[parser] find a better solution than pulling the entire wiki contents into memory

[Joe Talbott]

Facundo Batista recommends something along these lines for the part processing

def splitgen(text, sep):
    pos_from = 0
    while True:
        try:
            pos_to = text.index(sep, pos_from)
        except ValueError:
            yield text[pos_from:]
            break

        yield text[pos_from:pos_to]
        pos_from = pos_to + len(sep)

test = [
    'foo',
    'foo\n---\nbar',
    'foo\n---\n',
    'foo\n---\nbar\n---\n',
    'foo\n---\n\n---\nbar\n---\n',
]
for t in test:
    assert t.split("\n---\n") == list(splitgen(t, "\n---\n"))