Hacker News Reader

> You must sanitize ALL user input even if you don't think you're going to render it on a web page.

I'm not able to make sense of this. Sanitize it for what context? SQL? JSON? HTML? Inclusion as a command-line argument? All of these, and hope that sanitizing it for one context doesn't un-sanitize it for others?

Context here means the context of the output page.

Usually this means the HTML context. Different sanitization is needed depending on _where_ in the HTML document the input is used.

For instance, if the input is used in between HTML tags (let's say $foo is user input in this PHP example):

    ... <body><?php echo $foo ?></body>

Here, the input that you need to transition to JavaScript execution is a < character (among other things): <script>alert(1)</script>.

Therefore, to correctly sanitize this, you would call the PHP `htmlentities` function:

    ... <body><?php echo htmlentities($foo) ?></body>

Now, this XSS vulnerability is fixed.

What if foo is used in a different context?

    ... <body><a href='<?php echo htmlentities($foo) ?>'>...

Here, what we need to transition the HTML parser to executing JavaScript is a ' character, and this can be exploited by the following input (in between the double quotes): "' onclick='alert(1)"

The key problem is that `htmlentities` is not valid sanitization in the context of an HTML attribute value. In this example, you need to use `urlencode`

    ... <body><a href='<?php echo urlencode($foo) ?>'>...

The general idea also applies to CSS, JSON, and JavaScript. SQL is a different vulnerability class (SQL injection).

I highly recommend the following research paper from 2011 that discusses the context-sensitivity of JavaScript in depth: http://www.comp.nus.edu.sg/~prateeks/papers/scriptgard-ccs11...

In my mind, the context-sensitivity of XSS is one of the key reasons why it is so prevalent.

I've always had issue with this advice for this very reason. Worse, when people sanitize in and out of the database. Double encoded html just cracks me up.

I'm a big fan of not necessarily sanitizing, but treating it appropriate in context. This may mean removing characters, or mapping them, or just delimiting the entire thing.

To that end, I argue that you should not necessarily sanitize on the way to the storage mechanism. You should only sanitize at the boundaries. So, a web view should make sure any strings are treated as strings. A database layer should make sure query parameters are not able to alter the query. Etc.

(All of this is trying to simply reinforce your point.)

Right. It should be SQL sanitized (parameterized) going into the database, and and then HTML/PDF/whatever Santosh when it is put in one of those formats.

Honest and earnest question for you (as bluemonday's author): Where should gophers draw the line between Go's standard html/template library for rendering "safe" HTML, vs. needing to use bluemonday? The question assumes that any and all user inputs are checked/validated at the point of receipt. Is bluemonday just the "on steriods" version (e.g. provides more granular control, whitelist of specific tags, etc)?

Perhaps it is more of a use-case question - I haven't worked on comment or wiki systems but that seems like it would be much harder to validate user inputs, and thus maybe that is where the line should be drawn? When user inputs are expressly expected to have markup?

If you're using the Go (text|html)/template and you are not rendering any user or untrusted input... i.e. this is 100% a little blog where you are provably the only one providing content... you're gold. No need to do much else.

If you accept any user input and this will be displayed or processed anywhere (be it in HTML fragments, input to SQL, etc)... then sanitize it and use bluemonday or any other sanitizer applicable to your content.

For user generated content, like comments on Hacker News, there is a pretty rich policy provided by bluemonday so you don't have to think about it and are going to be alright.

That looks like this:

    p := bluemonday.UGCPolicy()
    htmlOut := p.Sanitize(htmlIn)

If you have custom needs beyond "I accept UGC"... then I'm afraid you're going to have to think about your needs and how to express your own policy, but even then, you can use bluemonday's built in policies as a starting point.

Nice. I use bluemonday, and I also took a lot of inspiration from how microcosm was written. Just wanted to say thanks for the effort!

>You must sanitize ALL user input even if you don't think you're going to render it on a web page

Not really.

Yes, really. Otherwise, you must take extra care not to reflect any input data back in any response to the user, whether it's in the HTML body or not. See: HTTP response splitting.

No, not really.

In fact, you have it totally backwards: you're not supposed to sanitize all user input before storing it. Instead you're supposed to sanitize any user input before you output it back to your webpage.

Even more so: it's the output that dictates what sanitization you should perform, not the input. You don't do input sanitization for HTML (for XSS etc) when you store your data in your DB. Instead you should sanitize the input for SQL Injection issues. And similarly for whatever other output -- if you take user input and run a shell command, you should sanitize for shell safety, not run html sanitization.

You've said nothing that contradicts my post.

As long as the data is sanitized before it can affect the storage/transport mechanism for its content type, you're good.

> As long as the data is sanitized before it can affect the storage/transport mechanism for its content type, you're good.

No, not really. Storing the user's data as is is almost always of paramount importance. The fact that it may be output as HTML/XML/MarkDown/whatever means that it really is at output-time that you must sanitize/escape/quote.

That's why the moral of the Bobby Tables story isn't: "Oh, just remove all semicolons". It's "use prepared queries".

I don't disagree with sanitizing data at output time when it's clear that A) the input won't affect anything else and B) output is going to happen. But realize not all input winds up in a SQL database, not all input will be considered valid in all contexts, and not all input eventually becomes output.

Sometimes, data really does need to be sanitized at the point of submission. If you disagree, that's more of a point about application design than appsec.

You're all saying the same thing.

I didn't specify whether the sanitize occurred on receiving user input or displaying it.

I only said, sanitize all user input.

> I didn't specify whether the sanitize occurred on receiving user input or displaying it.

I'm sorry, but you basically did. You said:

> You must sanitize ALL user input even if you don't think you're going to render it on a web page

Which implies that sanitizing input at display time, when you know you're rendering it to a web page, is too late. That's why people are jumping on you. Keeping a clean database is the absolute most important thing you can do. The database isn't contextual. The data it stores can find its way into HTML pages, REST responses, SQL queries, PDF reports, XML/JSON data exports and a ton of other formats. Each of these output formats will require a different form of sanitizing. Sanitizing before the data hits disk creates a nightmare for anyone displaying the data in a context other than the sanitization that was performed. So what you said originally is precisely incorrect. Only sanitize input when you know it's going to be rendered to a webpage. Otherwise, leave it alone.

Now, you should be using view-layer frameworks to make that sanitization easy, automatic and the default action. When rendering to HTML, the templating language should sanitize by default and give a way for template authors to opt-out when they know the data did not come from user input. Likewise, in the SQL context, prepared statements also make it easy for the developer to do the right thing. But at no point are you speculatively sanitizing all user input. You're getting user input to disk in as pristine a format as possible and sanitizing contextually depending on how the data is outputted.

> But realize not all input winds up in a SQL database, not all input will be considered valid in all contexts, and not all input eventually becomes output.

That was the point I was trying to make: Sanitizing input is fail-from-the-start. There's no way to know ahead of time what outputs you're going to be producing 5 years from now. Conclusion: Store all input exactly as received. (We can do that these days with form/url encodings and whatnot).

Ok, so now you have the data stored accurately.

Next step: You need to output to, let's say, HTML. Ok, so you just escape/quote everything appropriately and nobody gets hurt. If you just do the escaping/quoting properly there is no XSS attacks. It's really just that simple.

However, it is NOT about sanitizing at the "input" point. Do you get what I'm saying now?

(I realize that that sounds aggressive, but I really just want to force this point home. Please tell me if you disagree or find some detail in my explanation confusing. This is important for the security of the web and either I'm wrong or you're wrong or I didn't understand what you said. Let's figure out which is the case.)

[1] There are caveats here.

Sanitization does not refer to escaping HTML entities before rendering to avoid cross site scripting. That is just escaping of entities done when rendering templates.

I believe by sanitization most people mean processing content which will be rendered and not escaped, a good example is content from WYSIWYG editors. And this is where sanitization libraries would come into play.

You would sanitize HTML fragments before storing them in database because you don't escape them during rendering. Text content is not sanitized before saving to database as you can just escape it when rendering.

Also really it should be less about sanitising, more about escaping. If possible, the input should be stored essentially as-is.