dont double escape certain entities/character references, fix encodin…

…g test
markedjs · Nov 30, 2011 · b702e91 · b702e91
1 parent 9ab262b
commit b702e91
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 16 deletions.
diff --git a/lib/marked.js b/lib/marked.js
@@ -332,7 +332,7 @@ inline.lexer = function(str) {
     if (cap = inline.code.exec(str)) {
       str = str.substring(cap[0].length);
       out += '<code>'
-        + escape(cap[2] || cap[1])
+        + escape(cap[2] || cap[1], true)
         + '</code>';
       continue;
     }
@@ -413,7 +413,7 @@ var tok = function() {
     }
     case 'code': {
       return '<pre><code>'
-        + escape(token.text)
+        + escape(token.text, true)
         + '</code></pre>';
     }
     case 'blockquote_start': {
@@ -508,9 +508,11 @@ var parse = function(src) {
  * Helpers
  */
 
-var escape = function(html) {
+var escape = function(html, dbl) {
   return html
-    .replace(/&/g, '&amp;')
+    .replace(!dbl
+      ? /&(?!#?\w+;)/g
+      : /&/g, '&amp;')
     .replace(/</g, '&lt;')
     .replace(/>/g, '&gt;')
     .replace(/"/g, '&quot;')

diff --git a/test/fix.js b/test/fix.js
@@ -51,16 +51,11 @@ fs.readdirSync(dir).forEach(function(file) {
   fs.writeFileSync(file, text);
 });
 
-// markdown avoids double encoding half of the time
-// and does it the other half. this behavior will be
-// implemented eventually, but for now, this needs to 
-// be changed, because i want to see if the other tests
-// included in this file pass.
+// markdown is weird with encoding
 (function() {
   var file = dir + '/amps_and_angles_encoding.html';
   var html = fs.readFileSync(file, 'utf8')
     .replace('6 > 5.', '6 &gt; 5.')
-    .replace('AT&amp;T is another', 'AT&amp;amp;T is another');
 
   fs.writeFileSync(file, html);
 })();
@@ -73,7 +68,7 @@ fs.readdirSync(dir).forEach(function(file) {
   fs.writeFileSync(file, text);
 })();
 
-// fix strange markup that isnt likely 
+// fix strange markup that isnt likely
 // to exist in the reality
 (function _(ext) {
   var file = dir + '/inline_html_advanced.' + ext;
@@ -86,17 +81,17 @@ fs.readdirSync(dir).forEach(function(file) {
 ('html');
 
 // markdown parses backslashes in a very primitive
-// way because it's not a real parser. i cannot 
+// way because it's not a real parser. i cannot
 // include this test, because marked parses backslashes
 // in a very different way.
 (function(ext) {
-  fs.writeFileSync(dir + '/backslash_escapes.text', 
+  fs.writeFileSync(dir + '/backslash_escapes.text',
     'hello world \\[how](are you) today');
-  fs.writeFileSync(dir + '/backslash_escapes.html', 
+  fs.writeFileSync(dir + '/backslash_escapes.html',
     '<p>hello world [how](are you) today</p>');
 })();
 
-// can't do this for performance reasons 
+// can't do this for performance reasons
 // right now
 (function _(name) {
   fs.unlinkSync(dir + '/' + name + '.text');

diff --git a/test/tests/amps_and_angles_encoding.html b/test/tests/amps_and_angles_encoding.html
@@ -1,6 +1,6 @@
 <p>AT&amp;T has an ampersand in their name.</p>
 
-<p>AT&amp;amp;T is another way to write it.</p>
+<p>AT&amp;T is another way to write it.</p>
 
 <p>This &amp; that.</p>