Skip to content

Instantly share code, notes, and snippets.

@ergoithz

ergoithz/xpath_soup.py

Last active March 18, 2024 16:11
Show Gist options
  • Star 65 You must be signed in to star a gist
  • Fork 18 You must be signed in to fork a gist
  • Save ergoithz/6cf043e3fdedd1b94fcf to your computer and use it in GitHub Desktop.
Save ergoithz/6cf043e3fdedd1b94fcf to your computer and use it in GitHub Desktop.
Download ZIP
Generate unique XPATH for BeautifulSoup element
Raw
xpath_soup.py
#!/usr/bin/python
# -*- coding: utf-8 -*-
import bs4
def xpath_soup(element):
# type: (typing.Union[bs4.element.Tag, bs4.element.NavigableString]) -> str
"""
Generate xpath from BeautifulSoup4 element.
:param element: BeautifulSoup4 element.
:type element: bs4.element.Tag or bs4.element.NavigableString
:return: xpath as string
:rtype: str
Usage
-----
>>> import bs4
>>> html = (
... '<html><head><title>title</title></head>'
... '<body><p>p <i>1</i></p><p>p <i>2</i></p></body></html>'
... )
>>> soup = bs4.BeautifulSoup(html, 'html.parser')
>>> xpath_soup(soup.html.body.p.i)
'/html/body/p[1]/i'
>>> import bs4
>>> xml = (
... '<?xml version="1.0" encoding="UTF-8"?>'
... '<doc xmlns:ns1="http://localhost/ns1"'
... ' xmlns:ns2="http://localhost/ns2">'
... '<ns1:elm/><ns2:elm/><ns2:elm/></doc>'
... )
>>> soup = bs4.BeautifulSoup(xml, 'lxml-xml')
>>> xpath_soup(soup.doc.find('ns2:elm').next_sibling)
'/doc/ns2:elm[2]'
"""
components = []
target = element if element.name else element.parent
for node in (target, *target.parents)[-2::-1]: # type: bs4.element.Tag
tag = '%s:%s' % (node.prefix, node.name) if node.prefix else node.name
siblings = node.parent.find_all(tag, recursive=False)
components.append(tag if len(siblings) == 1 else '%s[%d]' % (tag, next(
index
for index, sibling in enumerate(siblings, 1)
if sibling is node
)))
return '/%s' % '/'.join(components)
if __name__ == '__main__':
import doctest
doctest.testmod(verbose=True, raise_on_error=True)
@vskritsky
Copy link

Smoothly generates xpaths from BS objects

@amoe
Copy link

amoe commented Nov 1, 2018

Freakin' sweet, thanks :)

@ilkerceng
Copy link

Thank you so much 💯

@eunorek
Copy link

eunorek commented Mar 21, 2019

This is amazing. Thanks a million!!

@httran13
Copy link

YOU ARE THE GENIUS!

@oookawesome
Copy link

So amazing:)

@ostashkevych
Copy link

grate tool! But still one bug

<div>
    <span>text</span>
    <p class="p14">identical tag</p>
    <p class="p14">identical tag</p>
</div>

will return the same xpath for both p tags
in row 35
siblings.index(child)

@callahantiff
Copy link

Awesome script, thanks! 🎉

@ergoithz
Copy link
Author

ergoithz commented Jul 12, 2019 •
edited

grate tool! But still one bug

<div>
    <span>text</span>
    <p class="p14">identical tag</p>
    <p class="p14">identical tag</p>
</div>

will return the same xpath for both p tags
in row 35
siblings.index(child)

Good point, I was relying on BeautifulSoup objects identity (which is probably __hash__ ) which looks like cannot be trusted in your case, so I updated the snippet (unfortunately, with a 3% performance hit on my tests).

@rdlou
Copy link

rdlou commented Jul 30, 2019

This is great. How would you include the attributes in this path. Like if the p had class="test" you'd get '/html/body/div/p.test' ? Thanks!

@ergoithz
Copy link
Author

This is great. How would you include the attributes in this path. Like if the p had class="test" you'd get '/html/body/div/p.test' ? Thanks!

This script relies purely on DOM tree node location, so no need for other filtering.

If your DOM tree is subject to a lot of changes, this script (and probably any automated script) will not suit your case.

@Zarmeena
Copy link

Zarmeena commented Aug 6, 2019

Really useful for generating XPath !
Currently it is generating absolute XPath like : /html/body/input. How can we generate a relative XPath for the same element?

@ergoithz
Copy link
Author

Currently it is generating absolute XPath like : /html/body/input. How can we generate a relative XPath for the same element?

Generate xpath for both parent and child elements, strip the parent xpath out.

@rosstex
Copy link

rosstex commented Aug 21, 2020 •
edited

Should this work for clicking with pyppeteer? Because it's unable to find the elements based on these selectors.

EDIT: One possible theory is that the DOM changes between parsing and clicking. Will investigate.

@xoxwgys56
Copy link

thanks it is super awesome

@eonnyhoney
Copy link

thank you this will be very useful

@cuskim
Copy link

cuskim commented Nov 13, 2020

it is very useful for me..
components.append(
child.name if siblings == [child] else
'%s[%d]' % (child.name, 1 + siblings.index(child))
)

    if (id_ := child.get('id')):
        components[-1] = f"/{child.name}[@id='{id_}']"
        break
       # like this ==>  //div[@id="MYC0101001S_mdlStszTbl_divButtons"]/span[2]/button[3]

    child = parent

@ergoithz
Copy link
Author

it is very useful for me..
components.append(
child.name if siblings == [child] else
'%s[%d]' % (child.name, 1 + siblings.index(child))
)

    if (id_ := child.get('id')):
        components[-1] = f"/{child.name}[@id='{id_}']"
        break
       # like this ==>  //div[@id="MYC0101001S_mdlStszTbl_divButtons"]/span[2]/button[3]

    child = parent

siblings.index(child) does not work: see this previous comment.

@rosstex
Copy link

rosstex commented Feb 5, 2021 •
edited

I've created a new version that uses attributes rather than sibling position. This is robust to actively changing DOMs, suitable for web crawling.
https://gist.github.com/rosstex/bc0df9db72833bcf6872f9ba8ec5db06

@sunil209
Copy link

sunil209 commented Feb 28, 2021 •
edited

Hi @ergoithz

This looks throwing wrong error. If not then can you please guide me the error in below HTML5 semantic .

 <footer id="copyright-container">
            <div id="copyright-content">
                <small class="copyright">Copyright 2021-21</small>
                <p> All Rights Reserved.</p>
            </div>
</footer>

Below Error I am getting

E AssertionError: current tag id: copyright-content (located at xpath: /html/body/footer/div) has wrong usage (xpath algorithm code: https://gist.github.com/ergoithz/6cf043e3fdedd1b94fcf)

I tried to change div to section/article but it does not work

@manisha-sharma
Copy link

Hi @ergoithz

This looks throwing wrong error. If not then can you please guide me the error in below HTML5 semantic .

 <footer id="copyright-container">
            <div id="copyright-content">
                <small class="copyright">Copyright 2021-21</small>
                <p> All Rights Reserved.</p>
            </div>
</footer>

Below Error I am getting

E AssertionError: current tag id: copyright-content (located at xpath: /html/body/footer/div) has wrong usage (xpath algorithm code: https://gist.github.com/ergoithz/6cf043e3fdedd1b94fcf)

I tried to change div to section/article but it does not work

Any Solution???

@ergoithz
Copy link
Author

ergoithz commented May 12, 2021 •
edited

Hi @ergoithz
This looks throwing wrong error. If not then can you please guide me the error in below HTML5 semantic .

 <footer id="copyright-container">
            <div id="copyright-content">
                <small class="copyright">Copyright 2021-21</small>
                <p> All Rights Reserved.</p>
            </div>
</footer>

Below Error I am getting
E AssertionError: current tag id: copyright-content (located at xpath: /html/body/footer/div) has wrong usage (xpath algorithm code: https://gist.github.com/ergoithz/6cf043e3fdedd1b94fcf)
I tried to change div to section/article but it does not work

Any Solution???

Sorry, but I do not have solutions to random poorly-reported errors triggered by third-party libraries well outside this snippet execution, but this probably caused by a lack of understanding on how XPATH works and what this snippet really does.

Things to try:

  • Other bs4 parsers.
  • Asking the correct people (whoever owns the code which triggers that assertion error).
  • Checking ids are not duplicated in the document (even tho my code does not relies on them).

@ACB-prgm
Copy link

Amazing! Thank you so much!

@midrare
Copy link

midrare commented Sep 28, 2021

Thanks a bunch! Do you have a license for the code?

@PttCodingMan
Copy link

You are my hero.

@ergoithz
Copy link
Author

ergoithz commented Aug 31, 2022 •
edited

I'm going to start removing every single Codility-related comment here. This is not the right place for cheaters/lazy people to share that kind of stuff.

@xandross389
Copy link

Wow it is awesome, you are a genius. Thanks!

@funway
Copy link

funway commented Dec 22, 2023

nice job 👍

@funway
Copy link

funway commented Dec 28, 2023

Hi @ergoithz,
I faced a little problem if the XML element had a namespace prefix, the xpath returned by this function would not include the prefix. Looks like that:

<?xml version="1.0" encoding="UTF-8"?>
<xfa:datasets xmlns:xfa="http://www.xfa.org/schema/xfa-data/1.0/">
    <data>ABC</data>
</xfa:datasets>

The output of tag will be /datasets/data, but I preferred it kept /xfa:datasets/data

So I modified the code to let it keep the XML namespace prefix. Hope you have some time to review it.

def xpath_soup(element, keep_prefix: bool=True):
    components = []
    child = element if element.name else element.parent
    for parent in child.parents:  # type: bs4.element.Tag
        siblings = parent.find_all(child.name, recursive=False)
        child_name = f'{child.prefix}:{child.name}' if child.prefix and keep_prefix else child.name
        components.append(
            child_name if 1 == len(siblings) else '%s[%d]' % (
                child_name,
                next(i for i, s in enumerate(siblings, 1) if s is child)
                )
            )
        child = parent
    components.reverse()
    return '/%s' % '/'.join(components)

@ergoithz
Copy link
Author

@funway thanks for the heads up, snippet updated with your suggestion.

(You might take a look at your code tho, your sibling search wasn't prefixed).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment