diff --git a/src/node/utils/ImportHtml.js b/src/node/utils/ImportHtml.js
index 48188dfd0..79217cb65 100644
--- a/src/node/utils/ImportHtml.js
+++ b/src/node/utils/ImportHtml.js
@@ -14,22 +14,18 @@
* limitations under the License.
*/
-var jsdom = require('jsdom-nocontextifiy').jsdom;
+var jsdom = require('jsdom').jsdom;
var log4js = require('log4js');
var Changeset = require("ep_etherpad-lite/static/js/Changeset");
var contentcollector = require("ep_etherpad-lite/static/js/contentcollector");
+var cheerio = require("cheerio");
function setPadHTML(pad, html, callback)
{
var apiLogger = log4js.getLogger("ImportHtml");
- // Parse the incoming HTML with jsdom
- try{
- var doc = jsdom(html.replace(/>\n+<'));
- }catch(e){
- apiLogger.warn("Error importing, possibly caused by malformed HTML");
- var doc = jsdom("Error during import, possibly malformed HTML
");
- }
+ var $ = cheerio.load(html);
+ var doc = $('html')[0];
apiLogger.debug('html:');
apiLogger.debug(html);
@@ -38,7 +34,7 @@ function setPadHTML(pad, html, callback)
// using the content collector object
var cc = contentcollector.makeContentCollector(true, null, pad.pool);
try{ // we use a try here because if the HTML is bad it will blow up
- cc.collectContent(doc.childNodes[0]);
+ cc.collectContent(doc);
}catch(e){
apiLogger.warn("HTML was not properly formed", e);
return; // We don't process the HTML because it was bad..
diff --git a/src/package.json b/src/package.json
index 4b0ae82b4..e69d97305 100644
--- a/src/package.json
+++ b/src/package.json
@@ -25,7 +25,7 @@
"formidable" : "1.0.9",
"log4js" : "0.6.6",
"nodemailer" : "0.3.x",
- "jsdom-nocontextifiy" : "0.2.10",
+ "cheerio" : "0.18.0",
"async-stacktrace" : "0.0.2",
"npm" : "1.4.x",
"ejs" : "0.6.1",
diff --git a/src/static/js/contentcollector.js b/src/static/js/contentcollector.js
index 5e393670f..048ed4915 100644
--- a/src/static/js/contentcollector.js
+++ b/src/static/js/contentcollector.js
@@ -54,10 +54,14 @@ function makeContentCollector(collectStyles, browser, apool, domInterface, class
},
nodeNumChildren: function(n)
{
+ if(n.childNodes == null) return 0;
return n.childNodes.length;
},
nodeChild: function(n, i)
{
+ if(n.childNodes.item == null){
+ return n.childNodes[i];
+ }
return n.childNodes.item(i);
},
nodeProp: function(n, p)
@@ -66,6 +70,7 @@ function makeContentCollector(collectStyles, browser, apool, domInterface, class
},
nodeAttr: function(n, a)
{
+ if(n.getAttribute == null) return null;
return n.getAttribute(a);
},
optNodeInnerHTML: function(n)