1 //<Script type="text/javascript">
3 Scope = imports.Scope.Scope;
4 TokenStream = imports.TokenStream.TokenStream;
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(">> 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());
256 print('SCOPE-VAR:' + 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:' + JSON.stringify(token,null,4));
317 //println("<i>"+token.data+"</i>");
318 var bn = this.braceNesting;
319 this.parseFunctionDeclaration();
320 this.braceNesting = bn;
323 case "PUNC.LEFT_CURLY": // {
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);
343 // it's an object literal..
344 // the values could be replaced..
348 print('SCOPE-CURLY:' + JSON.stringify(token,null,4));
350 token.items.forEach(function(expr) {
351 _this.ts = new TokenStream(expr);
352 _this.parseExpression(scope)
355 //print("NOT PROPS"); Seed.quit();
357 //isObjectLitAr.push(false);
358 //this.braceNesting++;
360 //print(">>>>>> OBJLIT PUSH(false)" + this.braceNesting);
363 case "PUNC.RIGHT_CURLY": // }
366 //println("<i>"+token.data+"</i>");
369 //print(">>>>>> OBJLIT POP"+ this.braceNesting);
370 //assert braceNesting >= scope.getBra ceNesting();
372 if (this.braceNesting < expressionBraceNesting) {
373 var ls = this.scopes.pop();
375 // eat symbol if we are currently at {
376 if (this.ts.look(0).data == '{') {
380 print("<<<<<<<EXIT SCOPE" +this.scopes.length);
386 print('SCOPE-WITH:' + token.toString());
387 //println("<i>"+token.data+"</i>");
388 if (this.mode == "BUILDING_SYMBOL_TREE") {
389 // Inside a 'with' block, it is impossible to figure out
390 // statically whether a symbol is a local variable or an
391 // object member. As a consequence, the only thing we can
392 // do is turn the obfuscation off for the highest scope
393 // containing the 'with' block.
394 this.protectScopeFromObfuscation(scope);
395 this.warn("Using 'with' is not recommended." + (this.munge ? " Moreover, using 'with' reduces the level of compression!" : ""), true);
400 print('SCOPE-CATCH:' + token.toString());
401 //println("<i>"+token.data+"</i>");
405 case Token.SPECIALCOMMENT:
406 if (mode == BUILDING_SYMBOL_TREE) {
407 protectScopeFromObfuscation(scope);
408 this.warn("Using JScript conditional comments is not recommended." + (munge ? " Moreover, using JScript conditional comments reduces the level of compression." : ""), true);
413 case "STRN.DOUBLE_QUOTE": // used for object lit detection..
414 case "STRN.SINGLE_QUOTE":
415 print('SCOPE-STRING:' + token.toString());
416 //println("<i>"+token.data+"</i>");
418 if (this.ts.lookTok(-1).data == '{' && this.ts.lookTok(1).data == ':') {
419 // then we are in an object lit.. -> we need to flag the brace as such...
421 isObjectLitAr.push(true);
422 //print(">>>>>> OBJLIT REPUSH(true)");
424 isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
426 if (isInObjectLitAr && this.ts.lookTok(1).data == ':' &&
427 ( this.ts.lookTok(-1).data == '{' || this.ts.lookTok(-1).data == ':' )) {
428 // see if we can replace..
429 // remove the quotes..
430 // should do a bit more checking!!!! (what about wierd char's in the string..
431 var str = token.data.substring(1,token.data.length-1);
432 if (/^[a-z_]+$/i.test(str) && ScopeParser.idents.indexOf(str) < 0) {
445 print('SCOPE-NAME:' + token.toString());
446 //print("DEAL WITH NAME:");
449 // look for { ** : <- indicates obj literal.. ** this could occur with numbers ..
450 if ((this.ts.lookTok(-1).data == "{") && (this.ts.lookTok(1).data == ":")) {
452 isObjectLitAr.push(true);
453 //print(">>>>>> OBJLIT REPUSH(true)");
454 //println("<i>"+token.data+"</i>");
457 // print("DEAL WITH obj lit:");
458 isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
460 if (isInObjectLitAr && (this.ts.lookTok(1).data == ":") && (this.ts.lookTok(-1).data == ",")) {
461 // skip, it's an object lit key..
462 //println("<i>"+token.data+"</i>");
467 // skip anyting with "." before it..!!
469 if (this.ts.lookTok(-1).data == ".") {
470 // skip, it's an object prop.
471 //println("<i>"+token.data+"</i>");
474 //print("SYMBOL: " + token.toString());
477 if (this.mode == 'PASS2_SYMBOL_TREE') {
479 //println("GOT IDENT: -2 : " + this.ts.lookT(-2).toString() + " <BR> ..... -1 : " + this.ts.lookT(-1).toString() + " <BR> ");
481 //print ("MUNGE?" + symbol);
483 //println("GOT IDENT: <B>" + symbol + "</B><BR/>");
485 //println("GOT IDENT (2): <B>" + symbol + "</B><BR/>");
486 identifier = this.getIdentifier(symbol, scope);
488 if (identifier == false) {
489 // BUG!find out where builtin is defined...
490 if (symbol.length <= 3 && Scope.builtin.indexOf(symbol) < 0) {
491 // Here, we found an undeclared and un-namespaced symbol that is
492 // 3 characters or less in length. Declare it in the global scope.
493 // We don't need to declare longer symbols since they won't cause
494 // any conflict with other munged symbols.
495 this.globalScope.declareIdentifier(symbol, token);
496 this.warn("Found an undeclared symbol: " + symbol + ' (line:' + token.line + ')', true);
499 //println("GOT IDENT IGNORE(3): <B>" + symbol + "</B><BR/>");
501 token.identifier = identifier;
502 identifier.refcount++;
507 //println("<B>SID</B>");
509 if (token.type != 'KEYW') {
512 print('SCOPE-KEYW:' + token.toString());
513 // print("Check eval:");
517 if (this.mode == 'BUILDING_SYMBOL_TREE') {
519 if (symbol == "eval") {
520 // look back one and see if we can find a comment!!!
521 if (this.ts.look(-1).type == "COMM") {
522 // look for eval:var:noreplace\n
524 this.ts.look(-1).data.replace(/eval:var:([a-z_]+)/ig, function(m, a) {
526 var hi = _t.getIdentifier(a, scope);
527 // println("PROTECT "+a+" from munge" + (hi ? "FOUND" : "MISSING"));
529 // println("PROTECT "+a+" from munge");
539 this.protectScopeFromObfuscation(scope);
540 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);
552 //print("parseScope TOK : " + token.toString());
553 token = this.ts.nextTok();
554 //if (this.ts.nextT()) break;
557 //print("<<<<<<<EXIT SCOPE ERR?" +this.scopes.length);
561 parseExpression : function() {
563 // Parse the expression until we encounter a comma or a semi-colon
564 // in the same brace nesting, bracket nesting and paren nesting.
565 // Parse functions if any...
566 //println("<i>EXP</i><BR/>");
567 !this.debug || print("PARSE EXPR");
570 // for printing stuff..
579 var expressionBraceNesting = this.braceNesting + 0;
580 var bracketNesting = 0;
581 var parensNesting = 0;
583 var isObjectLitAr = [ false ];
585 currentScope = this.scopes[this.scopes.length-1];
588 //print(scopeIndent + ">> ENTER EXPRESSION" + this.expN);
589 while (token = this.ts.nextTok()) {
594 // moved out of loop?
595 currentScope = this.scopes[this.scopes.length-1];
597 var scopeIndent = '';
598 this.scopes.forEach(function() {
603 //this.dumpToken(token, this.scopes, this.braceNesting );
604 print('EXP' + token.toString());
607 //println("<i>"+token.data+"</i>");
608 //this.log("EXP:" + token.data);
609 switch (token.type) {
611 print("EXPR-PUNC:" + JSON.stringify(token, null, 4));
617 if (this.braceNesting == expressionBraceNesting &&
618 bracketNesting == 0 &&
619 parensNesting == 0) {
620 print("<< EXIT EXPRESSION");
628 case '{': //Token.LC:
629 isObjectLitAr.push(false);
632 ///print(">>>>> EXP PUSH(false)"+this.braceNesting);
635 case '}': //Token.RC:
638 //print(">>>>> EXP POP" + this.braceNesting);
639 // assert braceNesting >= expressionBraceNesting;
642 case '[': //Token.LB:
646 token.items.forEach(function(ns) {
647 print("EXPR-PUNC-MAKE:" + JSON.stringify(ns, null, 4));
648 _this.ts = new TokenStream(ns);
650 _this.parseExpression();
656 case ']': //Token.RB:
660 case '(': //Token.LP:
664 case ')': //Token.RP:
670 case 'STRN': // used for object lit detection..
671 print("EXPR-STR:" + JSON.stringify(token, null, 4));
672 if (this.ts.lookTok(-1).data == "{" && this.ts.lookTok(1).data == ":" ) {
673 // then we are in an object lit.. -> we need to flag the brace as such...
675 isObjectLitAr.push(true);
676 //print(">>>>> EXP PUSH(true)");
681 isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
682 if (isInObjectLitAr && this.ts.lookTok(1).data == ":" &&
683 ( this.ts.lookTok(-1).data == "{" || this.ts.lookTok(-1).data == "," )) {
684 // see if we can replace..
685 // remove the quotes..
686 var str = token.data.substring(1,token.data.length-1);
687 if (/^[a-z_]+$/i.test(str) && ScopeParser.idents.indexOf(str) < 0) {
700 print("EXPR-NAME:" + JSON.stringify(token, null, 4));
702 //print("in NAME = " + token.toString());
703 //print("in NAME 0: " + this.ts.look(0).toString());
704 //print("in NAME 2: " + this.ts.lookTok(2).toString());
705 if (this.ts.look(0).data == "{" && this.ts.lookTok(2).data == ":") {
706 // then we are in an object lit.. -> we need to flag the brace as such...
708 isObjectLitAr.push(true);
709 //print(">>>>> EXP PUSH(true)");
713 isInObjectLitAr = isObjectLitAr[isObjectLitAr.length-1];
714 //print ("isInObjectLitAr : " + isInObjectLitAr + ' ' + token.toString());
716 if (isInObjectLitAr && this.ts.lookTok(0).data == "," && this.ts.lookTok(2).data == ":") {
719 //print(this.ts.lookTok(0).data);
720 if (this.ts.lookTok(0).data == ".") {
725 if (this.mode == 'PASS2_SYMBOL_TREE') {
727 identifier = this.getIdentifier(symbol, currentScope);
728 //println("<B>??</B>");
729 if (identifier == false) {
731 if (symbol.length <= 3 && Scope.builtin.indexOf(symbol) < 0) {
732 // Here, we found an undeclared and un-namespaced symbol that is
733 // 3 characters or less in length. Declare it in the global scope.
734 // We don't need to declare longer symbols since they won't cause
735 // any conflict with other munged symbols.
736 this.globalScope.declareIdentifier(symbol, token);
737 this.warn("Found an undeclared symbol: " + symbol + ' (line:' + token.line + ')', true);
739 //println("undeclared")
744 //println("<B>++</B>");
745 token.identifier = identifier;
746 identifier.refcount++;
755 //println("<B>EID</B>");
757 print("EXPR-KEYW:" + JSON.stringify(token, null, 4));
758 if (token.name == "FUNCTION") {
760 this.parseFunctionDeclaration();
767 if (this.mode == 'BUILDING_SYMBOL_TREE') {
769 if (symbol == "eval") {
770 if (this.ts.look(-1).type == 'COMM') {
771 // look for eval:var:noreplace\n
773 this.ts.look(-1).data.replace(/eval:var:([a-z]+)/ig, function(m, a) {
774 var hi = _t.getIdentifier(a, currentScope);
775 //println("PROTECT "+a+" from munge" + (hi ? "FOUND" : "MISSING"));
777 // println("PROTECT "+a+" from munge");
785 this.protectScopeFromObfuscation(currentScope);
786 this.warn("Using 'eval' is not recommended." + (this.munge ? " Moreover, using 'eval' reduces the level of compression!" : ""), true);
794 print("EXPR-SKIP:" + JSON.stringify(token, null, 4));
799 print("<< EXIT EXPRESSION");
804 parseCatch : function() {
811 //token = getToken(-1);
812 //assert token.getType() == Token.CATCH;
813 token = this.ts.nextTok();
814 //assert token.getType() == Token.LP; (
815 token = this.ts.nextTok();
816 //assert token.getType() == Token.NAME;
819 currentScope = this.scopes[this.scopes.length-1];
821 if (this.mode == 'BUILDING_SYMBOL_TREE') {
822 // We must declare the exception identifier in the containing function
823 // scope to avoid errors related to the obfuscation process. No need to
824 // display a warning if the symbol was already declared here...
825 currentScope.declareIdentifier(symbol, token);
827 //?? why inc the refcount?? - that should be set when building the tree???
828 identifier = this.getIdentifier(symbol, currentScope);
829 identifier.refcount++;
832 token = this.ts.nextTok();
833 //assert token.getType() == Token.RP; // )
836 parseFunctionDeclaration : function()
838 //print("PARSE FUNCTION");
841 var currentScope = false;
844 var b4braceNesting = this.braceNesting + 0;
846 //this.logR("<B>PARSING FUNCTION</B>");
847 currentScope = this.scopes[this.scopes.length-1];
849 token = this.ts.nextTok();
850 if (token.type == "NAME") {
851 if (this.mode == 'BUILDING_SYMBOL_TREE') {
852 // Get the name of the function and declare it in the current scope.
854 if (currentScope.getIdentifier(symbol) != false) {
855 this.warn("The function " + symbol + " has already been declared in the same scope...", true);
857 currentScope.declareIdentifier(symbol,token);
859 token = this.ts.nextTok();
862 //assert token.getType() == Token.LP;
863 if (this.mode == 'BUILDING_SYMBOL_TREE') {
864 fnScope = new Scope(this.braceNesting, currentScope, token.n, '');
866 //println("STORING SCOPE" + this.ts.cursor);
868 this.indexedScopes[this.ts.cursor] = fnScope;
871 //qln("FETCHING SCOPE" + this.ts.cursor);
872 fnScope = this.indexedScopes[this.ts.cursor];
876 print('FUNC-PARSE:' + JSON.stringify(token,null,4));
877 // Parse function arguments.
878 var args = token.items;
879 for (var argpos =0; argpos < args.length; argpos++) {
881 token = args[argpos][0];
882 // print ("FUNC ARGS: " + token.toString())
883 //assert token.getType() == Token.NAME ||
884 // token.getType() == Token.COMMA;
885 if (token.type == 'NAME' && this.mode == 'BUILDING_SYMBOL_TREE') {
887 identifier = fnScope.declareIdentifier(symbol,token);
888 if (symbol == "$super" && argpos == 0) {
889 // Exception for Prototype 1.6...
890 identifier.preventMunging();
896 token = this.ts.nextTok();
897 //print('FUNC-BODY:' + JSON.stringify(token.items,null,4));
899 //print(token.toString());
900 // assert token.getType() == Token.LC;
901 //this.braceNesting++;
903 //token = this.ts.nextTok();
904 //print(token.toString());
907 token.items.forEach(function(tar) {
908 _this.ts = new TokenStream(tar);
909 _this.parseScope(fnScope);
913 //print(JSON.stringify(this.ts,null,4));
914 //this.parseScope(fnScope);
916 // now pop it off the stack!!!
918 //this.braceNesting = b4braceNesting;
919 //print("ENDFN -1: " + this.ts.lookTok(-1).toString());
920 //print("ENDFN 0: " + this.ts.lookTok(0).toString());
921 //print("ENDFN 1: " + this.ts.lookTok(1).toString());
924 protectScopeFromObfuscation : function(scope) {
925 //assert scope != null;
927 if (scope == this.globalScope) {
928 // The global scope does not get obfuscated,
929 // so we don't need to worry about it...
933 // Find the highest local scope containing the specified scope.
934 while (scope && scope.parent != this.globalScope) {
935 scope = scope.parent;
938 //assert scope.getParentScope() == globalScope;
939 scope.preventMunging();
942 getIdentifier: function(symbol, scope) {
944 while (scope != false) {
945 identifier = scope.getIdentifier(symbol);
946 //println("ScopeParser.getIdentgetUsedSymbols("+symbol+")=" + scope.getUsedSymbols().join(','));
950 scope = scope.parent;