adaptive base54 digits depending on char frequency (WIP)

1 files changed, 110 insertions, 5 deletions

diff --git a/lib/scope.js b/lib/scope.js
index adf36493..e3433340 100644
--- a/lib/scope.js
+++ b/lib/scope.js
@@ -296,14 +296,19 @@ AST_Scope.DEFMETHOD("next_mangled", function(){
     }
 });
 
+AST_SymbolDeclaration.DEFMETHOD("unmangleable", function(){
+    return this.global || this.scope.uses_eval || this.scope.uses_with;
+});
+
+AST_Label.DEFMETHOD("unmangleable", function(){
+    return false;
+});
+
 AST_SymbolDeclaration.DEFMETHOD("mangle", function(){
-    if (this.uniq) {
+    if (this.uniq && this.uniq !== this) {
         this.uniq.mangle();
     }
-    else if (!(this.global
-               || this.scope.uses_eval
-               || this.scope.uses_with
-               || this.mangled_name)) {
+    else if (!(this.mangled_name || this.unmangleable())) {
         this.mangled_name = this.scope.next_mangled();
     }
 });
@@ -354,3 +359,103 @@ AST_Toplevel.DEFMETHOD("mangle_names", function(){
     });
     this.walk(tw);
 });
+
+AST_Toplevel.DEFMETHOD("compute_char_frequency", function(){
+    var tw = new TreeWalker(function(node){
+        if (node instanceof AST_Constant)
+            base54.consider(node.print_to_string());
+        else if (node instanceof AST_Debugger)
+            base54.consider("debugger");
+        else if (node instanceof AST_Directive)
+            base54.consider(node.value);
+        else if (node instanceof AST_While)
+            base54.consider("while");
+        else if (node instanceof AST_Do)
+            base54.consider("dowhile");
+        else if (node instanceof AST_If) {
+            base54.consider("if");
+            if (node.alternative) base54.consider("else");
+        }
+        else if (node instanceof AST_Var)
+            base54.consider("var");
+        else if (node instanceof AST_Const)
+            base54.consider("const");
+        else if (node instanceof AST_Lambda)
+            base54.consider("function");
+        else if (node instanceof AST_For)
+            base54.consider("for");
+        else if (node instanceof AST_ForIn)
+            base54.consider("forin");
+        else if (node instanceof AST_Switch)
+            base54.consider("switch");
+        else if (node instanceof AST_Case)
+            base54.consider("case");
+        else if (node instanceof AST_Default)
+            base54.consider("default");
+        else if (node instanceof AST_With)
+            base54.consider("with");
+        else if (node instanceof AST_ObjectSetter)
+            base54.consider("set" + node.key);
+        else if (node instanceof AST_ObjectGetter)
+            base54.consider("get" + node.key);
+        else if (node instanceof AST_ObjectKeyVal)
+            base54.consider(node.key);
+        else if (node instanceof AST_New)
+            base54.consider("new");
+        else if (node instanceof AST_This)
+            base54.consider("this");
+        else if (node instanceof AST_Try)
+            base54.consider("try");
+        else if (node instanceof AST_Catch)
+            base54.consider("catch");
+        else if (node instanceof AST_Finally)
+            base54.consider("finally");
+        else if (node instanceof AST_SymbolDeclaration && node.unmangleable())
+            base54.consider(node.name);
+        else if (node instanceof AST_SymbolRef && !node.uniq && !(node instanceof AST_LabelRef))
+            base54.consider(node.name);
+        else if (node instanceof AST_Unary || node instanceof AST_Binary)
+            base54.consider(node.operator);
+        else if (node instanceof AST_Dot)
+            base54.consider(node.property);
+    });
+    this.walk(tw);
+    base54.sort();
+});
+
+var base54 = (function() {
+    var string = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ$_0123456789";
+    var chars, frequency;
+    function reset() {
+        frequency = {};
+        chars = string.split("");
+        chars.map(function(ch){ frequency[ch] = 0 });
+    }
+    base54.consider = function(str){
+        for (var i = str.length; --i >= 0;) {
+            var ch = str.charAt(i);
+            if (string.indexOf(ch))
+                ++frequency[ch];
+        }
+    };
+    base54.sort = function() {
+        chars.sort(function(a, b){
+            if (is_digit(a) && !is_digit(b)) return 1;
+            if (is_digit(b) && !is_digit(a)) return -1;
+            return frequency[b] - frequency[a];
+        });
+    };
+    base54.reset = reset;
+    reset();
+    base54.get = function(){ return chars };
+    function base54(num) {
+        var ret = "", base = 54;
+        do {
+            ret += chars[num % base];
+            num = Math.floor(num / base);
+            base = 64;
+        } while (num > 0);
+        return ret;
+    };
+    return base54;
+})();