If you use the linky filter in ngSanitize any html will be rendered as escaped html

[karl-gustav]

Overview of the Issue

If you send the text <div>text</div>http://vg.no though the linky filter the resulting text will be:

&lt;div&gt;text&lt;/div&gt;<a href="http://vg.no">http://vg.no</a>

Motivation for or Use Case

What I would expect was

<div>text</div><a href="http://vg.no">http://vg.no</a>

because after all that's why you use the ngBindHtml directive, to put unescaped html into the page.

Angular Version(s)

I have reproduced the bug in angular version s 1.2.25 and 1.3.0-rc.5

Browsers and Operating System

I have only tested this bug in:

• Chrome version 38.0.2125.101
• Safari version 7.1 (9537.85.10.17.1)

Reproduce the Error

Plunker to prove the bug, and that it is sanitizeText() that causes it in angular-1.2.25 and angular-1.3.0-rc.5

Suggest a Fix

The bug seems to come from line 134 in the source code when calling the global function sanitizeText(). But since $santitize() is called on line 124 it seems redundant to call sanitizeText(). But I have not read the source code to confirm my suspicion, it's just that the funciton names seems to overlap.

Related Issues

I have not found any issues about the linky filter that looks like this bug.

[lorenzhs]

Uh, should linky really be called on HTML strings? I don't think it's equipped to handle DOM things at all.

[karl-gustav]

@lorenzhs : So the plugin/library ngSanitize, whose sole purpose is to enable html bindings, has a filter that don't support html? That makes no sense at all. And since it has been added to the backlog as a bug I'm guessing I'm not the only one that thinks that.

[lorenzhs]

@karl-gustav according to #3285 (comment), "notice that it doesn't take html as input, but text." -- I've had trouble with that in the past as well.

[karl-gustav]

@lorenzhs I see now that the bug is marked as "severity: broken expected use" so you might be right. Lets just hope they fix it to parse html in either case:-)

[lorenzhs]

@karl-gustav probably not going to happen, if what was written in the issue I linked still applies:

Now if you or others care about this functionality, then I suggest that you make it a project of its own and public is in bower so that others can reuse it.

[Narretz]

I labeled it as "broken expected use" because it seemed to me this way. But the docs also say it only takes text input. So I guess this is currently working as expected, but if it's possible to maintain a secure conversion with html input, then I don't see why it can't be a feature request at least.
As far as I can tell, the linked issue is slightly different, as the OP thought linky should sanitize the html. Imo, linky should only convert text links to html links, and leave the rest of the input alone.

[lorenzhs]

Yeah that's how it is specified at the moment.

Regarding the feature request, I implemented a "meta-filter" that applies a filter to all text nodes in an HTML string: https://github.com/glowing-bear/glowing-bear/blob/5df0ce21a26681fb30184ab1bc4c325aa1bcad79/js/filters.js#L62 -- this is probably what @karl-gustav wants (use it as {{ foo | DOMfilter:'linky' }}, doesn't support parameters to the filter but that shouldn't be too hard to add.

EDIT: Updated the URL because of an escaping issue

[amerikan]

Using linky filter on the following:

I like http://www.google.com & many others

gets converted to:

I like <a href="http://www.google.com">http://www.google.com</a> &amp; many others

as you can see the & gets encoded to & 😞

Anyone has a solution or know if this will be fixed sometime? Or a better filter that will convert URL's to links without this bug?

[lorenzhs]

@amerikan that is intended behaviour, ngSanitize does escaping of HTML characters. Why is this a problem for you?

[danieljsinclair]

I also believe it should leave existing HTML alone and just convert the links. I don't see why it needs to perform two functions and seems in conflict with the single responsibility principle IMHO.

I'm trying to use the linky filter to convert links entered by the user in a contenteditable where HTML is also allowed. I can sanitize the HTML just fine but in order to use linky on the content I would need to RegExp out the links minus the HTML first which defeats the purpose of using linky in the first place.

With it working this way I'll have to abandon it and go roll something else instead.