larlet-fr-david/widont.py

import re


def widont(text):
    """Replaces the space between the last two words in a string with `` ``
    Works in these block tags ``(h1-h6, p, li, dd, dt)`` and also accounts for
    potential closing inline elements ``a, em, strong, span, b, i, mark``

    Extracted from:
    https://github.com/mintchaos/typogrify/blob/
    20f693cbbb232ebc27733d9f1721a2cf1e7b25e3/typogrify/filters.py#L315-L368

    >>> widont('A very simple test')
    'A very simple test'
    Single word items shouldn't be changed
    >>> widont('Test')
    'Test'
    >>> widont(' Test')
    ' Test'
    >>> widont('<ul><li>Test</p></li><ul>')
    '<ul><li>Test</p></li><ul>'
    >>> widont('<ul><li> Test</p></li><ul>')
    '<ul><li> Test</p></li><ul>'
    >>> widont('<p>In a couple of paragraphs</p><p>paragraph two</p>')
    '<p>In a couple of&nbsp;paragraphs</p><p>paragraph&nbsp;two</p>'
    >>> widont('<h1><a href="#">In a link inside a heading</i> </a></h1>')
    '<h1><a href="#">In a link inside a&nbsp;heading</i> </a></h1>'
    >>> widont('<h1><a href="#">In a link</a> followed by other text</h1>')
    '<h1><a href="#">In a link</a> followed by other&nbsp;text</h1>'
    Empty HTMLs shouldn't error
    >>> widont('<h1><a href="#"></a></h1>')
    '<h1><a href="#"></a></h1>'
    >>> widont('<div>Divs get no love!</div>')
    '<div>Divs get no love!</div>'
    >>> widont('<pre>Neither do PREs</pre>')
    '<pre>Neither do PREs</pre>'
    >>> widont('<div><p>But divs with paragraphs do!</p></div>')
    '<div><p>But divs with paragraphs&nbsp;do!</p></div>'

    Adaptations:
    * add the mark element as a potential inline
    * avoid insertion of a nbsp if the sentence is a single word

    >>> widont("<p>Avec <mark>mon ami Marc.</mark></p>")
    '<p>Avec <mark>mon ami&nbsp;Marc.</mark></p>'
    >>> widont("Vraiment. Bien.")
    'Vraiment. Bien.'

    """

    widont_finder = re.compile(
        r"""((?:</?(?:a|em|span|strong|i|b|mark)[^>]*>)|[^<>\s\.]) # must be preceded by an approved inline opening or closing tag or a nontag/nonspace
            \s+                                                    # the space to replace
            ([^<>\s]+                                              # must be followed by non-tag non-space characters
            \s*                                                    # optional white space!
            (</(a|em|span|strong|i|b|mark)>\s*)*                   # optional closing inline tags with optional white space after each
            ((</(p|h[1-6]|li|dt|dd)>)|$))                          # end with a closing p, h1-6, li or the end of the string
        """,
        re.VERBOSE,
    )
    output = widont_finder.sub(r"\1&nbsp;\2", text)

    return output