How to get the raw text from lxml element with Python

user_1677041 0
Viewed 1829

I want to get the following inline text strings from the root element.

from lxml import etree

root = root = etree.fromstring(
'''<p>
    text-first
    <span>
        Child 1
    </span>
    text-middle
    <span>
        Child 2
    </span>
    text-last
</p>''')

root.text only returns the "text-first" including newlines

>>> build_text_list = etree.XPath("//text()")

>>> texts = build_text_list(root)
>>>
>>> texts
['\n    text-first\n    ', '\n        Child 1\n    ', '\n    text-middle\n    ', '\n        Child 2\n    ', '\n    text-last\n']
>>>
>>> for t in texts:
...     print t
...     print t.__dict__
...

    text-first

{'_parent': <Element p at 0x10140f638>, 'is_attribute': False, 'attrname': None, 'is_text': True, 'is_tail': False}

        Child 1

{'_parent': <Element span at 0x10140be18>, 'is_attribute': False, 'attrname': None, 'is_text': True, 'is_tail': False}

    text-middle

{'_parent': <Element span at 0x10140be18>, 'is_attribute': False, 'attrname': None, 'is_text': False, 'is_tail': True}

        Child 2

{'_parent': <Element span at 0x10140be60>, 'is_attribute': False, 'attrname': None, 'is_text': True, 'is_tail': False}

    text-last

{'_parent': <Element span at 0x10140be60>, 'is_attribute': False, 'attrname': None, 'is_text': False, 'is_tail': True}
>>>
>>> root.xpath("./p/following-sibling::text()") # following https://stackoverflow.com/a/39832753/1677041
[]

So, how can I get the text-first/middle/last parts from this?

lxml xml python
4 Answers

etree is perfectly capable of this:

from lxml import etree

root: etree.Element = etree.fromstring(
'''<p>
    text-first
    <span>
        Child 1
    </span>
    text-middle
    <span>
        Child 2
    </span>
    text-last
</p>''')

print(
    root.text,
    root[0].tail,
    root[1].tail,
)

All elements are lists of their subelements, so the indexing here refers to the 2 <span> elements. The tail attribute of any element contains the text directly after that element.

It will of course include newlines, so you might want to strip() the results: root.text.strip()

user_43318850

Your initial guess, //text() means: select all text nodes, no matter where in the document they are. What you actually would like to select are text nodes if they are an immediate child of p, or alternatively if they are not children of span.

The most precise answer given the input document you are showing is /p/text():

>>> root = etree.fromstring(
'''<p>
text-first
<span>
    Child 1
</span>
text-middle
<span>
    Child 2
</span>
text-last
</p>''')

>>> etree.XPath("/p/text()")(root)
['\n    text-first\n    ', '\n    text-middle\n    ', '\n    text-last\n']

Your own solution child::text() means: select text nodes if they are a child of the current context node. It works because XPath expressions are in this case evaluated using the root element, p, as context. That is why simply text() also works.

>>> etree.XPath("text()")(root)
['\n    text-first\n    ', '\n    text-middle\n    ', '\n    text-last\n']
user_19875980

My bad, xpath saves me in the end.

>>> root.xpath('child::text()')
['\n    text-first\n    ', '\n    text-middle\n    ', '\n    text-last\n']
user_16770410 
print(root.xpath('normalize-space(//*)'))
user_53904780
Related