class ContentTools.HTMLCleaner

    # A class for cleaning (sanitizing) HTML.
    #
    # This class is specifically designed for cleaning HTML so that it can be
    # safely convert to ContentEdit elements, it use outside of ContentTools/
    # Edit is probably limited.
    #
    # The class itself was inspired by Alok Menghrajani's code snippet for
    # sanitizing HTML which can be found here:
    #
    # ~ https://www.quaxio.com/html_white_listed_sanitizer/

    # A table of tags and the list of attributes we consider safe for them
    @DEFAULT_ATTRIBUTE_WHITELIST = {
        'a': ['href'],
        'td': ['colspan']
    }

    # A default list of tags we consider safe
    @DEFAULT_TAG_WHITELIST = [
        'a',
        'address',
        'b',
        'blockquote',
        'code',
        'del',
        'em',
        'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
        'i',
        'ins',
        'li',
        'ol',
        'p',
        'pre',
        'strong',
        'sup',
        'table',
        'tbody',
        'td',
        'tfoot',
        'th',
        'thead',
        'tr',
        'u',
        'ul',
        '#text'
    ]

    @NO_CONTENT_TAGS = [
        'head',
        'meta',
        'style',
        'script',
        'title'
    ]

    constructor: (tagWhitelist, attributeWhitelist) ->
        # List of tags we consider safe
        @tagWhitelist = tagWhitelist or @constructor.DEFAULT_TAG_WHITELIST

        # Table of tags and the attributes we consider safe
        @attributeWhitelist = attributeWhitelist or
                @constructor.DEFAULT_ATTRIBUTE_WHITELIST

    clean: (html) ->
        # Return a clean version of the given string of HTML

        # See https://github.com/GetmeUK/ContentTools/issues/542 for details,
        # thanks to https://github.com/cyclaero for the additional information
        # and workaround.
        html = html.replace(
            /<span( class="Apple-converted-space")?> <\/span>/g,
            ' '
        )

        sandbox = document.implementation.createHTMLDocument()
        wrapper = sandbox.createElement('div')
        wrapper.innerHTML = html

        stack = (c for c in wrapper.childNodes)
        while stack.length > 0
            node = stack.shift()
            nodeName = node.nodeName.toLowerCase()

            # Remove any non-whitelisted tags
            if @tagWhitelist.indexOf(nodeName) < 0

                if @constructor.NO_CONTENT_TAGS.indexOf(nodeName) > -1
                    node.remove()
                    continue

                # Add the children of the tag to the stack to be processed
                for childNode in node.childNodes
                    unless childNode
                        continue

                    childNode = childNode.cloneNode(true)
                    childNodeName = childNode.nodeName.toLowerCase()
                    node.parentNode.insertBefore(childNode, node)
                    stack.push(childNode)

                node.remove()
                continue

            # Remove any tag with no children or only whitespace children with
            # the exception of text or <br> tags.
            unless nodeName == '#text'
                if node.textContent.trim() == ''
                    if node.textContent == '' or node.parentNode == wrapper
                        # If the element is empty or top level remove it...
                        node.remove()

                    else
                        # ...replace it with a space
                        node.parentNode.replaceChild(
                            sandbox.createTextNode(' '),
                            node
                        )

                    continue

            # Remove paragraphs from table cells and list items (we see this
            # when copying in from LibreOffice office documents).
            #
            # We use a rather unforgiving approach to resolve the problem
            # simply setting the inner HTML of the table cell to the text
            # content value.
            if nodeName == 'td' or nodeName == 'th' or nodeName == 'li'
                if node.querySelector('p')
                    node.innerHTML = node.textContent

            # Remove any non-whilelisted attributes
            if node.attributes

                safeAttributes = @attributeWhitelist[nodeName] or []
                rawAttributes = (a for a in node.attributes)
                for attribute in rawAttributes
                    if safeAttributes.indexOf(attribute.name.toLowerCase()) < 0
                        node.removeAttribute(attribute.name)
                        continue

                    # Handle href attributes with JS actions
                    if attribute.name.toLowerCase() == 'href'
                        value = node.getAttribute(attribute.name)
                        if value.startsWith('javascript:')
                            node.removeAttribute(attribute.name)
                            continue

            # Add any children of the node to the stack to be cleaned
            stack.push.apply(stack, (c for c in node.childNodes))

        return wrapper.innerHTML