(function() { var BOOL, CALLABLE, CODE, COFFEE_ALIASES, COFFEE_ALIAS_MAP, COFFEE_KEYWORDS, COMMENT, COMPARE, COMPOUND_ASSIGN, HEREDOC, HEREDOC_ILLEGAL, HEREDOC_INDENT, HEREGEX, HEREGEX_OMIT, IDENTIFIER, INDEXABLE, INVERSES, JSTOKEN, JS_FORBIDDEN, JS_KEYWORDS, LINE_BREAK, LINE_CONTINUER, LOGIC, Lexer, MATH, MULTILINER, MULTI_DENT, NOT_REGEX, NOT_SPACED_REGEX, NUMBER, OPERATOR, REGEX, RELATION, RESERVED, Rewriter, SHIFT, SIMPLESTR, TRAILING_SPACES, UNARY, WHITESPACE, compact, count, key, last, starts, _ref, _ref2, __hasProp = Object.prototype.hasOwnProperty, __indexOf = Array.prototype.indexOf || function(item) { for (var i = 0, l = this.length; i < l; i++) { if (__hasProp.call(this, i) && this[i] === item) return i; } return -1; }; _ref = require('./rewriter'), Rewriter = _ref.Rewriter, INVERSES = _ref.INVERSES; _ref2 = require('./helpers'), count = _ref2.count, starts = _ref2.starts, compact = _ref2.compact, last = _ref2.last; exports.Lexer = Lexer = (function() { function Lexer() {} Lexer.prototype.tokenize = function(code, opts) { var i, tag; if (opts == null) opts = {}; if (WHITESPACE.test(code)) code = "\n" + code; code = code.replace(/\r/g, '').replace(TRAILING_SPACES, ''); this.code = code; this.line = opts.line || 0; this.indent = 0; this.indebt = 0; this.outdebt = 0; this.indents = []; this.ends = []; this.tokens = []; i = 0; while (this.chunk = code.slice(i)) { i += this.identifierToken() || this.commentToken() || this.whitespaceToken() || this.lineToken() || this.heredocToken() || this.stringToken() || this.numberToken() || this.regexToken() || this.jsToken() || this.literalToken(); } this.closeIndentation(); if (tag = this.ends.pop()) this.error("missing " + tag); if (opts.rewrite === false) return this.tokens; return (new Rewriter).rewrite(this.tokens); }; Lexer.prototype.identifierToken = function() { var colon, forcedIdentifier, id, input, match, prev, tag, _ref3, _ref4; if (!(match = IDENTIFIER.exec(this.chunk))) return 0; input = match[0], id = match[1], colon = match[2]; if (id === 'own' && this.tag() === 'FOR') { this.token('OWN', id); return id.length; } forcedIdentifier = colon || (prev = last(this.tokens)) && (((_ref3 = prev[0]) === '.' || _ref3 === '?.' || _ref3 === '::') || !prev.spaced && prev[0] === '@'); tag = 'IDENTIFIER'; if (!forcedIdentifier && (__indexOf.call(JS_KEYWORDS, id) >= 0 || __indexOf.call(COFFEE_KEYWORDS, id) >= 0)) { tag = id.toUpperCase(); if (tag === 'WHEN' && (_ref4 = this.tag(), __indexOf.call(LINE_BREAK, _ref4) >= 0)) { tag = 'LEADING_WHEN'; } else if (tag === 'FOR') { this.seenFor = true; } else if (tag === 'UNLESS') { tag = 'IF'; } else if (__indexOf.call(UNARY, tag) >= 0) { tag = 'UNARY'; } else if (__indexOf.call(RELATION, tag) >= 0) { if (tag !== 'INSTANCEOF' && this.seenFor) { tag = 'FOR' + tag; this.seenFor = false; } else { tag = 'RELATION'; if (this.value() === '!') { this.tokens.pop(); id = '!' + id; } } } } if (__indexOf.call(['eval', 'arguments'].concat(JS_FORBIDDEN), id) >= 0) { if (forcedIdentifier) { tag = 'IDENTIFIER'; id = new String(id); id.reserved = true; } else if (__indexOf.call(RESERVED, id) >= 0) { this.error("reserved word \"" + id + "\""); } } if (!forcedIdentifier) { if (__indexOf.call(COFFEE_ALIASES, id) >= 0) id = COFFEE_ALIAS_MAP[id]; tag = (function() { switch (id) { case '!': return 'UNARY'; case '==': case '!=': return 'COMPARE'; case '&&': case '||': return 'LOGIC'; case 'true': case 'false': case 'null': case 'undefined': return 'BOOL'; case 'break': case 'continue': return 'STATEMENT'; default: return tag; } })(); } this.token(tag, id); if (colon) this.token(':', ':'); return input.length; }; Lexer.prototype.numberToken = function() { var binaryLiteral, lexedLength, match, number; if (!(match = NUMBER.exec(this.chunk))) return 0; number = match[0]; lexedLength = number.length; if (binaryLiteral = /0b([01]+)/.exec(number)) { number = (parseInt(binaryLiteral[1], 2)).toString(); } this.token('NUMBER', number); return lexedLength; }; Lexer.prototype.stringToken = function() { var match, string; switch (this.chunk.charAt(0)) { case "'": if (!(match = SIMPLESTR.exec(this.chunk))) return 0; this.token('STRING', (string = match[0]).replace(MULTILINER, '\\\n')); break; case '"': if (!(string = this.balancedString(this.chunk, '"'))) return 0; if (0 < string.indexOf('#{', 1)) { this.interpolateString(string.slice(1, -1)); } else { this.token('STRING', this.escapeLines(string)); } break; default: return 0; } this.line += count(string, '\n'); return string.length; }; Lexer.prototype.heredocToken = function() { var doc, heredoc, match, quote; if (!(match = HEREDOC.exec(this.chunk))) return 0; heredoc = match[0]; quote = heredoc.charAt(0); doc = this.sanitizeHeredoc(match[2], { quote: quote, indent: null }); if (quote === '"' && 0 <= doc.indexOf('#{')) { this.interpolateString(doc, { heredoc: true }); } else { this.token('STRING', this.makeString(doc, quote, true)); } this.line += count(heredoc, '\n'); return heredoc.length; }; Lexer.prototype.commentToken = function() { var comment, here, match; if (!(match = this.chunk.match(COMMENT))) return 0; comment = match[0], here = match[1]; if (here) { this.token('HERECOMMENT', this.sanitizeHeredoc(here, { herecomment: true, indent: Array(this.indent + 1).join(' ') })); this.token('TERMINATOR', '\n'); } this.line += count(comment, '\n'); return comment.length; }; Lexer.prototype.jsToken = function() { var match, script; if (!(this.chunk.charAt(0) === '`' && (match = JSTOKEN.exec(this.chunk)))) { return 0; } this.token('JS', (script = match[0]).slice(1, -1)); return script.length; }; Lexer.prototype.regexToken = function() { var flags, length, match, prev, regex, _ref3, _ref4; if (this.chunk.charAt(0) !== '/') return 0; if (match = HEREGEX.exec(this.chunk)) { length = this.heregexToken(match); this.line += count(match[0], '\n'); return length; } prev = last(this.tokens); if (prev && (_ref3 = prev[0], __indexOf.call((prev.spaced ? NOT_REGEX : NOT_SPACED_REGEX), _ref3) >= 0)) { return 0; } if (!(match = REGEX.exec(this.chunk))) return 0; _ref4 = match, match = _ref4[0], regex = _ref4[1], flags = _ref4[2]; if (regex.slice(0, 2) === '/*') { this.error('regular expressions cannot begin with `*`'); } if (regex === '//') regex = '/(?:)/'; this.token('REGEX', "" + regex + flags); return match.length; }; Lexer.prototype.heregexToken = function(match) { var body, flags, heregex, re, tag, tokens, value, _i, _len, _ref3, _ref4, _ref5, _ref6; heregex = match[0], body = match[1], flags = match[2]; if (0 > body.indexOf('#{')) { re = body.replace(HEREGEX_OMIT, '').replace(/\//g, '\\/'); if (re.match(/^\*/)) { this.error('regular expressions cannot begin with `*`'); } this.token('REGEX', "/" + (re || '(?:)') + "/" + flags); return heregex.length; } this.token('IDENTIFIER', 'RegExp'); this.tokens.push(['CALL_START', '(']); tokens = []; _ref3 = this.interpolateString(body, { regex: true }); for (_i = 0, _len = _ref3.length; _i < _len; _i++) { _ref4 = _ref3[_i], tag = _ref4[0], value = _ref4[1]; if (tag === 'TOKENS') { tokens.push.apply(tokens, value); } else { if (!(value = value.replace(HEREGEX_OMIT, ''))) continue; value = value.replace(/\\/g, '\\\\'); tokens.push(['STRING', this.makeString(value, '"', true)]); } tokens.push(['+', '+']); } tokens.pop(); if (((_ref5 = tokens[0]) != null ? _ref5[0] : void 0) !== 'STRING') { this.tokens.push(['STRING', '""'], ['+', '+']); } (_ref6 = this.tokens).push.apply(_ref6, tokens); if (flags) this.tokens.push([',', ','], ['STRING', '"' + flags + '"']); this.token(')', ')'); return heregex.length; }; Lexer.prototype.lineToken = function() { var diff, indent, match, noNewlines, prev, size; if (!(match = MULTI_DENT.exec(this.chunk))) return 0; indent = match[0]; this.line += count(indent, '\n'); this.seenFor = false; prev = last(this.tokens, 1); size = indent.length - 1 - indent.lastIndexOf('\n'); noNewlines = this.unfinished(); if (size - this.indebt === this.indent) { if (noNewlines) { this.suppressNewlines(); } else { this.newlineToken(); } return indent.length; } if (size > this.indent) { if (noNewlines) { this.indebt = size - this.indent; this.suppressNewlines(); return indent.length; } diff = size - this.indent + this.outdebt; this.token('INDENT', diff); this.indents.push(diff); this.ends.push('OUTDENT'); this.outdebt = this.indebt = 0; } else { this.indebt = 0; this.outdentToken(this.indent - size, noNewlines); } this.indent = size; return indent.length; }; Lexer.prototype.outdentToken = function(moveOut, noNewlines) { var dent, len; while (moveOut > 0) { len = this.indents.length - 1; if (this.indents[len] === void 0) { moveOut = 0; } else if (this.indents[len] === this.outdebt) { moveOut -= this.outdebt; this.outdebt = 0; } else if (this.indents[len] < this.outdebt) { this.outdebt -= this.indents[len]; moveOut -= this.indents[len]; } else { dent = this.indents.pop() - this.outdebt; moveOut -= dent; this.outdebt = 0; this.pair('OUTDENT'); this.token('OUTDENT', dent); } } if (dent) this.outdebt -= moveOut; while (this.value() === ';') { this.tokens.pop(); } if (!(this.tag() === 'TERMINATOR' || noNewlines)) { this.token('TERMINATOR', '\n'); } return this; }; Lexer.prototype.whitespaceToken = function() { var match, nline, prev; if (!((match = WHITESPACE.exec(this.chunk)) || (nline = this.chunk.charAt(0) === '\n'))) { return 0; } prev = last(this.tokens); if (prev) prev[match ? 'spaced' : 'newLine'] = true; if (match) { return match[0].length; } else { return 0; } }; Lexer.prototype.newlineToken = function() { while (this.value() === ';') { this.tokens.pop(); } if (this.tag() !== 'TERMINATOR') this.token('TERMINATOR', '\n'); return this; }; Lexer.prototype.suppressNewlines = function() { if (this.value() === '\\') this.tokens.pop(); return this; }; Lexer.prototype.literalToken = function() { var match, prev, tag, value, _ref3, _ref4, _ref5, _ref6; if (match = OPERATOR.exec(this.chunk)) { value = match[0]; if (CODE.test(value)) this.tagParameters(); } else { value = this.chunk.charAt(0); } tag = value; prev = last(this.tokens); if (value === '=' && prev) { if (!prev[1].reserved && (_ref3 = prev[1], __indexOf.call(JS_FORBIDDEN, _ref3) >= 0)) { this.error("reserved word \"" + (this.value()) + "\" can't be assigned"); } if ((_ref4 = prev[1]) === '||' || _ref4 === '&&') { prev[0] = 'COMPOUND_ASSIGN'; prev[1] += '='; return value.length; } } if (value === ';') { this.seenFor = false; tag = 'TERMINATOR'; } else if (__indexOf.call(MATH, value) >= 0) { tag = 'MATH'; } else if (__indexOf.call(COMPARE, value) >= 0) { tag = 'COMPARE'; } else if (__indexOf.call(COMPOUND_ASSIGN, value) >= 0) { tag = 'COMPOUND_ASSIGN'; } else if (__indexOf.call(UNARY, value) >= 0) { tag = 'UNARY'; } else if (__indexOf.call(SHIFT, value) >= 0) { tag = 'SHIFT'; } else if (__indexOf.call(LOGIC, value) >= 0 || value === '?' && (prev != null ? prev.spaced : void 0)) { tag = 'LOGIC'; } else if (prev && !prev.spaced) { if (value === '(' && (_ref5 = prev[0], __indexOf.call(CALLABLE, _ref5) >= 0)) { if (prev[0] === '?') prev[0] = 'FUNC_EXIST'; tag = 'CALL_START'; } else if (value === '[' && (_ref6 = prev[0], __indexOf.call(INDEXABLE, _ref6) >= 0)) { tag = 'INDEX_START'; switch (prev[0]) { case '?': prev[0] = 'INDEX_SOAK'; } } } switch (value) { case '(': case '{': case '[': this.ends.push(INVERSES[value]); break; case ')': case '}': case ']': this.pair(value); } this.token(tag, value); return value.length; }; Lexer.prototype.sanitizeHeredoc = function(doc, options) { var attempt, herecomment, indent, match, _ref3; indent = options.indent, herecomment = options.herecomment; if (herecomment) { if (HEREDOC_ILLEGAL.test(doc)) { this.error("block comment cannot contain \"*/\", starting"); } if (doc.indexOf('\n') <= 0) return doc; } else { while (match = HEREDOC_INDENT.exec(doc)) { attempt = match[1]; if (indent === null || (0 < (_ref3 = attempt.length) && _ref3 < indent.length)) { indent = attempt; } } } if (indent) doc = doc.replace(RegExp("\\n" + indent, "g"), '\n'); if (!herecomment) doc = doc.replace(/^\n/, ''); return doc; }; Lexer.prototype.tagParameters = function() { var i, stack, tok, tokens; if (this.tag() !== ')') return this; stack = []; tokens = this.tokens; i = tokens.length; tokens[--i][0] = 'PARAM_END'; while (tok = tokens[--i]) { switch (tok[0]) { case ')': stack.push(tok); break; case '(': case 'CALL_START': if (stack.length) { stack.pop(); } else if (tok[0] === '(') { tok[0] = 'PARAM_START'; return this; } else { return this; } } } return this; }; Lexer.prototype.closeIndentation = function() { return this.outdentToken(this.indent); }; Lexer.prototype.balancedString = function(str, end) { var i, letter, match, prev, stack, _ref3; stack = [end]; for (i = 1, _ref3 = str.length; 1 <= _ref3 ? i < _ref3 : i > _ref3; 1 <= _ref3 ? i++ : i--) { switch (letter = str.charAt(i)) { case '\\': i++; continue; case end: stack.pop(); if (!stack.length) return str.slice(0, i + 1); end = stack[stack.length - 1]; continue; } if (end === '}' && (letter === '"' || letter === "'")) { stack.push(end = letter); } else if (end === '}' && letter === '/' && (match = HEREGEX.exec(str.slice(i)) || REGEX.exec(str.slice(i)))) { i += match[0].length - 1; } else if (end === '}' && letter === '{') { stack.push(end = '}'); } else if (end === '"' && prev === '#' && letter === '{') { stack.push(end = '}'); } prev = letter; } return this.error("missing " + (stack.pop()) + ", starting"); }; Lexer.prototype.interpolateString = function(str, options) { var expr, heredoc, i, inner, interpolated, len, letter, nested, pi, regex, tag, tokens, value, _len, _ref3, _ref4, _ref5; if (options == null) options = {}; heredoc = options.heredoc, regex = options.regex; tokens = []; pi = 0; i = -1; while (letter = str.charAt(i += 1)) { if (letter === '\\') { i += 1; continue; } if (!(letter === '#' && str.charAt(i + 1) === '{' && (expr = this.balancedString(str.slice(i + 1), '}')))) { continue; } if (pi < i) tokens.push(['NEOSTRING', str.slice(pi, i)]); inner = expr.slice(1, -1); if (inner.length) { nested = new Lexer().tokenize(inner, { line: this.line, rewrite: false }); nested.pop(); if (((_ref3 = nested[0]) != null ? _ref3[0] : void 0) === 'TERMINATOR') { nested.shift(); } if (len = nested.length) { if (len > 1) { nested.unshift(['(', '(', this.line]); nested.push([')', ')', this.line]); } tokens.push(['TOKENS', nested]); } } i += expr.length; pi = i + 1; } if ((i > pi && pi < str.length)) tokens.push(['NEOSTRING', str.slice(pi)]); if (regex) return tokens; if (!tokens.length) return this.token('STRING', '""'); if (tokens[0][0] !== 'NEOSTRING') tokens.unshift(['', '']); if (interpolated = tokens.length > 1) this.token('(', '('); for (i = 0, _len = tokens.length; i < _len; i++) { _ref4 = tokens[i], tag = _ref4[0], value = _ref4[1]; if (i) this.token('+', '+'); if (tag === 'TOKENS') { (_ref5 = this.tokens).push.apply(_ref5, value); } else { this.token('STRING', this.makeString(value, '"', heredoc)); } } if (interpolated) this.token(')', ')'); return tokens; }; Lexer.prototype.pair = function(tag) { var size, wanted; if (tag !== (wanted = last(this.ends))) { if ('OUTDENT' !== wanted) this.error("unmatched " + tag); this.indent -= size = last(this.indents); this.outdentToken(size, true); return this.pair(tag); } return this.ends.pop(); }; Lexer.prototype.token = function(tag, value) { return this.tokens.push([tag, value, this.line]); }; Lexer.prototype.tag = function(index, tag) { var tok; return (tok = last(this.tokens, index)) && (tag ? tok[0] = tag : tok[0]); }; Lexer.prototype.value = function(index, val) { var tok; return (tok = last(this.tokens, index)) && (val ? tok[1] = val : tok[1]); }; Lexer.prototype.unfinished = function() { var _ref3; return LINE_CONTINUER.test(this.chunk) || ((_ref3 = this.tag()) === '\\' || _ref3 === '.' || _ref3 === '?.' || _ref3 === 'UNARY' || _ref3 === 'MATH' || _ref3 === '+' || _ref3 === '-' || _ref3 === 'SHIFT' || _ref3 === 'RELATION' || _ref3 === 'COMPARE' || _ref3 === 'LOGIC' || _ref3 === 'COMPOUND_ASSIGN' || _ref3 === 'THROW' || _ref3 === 'EXTENDS'); }; Lexer.prototype.escapeLines = function(str, heredoc) { return str.replace(MULTILINER, heredoc ? '\\n' : ''); }; Lexer.prototype.makeString = function(body, quote, heredoc) { if (!body) return quote + quote; body = body.replace(/\\([\s\S])/g, function(match, contents) { if (contents === '\n' || contents === quote) { return contents; } else { return match; } }); body = body.replace(RegExp("" + quote, "g"), '\\$&'); return quote + this.escapeLines(body, heredoc) + quote; }; Lexer.prototype.error = function(message) { throw SyntaxError("" + message + " on line " + (this.line + 1)); }; return Lexer; })(); JS_KEYWORDS = ['true', 'false', 'null', 'this', 'new', 'delete', 'typeof', 'in', 'instanceof', 'return', 'throw', 'break', 'continue', 'debugger', 'if', 'else', 'switch', 'for', 'while', 'do', 'try', 'catch', 'finally', 'class', 'extends', 'super']; COFFEE_KEYWORDS = ['undefined', 'then', 'unless', 'until', 'loop', 'of', 'by', 'when']; COFFEE_ALIAS_MAP = { and: '&&', or: '||', is: '==', isnt: '!=', not: '!', yes: 'true', no: 'false', on: 'true', off: 'false' }; COFFEE_ALIASES = (function() { var _results; _results = []; for (key in COFFEE_ALIAS_MAP) { _results.push(key); } return _results; })(); COFFEE_KEYWORDS = COFFEE_KEYWORDS.concat(COFFEE_ALIASES); RESERVED = ['case', 'default', 'function', 'var', 'void', 'with', 'const', 'let', 'enum', 'export', 'import', 'native', '__hasProp', '__extends', '__slice', '__bind', '__indexOf']; JS_FORBIDDEN = JS_KEYWORDS.concat(RESERVED); exports.RESERVED = RESERVED.concat(JS_KEYWORDS).concat(COFFEE_KEYWORDS); IDENTIFIER = /^([$A-Za-z_\x7f-\uffff][$\w\x7f-\uffff]*)([^\n\S]*:(?!:))?/; NUMBER = /^0x[\da-f]+|^0b[01]+|^\d*\.?\d+(?:e[+-]?\d+)?/i; HEREDOC = /^("""|''')([\s\S]*?)(?:\n[^\n\S]*)?\1/; OPERATOR = /^(?:[-=]>|[-+*\/%<>&|^!?=]=|>>>=?|([-+:])\1|([&|<>])\2=?|\?\.|\.{2,3})/; WHITESPACE = /^[^\n\S]+/; COMMENT = /^###([^#][\s\S]*?)(?:###[^\n\S]*|(?:###)?$)|^(?:\s*#(?!##[^#]).*)+/; CODE = /^[-=]>/; MULTI_DENT = /^(?:\n[^\n\S]*)+/; SIMPLESTR = /^'[^\\']*(?:\\.[^\\']*)*'/; JSTOKEN = /^`[^\\`]*(?:\\.[^\\`]*)*`/; REGEX = /^(\/(?![\s=])[^[\/\n\\]*(?:(?:\\[\s\S]|\[[^\]\n\\]*(?:\\[\s\S][^\]\n\\]*)*])[^[\/\n\\]*)*\/)([imgy]{0,4})(?!\w)/; HEREGEX = /^\/{3}([\s\S]+?)\/{3}([imgy]{0,4})(?!\w)/; HEREGEX_OMIT = /\s+(?:#.*)?/g; MULTILINER = /\n/g; HEREDOC_INDENT = /\n+([^\n\S]*)/g; HEREDOC_ILLEGAL = /\*\//; LINE_CONTINUER = /^\s*(?:,|\??\.(?![.\d])|::)/; TRAILING_SPACES = /\s+$/; COMPOUND_ASSIGN = ['-=', '+=', '/=', '*=', '%=', '||=', '&&=', '?=', '<<=', '>>=', '>>>=', '&=', '^=', '|=']; UNARY = ['!', '~', 'NEW', 'TYPEOF', 'DELETE', 'DO']; LOGIC = ['&&', '||', '&', '|', '^']; SHIFT = ['<<', '>>', '>>>']; COMPARE = ['==', '!=', '<', '>', '<=', '>=']; MATH = ['*', '/', '%']; RELATION = ['IN', 'OF', 'INSTANCEOF']; BOOL = ['TRUE', 'FALSE', 'NULL', 'UNDEFINED']; NOT_REGEX = ['NUMBER', 'REGEX', 'BOOL', '++', '--', ']']; NOT_SPACED_REGEX = NOT_REGEX.concat(')', '}', 'THIS', 'IDENTIFIER', 'STRING'); CALLABLE = ['IDENTIFIER', 'STRING', 'REGEX', ')', ']', '}', '?', '::', '@', 'THIS', 'SUPER']; INDEXABLE = CALLABLE.concat('NUMBER', 'BOOL'); LINE_BREAK = ['INDENT', 'OUTDENT', 'TERMINATOR']; }).call(this);