pxx-13 Duplicate errors due to long element names

[mluis7]

UPDATE 2025-10-10

libxml2 issue 715 was released with version 2.15.0.
Latest lxml 6.0.2 uses libxml2 2.14.6 so the fix is not yet mainstream.

import sys
from lxml import etree

consts = [ ('Python', '.'.join([str(sys.version_info[i]) for i in range(3)])),
('lxml.etree', etree.LXML_VERSION),
('libxml used', etree.LIBXML_VERSION),
('libxml compiled', etree.LIBXML_COMPILED_VERSION),
('libxslt used', etree.LIBXSLT_VERSION),
('libxslt compiled', etree.LIBXSLT_COMPILED_VERSION)]

for c in consts:
    print(f"{c[0]:<20s}: {c[1]}")

Python              : 3.12.11
lxml.etree          : (6, 0, 2, 0)
libxml used         : (2, 14, 6)
libxml compiled     : (2, 14, 6)
libxslt used        : (1, 1, 43)
libxslt compiled    : (1, 1, 43)

UPDATE 2025-07-19:
libxml2 issue 715 is fixed upstream but it may take sometime until it's released.

UPDATE:
Reported as libxml2 bug since xmllint shows same behavior.
https://gitlab.gnome.org/GNOME/libxml2/-/issues/715

Long element names may cause some duplicate and count errors while parsing

ERROR: duplicated path: /xbrli:xbrl/in-bse-cg:AggregateValueOfSecurityProvidedDuringSixMonthsOfSecurityInConnectionWithLoanOrAnyOtherD
...
ERROR: 0 elements found with /xbrli:xbrl/in-bse-cg:AggregateAmountAdvancedDuringSixMonthsOfAnyLoanOrAnyOtherFormOfDebtToPromoterOrAnyOtherE xpath expression.
Original xpath: /xbrli:xbrl/in-bse-cg:AggregateAmountAdvancedDuringSixMonthsOfAnyLoanOrAnyOtherFormOfDebtToPromoterOrAnyOtherE

Cause:
tree.getpath(ele) truncates the returned path to 110 characters.

Possible fix:
tree.getelementpath(ele)

Reproduce
Given an XML with namespaces

<root xmlns:ns="http://example.com/ns">
    <ns:a01234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678z01234567890123456789012345678x>1</ns:a01234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678z01234567890123456789012345678x>
</root>

>>> from lxml import etree

>>> xtree = etree.parse('tmp.xml')
>>> root = xtree.getroot()

truncated path

>>> xtree.getpath(root.getchildren()[0])
'/root/ns:a0123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123'

complete path with tree.getelementpath()

>>> xtree.getelementpath(root.getchildren()[0])
'{http://example.com/ns}a01234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678z01234567890123456789012345678x'

While trying to report an lxml bug I found that it's an libxml2 bug as xmllint shows the same behavior

xmllint --shell tmp.xml
/ > du
/
root
  ns:a01234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678z01234567890123456789012345678x
/ > setrootns
/ > setns default=
/ > setns ns=http://example.com/ns
/ > whereis //ns:a01234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678z01234567890123456789012345678x
/root/ns:a0123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123