AST: Store rules in an array instead of an object

This simplifies the code a bit and makes the AST more regular (each node
type has a fixed set of properties). The latter may get useful later
when generalizing visitors.

src/emitter.js

@@ -281,7 +281,9 @@ PEG.compiler.emitter = function(ast) {
             '     */',
             '    parse: function(input, startRule) {',
             '      var parseFunctions = {',
-            '        #block parseFunctionTableItems.join(",\\n")',
+            '        #for rule in node.rules',
+            '          #{string(rule.name) + ": parse_" + rule.name},',
+            '        #end',
             '      };',
             '      ',
             '      if (startRule !== undefined) {',
@@ -340,8 +342,8 @@ PEG.compiler.emitter = function(ast) {
             '        rightmostFailuresExpected.push(failure);',
             '      }',
             '      ',
-            '      #for definition in parseFunctionDefinitions',
+            '      #for rule in node.rules',
-            '        #block definition',
+            '        #block emit(rule)',
             '        ',
             '      #end',
             '      ',
@@ -686,26 +688,7 @@ PEG.compiler.emitter = function(ast) {
   }
 
   var emit = buildNodeVisitor({
-    grammar: function(node) {
+    grammar: emitSimple("grammar"),
-      var name;
-
-      var parseFunctionTableItems = [];
-      for (name in node.rules) {
-        parseFunctionTableItems.push(quote(name) + ": parse_" + name);
-      }
-      parseFunctionTableItems.sort();
-
-      var parseFunctionDefinitions = [];
-      for (name in node.rules) {
-        parseFunctionDefinitions.push(emit(node.rules[name]));
-      }
-
-      return fill("grammar", {
-        node:                     node,
-        parseFunctionTableItems:  parseFunctionTableItems,
-        parseFunctionDefinitions: parseFunctionDefinitions
-      });
-    },
 
     initializer: function(node) { return node.code; },
 

src/parser.js

@@ -33,62 +33,62 @@ PEG.parser = (function(){
      */
     parse: function(input, startRule) {
       var parseFunctions = {
-        "__": parse___,
-        "action": parse_action,
-        "and": parse_and,
-        "braced": parse_braced,
-        "bracketDelimitedCharacter": parse_bracketDelimitedCharacter,
-        "choice": parse_choice,
-        "class": parse_class,
-        "classCharacter": parse_classCharacter,
-        "classCharacterRange": parse_classCharacterRange,
-        "colon": parse_colon,
-        "comment": parse_comment,
-        "digit": parse_digit,
-        "dot": parse_dot,
-        "doubleQuotedCharacter": parse_doubleQuotedCharacter,
-        "doubleQuotedString": parse_doubleQuotedString,
-        "eol": parse_eol,
-        "eolChar": parse_eolChar,
-        "eolEscapeSequence": parse_eolEscapeSequence,
-        "equals": parse_equals,
         "grammar": parse_grammar,
-        "hexDigit": parse_hexDigit,
-        "hexEscapeSequence": parse_hexEscapeSequence,
-        "identifier": parse_identifier,
         "initializer": parse_initializer,
+        "rule": parse_rule,
+        "choice": parse_choice,
+        "sequence": parse_sequence,
         "labeled": parse_labeled,
-        "letter": parse_letter,
-        "literal": parse_literal,
-        "lowerCaseLetter": parse_lowerCaseLetter,
-        "lparen": parse_lparen,
-        "multiLineComment": parse_multiLineComment,
-        "nonBraceCharacter": parse_nonBraceCharacter,
-        "nonBraceCharacters": parse_nonBraceCharacters,
-        "not": parse_not,
-        "plus": parse_plus,
         "prefixed": parse_prefixed,
+        "suffixed": parse_suffixed,
         "primary": parse_primary,
-        "question": parse_question,
+        "action": parse_action,
-        "rparen": parse_rparen,
+        "braced": parse_braced,
-        "rule": parse_rule,
+        "nonBraceCharacters": parse_nonBraceCharacters,
+        "nonBraceCharacter": parse_nonBraceCharacter,
+        "equals": parse_equals,
+        "colon": parse_colon,
         "semicolon": parse_semicolon,
-        "sequence": parse_sequence,
-        "simpleBracketDelimitedCharacter": parse_simpleBracketDelimitedCharacter,
-        "simpleDoubleQuotedCharacter": parse_simpleDoubleQuotedCharacter,
-        "simpleEscapeSequence": parse_simpleEscapeSequence,
-        "simpleSingleQuotedCharacter": parse_simpleSingleQuotedCharacter,
-        "singleLineComment": parse_singleLineComment,
-        "singleQuotedCharacter": parse_singleQuotedCharacter,
-        "singleQuotedString": parse_singleQuotedString,
         "slash": parse_slash,
+        "and": parse_and,
+        "not": parse_not,
+        "question": parse_question,
         "star": parse_star,
+        "plus": parse_plus,
+        "lparen": parse_lparen,
+        "rparen": parse_rparen,
+        "dot": parse_dot,
+        "identifier": parse_identifier,
+        "literal": parse_literal,
         "string": parse_string,
-        "suffixed": parse_suffixed,
+        "doubleQuotedString": parse_doubleQuotedString,
+        "doubleQuotedCharacter": parse_doubleQuotedCharacter,
+        "simpleDoubleQuotedCharacter": parse_simpleDoubleQuotedCharacter,
+        "singleQuotedString": parse_singleQuotedString,
+        "singleQuotedCharacter": parse_singleQuotedCharacter,
+        "simpleSingleQuotedCharacter": parse_simpleSingleQuotedCharacter,
+        "class": parse_class,
+        "classCharacterRange": parse_classCharacterRange,
+        "classCharacter": parse_classCharacter,
+        "bracketDelimitedCharacter": parse_bracketDelimitedCharacter,
+        "simpleBracketDelimitedCharacter": parse_simpleBracketDelimitedCharacter,
+        "simpleEscapeSequence": parse_simpleEscapeSequence,
+        "zeroEscapeSequence": parse_zeroEscapeSequence,
+        "hexEscapeSequence": parse_hexEscapeSequence,
         "unicodeEscapeSequence": parse_unicodeEscapeSequence,
+        "eolEscapeSequence": parse_eolEscapeSequence,
+        "digit": parse_digit,
+        "hexDigit": parse_hexDigit,
+        "letter": parse_letter,
+        "lowerCaseLetter": parse_lowerCaseLetter,
         "upperCaseLetter": parse_upperCaseLetter,
+        "__": parse___,
+        "comment": parse_comment,
+        "singleLineComment": parse_singleLineComment,
+        "multiLineComment": parse_multiLineComment,
+        "eol": parse_eol,
+        "eolChar": parse_eolChar,
         "whitespace": parse_whitespace,
-        "zeroEscapeSequence": parse_zeroEscapeSequence
       };
       
       if (startRule !== undefined) {
@@ -189,13 +189,10 @@ PEG.parser = (function(){
         }
         if (result0 !== null) {
           result0 = (function(initializer, rules) {
-              var rulesConverted = {};
-              each(rules, function(rule) { rulesConverted[rule.name] = rule; });
-        
               return {
                 type:        "grammar",
                 initializer: initializer !== "" ? initializer : null,
-                rules:       rulesConverted,
+                rules:       rules,
                 startRule:   rules[0].name
               };
             })(result0[1], result0[2]);

src/parser.pegjs

@@ -1,12 +1,9 @@
 grammar
   = __ initializer:initializer? rules:rule+ {
-      var rulesConverted = {};
-      each(rules, function(rule) { rulesConverted[rule.name] = rule; });
-
       return {
         type:        "grammar",
         initializer: initializer !== "" ? initializer : null,
-        rules:       rulesConverted,
+        rules:       rules,
         startRule:   rules[0].name
       };
     }

src/passes.js

@@ -17,13 +17,7 @@ PEG.compiler.passes = {
     }
 
     var check = buildNodeVisitor({
-      grammar:
+      grammar:      checkSubnodes("rules"),
-        function(node) {
-          for (var name in node.rules) {
-            check(node.rules[name]);
-          }
-        },
-
       rule:         checkExpression,
       choice:       checkSubnodes("alternatives"),
       sequence:     checkSubnodes("elements"),
@@ -39,7 +33,7 @@ PEG.compiler.passes = {
 
       rule_ref:
         function(node) {
-          if (ast.rules[node.name] === undefined) {
+          if (!findRuleByName(ast, node.name)) {
             throw new PEG.GrammarError(
               "Referenced rule \"" + node.name + "\" does not exist."
             );
@@ -62,25 +56,23 @@ PEG.compiler.passes = {
       check(node.expression, appliedRules);
     }
 
+    function checkSubnodes(propertyName) {
+      return function(node, appliedRules) {
+        each(node[propertyName], function(subnode) {
+          check(subnode, appliedRules);
+        });
+      };
+    }
+
     var check = buildNodeVisitor({
-      grammar:
+      grammar:     checkSubnodes("rules"),
-        function(node, appliedRules) {
-          for (var name in node.rules) {
-            check(node.rules[name], appliedRules);
-          }
-        },
 
       rule:
         function(node, appliedRules) {
           check(node.expression, appliedRules.concat(node.name));
         },
 
-      choice:
+      choice:      checkSubnodes("alternatives"),
-        function(node, appliedRules) {
-          each(node.alternatives, function(alternative) {
-            check(alternative, appliedRules);
-          });
-        },
 
       sequence:
         function(node, appliedRules) {
@@ -106,7 +98,7 @@ PEG.compiler.passes = {
               "Left recursion detected for rule \"" + node.name + "\"."
             );
           }
-          check(ast.rules[node.name], appliedRules);
+          check(findRuleByName(ast, node.name), appliedRules);
         },
 
       literal:      nop,
@@ -141,13 +133,7 @@ PEG.compiler.passes = {
       }
 
       var replace = buildNodeVisitor({
-        grammar:
+        grammar:      replaceInSubnodes("rules"),
-          function(node, from, to) {
-            for (var name in node.rules) {
-              replace(node.rules[name], from, to);
-            }
-          },
-
         rule:         replaceInExpression,
         choice:       replaceInSubnodes("alternatives"),
         sequence:     replaceInSubnodes("elements"),
@@ -176,15 +162,23 @@ PEG.compiler.passes = {
       replace(ast, from, to);
     }
 
-    for (var name in ast.rules) {
+    var indices = [];
-      if (isProxyRule(ast.rules[name])) {
+
-        replaceRuleRefs(ast, ast.rules[name].name, ast.rules[name].expression.name);
+    each(ast.rules, function(rule, i) {
-        if (name === ast.startRule) {
+      if (isProxyRule(rule)) {
-          ast.startRule = ast.rules[name].expression.name;
+        replaceRuleRefs(ast, rule.name, rule.expression.name);
+        if (rule.name === ast.startRule) {
+          ast.startRule = rule.expression.name;
         }
-        delete ast.rules[name];
+        indices.push(i);
       }
-    }
+    });
+
+    indices.reverse();
+
+    each(indices, function(index) {
+      ast.rules.splice(index, 1);
+    });
   },
 
   /*
@@ -240,11 +234,9 @@ PEG.compiler.passes = {
     var compute = buildNodeVisitor({
       grammar:
         function(node, index) {
-          var name;
+          each(node.rules, function(node) {
-
+            compute(node, index);
-          for (name in node.rules) {
+          });
-            compute(node.rules[name], index);
-          }
         },
 
       rule:
@@ -356,11 +348,7 @@ PEG.compiler.passes = {
     var compute = buildNodeVisitor({
       grammar:
         function(node) {
-          var name;
+          each(node.rules, compute);
-
-          for (name in node.rules) {
-            compute(node.rules[name]);
-          }
         },
 
       rule:         computeForScopedExpression,

src/utils.js

@@ -12,6 +12,15 @@ function range(start, stop) {
   return result;
 }
 
+function find(array, callback) {
+  var length = array.length;
+  for (var i = 0; i < length; i++) {
+    if (callback(array[i])) {
+      return array[i];
+    }
+  }
+}
+
 function contains(array, value) {
   /*
    * Stupid IE does not have Array.prototype.indexOf, otherwise this function
@@ -168,3 +177,7 @@ function buildNodeVisitor(functions) {
     return functions[node.type].apply(null, arguments);
   };
 }
+
+function findRuleByName(ast, name) {
+  return find(ast.rules, function(r) { return r.name === name; });
+}

test/parser-test.js

@@ -125,7 +125,7 @@ function oneRuleGrammar(expression) {
   return {
     type:        "grammar",
     initializer: null,
-    rules:       { start: rule("start", null, expression) },
+    rules:       [rule("start", null, expression)],
     startRule:   "start"
   };
 }
@@ -155,16 +155,14 @@ function actionGrammar(action) {
 var initializerGrammar = {
   type:        "grammar",
   initializer: initializer(" code "),
-  rules: {
+  rules:       [rule("a", null, literalAbcd)],
-    a: rule("a", null, literalAbcd)
-  },
   startRule:   "a"
 };
 
 var namedRuleGrammar = {
   type:        "grammar",
   initializer: null,
-  rules:       { start: rule("start", "abcd", literalAbcd) },
+  rules:       [rule("start", "abcd", literalAbcd)],
   startRule:   "start"
 };
 
@@ -175,7 +173,7 @@ test("parses grammar", function() {
     {
       type:        "grammar",
       initializer: null,
-      rules:       { a: rule("a", null, literalAbcd) },
+      rules:       [rule("a", null, literalAbcd)],
       startRule:   "a"
     }
   );
@@ -185,11 +183,11 @@ test("parses grammar", function() {
     {
       type:        "grammar",
       initializer: null,
-      rules: {
+      rules: [
-        a: rule("a", null, literalAbcd),
+        rule("a", null, literalAbcd),
-        b: rule("b", null, literalEfgh),
+        rule("b", null, literalEfgh),
-        c: rule("c", null, literalIjkl)
+        rule("c", null, literalIjkl)
-      },
+      ],
       startRule:   "a"
     }
   );
@@ -212,7 +210,7 @@ test("parses rule", function() {
     {
       type:        "grammar",
       initializer: null,
-      rules:       { start: rule("start", "start rule", choiceLiterals) },
+      rules:       [rule("start", "start rule", choiceLiterals)],
       startRule:   "start"
     }
   );

test/passes-test.js

@@ -93,15 +93,15 @@ test("removes proxy rules", function() {
 
   function simpleGrammarWithStartAndProxied(startRuleExpression) {
     return simpleGrammar(
-      {
+      [
-        start: {
+        {
           type:        "rule",
           name:        "start",
           displayName: null,
           expression:  startRuleExpression
         },
-        proxied: proxiedRule
+        proxiedRule
-      },
+      ],
       "start"
     );
   }
@@ -109,7 +109,7 @@ test("removes proxy rules", function() {
   var cases = [
     {
       grammar: 'start = proxy; proxy = proxied; proxied = "a"',
-      ast:     simpleGrammar({ proxied: proxiedRule }, "proxied")
+      ast:     simpleGrammar([proxiedRule], "proxied")
     },
     {
       grammar: 'start = proxy / "a" / "b"; proxy = proxied; proxied = "a"',
@@ -436,9 +436,9 @@ test("computes variable names", function() {
     var ast = PEG.parser.parse(cases[i].grammar);
     PEG.compiler.passes.computeVarNames(ast);
 
-    deepEqual(ast.rules["start"].resultVars, cases[i].resultVars);
+    deepEqual(ast.rules[0].resultVars, cases[i].resultVars);
-    deepEqual(ast.rules["start"].posVars,    cases[i].posVars);
+    deepEqual(ast.rules[0].posVars,    cases[i].posVars);
-    checkDetails(ast.rules["start"].expression, cases[i].details);
+    checkDetails(ast.rules[0].expression, cases[i].details);
   }
 });
 
@@ -588,7 +588,7 @@ test("computes params", function() {
     PEG.compiler.passes.computeParams(ast);
 
     deepEqual(
-      cases[i].extractor(ast.rules["start"].expression).params,
+      cases[i].extractor(ast.rules[0].expression).params,
       cases[i].params
     );
   }