Select fails when XML contains leading comment

[sean-moore3]

Hi,

I stumbled upon this issue while upgrading from 4.1.5. I can reproduce in 4.2.0 and later.

import lxml.etree
import elementpath


root = lxml.etree.XML("<!--comment--><root><trunk><branch></branch></trunk></root>")
trunk = elementpath.select(root, "trunk")
print(trunk)
"""[]"""
root = lxml.etree.XML("<root><trunk><branch></branch></trunk></root>")
trunk = elementpath.select(root, "trunk")
print(trunk)
"""[<Element trunk at 0x102b862c0>]"""

[brunato]

Hi, with v4.2.0 something is changed with node tree build, see:

https://elementpath.readthedocs.io/en/latest/advanced.html#the-context-root-and-the-context-item

The change has been necessary to handle XML document and fragments. A root node without siblings can skip the document position if not explicitly selected by the XPath expression (e.g. /root/trunk). A comment sibling of the root element can't be ignored so the initial position is set to the document.

The keyword arguments item and fragment can be used to set the initial node or to skip the dummy document creation.

[sean-moore3]

Thanks! I found fragment to be a good fit for my use case.