A module for querying the DOM tree and writing XPath expressions using native Python syntax.
>>> from xpyth import xpath, DOM, X
>>> xpath(X for X in DOM if X.name == 'main')
"//*[@name='main']"
>>> xpath(span for div in DOM for span in div if div.id == 'main')
"//div[@id='main']//span"
>>> xpath(a for a in DOM if '.com' not in a.href)
"//a[not(contains(@href, '.com'))]"
>>> xpath(a.href for a in DOM if any(p for p in a.ancestors if p.id))
"//a[./ancestor::p[@id]]/@href"
>>> xpath(X.data-bind for X in DOM if X.data-bind == '1')
"//*[@data-bind='1']/@data-bind"
>>> xpath(
... form.action
... for form in DOM
... if all(
... input
... for input in form.children
... if input.value == 'a'
... )
... )
"//form[not(./input[not(@value='a')])]/@action"
>>> allowed_ids = list('abc')
>>> xpath(X for X in DOM if X.id in allowed_ids)
"//*[@id='a' or @id='b' or @id='c']"XPath is the de facto standard in querying XML and HTML documents. In Python (and most other languages), XPath expressions are represented as strings; this not only constitutes a potential security threat, but also means that developers are denied standard text-editor and IDE features such as syntax highlighting and autocomplete when writing XPaths. Furthermore, having to become familiar with XPath (or CSS selectors) presents a barrier to entry for developers who want to interact with the web.
Great inroads have been made in various programming languages in allowing the use of native list-comprehension-like syntax to generate SQL queries. xpyth piggybacks off one such effort, Pony, to extend this functionality to XPath. Now anyone familiar with Python comprehension syntax can query XML/HTML documents quickly and easily. Moreover, xpyth integrates with the popular lxml library to enable developers to go beyond the querying capabilities of XPath (when necessary).
pip install xpythxpyth supports querying lxml ElementTrees using the query function. For example, given a document
<html>
<div id='main' class='main'>
<a href='https://www.google.com'>Google</a>
<a href='https://www.chasestevens.com'>Not Google</a>
<p>Lorem ipsum</p>
<p id='123'>no numbers here</p>
<p id='numbers_only'>123</p>
</div>
<div id='123' class='secondary'>
<a href='https://www.google.org'>Google Charity</a>
<a href='https://www.chasestevens.org'>Broken link!</a>
</div>
</html>accessible as the ElementTree tree, the following can be executed:
>>> len(query(a for a in tree))
4
>>> query(a for a in tree if 'Not Google' not in a.text)[0].attrib.get('href')
"https://www.google.com"
>>> next(
... node
... for node in
... query(
... p
... for p in
... tree
... if p.id
... )
... if re.match(r'\D+', node.attrib.get('id'))
... ).text
"123"-
HTML tag names that contain special characters (dashes) cannot be selected, as they violate Python's generator comprehension syntax. HTML attributes containing dashes, e.g.
data-bind, work normally. -
The use of
allis quite buggy, e.g. the following return incorrect expressions:>>> xpath(X for X in DOM if all(p.id in ('a', 'b') for p in X)) "//*[not(.//p/@id='a' or //p/@id='b')]" # expected "//*[not(.//p[./@id!='a' and ./@id!='b'])]" >>> xpath(X for X in DOM if all('x' in p.id for p in X)) "//*[not(.contains(@id, //p))]" # expected "//*[not(.//p[not(contains(@id, 'x'))])]"
- Name: H. Chase Stevens
- Twitter: @hchasestevens