tagName and nodeName getters overridden when parsed with jsdom.jsdom, causing various problems with XML DOM #393
Comments
|
Also, I think I found some dead code: https://github.com/tmpvar/jsdom/blob/master/lib/jsdom.js#L58 It looks like browser.HTMLDocument will always be populated in browserAugmentation: https://github.com/tmpvar/jsdom/blob/master/lib/jsdom/browser/index.js#L325 This means that the else branch ("new browser.Document(options);") will never be called: https://github.com/tmpvar/jsdom/blob/master/lib/jsdom.js#L60 This means that an HTMLDocument will always been initialized, and a Document will never been initialized, even if XML is passed in. I'm not sure what implications this has. |
…n, as they override getters defined in dom level1, which normalize case only when working with HTML (as opposed to XML). This solves issue jsdom#393.
|
I have the following problem: var jsdom = require("jsdom");
var dom = jsdom.level(3, 'core');
var doc, elem;
doc = new dom.Document();
elem = doc.createElement('xml');
doc.appendChild(elem);
console.log(elem.nodeName);
jsdom.jsdom("<html><head></head><body></body></html>");
// After HTML parsing, the same fragment as above returns upper case nodeName
doc = new dom.Document();
elem = doc.createElement('xml');
doc.appendChild(elem);
console.log(elem.nodeName);Output: Is this related to the case given in this issue? |
|
Yes, I think so, because when you call jsdom.jsdom, it calls browserAugmentation, which overrides the getters for tagName and nodeName, such that they always return uppercase. |
|
Agreed, we need to properly handle xml. Until now, there have been few users of the xml portion of jsdom. I sense a change a brewin' |
|
If you can get some tests in for this, it seems like a good thing to merge in. |
There are various problems when parsing HTML using jsdom.jsdom. Here is a reduced test case:
Basically, document.documentElement will be undefined, and tagName and nodeName getters will always return uppercase form.
I traced the code, and found that the problem is most likely in jsdom/browser/index.js, lines 550 - 561, as tagName and nodeName get overriden:
https://github.com/tmpvar/jsdom/blob/master/lib/jsdom/browser/index.js#L550
This will override the getters defined in jsdom.dom.level1.core module, which will return lowercase form for non-HTML documents.
Deleting lines 550 - 561 seems to fix both of these problems: document.documentElement is set to element foo, and f.tagName is "foo" (lowercase).
The text was updated successfully, but these errors were encountered: