1 //<Script type="text/javascript">
3 Scope = imports['JSDOC/Scope.js'].Scope;
8 * // FIXME - I need this to do next() without doccomments..
12 * Need to make this alot simpler...
14 * so debugging is possible.
17 * at present it just runs along the stream and finds stuff then calls parseExpr .. etc,,
20 * It would be better to parse blocks of code rather than the whole stream..
28 ScopeParser = function(ts) {
29 this.ts = ts; // {TokenStream}
32 this.indexedScopes = {};
33 this.timer = new Date() * 1;
37 // list of keywords that should not be used in object literals.
38 ScopeParser.idents = [
104 ScopeParser.prototype = {
106 timerPrint: function (str) {
107 var ntime = new Date() * 1;
108 var tdif = ntime -this.timer;
111 if (tdif > 100) { //slower ones..
114 println(pref+'['+tdif+']'+str);
118 //print('****************' + s);
119 this.warnings.push(s);
120 //println("WARNING:" + htmlescape(s) + "<BR>");
122 // defaults should not be initialized here =- otherwise they get duped on new, rather than initalized..
127 mode : "", //"BUILDING_SYMBOL_TREE",
129 indexedScopes : false,
136 buildSymbolTree : function()
141 this.braceNesting = 0;
147 this.globalScope = new Scope(-1, false, -1, '');
148 indexedScopes = { 0 : this.globalScope };
150 this.mode = 'BUILDING_SYMBOL_TREE';
151 this.parseScope(this.globalScope);
153 //print("---------------END PASS 1 ---------------- ");
156 mungeSymboltree : function()
163 // One problem with obfuscation resides in the use of undeclared
164 // and un-namespaced global symbols that are 3 characters or less
165 // in length. Here is an example:
167 // var declaredGlobalVar;
169 // function declaredGlobalFn() {
171 // localvar = abc; // abc is an undeclared global symbol
174 // In the example above, there is a slim chance that localvar may be
175 // munged to 'abc', conflicting with the undeclared global symbol
176 // abc, creating a potential bug. The following code detects such
177 // global symbols. This must be done AFTER the entire file has been
178 // parsed, and BEFORE munging the symbol tree. Note that declaring
179 // extra symbols in the global scope won't hurt.
181 // Note: Since we go through all the tokens to do this, we also use
182 // the opportunity to count how many times each identifier is used.
185 this.braceNesting = 0;
187 this.mode = 'PASS2_SYMBOL_TREE';
189 //println("MUNGING?");
191 this.parseScope(this.globalScope);
192 this.globalScope.munge();
198 print (" ".substring(0, this.braceNesting*2) + str);
200 //println("<B>LOG:</B>" + htmlescape(str) + "<BR/>\n");
204 //println("<B>LOG:</B>" + str + "<BR/>");
211 parseScope : function(scope) // parse a token stream..
213 //this.timerPrint("parseScope EnterScope");
214 //this.log(">>> ENTER SCOPE" + this.scopes.length);
220 var expressionBraceNesting = this.braceNesting + 0;
222 var parensNesting = 0;
224 var isObjectLitAr = [ false ];
226 this.scopes.push(scope);
228 //var scopeIndent = '';
229 //this.scopes.forEach(function() {
230 // scopeIndent += ' ';
232 //print(scopeIndent + ">> ENTER SCOPE");
237 token = this.ts.lookTok(1);
239 // this.timerPrint("parseScope AFTER lookT: " + token.toString());
240 //this.dumpToken(token , this.scopes, this.braceNesting);
241 //print('SCOPE:' + token.toString());
242 //this.log(token.data);
243 if (token.type == 'NAME') {
244 // print('*' + token.data);
246 switch(token.type + '.' + token.name) {
248 case "KEYW.CONST": // not really relivant as it's only mozzy that does this.
249 print('SCOPE-VAR:' + token.toString());
250 var vstart = this.ts.cursor +1;
252 //this.log("parseScope GOT VAR/CONST : " + token.toString());
254 token = this.ts.nextTok();
255 !this.debug|| print( token.toString());
257 if (!token) { // can return false at EOF!
260 if (token.name == "VAR" || token.data == ',') { // kludge..
263 //this.logR("parseScope GOT VAR : <B>" + token.toString() + "</B>");
264 if (token.type !="NAME") {
265 for(var i = Math.max(this.ts.cursor-10,0); i < this.ts.cursor+1; i++) {
266 print(this.ts.tokens[i].toString());
269 print( "var without ident");
274 if (this.mode == "BUILDING_SYMBOL_TREE") {
275 identifier = scope.getIdentifier(token.data) ;
277 if (identifier == false) {
278 scope.declareIdentifier(token.data, token);
280 token.identifier = identifier;
281 this.warn("(SCOPE) The variable " + token.data + ' (line:' + token.line + ") has already been declared in the same scope...");
285 token = this.ts.nextTok();
286 !this.debug|| print(token.toString());
288 assert token.getType() == Token.SEMI ||
289 token.getType() == Token.ASSIGN ||
290 token.getType() == Token.COMMA ||
291 token.getType() == Token.IN;
293 if (token.name == "IN") {
296 //var bn = this.braceNesting;
297 var bn = this.braceNesting;
298 this.parseExpression();
299 this.braceNesting = bn;
300 //this.braceNesting = bn;
301 //this.logR("parseScope DONE : <B>ParseExpression</B> - tok is:" + this.ts.lookT(0).toString());
303 token = this.ts.lookTok(1);
304 !this.debug|| print("AFTER EXP: " + token.toString());
305 if (token.data == ';') {
312 //this.ts.dump(vstart , this.ts.cursor);
315 case "KEYW.FUNCTION":
316 print('SCOPE-FUNC:' + token.toString());
317 //println("<i>"+token.data+"</i>");
318 var bn = this.braceNesting;
319 this.parseFunctionDeclaration();
320 this.braceNesting = bn;
323 case "PUNC.LEFT_CURLY": // {
324 print('SCOPE-CURLY:' + JSON.stringify(token,null,4));
325 //println("<i>"+token.data+"</i>");
329 for (var prop in token.props) {
330 if (token.props[prop].val.data == 'function') {
331 // parse a function..
332 this.parseFunctProp(token.props[prop]);
337 this.ts = new TokenStream(token.props[prop].val);
338 this.parseScope(scope);
344 // it's an object literal..
345 // the values could be replaced..
349 isObjectLitAr.push(false);
352 //print(">>>>>> OBJLIT PUSH(false)" + this.braceNesting);
355 case "PUNC.RIGHT_CURLY": // }
356 //println("<i>"+token.data+"</i>");
359 //print(">>>>>> OBJLIT POP"+ this.braceNesting);
360 //assert braceNesting >= scope.getBra ceNesting();
362 if (this.braceNesting < expressionBraceNesting) {
363 var ls = this.scopes.pop();
365 // eat symbol if we are currently at {
366 if (this.ts.look(0).data == '{') {
370 print("<<<<<<<EXIT SCOPE" +this.scopes.length);
376 print('SCOPE-WITH:' + token.toString());
377 //println("<i>"+token.data+"</i>");
378 if (this.mode == "BUILDING_SYMBOL_TREE") {
379 // Inside a 'with' block, it is impossible to figure out
380 // statically whether a symbol is a local variable or an
381 // object member. As a consequence, the only thing we can
382 // do is turn the obfuscation off for the highest scope
383 // containing the 'with' block.
384 this.protectScopeFromObfuscation(scope);
385 this.warn("Using 'with' is not recommended." + (this.munge ? " Moreover, using 'with' reduces the level of compression!" : ""), true);
390 print('SCOPE-CATCH:' + token.toString());
391 //println("<i>"+token.data+"</i>");
395 case Token.SPECIALCOMMENT:
396 if (mode == BUILDING_SYMBOL_TREE) {
397 protectScopeFromObfuscation(scope);
398 this.warn("Using JScript conditional comments is not recommended." + (munge ? " Moreover, using JScript conditional comments reduces the level of compression." : ""), true);
403 case "STRN.DOUBLE_QUOTE": // used for object lit detection..
404 case "STRN.SINGLE_QUOTE":
405 print('SCOPE-STRING:' + token.toString());
406 //println("<i>"+token.data+"</i>");
408 if (this.ts.lookTok(-1).data == '{' && this.ts.lookTok(1).data == ':') {
409 // then we are in an object lit.. -> we need to flag the brace as such...
411 isObjectLitAr.push(true);
412 //print(">>>>>> OBJLIT REPUSH(true)");
414 isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
416 if (isInObjectLitAr && this.ts.lookTok(1).data == ':' &&
417 ( this.ts.lookTok(-1).data == '{' || this.ts.lookTok(-1).data == ':' )) {
418 // see if we can replace..
419 // remove the quotes..
420 // should do a bit more checking!!!! (what about wierd char's in the string..
421 var str = token.data.substring(1,token.data.length-1);
422 if (/^[a-z_]+$/i.test(str) && ScopeParser.idents.indexOf(str) < 0) {
435 print('SCOPE-NAME:' + token.toString());
436 //print("DEAL WITH NAME:");
439 // look for { ** : <- indicates obj literal.. ** this could occur with numbers ..
440 if ((this.ts.lookTok(-1).data == "{") && (this.ts.lookTok(1).data == ":")) {
442 isObjectLitAr.push(true);
443 //print(">>>>>> OBJLIT REPUSH(true)");
444 //println("<i>"+token.data+"</i>");
447 // print("DEAL WITH obj lit:");
448 isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
450 if (isInObjectLitAr && (this.ts.lookTok(1).data == ":") && (this.ts.lookTok(-1).data == ",")) {
451 // skip, it's an object lit key..
452 //println("<i>"+token.data+"</i>");
457 // skip anyting with "." before it..!!
459 if (this.ts.lookTok(-1).data == ".") {
460 // skip, it's an object prop.
461 //println("<i>"+token.data+"</i>");
464 //print("SYMBOL: " + token.toString());
467 if (this.mode == 'PASS2_SYMBOL_TREE') {
469 //println("GOT IDENT: -2 : " + this.ts.lookT(-2).toString() + " <BR> ..... -1 : " + this.ts.lookT(-1).toString() + " <BR> ");
471 //print ("MUNGE?" + symbol);
473 //println("GOT IDENT: <B>" + symbol + "</B><BR/>");
475 //println("GOT IDENT (2): <B>" + symbol + "</B><BR/>");
476 identifier = this.getIdentifier(symbol, scope);
478 if (identifier == false) {
479 // BUG!find out where builtin is defined...
480 if (symbol.length <= 3 && Scope.builtin.indexOf(symbol) < 0) {
481 // Here, we found an undeclared and un-namespaced symbol that is
482 // 3 characters or less in length. Declare it in the global scope.
483 // We don't need to declare longer symbols since they won't cause
484 // any conflict with other munged symbols.
485 this.globalScope.declareIdentifier(symbol, token);
486 this.warn("Found an undeclared symbol: " + symbol + ' (line:' + token.line + ')', true);
489 //println("GOT IDENT IGNORE(3): <B>" + symbol + "</B><BR/>");
491 token.identifier = identifier;
492 identifier.refcount++;
497 //println("<B>SID</B>");
499 if (token.type != 'KEYW') {
502 print('SCOPE-KEYW:' + token.toString());
503 // print("Check eval:");
507 if (this.mode == 'BUILDING_SYMBOL_TREE') {
509 if (symbol == "eval") {
510 // look back one and see if we can find a comment!!!
511 if (this.ts.look(-1).type == "COMM") {
512 // look for eval:var:noreplace\n
514 this.ts.look(-1).data.replace(/eval:var:([a-z_]+)/ig, function(m, a) {
516 var hi = _t.getIdentifier(a, scope);
517 // println("PROTECT "+a+" from munge" + (hi ? "FOUND" : "MISSING"));
519 // println("PROTECT "+a+" from munge");
529 this.protectScopeFromObfuscation(scope);
530 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);
542 //print("parseScope TOK : " + token.toString());
543 token = this.ts.nextTok();
544 //if (this.ts.nextT()) break;
547 //print("<<<<<<<EXIT SCOPE ERR?" +this.scopes.length);
551 parseExpression : function() {
553 // Parse the expression until we encounter a comma or a semi-colon
554 // in the same brace nesting, bracket nesting and paren nesting.
555 // Parse functions if any...
556 //println("<i>EXP</i><BR/>");
557 !this.debug || print("PARSE EXPR");
560 // for printing stuff..
569 var expressionBraceNesting = this.braceNesting + 0;
570 var bracketNesting = 0;
571 var parensNesting = 0;
573 var isObjectLitAr = [ false ];
575 currentScope = this.scopes[this.scopes.length-1];
577 var scopeIndent = '';
578 this.scopes.forEach(function() {
581 //print(scopeIndent + ">> ENTER EXPRESSION" + this.expN);
582 while (token = this.ts.lookTok()) {
587 // moved out of loop?
588 currentScope = this.scopes[this.scopes.length-1];
590 var scopeIndent = '';
591 this.scopes.forEach(function() {
596 //this.dumpToken(token, this.scopes, this.braceNesting );
597 //print('EXP' + this.expN + ':' + token.toString());
600 //println("<i>"+token.data+"</i>");
601 //this.log("EXP:" + token.data);
602 switch (token.type) {
608 if (this.braceNesting == expressionBraceNesting &&
609 bracketNesting == 0 &&
610 parensNesting == 0) {
611 print(scopeIndent + "<< EXIT EXPRESSION");
619 case '{': //Token.LC:
620 isObjectLitAr.push(false);
623 ///print(">>>>> EXP PUSH(false)"+this.braceNesting);
626 case '}': //Token.RC:
629 //print(">>>>> EXP POP" + this.braceNesting);
630 // assert braceNesting >= expressionBraceNesting;
633 case '[': //Token.LB:
637 case ']': //Token.RB:
641 case '(': //Token.LP:
645 case ')': //Token.RP:
651 case 'STRN': // used for object lit detection..
652 if (this.ts.lookTok(-1).data == "{" && this.ts.lookTok(1).data == ":" ) {
653 // then we are in an object lit.. -> we need to flag the brace as such...
655 isObjectLitAr.push(true);
656 //print(">>>>> EXP PUSH(true)");
661 isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
662 if (isInObjectLitAr && this.ts.lookTok(1).data == ":" &&
663 ( this.ts.lookTok(-1).data == "{" || this.ts.lookTok(-1).data == "," )) {
664 // see if we can replace..
665 // remove the quotes..
666 var str = token.data.substring(1,token.data.length-1);
667 if (/^[a-z_]+$/i.test(str) && ScopeParser.idents.indexOf(str) < 0) {
682 //print("in NAME = " + token.toString());
683 //print("in NAME 0: " + this.ts.look(0).toString());
684 //print("in NAME 2: " + this.ts.lookTok(2).toString());
685 if (this.ts.look(0).data == "{" && this.ts.lookTok(2).data == ":") {
686 // then we are in an object lit.. -> we need to flag the brace as such...
688 isObjectLitAr.push(true);
689 //print(">>>>> EXP PUSH(true)");
693 isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
694 //print ("isInObjectLitAr : " + isInObjectLitAr + ' ' + token.toString());
696 if (isInObjectLitAr && this.ts.lookTok(0).data == "," && this.ts.lookTok(2).data == ":") {
699 //print(this.ts.lookTok(0).data);
700 if (this.ts.lookTok(0).data == ".") {
705 if (this.mode == 'PASS2_SYMBOL_TREE') {
707 identifier = this.getIdentifier(symbol, currentScope);
708 //println("<B>??</B>");
709 if (identifier == false) {
711 if (symbol.length <= 3 && Scope.builtin.indexOf(symbol) < 0) {
712 // Here, we found an undeclared and un-namespaced symbol that is
713 // 3 characters or less in length. Declare it in the global scope.
714 // We don't need to declare longer symbols since they won't cause
715 // any conflict with other munged symbols.
716 this.globalScope.declareIdentifier(symbol, token);
717 this.warn("Found an undeclared symbol: " + symbol + ' (line:' + token.line + ')', true);
719 //println("undeclared")
724 //println("<B>++</B>");
725 token.identifier = identifier;
726 identifier.refcount++;
735 //println("<B>EID</B>");
738 if (token.name == "FUNCTION") {
740 this.parseFunctionDeclaration();
747 if (this.mode == 'BUILDING_SYMBOL_TREE') {
749 if (symbol == "eval") {
750 if (this.ts.look(-1).type == 'COMM') {
751 // look for eval:var:noreplace\n
753 this.ts.look(-1).data.replace(/eval:var:([a-z]+)/ig, function(m, a) {
754 var hi = _t.getIdentifier(a, currentScope);
755 //println("PROTECT "+a+" from munge" + (hi ? "FOUND" : "MISSING"));
757 // println("PROTECT "+a+" from munge");
765 this.protectScopeFromObfuscation(currentScope);
766 this.warn("Using 'eval' is not recommended." + (this.munge ? " Moreover, using 'eval' reduces the level of compression!" : ""), true);
775 if (!this.ts.nextTok()) break;
777 print(scopeIndent + "<< EXIT EXPRESSION");
782 parseCatch : function() {
789 //token = getToken(-1);
790 //assert token.getType() == Token.CATCH;
791 token = this.ts.nextTok();
792 //assert token.getType() == Token.LP; (
793 token = this.ts.nextTok();
794 //assert token.getType() == Token.NAME;
797 currentScope = this.scopes[this.scopes.length-1];
799 if (this.mode == 'BUILDING_SYMBOL_TREE') {
800 // We must declare the exception identifier in the containing function
801 // scope to avoid errors related to the obfuscation process. No need to
802 // display a warning if the symbol was already declared here...
803 currentScope.declareIdentifier(symbol, token);
805 //?? why inc the refcount?? - that should be set when building the tree???
806 identifier = this.getIdentifier(symbol, currentScope);
807 identifier.refcount++;
810 token = this.ts.nextTok();
811 //assert token.getType() == Token.RP; // )
814 parseFunctionDeclaration : function()
816 //print("PARSE FUNCTION");
819 var currentScope = false;
822 var b4braceNesting = this.braceNesting + 0;
824 //this.logR("<B>PARSING FUNCTION</B>");
825 currentScope = this.scopes[this.scopes.length-1];
827 token = this.ts.nextTok();
828 if (token.type == "NAME") {
829 if (this.mode == 'BUILDING_SYMBOL_TREE') {
830 // Get the name of the function and declare it in the current scope.
832 if (currentScope.getIdentifier(symbol) != false) {
833 this.warn("The function " + symbol + " has already been declared in the same scope...", true);
835 currentScope.declareIdentifier(symbol,token);
837 token = this.ts.nextTok();
840 //assert token.getType() == Token.LP;
841 if (this.mode == 'BUILDING_SYMBOL_TREE') {
842 fnScope = new Scope(this.braceNesting, currentScope, token.n, '');
844 //println("STORING SCOPE" + this.ts.cursor);
846 this.indexedScopes[this.ts.cursor] = fnScope;
849 //qln("FETCHING SCOPE" + this.ts.cursor);
850 fnScope = this.indexedScopes[this.ts.cursor];
854 // Parse function arguments.
856 while (this.ts.lookTok().data != ')') { //(token = consumeToken()).getType() != Token.RP) {
857 token = this.ts.nextTok();
858 // print ("FUNC ARGS: " + token.toString())
859 //assert token.getType() == Token.NAME ||
860 // token.getType() == Token.COMMA;
861 if (token.type == 'NAME' && this.mode == 'BUILDING_SYMBOL_TREE') {
863 identifier = fnScope.declareIdentifier(symbol,token);
864 if (symbol == "$super" && argpos == 0) {
865 // Exception for Prototype 1.6...
866 identifier.preventMunging();
872 token = this.ts.nextTok();
873 //print(token.toString());
874 // assert token.getType() == Token.LC;
877 token = this.ts.nextTok();
878 //print(token.toString());
881 this.parseScope(fnScope);
882 // now pop it off the stack!!!
884 this.braceNesting = b4braceNesting;
885 //print("ENDFN -1: " + this.ts.lookTok(-1).toString());
886 //print("ENDFN 0: " + this.ts.lookTok(0).toString());
887 //print("ENDFN 1: " + this.ts.lookTok(1).toString());
890 protectScopeFromObfuscation : function(scope) {
891 //assert scope != null;
893 if (scope == this.globalScope) {
894 // The global scope does not get obfuscated,
895 // so we don't need to worry about it...
899 // Find the highest local scope containing the specified scope.
900 while (scope && scope.parent != this.globalScope) {
901 scope = scope.parent;
904 //assert scope.getParentScope() == globalScope;
905 scope.preventMunging();
908 getIdentifier: function(symbol, scope) {
910 while (scope != false) {
911 identifier = scope.getIdentifier(symbol);
912 //println("ScopeParser.getIdentgetUsedSymbols("+symbol+")=" + scope.getUsedSymbols().join(','));
916 scope = scope.parent;