From 636ceb2719680528ea355b1d72fa9d942b80b3c0 Mon Sep 17 00:00:00 2001
From: David Majda <david@majda.cz>
Date: Sat, 13 Mar 2010 09:49:06 +0100
Subject: [PATCH] Metagrammar recognizes JavaScript-like comments.

---
 lib/compiler.js       | 42 +++++++++++++++++++++++++++++++++++++++++-
 test/compiler-test.js | 36 ++++++++++++++++++++++++++++++++----
 2 files changed, 73 insertions(+), 5 deletions(-)

diff --git a/lib/compiler.js b/lib/compiler.js
index 12ee525..7520f08 100644
--- a/lib/compiler.js
+++ b/lib/compiler.js
@@ -1274,10 +1274,50 @@ with (PEG.Grammar) {
         "__",
         null,
         new ZeroOrMore(
-          new Choice([new RuleRef("whitespace"), new RuleRef("eol")])
+          new Choice([
+            new RuleRef("whitespace"),
+            new RuleRef("eol"),
+            new RuleRef("comment")
+          ])
         )
       ),
 
+    /* Modelled after ECMA-262, 5th ed., 7.4. */
+    comment:
+      new Rule(
+        "comment",
+        "comment",
+        new Choice([
+          new RuleRef("singleLineComment"),
+          new RuleRef("multiLineComment")
+        ])
+      ),
+
+    singleLineComment:
+      new Rule(
+        "singleLineComment",
+        null,
+        new Sequence([
+          new Literal("//"),
+          new ZeroOrMore(
+            new Sequence([new NotPredicate(new RuleRef("eolChar")), new Any()])
+          )
+        ])
+      ),
+
+    multiLineComment:
+      new Rule(
+        "multiLineComment",
+        null,
+        new Sequence([
+          new Literal("/*"),
+          new ZeroOrMore(
+            new Sequence([new NotPredicate(new Literal("*/")), new Any()])
+          ),
+          new Literal("*/")
+        ])
+      ),
+
     /* Modelled after ECMA-262, 5th ed., 7.3. */
     eol:
       new Rule(
diff --git a/test/compiler-test.js b/test/compiler-test.js
index 9c3b536..6173aa2 100644
--- a/test/compiler-test.js
+++ b/test/compiler-test.js
@@ -892,14 +892,42 @@ with (PEG.Grammar) {
 
   /* Canonical __ is "\n". */
   test("parses __", function() {
-    grammarParserParses('start:"abcd"',    simpleGrammar);
-    grammarParserParses('start: "abcd"',   simpleGrammar);
-    grammarParserParses('start:\n"abcd"',  simpleGrammar);
-    grammarParserParses('start:   "abcd"', simpleGrammar);
+    grammarParserParses('start:"abcd"',              simpleGrammar);
+    grammarParserParses('start: "abcd"',             simpleGrammar);
+    grammarParserParses('start:\n"abcd"',            simpleGrammar);
+    grammarParserParses('start:/* comment */"abcd"', simpleGrammar);
+    grammarParserParses('start:   "abcd"',           simpleGrammar);
   });
 
   /* Trivial character class rules are not tested. */
 
+  /* Canonical comment is "\/* comment *\/". */
+  test("parses comment", function() {
+    grammarParserParses('start:// comment\n"abcd"',  simpleGrammar);
+    grammarParserParses('start:/* comment */"abcd"', simpleGrammar);
+  });
+  /* Canonical singleLineComment is "// comment". */
+  test("parses singleLineComment", function() {
+    grammarParserParses('start://\n"abcd"',    simpleGrammar);
+    grammarParserParses('start://a\n"abcd"',   simpleGrammar);
+    grammarParserParses('start://aaa\n"abcd"', simpleGrammar);
+    grammarParserParses('start: "abcd"//',     simpleGrammar);
+  });
+
+  /* Canonical multiLineComment is "\/* comment *\/". */
+  test("parses multiLineComment", function() {
+    grammarParserParses('start:/**/"abcd"',    simpleGrammar);
+    grammarParserParses('start:/*a*/"abcd"',   simpleGrammar);
+    grammarParserParses('start:/*aaa*/"abcd"', simpleGrammar);
+    grammarParserParses('start:/*\n*/"abcd"',  simpleGrammar);
+    grammarParserParses('start:/***/"abcd"',   simpleGrammar);
+    grammarParserParses('start:/*a/*/"abcd"',  simpleGrammar);
+
+    grammarParserDoesNotParse('start:/*"abcd"');
+    grammarParserDoesNotParse('start:/*/"abcd"');
+    grammarParserDoesNotParse('start:/*/**/*/"abcd"');
+  });
+
   /* Canonical eol is "\n". */
   test("parses eol", function() {
     grammarParserParses('start:\n"abcd"',     simpleGrammar);