lxml.html.fragment_fromstring with create_parent fails

[Tes]

When feeding fragment_fromstring some broken html, it may brake with a ParserError: Multiple elements found.

For example:

s = '<i>This wil</i></div><b>fail</b>'
lxml.html.fragment_fromstring(s, create_parent='div')

I expected this to work, and just drop the </div>.

I propose the following (not tested):

in lxml/html/html5parser.py:

def fragment_fromstring(html, create_parent=False,
                        guess_charset=False, parser=None):
    """Parses a single HTML element; it is an error if there is more than
    one element, or if anything but whitespace precedes or follows the
    element.

    If create_parent is true (or is a tag name) then a parent node
    will be created to encapsulate the HTML in a single element.
    """
    if not isinstance(html, _strings):
        raise TypeError('string required')

    children = fragments_fromstring(html, True, guess_charset, parser)
    if not children:
        raise etree.ParserError('No elements found')
    if len(children) > 1:
        if not create_parent:
            raise etree.ParserError('Multiple elements found')
        else:
            container = etree.Element(create_parent or 'div')
            for element in children:
                if isinstance(element, _strings):
                    if len(container) == 0:
                        container.text = element
                    else:
                        container[-1].tail += element
                else:
                    container.append(element)
            children = container

    result = children[0]
    if result.tail and result.tail.strip():
        raise etree.ParserError('Element followed by text: %r' % result.tail)
    result.tail = None
    return result

[scoder (scoder)]

Fixed in trunk rev 71010/71011.

https://codespeak.net/viewvc/?view=rev&revision=71010
https://codespeak.net/viewvc/?view=rev&revision=71011

[scoder (scoder)]

Fixed in lxml 2.3alpha1.