A dead simple Go HTML whitelist-sanitization library.

Efficiently support the following types of HTML sanitization through simple programmatic or JSON configuration:

• Removal of all non-whitelisted elements
• Unwrapping of all non-whitelisted elements

Given a whitelist configuration

{
    "elements": {
        "div": ["id", "class"],
        "b": [],
        "i": []
    }
}

and basic input

<div class="my-class" style="position:relative;">
    <i>Something emphasized</i>
    <p>
        here is a
        <i>paragraph</i>
    </p>
    <b>Something bold</b> 
</div>

Removal

Removal of non-whitelisted elements in the provided example would yield

<div class="my-class">
    <i>Something emphasized</i>
    <b>Something bold</b> 
</div>

Note how the style attribute was removed from the div element and the p element was removed entirely

Unwrapping

Unwrapping of non-whitelisted elements in the provided example would yield

<div class="my-class">
    <i>Something emphasized</i>
    here is a
    <i>paragraph</i>
    <b>Something bold</b> 
</div>

Note how the style attribute was still removed from the div element, while the p element was 'unwrapped' (ie. it's children were attached to it's parent)

Create JSON configuration. Below are the currently supported options

{
    "stripComments": true,
    "stripWhitespace": true,
    "elements": {
        "html": ["xmlns"],
        "head": [],
        "body": [],
        "div": ["id", "class"],
    }
}

Create a sanitize.Whitelist object from a json file with sanitize.WhitelistFromFile(filepath string) or from a []byte with sanitize.NewWhitelist(byteArray []byte) and use it to parse some HTML:

whitelist, err := sanitize.WhitelistFromFile("./path/to/file.json")
// or create from a json []byte
// whitelist, err := sanitize.NewWhitelist(byteArray)

f, _ := os.Open("./path/to/example.html")
sanitized, _ := whitelist.SanitizeRemove(f) // takes any io.Reader

fmt.Printf("sanitized html: %d", sanitized)

whitelist, err := sanitize.WhitelistFromFile("./path/to/file.json")
f, _ := os.Open("./path/to/example.html")

// sanitize a full HTML document by removing
// non-whitelisted elements and attributes
sanitized, _ := whitelist.SanitizeRemove(f)

// sanitize a full HTML document by reattaching
// the children of non-whitelisted elements to the
// non-whitelisted parent; also removes non whitelisted
// attributes for any element
sanitized, _ := whitelist.SanitizeUnwrap(f)

// sanitize an HTML document fragment (ie no html,
// head, or body tags) by removing
// non-whitelisted elements and attributes
sanitized, _ := whitelist.SanitizeRemoveFragment(f)

// sanitize an HTML document fragment (ie no html,
// head, or body tags) by reattaching
// the children of non-whitelisted elements to the
// non-whitelisted parent; also removes non whitelisted
// attributes for any element
sanitized, _ := whitelist.SanitizeUnwrapFragment(f)

• Support sanitization that unwraps non-whitelisted nodes, allowing the text and/or whitelisted subtree through
• Whitelist-level configuration options (eg. stripWhitespace)
• Efficient attribute checking by not allocating a new slice on every whitelisted attribute for an element
• Support sanitization of HTML fragments (instead of just full documents)
• Support non string type attribute values
• Refactor configuration parsing to have []byte interface instead of expecting a filepath
• Create sane defaults
• Usable godoc documentation

Head over to the issues page or open a pull request. Please ensure your code is documented, all existing tests pass, and any new features have tests before submitting a pull request. If you want to check in whether a pull request for a new feature would be accepted, feel free to open an issue.

MIT