From 98d0954e32ce230ff6e12c465a5eadf1378e54c4 Mon Sep 17 00:00:00 2001
From: Mo Valipour <valipour@gmail.com>
Date: Fri, 12 Sep 2014 15:40:34 +0100
Subject: [PATCH] Fix URL replacement in RSS feed mucking with content

closes #3983
- removed naive regex implementation
- added cheerio to project dependencies
- used cheerio to parse RSS content
- use attribute getter/setter to replace href/src attribute urls with the resolved version
---
 core/server/controllers/frontend.js | 27 +++++++++++++--------------
 package.json                        |  1 +
 2 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/core/server/controllers/frontend.js b/core/server/controllers/frontend.js
index 497f870d0f..8723ca414f 100644
--- a/core/server/controllers/frontend.js
+++ b/core/server/controllers/frontend.js
@@ -14,6 +14,7 @@ var moment      = require('moment'),
     filters     = require('../../server/filters'),
     template    = require('../helpers/template'),
     errors      = require('../errors'),
+    cheerio     = require('cheerio'),
 
     frontendControllers,
     staticPostPermalink,
@@ -485,23 +486,21 @@ frontendControllers = {
                                 categories: _.pluck(post.tags, 'name'),
                                 author: post.author ? post.author.name : null
                             },
-                            content = post.html;
+                            htmlContent = cheerio.load(post.html, { decodeEntities: false });
 
-                        //set img src to absolute url
-                        content = content.replace(/src=["|'|\s]?([\w\/\?\$\.\+\-;%:@&=,_]+)["|'|\s]?/gi, function (match, p1) {
-                            /*jslint unparam:true*/
-                            p1 = url.resolve(siteUrl, p1);
-                            return "src='" + p1 + "' ";
+                        // convert relative resource urls to absolute
+                        ['href', 'src'].forEach(function (attributeName) {
+                            htmlContent('[' + attributeName + ']').each(function (ix, el) {
+                                el = htmlContent(el);
+
+                                var attributeValue = el.attr(attributeName);
+                                attributeValue = url.resolve(siteUrl, attributeValue);
+
+                                el.attr(attributeName, attributeValue);
+                            });
                         });
 
-                        //set a href to absolute url
-                        content = content.replace(/href=["|'|\s]?([\w\/\?\$\.\+\-;%:@&=,_]+)["|'|\s]?/gi, function (match, p1) {
-                            /*jslint unparam:true*/
-                            p1 = url.resolve(siteUrl, p1);
-                            return "href='" + p1 + "' ";
-                        });
-
-                        item.description = content;
+                        item.description = htmlContent.html();
                         feed.item(item);
                     });
                 }).then(function () {
diff --git a/package.json b/package.json
index 874f6e4b95..0285dbb267 100644
--- a/package.json
+++ b/package.json
@@ -36,6 +36,7 @@
         "body-parser": "1.6.3",
         "bookshelf": "0.7.6",
         "busboy": "0.2.3",
+        "cheerio": "0.17.0",
         "colors": "0.6.2",
         "compression": "^1.0.2",
         "connect": "3.0.0-rc.1",