etree Clone Node

Question

How to clone Element objects in Python xml.etree? I'm trying to procedurally move and copy (then modify their attributes) nodes.

Geehan

Answer 1

You can just use copy.deepcopy() to make a copy of the element. (this will also work with lxml by the way).

Luxurious

That makes a copy, but it's not added to the tree. You'll need to do an append() or insert() to do that. – Aitken

Answer 2

A different, and somewhat disturbing solution:

new_element = lxml.etree.fromstring(lxml.etree.tostring(elem))

Hectograph

bahm, you just saved my life! This is really useful when replacing values – Benzel

This is indeed a disturbing solution :) – Pendley

Hah, love it! Indeed disturbing. – Hoelscher

Answer 3

If you have a handle on the Element elem's parent you can call

new_element = SubElement(parent, elem.tag, elem.attrib)

Otherwise you might want to try

new_element = makeelement(elem.tag, elem.attrib)

but this is not advised.

Messinger

@SHiNKiROU You can compare id(old_element) with id(new_element) to see if it actually creates a different object in memory. Does this help? – Messinger

As @Geehan mentions, this does not copy the children. – Fleshings

This is useful when you want to copy an Element and its attributes, but you do not want to copy the children (for example when reconstructing a strict subtree by iterating through an element's ancestors). It is portable to lxml.etree, because unfortunately with lxml.etree, copy.copy() also copies children (documented, but how is this different from a deepcopy?). – Shellashellac

Answer 4

At least in Python 2.7 etree Element has a copy method: http://hg.python.org/cpython/file/2.7/Lib/xml/etree/ElementTree.py#l233

It is a shallow copy, but that is preferable in some cases.

In my case I am duplicating some SVG Elements and adding a transform. Duplicating children wouldn't serve any purpose since where relevant they already inherit their parent's transform.

Hetero

For anyone using this and thinking of replacing xml.etree.ElementTree with lxml.etree in the future, note that Element.copy() does not exist in lxml.etree, and copy.copy() copies children too, when applied to an lxml.etree.Element. – Shellashellac

Does not work either with cElementTree (Python 2.7). So prefer copy.copy() (shallow copy) or copy.deepcopy() for code evolutivity. – Adjacent

Answer 5

If you procedurally move through your tree with loops, you can use insert to clone directly ( insert(index, subelement) ) and tree indexing (both in the documentation):

import xml.etree.ElementTree as ET
mytree = ET.parse('some_xml_file.xml')  # parse tree from xml file
root = mytree.getroot()  # get the tree root
    
for elem in root:  # iterate over children of root
   if condition_for_cloning(elem) == True:      
      elem.insert(len(elem), elem[3])  # insert the 4th child of elem to the end of the element (clone an element)  

or for children with some tag:

for elem in root:
   children_of_interest = elem.findall("tag_of_element_to_clone")
   elem.insert(len(elem), children_of_interest[1])

Virgy

Answer 6

For anyone visiting from the future:

If you want to clone the entire element, use append.

new_tree = ET.Element('root')
for elem in a_different_tree:
    new_tree.append(elem)

@dennis-williamson made a comment about it which I overlooked and eventually stumbled on the answer here https://mcmap.net/q/340005/-can-39-t-dump-or-write-an-elementtree-element

Alkalinity

To be clear, this will remove the element from the original tree! – Hoelscher

@Hoelscher it won't remove anything.. but it will make a copy-by-reference, not a clone. So any changes to the new element will alter the original – Expressivity

Answer 7

For future reference.

Simplest way to copy a node (or tree) and keep it's children, without having to import ANOTHER library ONLY for that:

def copy_tree( tree_root ):
    return et.ElementTree( tree_root );

duplicated_node_tree = copy_tree ( node );    # type(duplicated_node_tree) is ElementTree
duplicated_tree_root_element = new_tree.getroot();  # type(duplicated_tree_root_element) is Element

Claudclauddetta

To be clear, this is not a deep copy. (Yes, the post says "and keep its children," but I still felt the need to test what it meant.) – Watkins