1 //<Script type="text/javascript">
3 Scope = imports['JSDOC/Scope.js'].Scope;
8 * // FIXME - I need this to do next() without doccomments..
11 ScopeParser = function(ts) {
12 this.ts = ts; // {TokenStream}
15 this.indexedScopes = {};
16 this.timer = new Date() * 1;
20 // list of keywords that should not be used in object literals.
21 ScopeParser.idents = [
87 ScopeParser.prototype = {
89 timerPrint: function (str) {
90 var ntime = new Date() * 1;
91 var tdif = ntime -this.timer;
94 if (tdif > 100) { //slower ones..
97 println(pref+'['+tdif+']'+str);
101 this.warnings.push(s);
102 //println("WARNING:" + htmlescape(s) + "<BR>");
104 // defaults should not be initialized here =- otherwise they get duped on new, rather than initalized..
109 mode : "", //"BUILDING_SYMBOL_TREE",
111 indexedScopes : false,
118 buildSymbolTree : function()
123 this.braceNesting = 0;
129 this.globalScope = new Scope(-1, false, -1, '');
130 indexedScopes = { 0 : this.globalScope };
132 this.mode = 'BUILDING_SYMBOL_TREE';
133 this.parseScope(this.globalScope);
135 mungeSymboltree : function()
142 // One problem with obfuscation resides in the use of undeclared
143 // and un-namespaced global symbols that are 3 characters or less
144 // in length. Here is an example:
146 // var declaredGlobalVar;
148 // function declaredGlobalFn() {
150 // localvar = abc; // abc is an undeclared global symbol
153 // In the example above, there is a slim chance that localvar may be
154 // munged to 'abc', conflicting with the undeclared global symbol
155 // abc, creating a potential bug. The following code detects such
156 // global symbols. This must be done AFTER the entire file has been
157 // parsed, and BEFORE munging the symbol tree. Note that declaring
158 // extra symbols in the global scope won't hurt.
160 // Note: Since we go through all the tokens to do this, we also use
161 // the opportunity to count how many times each identifier is used.
164 this.braceNesting = 0;
166 this.mode = 'CHECKING_SYMBOL_TREE';
168 //println("MUNGING?");
170 this.parseScope(this.globalScope);
171 this.globalScope.munge();
177 //println("<B>LOG:</B>" + htmlescape(str) + "<BR/>\n");
181 //println("<B>LOG:</B>" + str + "<BR/>");
189 parseScope : function(scope) // parse a token stream..
191 //this.timerPrint("parseScope EnterScope");
197 var expressionBraceNesting = this.braceNesting;
198 var bracketNesting = 0;
199 var parensNesting = 0;
201 var isObjectLitAr = [ false ];
203 this.scopes.push(scope);
204 token = this.ts.lookTok();
206 // this.timerPrint("parseScope AFTER lookT: " + token.toString());
208 print(token.type + '.' + token.name+ ":" + token.data);
209 switch(token.type + '.' + token.name) {
211 case "KEYW.CONST": // not really relivant as it's only mozzy that does this.
213 //this.log("parseScope GOT VAR/CONST : " + token.toString());
215 token = this.ts.nextTok();
217 if (token.name == "var") { // kludge..
220 if (!token) { // can return false at EOF!
223 //this.logR("parseScope GOT VAR : <B>" + token.toString() + "</B>");
224 if (!token.type =="NAME") {
225 println(token.toString());
226 throw "var without ident";
230 if (this.mode == "BUILDING_SYMBOL_TREE") {
231 identifier = scope.getIdentifier(token.data) ;
233 if (identifier == false) {
234 scope.declareIdentifier(token.data, token);
236 token.identifier = identifier;
237 this.warn("The variable " + symbol + " has already been declared in the same scope...");
241 token = this.ts.nextTok();
243 assert token.getType() == Token.SEMI ||
244 token.getType() == Token.ASSIGN ||
245 token.getType() == Token.COMMA ||
246 token.getType() == Token.IN;
248 if (token.name == "IN") {
251 this.parseExpression();
252 //this.logR("parseScope DONE : <B>ParseExpression</B> - tok is:" + this.ts.lookT(0).toString());
255 if (this.ts.lookTok(0).data == ';') {
261 case "KEYW.FUNCTION":
262 //println("<i>"+token.data+"</i>");
263 this.parseFunctionDeclaration();
266 case "KEYW.LEFT_CURLY": // {
267 //println("<i>"+token.data+"</i>");
268 isObjectLitAr.push(false);
272 case "KEYW.RIGHT_CURLY": // }
273 //println("<i>"+token.data+"</i>");
276 //assert braceNesting >= scope.getBraceNesting();
277 if (this.braceNesting == scope.braceN) {
278 var ls = this.scopes.pop();
285 //println("<i>"+token.data+"</i>");
286 if (this.mode == "BUILDING_SYMBOL_TREE") {
287 // Inside a 'with' block, it is impossible to figure out
288 // statically whether a symbol is a local variable or an
289 // object member. As a consequence, the only thing we can
290 // do is turn the obfuscation off for the highest scope
291 // containing the 'with' block.
292 this.protectScopeFromObfuscation(scope);
293 this.warn("Using 'with' is not recommended." + (this.munge ? " Moreover, using 'with' reduces the level of compression!" : ""), true);
298 //println("<i>"+token.data+"</i>");
302 case Token.SPECIALCOMMENT:
303 if (mode == BUILDING_SYMBOL_TREE) {
304 protectScopeFromObfuscation(scope);
305 this.warn("Using JScript conditional comments is not recommended." + (munge ? " Moreover, using JScript conditional comments reduces the level of compression." : ""), true);
310 case "STRN.DOUBLE_QUOTE": // used for object lit detection..
311 case "STRN.SINGLE_QUOTE":
312 //println("<i>"+token.data+"</i>");
313 if (this.ts.lookTok(-1).data == '{' && this.ts.lookTok(1).data == ':') {
314 // then we are in an object lit.. -> we need to flag the brace as such...
316 isObjectLitAr.push(true);
318 var isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
320 if (isInObjectLitAr && this.ts.lookTok(1).data == ':' &&
321 ( this.ts.lookTok(-1).data == '{' || this.ts.lookTok(-1).data == ':' )) {
322 // see if we can replace..
323 // remove the quotes..
324 // should do a bit more checking!!!! (what about wierd char's in the string..
325 var str = token.data.substring(1,token.data.length-1);
326 if (/^[a-z_]+$/i.test(str) && ScopeParser.idents.indexOf(str) < 0) {
340 if ((token.type != "NAME") && (token.type != "KEYW")){
345 // look for { ** : <- indicates obj literal.. ** this could occur with numbers ..
346 if ((this.ts.lookTok(-1).data == "{") && (this.ts.lookTok(1).data == ":")) {
348 isObjectLitAr.push(true);
349 //println("<i>"+token.data+"</i>");
352 var isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
354 if (isInObjectLitAr && (this.ts.lookTok(1).data == ":") && (this.ts.lookTok(-1).data == ",")) {
355 // skip, it's an object lit key..
356 //println("<i>"+token.data+"</i>");
361 // skip anyting with "." before it..!!
363 if (this.ts.lookTok(-1).data(".")) {
364 // skip, it's an object prop.
365 //println("<i>"+token.data+"</i>");
369 //println("<B>SID</B>");
375 if (this.mode == 'BUILDING_SYMBOL_TREE') {
377 if (symbol == "eval") {
378 // look back one and see if we can find a comment!!!
379 if (this.ts.look(-1).type == "COMM") {
380 // look for eval:var:noreplace\n
382 this.ts.look(-1).data.replace(/eval:var:([a-z_]+)/ig, function(m, a) {
384 var hi = _t.getIdentifier(a, scope);
385 // println("PROTECT "+a+" from munge" + (hi ? "FOUND" : "MISSING"));
387 // println("PROTECT "+a+" from munge");
397 this.protectScopeFromObfuscation(scope);
398 this.warn("Using 'eval' is not recommended. (use eval:var:noreplace in comments to optimize) " + (this.munge ? " Moreover, using 'eval' reduces the level of compression!" : ""), true);
403 } else if (this.mode == 'CHECKING_SYMBOL_TREE') {
405 //println("GOT IDENT: -2 : " + this.ts.lookT(-2).toString() + " <BR> ..... -1 : " + this.ts.lookT(-1).toString() + " <BR> ");
409 //println("GOT IDENT: <B>" + symbol + "</B><BR/>");
411 //println("GOT IDENT (2): <B>" + symbol + "</B><BR/>");
412 identifier = this.getIdentifier(symbol, scope);
414 if (identifier == false) {
415 // BUG!find out where builtin is defined...
416 if (symbol.length <= 3 && Scope.builtin.indexOf(symbol) < 0) {
417 // Here, we found an undeclared and un-namespaced symbol that is
418 // 3 characters or less in length. Declare it in the global scope.
419 // We don't need to declare longer symbols since they won't cause
420 // any conflict with other munged symbols.
421 this.globalScope.declareIdentifier(symbol, token);
422 this.warn("Found an undeclared symbol: " + symbol, true);
425 //println("GOT IDENT IGNORE(3): <B>" + symbol + "</B><BR/>");
427 token.identifier = identifier;
428 identifier.refcount++;
438 //this.timerPrint("parseScope TOK : " + token.toString());
439 token = this.ts.nextTok();
440 //if (this.ts.nextT()) break;
446 parseExpression : function() {
448 // Parse the expression until we encounter a comma or a semi-colon
449 // in the same brace nesting, bracket nesting and paren nesting.
450 // Parse functions if any...
451 //println("<i>EXP</i><BR/>");
457 var expressionBraceNesting = this.braceNesting;
458 var bracketNesting = 0;
459 var parensNesting = 0;
461 var isObjectLitAr = [ false ];
462 while (token = this.ts.lookTok()) {
466 currentScope = this.scopes[this.scopes.length-1];
468 //println("<i>"+token.data+"</i>");
470 switch (token.data.toUpperCase()) {
474 if (this.braceNesting == expressionBraceNesting &&
475 bracketNesting == 0 &&
476 parensNesting == 0) {
482 this.parseFunctionDeclaration();
485 case '{': //Token.LC:
486 isObjectLitAr.push(false);
491 case '}': //Token.RC:
495 // assert braceNesting >= expressionBraceNesting;
498 case '[': //Token.LB:
502 case ']': //Token.RB:
506 case '(': //Token.LP:
510 case ')': //Token.RP:
516 case 'STRN': // used for object lit detection..
517 if (this.ts.lookTok(-1).data == "{" && this.ts.lookTok(1).data == ":" ) {
518 // then we are in an object lit.. -> we need to flag the brace as such...
520 isObjectLitAr.push(true);
525 var isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
526 if (isInObjectLitAr && this.ts.lookTok(1).data == ":" &&
527 ( this.ts.lookTok(-1).data == "{" || this.ts.lookTok(-1).data == "," )) {
528 // see if we can replace..
529 // remove the quotes..
530 var str = token.data.substring(1,token.data.length-1);
531 if (/^[a-z_]+$/i.test(str) && ScopeParser.idents.indexOf(str) < 0) {
544 case Token.SPECIALCOMMENT:
545 if (mode == BUILDING_SYMBOL_TREE) {
546 protectScopeFromObfuscation(currentScope);
547 this.warn("Using JScript conditional comments is not recommended." + (munge ? " Moreover, using JScript conditional comments reduces the level of compression!" : ""), true);
554 if (this.ts.lookTok(-1).data == "{" && this.ts.lookTok(1).data == ":") {
555 // then we are in an object lit.. -> we need to flag the brace as such...
557 isObjectLitAr.push(true);
560 var isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
561 if (isInObjectLitAr && this.ts.lookTok(-1).data == "," && this.ts.lookTok(1).data == ":") {
565 if (this.ts.lookTok(-1).data == ".") {
572 //println("<B>EID</B>");
575 if (this.mode == 'BUILDING_SYMBOL_TREE') {
577 if (symbol == "eval") {
578 if (this.ts.look(-1).type == 'COMM') {
579 // look for eval:var:noreplace\n
581 this.ts.look(-1).data.replace(/eval:var:([a-z]+)/ig, function(m, a) {
582 var hi = _t.getIdentifier(a, currentScope);
583 //println("PROTECT "+a+" from munge" + (hi ? "FOUND" : "MISSING"));
585 // println("PROTECT "+a+" from munge");
593 this.protectScopeFromObfuscation(currentScope);
594 this.warn("Using 'eval' is not recommended." + (this.munge ? " Moreover, using 'eval' reduces the level of compression!" : ""), true);
601 if (this.mode == 'CHECKING_SYMBOL_TREE') {
603 identifier = this.getIdentifier(symbol, currentScope);
604 //println("<B>??</B>");
605 if (identifier == false) {
607 if (symbol.length <= 3 && Scope.builtin.indexOf(symbol) < 0) {
608 // Here, we found an undeclared and un-namespaced symbol that is
609 // 3 characters or less in length. Declare it in the global scope.
610 // We don't need to declare longer symbols since they won't cause
611 // any conflict with other munged symbols.
612 this.globalScope.declareIdentifier(symbol, token);
613 this.warn("Found an undeclared symbol: " + symbol, true);
615 //println("undeclared")
620 //println("<B>++</B>");
621 token.identifier = identifier;
622 identifier.refcount++;
628 if (!this.ts.nextTok()) break;
633 parseCatch : function() {
640 //token = getToken(-1);
641 //assert token.getType() == Token.CATCH;
642 token = this.ts.nextTok();
643 //assert token.getType() == Token.LP; (
644 token = this.ts.nextTok();
645 //assert token.getType() == Token.NAME;
648 currentScope = this.scopes[this.scopes.length-1];
650 if (this.mode == 'BUILDING_SYMBOL_TREE') {
651 // We must declare the exception identifier in the containing function
652 // scope to avoid errors related to the obfuscation process. No need to
653 // display a warning if the symbol was already declared here...
654 currentScope.declareIdentifier(symbol, token);
656 //?? why inc the refcount?? - that should be set when building the tree???
657 identifier = this.getIdentifier(symbol, currentScope);
658 identifier.refcount++;
661 token = this.ts.nextTok();
662 //assert token.getType() == Token.RP; // )
665 parseFunctionDeclaration : function()
670 var currentScope = false;
673 //this.logR("<B>PARSING FUNCTION</B>");
674 currentScope = this.scopes[this.scopes.length-1];
676 token = this.ts.nextTok();
677 if (token.type == "NAME") {
678 if (this.mode == 'BUILDING_SYMBOL_TREE') {
679 // Get the name of the function and declare it in the current scope.
681 if (currentScope.getIdentifier(symbol) != false) {
682 this.warn("The function " + symbol + " has already been declared in the same scope...", true);
684 currentScope.declareIdentifier(symbol,token);
686 token = this.ts.nextTok();
689 //assert token.getType() == Token.LP;
690 if (this.mode == 'BUILDING_SYMBOL_TREE') {
691 fnScope = new Scope(this.braceNesting, currentScope, token.n, '');
693 //println("STORING SCOPE" + this.ts.cursor);
695 this.indexedScopes[this.ts.cursor] = fnScope;
698 //println("FETCHING SCOPE" + this.ts.cursor);
699 fnScope = this.indexedScopes[this.ts.cursor];
702 // Parse function arguments.
704 while (!this.ts.lookTok().data == ')') { //(token = consumeToken()).getType() != Token.RP) {
705 token = this.ts.nextTok();
707 //assert token.getType() == Token.NAME ||
708 // token.getType() == Token.COMMA;
709 if (token.type == 'NAME' && this.mode == 'BUILDING_SYMBOL_TREE') {
711 identifier = fnScope.declareIdentifier(symbol,token);
712 if (symbol == "$super" && argpos == 0) {
713 // Exception for Prototype 1.6...
714 identifier.preventMunging();
720 token = this.ts.nextTok();
721 // assert token.getType() == Token.LC;
724 token = this.ts.nextTok();
725 if (token.type == "STRN" && this.ts.lookTok(1).data == ';') {
728 NOT SUPPORTED YET!?!!?!
730 // This is a hint. Hints are empty statements that look like
731 // "localvar1:nomunge, localvar2:nomunge"; They allow developers
732 // to prevent specific symbols from getting obfuscated (some heretic
733 // implementations, such as Prototype 1.6, require specific variable
734 // names, such as $super for example, in order to work appropriately.
735 // Note: right now, only "nomunge" is supported in the right hand side
736 // of a hint. However, in the future, the right hand side may contain
739 String hints = token.getValue();
740 // Remove the leading and trailing quotes...
741 hints = hints.substring(1, hints.length() - 1).trim();
742 StringTokenizer st1 = new StringTokenizer(hints, ",");
743 while (st1.hasMoreTokens()) {
744 String hint = st1.nextToken();
745 int idx = hint.indexOf(':');
746 if (idx <= 0 || idx >= hint.length() - 1) {
747 if (mode == BUILDING_SYMBOL_TREE) {
748 // No need to report the error twice, hence the test...
749 this.warn("Invalid hint syntax: " + hint, true);
753 String variableName = hint.substring(0, idx).trim();
754 String variableType = hint.substring(idx + 1).trim();
755 if (mode == BUILDING_SYMBOL_TREE) {
756 fnScope.addHint(variableName, variableType);
757 } else if (mode == CHECKING_SYMBOL_TREE) {
758 identifier = fnScope.getIdentifier(variableName);
759 if (identifier != null) {
760 if (variableType.equals("nomunge")) {
761 identifier.preventMunging();
763 this.warn("Unsupported hint value: " + hint, true);
766 this.warn("Hint refers to an unknown identifier: " + hint, true);
773 this.parseScope(fnScope);
774 // now pop it off the stack!!!
780 protectScopeFromObfuscation : function(scope) {
781 //assert scope != null;
783 if (scope == this.globalScope) {
784 // The global scope does not get obfuscated,
785 // so we don't need to worry about it...
789 // Find the highest local scope containing the specified scope.
790 while (scope && scope.parent != this.globalScope) {
791 scope = scope.parent;
794 //assert scope.getParentScope() == globalScope;
795 scope.preventMunging();
798 getIdentifier: function(symbol, scope) {
800 while (scope != false) {
801 identifier = scope.getIdentifier(symbol);
802 //println("ScopeParser.getIdentgetUsedSymbols("+symbol+")=" + scope.getUsedSymbols().join(','));
806 scope = scope.parent;