Reliably make a clean copy of an XML node

TimG · May 31, 2025, 7:57am

I am using XML.jl.

I have a basic XML node that I use as a base from which I wish to construct many slightly different instances. On each occasion, I want to copy the basic node and build from there. The Readme page for XML.jl says:

Node is an immutable type. However, you can easily create a copy with one or more field values changed by using the Node(::Node; kw...) constructor where kw are the fields you want to change.

I don’t want to change any fields, So I just tried new_node=XML.Node(old_node).

I then build up the rest of the new_node as I want it.

The consequence of this is that the changes I make to new_node also propagate back to old_node - specfically to its child nodes.

I have experimented with many different ways of specifying the kw... options in XML.Node but to no avail. For example, I tried variations on new_node=XML.Node(old_node, children=[x for x in XML.children(old_node)]) to try to copy the child nodes, too.

Two ways I have found to make this work are to try deepcopy(old_node) and XML.parse(Node, XML.write(old_node))[1].

The first of these seems not to be recommended (“A brief conversation of deepcopy vs copy in Julia and why deepcopy should be avoided unless you know you have a good reason to use it (serialization-like)” (here)).

The second is very inefficient.

MWE below the fold

using XML
copynode(o)=Node(o)
deepcopynode(o) = deepcopy(o)
parsenode(o) = XML.parse(Node, XML.write(o))[1]
function createRule()
    rule = XML.Element("x14:cfRule", type="iconSet", priority="1", id="XXXX-xxxx-XXXX")
    icon = XML.Element("x14:iconSet", iconSet="3Arrows", custom="1")
    cfvo = XML.Element("x14:cfvo", type="percent")
    push!(cfvo, XML.Element("xm:f", XML.Text("0")))
    push!(icon, cfvo)
    push!(rule, icon)
    return rule
end
function createCfx(rule, f)
    cfx=f(rule)
    cfvo = XML.Element("x14:cfvo", type="percent")
    push!(cfvo, XML.Element("xm:f", XML.Text("dummy")))
    push!(cfx[1], f(cfvo))
    push!(cfx[1], f(cfvo))
    push!(cfx[1], f(cfvo))
    cfx[1]["iconSet"] = "4Arrows"
    return cfx
end

println("\nusing XML.Node()")
rule = createRule()
println("\n `rule` before:\n",XML.write(rule))
cfx=createCfx(rule, copynode)
println("\n `cfx`:\n",XML.write(cfx))
println("\n `rule` after:\n",XML.write(rule))

println("\nusing deepcopy()")
rule = createRule()
println("\n `rule` before:\n",XML.write(rule))
cfx=createCfx(rule, deepcopynode)
println("\n `cfx`:\n",XML.write(cfx))
println("\n `rule` after:\n",XML.write(rule))

println("\nusing XML.parse()")
rule = createRule()
println("\n `rule` before:\n",XML.write(rule))
cfx=createCfx(rule, parsenode)
println("\n `cfx`:\n",XML.write(cfx))
println("\n `rule` after:\n",XML.write(rule))

using XML.Node()

 `rule` before:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="3Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

 `cfx`:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="4Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

 `rule` after:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="4Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

using deepcopy()

 `rule` before:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="3Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

 `cfx`:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="4Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

 `rule` after:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="3Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

using XML.parse()

 `rule` before:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="3Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

 `cfx`:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="4Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

 `rule` after:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="3Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

Does anyone have any suggestions?

TimG · May 31, 2025, 9:19am

Thank you for the suggestion @Pamela329Lac.

I did try copynode(o)=XML.Node(o, children=[XML.Node(x) for x in o.children]) but I this gives

 `rule` after:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="4Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
    <x14:cfvo children="Node[Node Element <xm:f> (1 child)]" type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo children="Node[Node Element <xm:f> (1 child)]" type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
    <x14:cfvo children="Node[Node Element <xm:f> (1 child)]" type="percent">
      <xm:f>dummy</xm:f>
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

This suggests that Node thinks that children=... is a keyword argument specifying a children attribute - which it isn’t.

copynode(o)=XML.Node(o, [XML.Node(x) for x in o.children]) does no better:

 `rule` after:
<x14:cfRule type="iconSet" priority="1" id="XXXX-xxxx-XXXX">
  <x14:iconSet iconSet="3Arrows" custom="1">
    <x14:cfvo type="percent">
      <xm:f>0</xm:f>
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
      Node[Node Element <xm:f> (1 child)]
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
      Node[Node Element <xm:f> (1 child)]
    </x14:cfvo>
    <x14:cfvo type="percent">
      <xm:f>dummy</xm:f>
      Node[Node Element <xm:f> (1 child)]
    </x14:cfvo>
  </x14:iconSet>
</x14:cfRule>

nilshg · May 31, 2025, 10:46am

Tim, I have no answer to your question but because this has happened to you for the second time in the last couple of days I thought I’d mention that these plausible-sounding replies from Firstname[Number]Lastname users are GenAI spam, so there’s little point engaging with them.

(If you do find them helpful of course you might want to try simply talking to one of the better models directly before posting here!)

TimG · May 31, 2025, 11:24am

Understood! The like was to try to be welcoming

TimG · May 31, 2025, 12:17pm

On a random walk through options, I seem to have hit upon

copynode(o)=XML.Node(o.nodetype, o.tag, o.attributes, o.value, isnothing(o.children) ? nothing : [copynode(x) for x in o.children])

Which recursively copies a node and all its children.

As far as I can tell, this gives the same result as deepcopy but is quite a bit faster. Here, I’m just using an XML file representing an Excel worksheet (nested a few levels deep).

julia> using XML, BenchmarkTools, XLSX

julia> copynode(o)=XML.Node(o.nodetype, o.tag, o.attributes, o.value, isnothing(o.children) ? nothing : [copynode(x) for x in o.children])
copynode (generic function with 1 method)

julia> deepcopynode(o) = deepcopy(o)
deepcopynode (generic function with 1 method)

julia> function createCfx(rule, f)                                                                                                                                                                     
           cfx=f(rule)                                                                                                                                                                                 
           cfvo = XML.Element("x14:cfvo", type="percent")                                                                                                                                              
           push!(cfvo, XML.Element("xm:f", XML.Text("dummy")))                                                                                                                                     
           push!(cfx[end], f(cfvo))                                                                                                                                                            
           push!(cfx[end], f(cfvo))                                                                                                                                                        
           push!(cfx[end], f(cfvo))                                                                                                                                                                    
           cfx[1]["iconSet"] = "4Arrows"                                                                                                                                                               
           return cfx                                                                                                                                                                              
       end
createCfx (generic function with 1 method)

julia> f=XLSX.opentemplate(raw"C:\Users\tim\OneDrive\Documents\Julia\XLSX\iconKey.xlsx")
XLSXFile("C:\Users\tim\OneDrive\Documents\Julia\XLSX\iconKey.xlsx") containing 1 Worksheet
            sheetname size          range
-------------------------------------------------
               Sheet1 4x13          A1:M4

julia> rule = XLSX.get_worksheet_xml_document(f[1])
Node Document (2 children)

julia> @benchmark cfx=createCfx($rule, $copynode)
BenchmarkTools.Trial: 10000 samples with 1 evaluation per sample.
 Range (min … max):   86.100 μs …   7.979 ms  ┊ GC (min … max): 0.00% … 97.99%
 Time  (median):      93.900 μs               ┊ GC (median):    0.00%
 Time  (mean ± σ):   109.910 μs ± 153.160 μs  ┊ GC (mean ± σ):  6.75% ±  5.06%

     ▁▇█▃
  ▁▂▄████▇▄▂▂▂▁▁▁▁▂▂▂▂▃▃▂▂▂▂▂▂▁▂▂▃▃▃▃▂▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ ▂
  86.1 μs          Histogram: frequency by time          159 μs <

 Memory estimate: 205.73 KiB, allocs estimate: 4520.

julia> rule2 = XLSX.get_worksheet_xml_document(f[1])
Node Document (2 children)

julia> cfx2=createCfx(rule2, deepcopynode)
Node Document (2 children)

julia> rule2 = XLSX.get_worksheet_xml_document(f[1])
Node Document (2 children)

julia> @benchmark cfx2=createCfx($rule2, $deepcopynode)
BenchmarkTools.Trial: 9872 samples with 1 evaluation per sample.
 Range (min … max):  438.400 μs …   4.798 ms  ┊ GC (min … max): 0.00% … 87.00%
 Time  (median):     456.900 μs               ┊ GC (median):    0.00%
 Time  (mean ± σ):   505.863 μs ± 259.013 μs  ┊ GC (mean ± σ):  4.52% ±  7.66%

   ▅▇█▇▆▆▅▄▂▁▁▁            ▁▂▂▂▂▁          ▁▁▁▁▁▁▁▂▁▁           ▂
  ▆██████████████▇▆▆▅▆▇▇▆▇██████████▇▆▇▇▇▇████████████▇▇▆▅▅▄▄▄▄ █
  438 μs        Histogram: log(frequency) by time        698 μs <

 Memory estimate: 537.81 KiB, allocs estimate: 12190.

julia> cfx=createCfx(rule, copynode)
Node Document (2 children)

julia> rule = XLSX.get_worksheet_xml_document(f[1])
Node Document (2 children)

julia> cfx=createCfx(rule, copynode)
Node Document (2 children)

julia> rule2 = XLSX.get_worksheet_xml_document(f[1])
Node Document (2 children)

julia> cfx2=createCfx(rule2, copynode)
Node Document (2 children)

julia> cfx==cfx2
true

Topic		Replies	Views
Convert LightXML array to XMLElement? New to Julia	3	369	September 7, 2020
[ANN] Announcing XML.jl Package Announcements xml	29	2127	June 9, 2023
Sorted Attributes in XML.jl? General Usage	8	242	June 14, 2023
Help with data movement and commuication Julia at Scale	1	513	September 8, 2018
Copy of immutable with multiple fields changed General Usage question	3	925	October 10, 2019

Reliably make a clean copy of an XML node

Related topics