Python, Lxml - Access Text

I m currently a bit out of ideas, and I really hope that you can give me a hint: Its probably best to explain my question with a small piece of sample code: from lxml import etree

Solution 1:

You could just reimplement itertext() function and insert special handlers for ul, table if necessary:

from lxml import html

def itertext(root, handlers=dict(ul=lambda el: (list(el.itertext()),
                                                el.tail))):
    if root.text:
        yield root.text
    for el in root:
        yield from handlers.get(el.tag, itertext)(el)
    if root.tail:
        yield root.tail

print(list(itertext(html.fromstring(
                "<b>text0<i>text1</i><ul><li>item1</li>"
                "<li>item2</li></ul>text2<b/><b>sib</b>"))))

Output

['text0', 'text1', ['item1', 'item2'], 'text2', 'sib']

Note: yield from X could be replaced by for x in X: yield x on older than Python 3.3 versions.

To join adjacent strings:

def joinadj(iterable, join=' '.join):
    adj = []
    for item in iterable:
        if isinstance(item, str):
            adj.append(item) # save for later
        else:
            if adj: # yield items accumulated so far
                yield join(adj)
                del adj[:] # remove yielded items
            yield item # not a string, yield as is

    if adj: # yield the rest
        yield join(adj)

print(list(joinadj(itertext(html.fromstring(
                "<b>text0<i>text1</i><ul><li>item1</li>"
                "<li>item2</li></ul>text2<b/><b>sib</b>")))))

Output

['text0 text1', ['item1', 'item2'], 'text2 sib']

To allow tables, nested list in <ul> the handler should call itertext() recursively:

def ul_handler(el):
    yield list(itertext(el, with_tail=False))
    if el.tail:
        yield el.tail

def itertext(root, handlers=dict(ul=ul_handler), with_tail=True):
    if root.text:
        yield root.text
    for el in root:
        yield from handlers.get(el.tag, itertext)(el)
    if with_tail and root.tail:
        yield root.tail

print(list(joinadj(itertext(html.fromstring(
                    "<b>text0<i>text1</i><ul><li>item1</li>"
                    "<li>item2<ul><li>sub1<li>sub2</li></ul></ul>"
                    "text2<b/><b>sib</b>")))))

Output

['text0 text1', ['item1', 'item2', ['sub1', 'sub2']], 'text2 sib']