Rewrote implementation of classes to be regexp-based.

lib/compiler.js

@@ -82,6 +82,8 @@ PEG.Grammar.Rule.prototype = {
 
 PEG.Grammar.Literal = function(value) { this._value = value; };
 
+PEG.Grammar.Class = function(characters) { this._characters = characters; };
+
 PEG.Grammar.Any = function() {};
 
 PEG.Grammar.Sequence = function(elements) { this._elements = elements; };
@@ -111,6 +113,8 @@ PEG.Grammar.Rule.prototype.checkReferencedRulesExist = function(grammar) {
 
 PEG.Grammar.Literal.prototype.checkReferencedRulesExist = function(grammar) {};
 
+PEG.Grammar.Class.prototype.checkReferencedRulesExist = function(grammar) {};
+
 PEG.Grammar.Any.prototype.checkReferencedRulesExist = function(grammar) {};
 
 PEG.Grammar.Sequence.prototype.checkReferencedRulesExist = function(grammar) {
@@ -151,6 +155,8 @@ PEG.Grammar.Rule.prototype.checkNoLeftRecursion = function(grammar, appliedRules
 
 PEG.Grammar.Literal.prototype.checkNoLeftRecursion = function(grammar, appliedRules) {};
 
+PEG.Grammar.Class.prototype.checkNoLeftRecursion = function(grammar, appliedRules) {};
+
 PEG.Grammar.Any.prototype.checkNoLeftRecursion = function(grammar, appliedRules) {};
 
 PEG.Grammar.Sequence.prototype.checkNoLeftRecursion = function(grammar, appliedRules) {
@@ -418,6 +424,25 @@ PEG.Grammar.Literal.prototype.compile = function(resultVar) {
   );
 };
 
+PEG.Grammar.Class.prototype.compile = function(resultVar) {
+  return PEG.Compiler.formatCode(
+    "if (this._input.substr(this._pos).match(${regexp}) !== null) {",
+    "  var ${resultVar} = this._input[this._pos];",
+    "  this._pos++;",
+    "} else {",
+    "  var ${resultVar} = null;",
+    "  if (context.reportMatchFailures) {",
+    "    this._matchFailed(new PEG.Parser.ClassMatchFailure(${characters|string}));",
+    "  }",
+    "}",
+    {
+      characters: this._characters,
+      regexp:     "/^[" + this._characters + "]/",
+      resultVar:  resultVar
+    }
+  );
+};
+
 PEG.Grammar.Any.prototype.compile = function(resultVar) {
   return PEG.Compiler.formatCode(
     "if (this._input.length > this._pos) {",

lib/metagrammar.pegjs

@@ -52,15 +52,8 @@ primary
   : identifier !(( literal / "") colon) { return new PEG.Grammar.RuleRef($1); }
   / literal                             { return new PEG.Grammar.Literal($1); }
   / dot                                 { return new PEG.Grammar.Any();       }
-  / class {
-      return new PEG.Grammar.Choice(
-        PEG.ArrayUtils.map(
-          $1.split(""),
-          function(character) { return new PEG.Grammar.Literal(character); }
-        )
-      );
-    }
-  / lparen expression rparen { return $2; }
+  / class                               { return new PEG.Grammar.Class($1);   }
+  / lparen expression rparen            { return $2; }
 
 /* "Lexical" elements */
 
@@ -137,22 +130,24 @@ class "character class": "[" (classCharacterRange / classCharacter)* "]" __ {
 }
 
 classCharacterRange: bracketDelimitedCharacter "-" bracketDelimitedCharacter {
-  var beginCharCode = $1.charCodeAt(0);
-  var endCharCode = $3.charCodeAt(0);
-  if (beginCharCode > endCharCode) {
+  if ($1.charCodeAt(0) > $3.charCodeAt(0)) {
     throw new PEG.Parser.SyntaxError(
-      "Invalid character range: " + $1 + "-" + $3 + "."
+      "Invalid character range: "
+        + PEG.RegExpUtils.quoteForClass($1)
+        + "-"
+        + PEG.RegExpUtils.quoteForClass($3)
+        + "."
     );
   }
 
-  var result = "";
-  for (var charCode = beginCharCode; charCode <= endCharCode; charCode++) {
-    result += String.fromCharCode(charCode);
-  }
-  return result;
+  return PEG.RegExpUtils.quoteForClass($1)
+    + "-"
+    + PEG.RegExpUtils.quoteForClass($3);
 }
 
-classCharacter: bracketDelimitedCharacter
+classCharacter: bracketDelimitedCharacter {
+  return PEG.RegExpUtils.quoteForClass($1);
+}
 
 bracketDelimitedCharacter
   : simpleBracketDelimitedCharacter

lib/runtime.js

@@ -58,6 +58,29 @@ PEG.StringUtils = {
 
 };
 
+/* ===== PEG.RegExpUtils ===== */
+
+/* RegExp manipulation utility functions. */
+
+PEG.RegExpUtils = {
+  /*
+   * Escapes characters inside the string so that it can be used as a list of
+   * characters in a character class of a regular expresion.
+   */
+  quoteForClass: function(s) {
+    /* Based on ECMA-262, 5th ed., 7.8.5 & 15.10.1. */
+    return s
+      .replace(/\\/g, '\\\\')        // backslash
+      .replace(/\//g, '\\/')         // closing slash
+      .replace(/]/g, '\\]')          // closing bracket
+      .replace(/-/g, '\\-')          // dash
+      .replace(/\r/g, '\\r')         // carriage return
+      .replace(/\u2028/g, '\\u2028') // line separator
+      .replace(/\u2029/g, '\\u2029') // paragraph separator
+      .replace(/\n/g, '\\n')         // line feed
+  }
+};
+
 /* ===== PEG.Parser ===== */
 
 /* Prototype of all parsers generated by PEG.js. */
@@ -211,6 +234,18 @@ PEG.Parser.LiteralMatchFailure.prototype = {
   toString: function() { return PEG.StringUtils.quote(this._value); }
 };
 
+/* ===== PEG.Parser.ClassMatchFailure ===== */
+
+/* Stores information about a class match failure. */
+
+PEG.Parser.ClassMatchFailure = function(characters) {
+  this._characters = characters;
+};
+
+PEG.Parser.ClassMatchFailure.prototype = {
+  toString: function() { return "[" + this._characters + "]"; }
+};
+
 /* ===== PEG.Parser.AnyMatchFailure ===== */
 
 /* Stores information about a failure to match a "." expression. */

test/metagrammar-test.js

@@ -47,12 +47,7 @@ with (PEG.Grammar) {
   }
 
   function classGrammar(chars) {
-    return oneRuleGrammar(new PEG.Grammar.Choice(
-      PEG.ArrayUtils.map(
-        chars.split(""),
-        function(char) { return new PEG.Grammar.Literal(char); }
-      )
-    ));
+    return oneRuleGrammar(new PEG.Grammar.Class(chars));
   }
 
   var anyGrammar = oneRuleGrammar(new Any());
@@ -169,7 +164,7 @@ with (PEG.Grammar) {
     grammarParserParses('start: a',        identifierGrammar("a"));
     grammarParserParses('start: "abcd"',   literalGrammar("abcd"));
     grammarParserParses('start: .',        anyGrammar);
-    grammarParserParses('start: [a-d]',    classGrammar("abcd"));
+    grammarParserParses('start: [a-d]',    classGrammar("a-d"));
     grammarParserParses('start: ("abcd")', literalGrammar("abcd"));
   });
 
@@ -280,17 +275,17 @@ with (PEG.Grammar) {
   /* Canonical class is "[a-d]". */
   test("parses classCharacterRange", function() {
     grammarParserParses("start: []",          classGrammar(""));
-    grammarParserParses("start: [a-d]",       classGrammar("abcd"));
+    grammarParserParses("start: [a-d]",       classGrammar("a-d"));
     grammarParserParses("start: [a]",         classGrammar("a"));
-    grammarParserParses("start: [a-de-hi-l]", classGrammar("abcdefghijkl"));
+    grammarParserParses("start: [a-de-hi-l]", classGrammar("a-de-hi-l"));
 
-    grammarParserParses("start: [a-d]\n", classGrammar("abcd"));
+    grammarParserParses("start: [a-d]\n", classGrammar("a-d"));
   });
 
   /* Canonical classCharacterRange is "a-d". */
   test("parses classCharacterRange", function() {
-    grammarParserParses("start: [a-d]", classGrammar("abcd"));
-    grammarParserParses("start: [a-a]", classGrammar("a"));
+    grammarParserParses("start: [a-d]", classGrammar("a-d"));
+    grammarParserParses("start: [a-a]", classGrammar("a-a"));
     grammarParserDoesNotParse("start: [b-a]");
   });
 
@@ -302,11 +297,11 @@ with (PEG.Grammar) {
   /* Canonical bracketDelimitedCharacter is "a". */
   test("parses bracketDelimitedCharacter", function() {
     grammarParserParses("start: [a]",       classGrammar("a"));
-    grammarParserParses("start: [\\n]",     classGrammar("\n"));
+    grammarParserParses("start: [\\n]",     classGrammar("\\n"));
     grammarParserParses("start: [\\0]",     classGrammar("\0"));
     grammarParserParses("start: [\\x00]",   classGrammar("\x00"));
     grammarParserParses("start: [\\u0120]", classGrammar("\u0120"));
-    grammarParserParses("start: [\\\n]",    classGrammar("\n"));
+    grammarParserParses("start: [\\\n]",    classGrammar("\\n"));
   });
 
   /* Canonical simpleBracketDelimiedCharacter is "a". */

test/runtime-test.js

@@ -37,4 +37,17 @@ test("quote", function() {
   );
 });
 
+/* ===== PEG.RegExpUtils ===== */
+
+module("PEG.RegExpUtils");
+
+test("quoteForClass", function() {
+  strictEqual(PEG.RegExpUtils.quoteForClass(""), '');
+  strictEqual(PEG.RegExpUtils.quoteForClass("abcd"), 'abcd');
+  strictEqual(
+    PEG.RegExpUtils.quoteForClass("\\/]-\r\u2028\u2029\n\\/]-\r\u2028\u2029\n"),
+    '\\\\\\/\\]\\-\\r\\u2028\\u2029\\n\\\\\\/\\]\\-\\r\\u2028\\u2029\\n'
+  );
+});
+
 })();