Support HTML5 comment syntax (enabled by default!)

See http://javascript.spec.whatwg.org/#comment-syntax
    https://github.com/mishoo/UglifyJS/issues/503
    https://github.com/marijnh/acorn/issues/62

2 files changed, 45 insertions, 21 deletions

diff --git a/lib/output.js b/lib/output.js
index b7bcd1e3..df2692ad 100644
--- a/lib/output.js
+++ b/lib/output.js
@@ -399,7 +399,7 @@ function OutputStream(options) {
                     });
                 }
                 comments.forEach(function(c){
-                    if (c.type == "comment1") {
+                    if (/comment[134]/.test(c.type)) {
                         output.print("//" + c.value + "\n");
                         output.indent();
                     }
diff --git a/lib/parse.js b/lib/parse.js
index 6a9f1ec9..a423aa0a 100644
--- a/lib/parse.js
+++ b/lib/parse.js
@@ -210,7 +210,7 @@ function is_token(token, type, val) {
 
 var EX_EOF = {};
 
-function tokenizer($TEXT, filename) {
+function tokenizer($TEXT, filename, html5_comments) {
 
     var S = {
         text            : $TEXT.replace(/\r\n?|[\n\u2028\u2029]/g, "\n").replace(/\uFEFF/g, ''),
@@ -242,6 +242,14 @@ function tokenizer($TEXT, filename) {
         return ch;
     };
 
+    function forward(i) {
+        while (i-- > 0) next();
+    };
+
+    function looking_at(str) {
+        return S.text.substr(S.pos, str.length) == str;
+    };
+
     function find(what, signal_eof) {
         var pos = S.text.indexOf(what, S.pos);
         if (signal_eof && pos == -1) throw EX_EOF;
@@ -381,8 +389,8 @@ function tokenizer($TEXT, filename) {
         return token("string", ret);
     });
 
-    function read_line_comment() {
-        next();
+    function skip_line_comment(type) {
+        var regex_allowed = S.regex_allowed;
         var i = find("\n"), ret;
         if (i == -1) {
             ret = S.text.substr(S.pos);
@@ -391,11 +399,13 @@ function tokenizer($TEXT, filename) {
             ret = S.text.substring(S.pos, i);
             S.pos = i;
         }
-        return token("comment1", ret, true);
+        S.comments_before.push(token(type, ret, true));
+        S.regex_allowed = regex_allowed;
+        return next_token();
     };
 
-    var read_multiline_comment = with_eof_error("Unterminated multiline comment", function(){
-        next();
+    var skip_multiline_comment = with_eof_error("Unterminated multiline comment", function(){
+        var regex_allowed = S.regex_allowed;
         var i = find("*/", true);
         var text = S.text.substring(S.pos, i);
         var a = text.split("\n"), n = a.length;
@@ -405,8 +415,11 @@ function tokenizer($TEXT, filename) {
         if (n > 1) S.col = a[n - 1].length;
         else S.col += a[n - 1].length;
         S.col += 2;
-        S.newline_before = S.newline_before || text.indexOf("\n") >= 0;
-        return token("comment2", text, true);
+        var nlb = S.newline_before = S.newline_before || text.indexOf("\n") >= 0;
+        S.comments_before.push(token("comment2", text, true));
+        S.regex_allowed = regex_allowed;
+        S.newline_before = nlb;
+        return next_token();
     });
 
     function read_name() {
@@ -470,16 +483,13 @@ function tokenizer($TEXT, filename) {
 
     function handle_slash() {
         next();
-        var regex_allowed = S.regex_allowed;
         switch (peek()) {
           case "/":
-            S.comments_before.push(read_line_comment());
-            S.regex_allowed = regex_allowed;
-            return next_token();
+            next();
+            return skip_line_comment("comment1");
           case "*":
-            S.comments_before.push(read_multiline_comment());
-            S.regex_allowed = regex_allowed;
-            return next_token();
+            next();
+            return skip_multiline_comment();
         }
         return S.regex_allowed ? read_regexp("") : read_operator("/");
     };
@@ -516,6 +526,16 @@ function tokenizer($TEXT, filename) {
             return read_regexp(force_regexp);
         skip_whitespace();
         start_token();
+        if (html5_comments) {
+            if (looking_at("<!--")) {
+                forward(4);
+                return skip_line_comment("comment3");
+            }
+            if (looking_at("-->") && S.newline_before) {
+                forward(3);
+                return skip_line_comment("comment4");
+            }
+        }
         var ch = peek();
         if (!ch) return token("eof");
         var code = ch.charCodeAt(0);
@@ -591,14 +611,18 @@ var ATOMIC_START_TOKEN = array_to_hash([ "atom", "num", "string", "regexp", "nam
 function parse($TEXT, options) {
 
     options = defaults(options, {
-        strict     : false,
-        filename   : null,
-        toplevel   : null,
-        expression : false
+        strict         : false,
+        filename       : null,
+        toplevel       : null,
+        expression     : false,
+        html5_comments : true,
     });
 
     var S = {
-        input         : typeof $TEXT == "string" ? tokenizer($TEXT, options.filename) : $TEXT,
+        input         : (typeof $TEXT == "string"
+                         ? tokenizer($TEXT, options.filename,
+                                     options.html5_comments)
+                         : $TEXT),
         token         : null,
         prev          : null,
         peeked        : null,