loc)) { return; } const errorMessage = std.fmt.allocPrint(self.allocator, format, args) catch unreachable; var msg = self.log.addRangeError(self.source, r, errorMessage); self.prev_error_loc = r.loc; // if (panic) { // return Error.ParserError; // } } pub fn codePointEql(self: *LexerType, a: u8) bool { return @intCast(CodePoint, a) == self.code_point; } /// Look ahead at the next n codepoints without advancing the iterator. /// If fewer than n codepoints are available, then return the remainder of the string. fn peek(it: *LexerType, n: usize) string { const original_i = it.current; defer it.current = original_i; var end_ix = original_i; var found: usize = 0; while (found < n) : (found += 1) { const next_codepoint = it.nextCodepointSlice(); if (next_codepoint.len == 0) break; end_ix += next_codepoint.len; } return it.source.contents[original_i..end_ix]; } pub fn isIdentifierOrKeyword(lexer: LexerType) bool { return @enumToInt(lexer.token) >= @enumToInt(T.t_identifier); } pub fn stringLiteralUTF16(lexer: *LexerType) JavascriptString { if (lexer.string_literal_is_ascii) { return lexer.stringToUTF16(lexer.string_literal_slice); } else { return lexer.allocator.dupe(u16, lexer.string_literal) catch unreachable; } } pub fn decodeEscapeSequences(lexer: *LexerType, start: usize, text: string, comptime BufType: type, buf_: *BufType) !void { var buf = buf_.*; defer buf_.* = buf; lexer.is_ascii_only = false; var iter = CodepointIterator{ .bytes = text[start..], .i = 0 }; const start_length = buf.items.len; while (iter.nextCodepoint()) |c| { const width = iter.width; @setRuntimeSafety(false); switch (c) { '\r' => { // From the specification: // // 11.8.6.1 Static Semantics: TV and TRV // // TV excludes the code units of LineContinuation while TRV includes // them. and LineTerminatorSequences are normalized to // for both TV and TRV. An explicit EscapeSequence is needed to // include a or sequence. // Convert '\r\n' into '\n' if (iter.i < text.len and text[iter.i] == '\n') { iter.i += 1; } // Convert '\r' into '\n' buf.append('\n') catch unreachable; continue; }, '\\' => { @setRuntimeSafety(false); const c2 = iter.nextCodepoint() orelse return; const width2 = iter.width; switch (c2) { 'b' => { buf.append(std.mem.readIntNative(u16, "\\b")) catch unreachable; continue; }, 'f' => { buf.append(std.mem.readIntNative(u16, "\\f")) catch unreachable; continue; }, 'n' => { buf.append(std.mem.readIntNative(u16, "\\n")) catch unreachable; continue; }, 'r' => { buf.append(std.mem.readIntNative(u16, "\\r")) catch unreachable; continue; }, 't' => { buf.append(std.mem.readIntNative(u16, "\\t")) catch unreachable; continue; }, 'v' => { if (lexer.json_options != null) { lexer.end = start + iter.i - width2; try lexer.syntaxError(); } buf.append(std.mem.readIntNative(u16, "\\v")) catch unreachable; continue; }, // legacy octal literals '0'...'7' => { const octal_start = iter.i - 2; if (lexer.json_options != null) { lexer.end = start + iter.i - width2; try lexer.syntaxError(); } // 1-3 digit octal var is_bad = false; var value: i64 = c2 - '0'; const c3: CodePoint = iter.nextCodepoint() orelse return lexer.syntaxError(); const width3 = iter.width; switch (c3) { '0'...'7' => { value = value * 8 + c3 - '0'; iter.i += width3; const c4 = iter.nextCodepoint() orelse return lexer.syntaxError(); const width4 = iter.width; switch (c4) { '0'...'7' => { const temp = value * 8 + c4 - '0'; if (temp < 256) { value = temp; iter.i += width4; } }, '8', '9' => { is_bad = true; }, else => {}, } }, '8', '9' => { is_bad = true; }, else => {}, } iter.c = @intCast(i32, value); if (is_bad) { lexer.addRangeError( logger.Range{ .loc = .{ .start = @intCast(i32, octal_start) }, .len = @intCast(i32, iter.i - octal_start) }, "Invalid legacy octal literal", .{}, false, ) catch unreachable; } }, '8', '9' => { iter.c = c2; }, // 2-digit hexadecimal 'x' => { if (lexer.json_options != null) { lexer.end = start + iter.i - width2; try lexer.syntaxError(); } var value: CodePoint = 0; var c3: CodePoint = 0; var width3: u3 = 0; c3 = iter.nextCodepoint() orelse return lexer.syntaxError(); width3 = iter.width; switch (c3) { '0'...'9' => { value = value * 16 | (c3 - '0'); }, 'a'...'f' => { value = value * 16 | (c3 + 10 - 'a'); }, 'A'...'F' => { value = value * 16 | (c3 + 10 - 'A'); }, else => { lexer.end = start + iter.i - width3; return lexer.syntaxError(); }, } c3 = iter.nextCodepoint() orelse return lexer.syntaxError(); width3 = iter.width; switch (c3) { '0'...'9' => { value = value * 16 | (c3 - '0'); }, 'a'...'f' => { value = value * 16 | (c3 + 10 - 'a'); }, 'A'...'F' => { value = value * 16 | (c3 + 10 - 'A'); }, else => { lexer.end = start + iter.i - width3; return lexer.syntaxError(); }, } iter.c = value; }, 'u' => { // We're going to make this an i64 so we don't risk integer overflows // when people do weird things var value: i64 = 0; var c3 = iter.nextCodepoint() orelse return lexer.syntaxError(); var width3 = iter.width; // variable-length if (c3 == '{') { if (lexer.json_options != null) { lexer.end = start + iter.i - width2; try lexer.syntaxError(); } const hex_start = iter.i - width - width2 - width3; var is_first = true; var is_out_of_range = false; variableLength: while (true) { c3 = iter.nextCodepoint() orelse break :variableLength; switch (c3) { '0'...'9' => { value = value * 16 | (c3 - '0'); }, 'a'...'f' => { value = value * 16 | (c3 + 10 - 'a'); }, 'A'...'F' => { value = value * 16 | (c3 + 10 - 'A'); }, '}' => { if (is_first) { lexer.end = start + iter.i - width3; return lexer.syntaxError(); } break :variableLength; }, else => { lexer.end = start + iter.i - width3; return lexer.syntaxError(); }, } // '\U0010FFFF // copied from golang utf8.MaxRune if (value > 1114111) { is_out_of_range = true; } is_first = false; } if (is_out_of_range) { try lexer.addRangeError( .{ .loc = .{ .start = @intCast(i32, start + hex_start) }, .len = @intCast(i32, (iter.i - hex_start)) }, "Unicode escape sequence is out of range", .{}, true, ); return; } // fixed-length } else { // Fixed-length // comptime var j: usize = 0; var j: usize = 0; while (j < 4) : (j += 1) { switch (c3) { '0'...'9' => { value = value * 16 | (c3 - '0'); }, 'a'...'f' => { value = value * 16 | (c3 + 10 - 'a'); }, 'A'...'F' => { value = value * 16 | (c3 + 10 - 'A'); }, else => { lexer.end = start + iter.i - width3; return lexer.syntaxError(); }, } if (j < 3) { c3 = iter.nextCodepoint() orelse return lexer.syntaxError(); width3 = iter.width; } } } iter.c = @truncate(CodePoint, value); }, '\r' => { if (lexer.json_options != null) { lexer.end = start + iter.i - width2; try lexer.syntaxError(); } // Ignore line continuations. A line continuation is not an escaped newline. if (iter.i < text.len and text[iter.i + 1] == '\n') { // Make sure Windows CRLF counts as a single newline iter.i += 1; } continue; }, '\n', 0x2028, 0x2029 => { if (lexer.json_options != null) { lexer.end = start + iter.i - width2; try lexer.syntaxError(); } // Ignore line continuations. A line continuation is not an escaped newline. continue; }, else => { if (lexer.json_options != null) { switch (c2) { '"', '\\', '/' => {}, else => { lexer.end = start + iter.i - width2; try lexer.syntaxError(); }, } } iter.c = c2; }, } }, else => {}, } switch (iter.c) { 0...0xFFFF => { buf.append(@intCast(u16, iter.c)) catch unreachable; }, else => { iter.c -= 0x10000; buf.ensureUnusedCapacity(2) catch unreachable; buf.appendAssumeCapacity(@intCast(u16, 0xD800 + ((iter.c >> 10) & 0x3FF))); buf.appendAssumeCapacity(@intCast(u16, 0xDC00 + (iter.c & 0x3FF))); }, } } } pub const InnerStringLiteral = packed struct { suffix_len: u3, needs_slow_path: bool }; fn parseStringLiteralInnter(lexer: *LexerType, comptime quote: CodePoint) !InnerStringLiteral { var needs_slow_path = false; var suffix_len: u3 = if (comptime quote == 0) 0 else 1; stringLiteral: while (true) { @setRuntimeSafety(false); switch (lexer.code_point) { '\\' => { try lexer.step(); // Handle Windows CRLF if (lexer.code_point == 'r' and lexer.json_options != null) { try lexer.step(); if (lexer.code_point == '\n') { try lexer.step(); } continue :stringLiteral; } switch (lexer.code_point) { 'f', 't', 'r', 'n', '`', '\'', '0', '"', 0x2028, 0x2029 => { try lexer.step(); continue :stringLiteral; }, else => { needs_slow_path = true; }, } }, // This indicates the end of the file -1 => { if (comptime quote != 0) { try lexer.addDefaultError("Unterminated string literal"); } break :stringLiteral; }, '\r' => { if (comptime quote != '`') { try lexer.addDefaultError("Unterminated string literal"); } // Template literals require newline normalization needs_slow_path = true; }, '\n' => { // Implicitly-quoted strings end when they reach a newline OR end of file // This only applies to .env switch (comptime quote) { 0 => { break :stringLiteral; }, '`' => {}, else => { try lexer.addDefaultError("Unterminated string literal"); }, } }, '$' => { if (comptime quote == '`') { try lexer.step(); if (lexer.code_point == '{') { suffix_len = 2; try lexer.step(); if (lexer.rescan_close_brace_as_template_token) { lexer.token = T.t_template_middle; } else { lexer.token = T.t_template_head; } break :stringLiteral; } continue :stringLiteral; } }, // exit condition quote => { try lexer.step(); break; }, else => { // Non-ASCII strings need the slow path if (lexer.code_point >= 0x80) { needs_slow_path = true; } else if (lexer.json_options != null and lexer.code_point < 0x20) { try lexer.syntaxError(); } }, } try lexer.step(); } return InnerStringLiteral{ .needs_slow_path = needs_slow_path, .suffix_len = suffix_len }; } pub fn parseStringLiteral(lexer: *LexerType, comptime quote: CodePoint) !void { if (comptime quote != '`') { lexer.token = T.t_string_literal; } else if (lexer.rescan_close_brace_as_template_token) { lexer.token = T.t_template_tail; } else { lexer.token = T.t_no_substitution_template_literal; } // quote is 0 when parsing JSON from .env // .env values may not always be quoted. try lexer.step(); var string_literal_details = try lexer.parseStringLiteralInnter(quote); // Reset string literal const base = if (comptime quote == 0) lexer.start else lexer.start + 1; lexer.string_literal_slice = lexer.source.contents[base .. lexer.end - string_literal_details.suffix_len]; lexer.string_literal_is_ascii = !string_literal_details.needs_slow_path; lexer.string_literal_buffer.shrinkRetainingCapacity(0); if (string_literal_details.needs_slow_path) { lexer.string_literal_buffer.ensureUnusedCapacity(lexer.string_literal_slice.len) catch unreachable; try lexer.decodeEscapeSequences(0, lexer.string_literal_slice, @TypeOf(lexer.string_literal_buffer), &lexer.string_literal_buffer); lexer.string_literal = lexer.string_literal_buffer.items; } lexer.is_ascii_only = lexer.is_ascii_only and lexer.string_literal_is_ascii; if (comptime !FeatureFlags.allow_json_single_quotes) { if (quote == '\'' and lexer.json_options != null) { try lexer.addRangeError(lexer.range(), "JSON strings must use double quotes", .{}, true); } } // for (text) // // if (needs_slow_path) { // // // Slow path // // // lexer.string_literal = lexer.(lexer.start + 1, text); // // } else { // // // Fast path // // } } inline fn nextCodepointSlice(it: *LexerType) []const u8 { @setRuntimeSafety(false); const cp_len = strings.utf8ByteSequenceLength(it.source.contents[it.current]); it.end = it.current; it.current += cp_len; return if (!(it.current > it.source.contents.len)) it.source.contents[it.current - cp_len .. it.current] else ""; } inline fn nextCodepoint(it: *LexerType) !CodePoint { const slice = it.nextCodepointSlice(); return switch (slice.len) { 0 => -1, 1 => @as(CodePoint, slice[0]), 2 => @as(CodePoint, unicode.utf8Decode2(slice) catch unreachable), 3 => @as(CodePoint, unicode.utf8Decode3(slice) catch unreachable), 4 => @as(CodePoint, unicode.utf8Decode4(slice) catch unreachable), else => unreachable, }; } fn step(lexer: *LexerType) !void { lexer.code_point = try lexer.nextCodepoint(); // Track the approximate number of newlines in the file so we can preallocate // the line offset table in the printer for source maps. The line offset table // is the #1 highest allocation in the heap profile, so this is worth doing. // This count is approximate because it handles "\n" and "\r\n" (the common // cases) but not "\r" or "\u2028" or "\u2029". Getting this wrong is harmless // because it's only a preallocation. The array will just grow if it's too small. lexer.approximate_newline_count += @boolToInt(lexer.code_point == '\n'); } pub fn expect(self: *LexerType, comptime token: T) !void { if (self.token != token) { try self.expected(token); } try self.next(); } pub fn expectOrInsertSemicolon(lexer: *LexerType) !void { if (lexer.token == T.t_semicolon or (!lexer.has_newline_before and lexer.token != T.t_close_brace and lexer.token != T.t_end_of_file)) { try lexer.expect(T.t_semicolon); } } pub fn addUnsupportedSyntaxError(self: *LexerType, msg: []const u8) !void { self.addError(self.end, "Unsupported syntax: {s}", .{msg}, true); return Error.SyntaxError; } pub const IdentifierKind = enum { normal, private }; pub const ScanResult = struct { token: T, contents: string }; threadlocal var small_escape_sequence_buffer: [4096]u16 = undefined; const FakeArrayList16 = struct { items: []u16, i: usize = 0, pub fn append(fake: *FakeArrayList16, value: u16) !void { std.debug.assert(fake.items.len < fake.i); fake.items[fake.i] = value; fake.i += 1; } pub fn appendAssumeCapacity(fake: *FakeArrayList16, value: u16) void { std.debug.assert(fake.items.len < fake.i); fake.items[fake.i] = value; fake.i += 1; } pub fn ensureUnusedCapacity(fake: *FakeArrayList16, int: anytype) !void { std.debug.assert(fake.items.len < fake.i + int); } }; threadlocal var large_escape_sequence_list: std.ArrayList(u16) = undefined; threadlocal var large_escape_sequence_list_loaded: bool = false; // This is an edge case that doesn't really exist in the wild, so it doesn't // need to be as fast as possible. pub fn scanIdentifierWithEscapes(lexer: *LexerType, comptime kind: IdentifierKind) anyerror!ScanResult { var result = ScanResult{ .token = .t_end_of_file, .contents = "" }; // First pass: scan over the identifier to see how long it is while (true) { // Scan a unicode escape sequence. There is at least one because that's // what caused us to get on this slow path in the first place. if (lexer.code_point == '\\') { try lexer.step(); if (lexer.code_point != 'u') { try lexer.syntaxError(); } try lexer.step(); if (lexer.code_point == '{') { // Variable-length try lexer.step(); while (lexer.code_point != '}') { switch (lexer.code_point) { '0'...'9', 'a'...'f', 'A'...'F' => { try lexer.step(); }, else => { try lexer.syntaxError(); }, } } try lexer.step(); } else { // Fixed-length // comptime var j: usize = 0; switch (lexer.code_point) { '0'...'9', 'a'...'f', 'A'...'F' => { try lexer.step(); }, else => { try lexer.syntaxError(); }, } switch (lexer.code_point) { '0'...'9', 'a'...'f', 'A'...'F' => { try lexer.step(); }, else => { try lexer.syntaxError(); }, } switch (lexer.code_point) { '0'...'9', 'a'...'f', 'A'...'F' => { try lexer.step(); }, else => { try lexer.syntaxError(); }, } switch (lexer.code_point) { '0'...'9', 'a'...'f', 'A'...'F' => { try lexer.step(); }, else => { try lexer.syntaxError(); }, } } continue; } if (!isIdentifierContinue(lexer.code_point)) { break; } try lexer.step(); } // Second pass: re-use our existing escape sequence parser var original_text = lexer.raw(); if (original_text.len < 1024) { var buf = FakeArrayList16{ .items = &small_escape_sequence_buffer, .i = 0 }; try lexer.decodeEscapeSequences(lexer.start, original_text, FakeArrayList16, &buf); result.contents = lexer.utf16ToString(buf.items[0..buf.i]); } else { if (!large_escape_sequence_list_loaded) { large_escape_sequence_list = try std.ArrayList(u16).initCapacity(lexer.allocator, original_text.len); large_escape_sequence_list_loaded = true; } large_escape_sequence_list.shrinkRetainingCapacity(0); try lexer.decodeEscapeSequences(lexer.start, original_text, std.ArrayList(u16), &large_escape_sequence_list); result.contents = lexer.utf16ToString(large_escape_sequence_list.items); } var identifier = result.contents; if (kind == .private) { identifier = result.contents[1..]; } if (!isIdentifier(identifier)) { try lexer.addRangeError( .{ .loc = logger.usize2Loc(lexer.start), .len = @intCast(i32, lexer.end - lexer.start) }, "Invalid identifier: \"{s}\"", .{result.contents}, true, ); } result.contents = identifier; // Escaped keywords are not allowed to work as actual keywords, but they are // allowed wherever we allow identifiers or keywords. For example: // // // This is an error (equivalent to "var var;") // var \u0076\u0061\u0072; // // // This is an error (equivalent to "var foo;" except for this rule) // \u0076\u0061\u0072 foo; // // // This is an fine (equivalent to "foo.var;") // foo.\u0076\u0061\u0072; // result.token = if (Keywords.has(result.contents)) .t_escaped_keyword else .t_identifier; // const text = lexer.decodeEscapeSequences(lexer.start, lexer.raw(), ) return result; } pub fn expectContextualKeyword(self: *LexerType, comptime keyword: string) !void { if (!self.isContextualKeyword(keyword)) { if (std.builtin.mode == std.builtin.Mode.Debug) { self.addError(self.start, "Expected \"{s}\" but found \"{s}\" (token: {s})", .{ keyword, self.raw(), self.token, }, true); } else { self.addError(self.start, "Expected \"{s}\" but found \"{s}\"", .{ keyword, self.raw() }, true); } return Error.UnexpectedSyntax; } try self.next(); } pub fn maybeExpandEquals(lexer: *LexerType) !void { switch (lexer.code_point) { '>' => { // "=" + ">" = "=>" lexer.token = .t_equals_greater_than; try lexer.step(); }, '=' => { // "=" + "=" = "==" lexer.token = .t_equals_equals; try lexer.step(); if (lexer.code_point == '=') { // "=" + "==" = "===" lexer.token = .t_equals_equals_equals; try lexer.step(); } }, else => {}, } } pub fn expectLessThan(lexer: *LexerType, comptime is_inside_jsx_element: bool) !void { switch (lexer.token) { .t_less_than => { if (is_inside_jsx_element) { try lexer.nextInsideJSXElement(); } else { try lexer.next(); } }, .t_less_than_equals => { lexer.token = .t_equals; lexer.start += 1; try lexer.maybeExpandEquals(); }, .t_less_than_less_than => { lexer.token = .t_less_than; lexer.start += 1; }, .t_less_than_less_than_equals => { lexer.token = .t_less_than_equals; lexer.start += 1; }, else => { try lexer.expected(.t_less_than); }, } } pub fn expectGreaterThan(lexer: *LexerType, comptime is_inside_jsx_element: bool) !void { switch (lexer.token) { .t_greater_than => { if (is_inside_jsx_element) { try lexer.nextInsideJSXElement(); } else { try lexer.next(); } }, .t_greater_than_greater_than => { lexer.token = .t_greater_than; lexer.start += 1; }, .t_greater_than_greater_than_greater_than => { lexer.token = .t_greater_than_greater_than; lexer.start += 1; }, .t_greater_than_equals => { lexer.token = .t_equals; lexer.start += 1; try lexer.maybeExpandEquals(); }, .t_greater_than_greater_than_equals => { lexer.token = .t_greater_than_greater_than; lexer.start += 1; }, .t_greater_than_greater_than_greater_than_equals => { lexer.token = .t_greater_than_greater_than_equals; lexer.start += 1; }, else => { try lexer.expected(.t_greater_than); }, } } pub fn next(lexer: *LexerType) !void { lexer.has_newline_before = lexer.end == 0; while (true) { lexer.start = lexer.end; lexer.token = T.t_end_of_file; @setRuntimeSafety(false); switch (lexer.code_point) { -1 => { lexer.token = T.t_end_of_file; }, '#' => { @setRuntimeSafety(false); if (lexer.start == 0 and lexer.source.contents[1] == '!') { // "#!/usr/bin/env node" lexer.token = .t_hashbang; hashbang: while (true) { @setRuntimeSafety(false); try lexer.step(); switch (lexer.code_point) { '\r', '\n', 0x2028, 0x2029 => { break :hashbang; }, -1 => { break :hashbang; }, else => {}, } } lexer.identifier = lexer.raw(); } else { try lexer.step(); if (lexer.code_point == '\\') { @setRuntimeSafety(false); lexer.identifier = (try lexer.scanIdentifierWithEscapes(.private)).contents; lexer.token = T.t_private_identifier; } else { @setRuntimeSafety(false); if (!isIdentifierStart(lexer.code_point)) { try lexer.syntaxError(); } try lexer.step(); while (isIdentifierContinue(lexer.code_point)) { try lexer.step(); } if (lexer.code_point == '\\') { lexer.identifier = (try lexer.scanIdentifierWithEscapes(.private)).contents; lexer.token = T.t_private_identifier; } else { lexer.token = T.t_private_identifier; lexer.identifier = lexer.raw(); } break; } } }, '\r', '\n', 0x2028, 0x2029 => { try lexer.step(); lexer.has_newline_before = true; continue; }, '\t', ' ' => { try lexer.step(); continue; }, '(' => { try lexer.step(); lexer.token = T.t_open_paren; }, ')' => { try lexer.step(); lexer.token = T.t_close_paren; }, '[' => { try lexer.step(); lexer.token = T.t_open_bracket; }, ']' => { try lexer.step(); lexer.token = T.t_close_bracket; }, '{' => { try lexer.step(); lexer.token = T.t_open_brace; }, '}' => { try lexer.step(); lexer.token = T.t_close_brace; }, ',' => { try lexer.step(); lexer.token = T.t_comma; }, ':' => { try lexer.step(); lexer.token = T.t_colon; }, ';' => { try lexer.step(); lexer.token = T.t_semicolon; }, '@' => { try lexer.step(); lexer.token = T.t_at; }, '~' => { try lexer.step(); lexer.token = T.t_tilde; }, '?' => { // '?' or '?.' or '??' or '??=' try lexer.step(); switch (lexer.code_point) { '?' => { try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_question_question_equals; }, else => { lexer.token = T.t_question_question; }, } }, '.' => { lexer.token = T.t_question; const current = lexer.current; const contents = lexer.source.contents; // Lookahead to disambiguate with 'a?.1:b' if (current < contents.len) { const c = contents[current]; if (c < '0' or c > '9') { try lexer.step(); lexer.token = T.t_question_dot; } } }, else => { lexer.token = T.t_question; }, } }, '%' => { // '%' or '%=' try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_percent_equals; }, else => { lexer.token = T.t_percent; }, } }, '&' => { // '&' or '&=' or '&&' or '&&=' try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_ampersand_equals; }, '&' => { try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_ampersand_ampersand_equals; }, else => { lexer.token = T.t_ampersand_ampersand; }, } }, else => { lexer.token = T.t_ampersand; }, } }, '|' => { // '|' or '|=' or '||' or '||=' try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_bar_equals; }, '|' => { try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_bar_bar_equals; }, else => { lexer.token = T.t_bar_bar; }, } }, else => { lexer.token = T.t_bar; }, } }, '^' => { // '^' or '^=' try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_caret_equals; }, else => { lexer.token = T.t_caret; }, } }, '+' => { // '+' or '+=' or '++' try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_plus_equals; }, '+' => { try lexer.step(); lexer.token = T.t_plus_plus; }, else => { lexer.token = T.t_plus; }, } }, '-' => { // '+' or '+=' or '++' try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_minus_equals; }, '-' => { try lexer.step(); if (lexer.code_point == '>' and lexer.has_newline_before) { try lexer.step(); lexer.log.addRangeWarning(lexer.source, lexer.range(), "Treating \"-->\" as the start of a legacy HTML single-line comment") catch unreachable; singleLineHTMLCloseComment: while (true) { switch (lexer.code_point) { '\r', '\n', 0x2028, 0x2029 => { break :singleLineHTMLCloseComment; }, -1 => { break :singleLineHTMLCloseComment; }, else => {}, } try lexer.step(); } continue; } lexer.token = T.t_minus_minus; }, else => { lexer.token = T.t_minus; }, } }, '*' => { // '*' or '*=' or '**' or '**=' try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = .t_asterisk_equals; }, '*' => { try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = .t_asterisk_asterisk_equals; }, else => { lexer.token = .t_asterisk_asterisk; }, } }, else => { lexer.token = .t_asterisk; }, } }, '/' => { // '/' or '/=' or '//' or '/* ... */' try lexer.step(); if (lexer.for_global_name) { lexer.token = .t_slash; break; } switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = .t_slash_equals; }, '/' => { singleLineComment: while (true) { try lexer.step(); switch (lexer.code_point) { '\r', '\n', 0x2028, 0x2029 => { break :singleLineComment; }, -1 => { break :singleLineComment; }, else => {}, } } if (lexer.json_options) |json| { if (!json.allow_comments) { try lexer.addRangeError(lexer.range(), "JSON does not support comments", .{}, true); return; } } lexer.scanCommentText(); continue; }, '*' => { try lexer.step(); multiLineComment: while (true) { switch (lexer.code_point) { '*' => { try lexer.step(); if (lexer.code_point == '/') { try lexer.step(); break :multiLineComment; } }, '\r', '\n', 0x2028, 0x2029 => { try lexer.step(); lexer.has_newline_before = true; }, -1 => { lexer.start = lexer.end; try lexer.addSyntaxError( lexer.start, "Expected \"*/\" to terminate multi-line comment", .{}, ); }, else => { try lexer.step(); }, } } if (lexer.json_options) |json| { if (!json.allow_comments) { try lexer.addRangeError(lexer.range(), "JSON does not support comments", .{}, true); return; } } lexer.scanCommentText(); continue; }, else => { lexer.token = .t_slash; }, } }, '=' => { // '=' or '=>' or '==' or '===' try lexer.step(); switch (lexer.code_point) { '>' => { try lexer.step(); lexer.token = T.t_equals_greater_than; }, '=' => { try lexer.step(); switch (lexer.code_point) { '=' => { try lexer.step(); lexer.token = T.t_equals_equals_equals; }, else => { lexer.token = T.t_equals_equals; }, } }, else => { lexer.token = T.t_equals; }, } }, '<' => { // '<' or '<<' or '<=' or '<<=' or '