如何使用 Python ElementTree 从 XML 文件中的不同子级元素中提取相关属性

Question

I am new to Python and somewhat new to XML parsing and am struggling to find the right code algorithm to extract the data I need whilst maintaining the relationship between the child element\attributes.

The XML is from an Audio Recording software application. It defines configuration for multiple aspects of the product (so is large). I am wanting to extract some config items from a small part of the file for external use.

The parent path to the below sample is: ./Presets/rootObjects/root/list/item[]

<item>
   <string name="ID" value="InBusConfig" wide="true"/>
   <string name="Name" value="InBusConfig" wide="true"/>
   <list name="Items" type="obj">
      <obj class="FPreset" ID="1511747008">
         <string name="Name" value="Drum &amp; Bass Beds" wide="true"/>
         <member name="Object">
            <list name="Busses" type="list">
               <item>
                  <string ***name="BusName" value="Kick In"*** wide="true"/>
                  <int name="SpeakerArr" value="0"/>
                  <list name="Connections" type="list">
                     <item>
                        <string ***name="PortId" value="I|Focusrite USB ASIO|Input 3"*** wide="true"/>
                        <int name="Speaker" value="0"/>
                     </item>
                  </list>
               </item>
...

               <item>
                  <string name="BusName" value="Live Room" wide="true"/>
                  <int name="SpeakerArr" value="0"/>
                  <list name="Connections" type="list">
                     <item>
                        <string name="PortId" value="I|Focusrite USB ASIO|Digital 10" wide="true"/>
                        <int name="Speaker" value="0"/>
                     </item>
                  </list>
               </item>
            </list>
            <int name="Default Bus Index" value="0"/>
         </member>
         <int name="Unrenamed" value="1"/>
      </obj>
   </list>
</item>

The items I need to extract (and keep related) are the "BusName" attribute and the respective "PortId" attribute which is a child of the "list" element named "connections".

For the 13 items that exist in my test configuration I want to output this data as a csv (or JSON) file for use in another tool.

The format I want this output in would ideally be similar to: (I have replaced the pipe char with a "-" as it was messing with the table formatting)

BusName	PortId
Kick In	I - Focusrite USB ASIO - Input 3
Live Room	I - Focusrite USB ASIO - Digital 10

I have not provided any py code as nothing I have tried so far is close enough to ask a specific question. So I am requesting help of a more general nature as to how to approach this problem.

I'd be happy with an algorithm, psuedocode or even some python specific code/functions etc.

Any direction really would be more than I have right now.

Thanks in advance.

Answer 1

I recommend not using the built-in ElementTree, but rather work with a package like lxml or BeautifulSoup4 .

Edit: This is purely a matter of preference. As pointed out correctly by @barny, the exact same thing can be achieved with xml.etree.ElementTree .

Here's an attempt with lxml :

from lxml import etree

data = """
<item>
   <string name="ID" value="InBusConfig" wide="true"/>
   <string name="Name" value="InBusConfig" wide="true"/>
   <list name="Items" type="obj">
      <obj class="FPreset" ID="1511747008">
         <string name="Name" value="Drum &amp; Bass Beds" wide="true"/>
         <member name="Object">
            <list name="Busses" type="list">
               <item>
                  <string name="BusName" value="Kick In" wide="true"/>
                  <int name="SpeakerArr" value="0"/>
                  <list name="Connections" type="list">
                     <item>
                        <string name="PortId" value="I|Focusrite USB ASIO|Input 3" wide="true"/>
                        <int name="Speaker" value="0"/>
                     </item>
                  </list>
               </item>
               <item>
                  <string name="BusName" value="Live Room" wide="true"/>
                  <int name="SpeakerArr" value="0"/>
                  <list name="Connections" type="list">
                     <item>
                        <string name="PortId" value="I|Focusrite USB ASIO|Digital 10" wide="true"/>
                        <int name="Speaker" value="0"/>
                     </item>
                  </list>
               </item>
            </list>
            <int name="Default Bus Index" value="0"/>
         </member>
         <int name="Unrenamed" value="1"/>
      </obj>
   </list>
</item>

"""

root = etree.fromstring(data)

buses = root.xpath('//item[string/@name="BusName"]')

for bus in buses:
    bus_name = bus.find('string').get('value')
    port_id = bus.xpath('list/item/string/@value')[0]
    pair = (bus_name, port_id,)
    print(pair)

The general idea here is that it uses xpath to find a <item> , where there is a <string name="BusName"> .
From that item it takes:

• the value attribute of the <string> element.
• the first match (because xpath returns a list) of the value attribute of the list/item/string child.
  (If there were more items in this list, you'd want to adjust that here)

Note: I'm just creating a tuple called pair , but obviously you can also store these variables in a dataframe ( pandas ), or wirte it directly to json or csv (see eg here: https://realpython.com/python-csv/ ).

Answer 2

Not sure why @balduin recommends lxml/avoids ElementTree - it works perfectly well.

ElementTree has limited xpath support compared to lxml but it's perfectly sufficient for many tasks and avoiding need for an external package is often helpful. Docs are here: https://docs.python.org/3/library/xml.etree.elementtree.html?highlight=findall#xml.etree.ElementTree.Element.findall

There are some xpath examples further up that page

import lxml.etree as ETL
import xml.etree.ElementTree as ET

data = """
<item>
   <string name="ID" value="InBusConfig" wide="true"/>
   <string name="Name" value="InBusConfig" wide="true"/>
   <list name="Items" type="obj">
      <obj class="FPreset" ID="1511747008">
         <string name="Name" value="Drum &amp; Bass Beds" wide="true"/>
         <member name="Object">
            <list name="Busses" type="list">
               <item>
                  <string name="BusName" value="Kick In" wide="true"/>
                  <int name="SpeakerArr" value="0"/>
                  <list name="Connections" type="list">
                     <item>
                        <string name="PortId" value="I|Focusrite USB ASIO|Input 3" wide="true"/>
                        <int name="Speaker" value="0"/>
                     </item>
                  </list>
               </item>
               <item>
                  <string name="BusName" value="Live Room" wide="true"/>
                  <int name="SpeakerArr" value="0"/>
                  <list name="Connections" type="list">
                     <item>
                        <string name="PortId" value="I|Focusrite USB ASIO|Digital 10" wide="true"/>
                        <int name="Speaker" value="0"/>
                     </item>
                  </list>
               </item>
            </list>
            <int name="Default Bus Index" value="0"/>
         </member>
         <int name="Unrenamed" value="1"/>
      </obj>
   </list>
</item>

"""

# lxml
root = ETL.fromstring(data)

buses = root.xpath('//item[string/@name="BusName"]')

for bus in buses:
    bus_name = bus.find('string').get('value')
    port_id = bus.xpath('list/item/string/@value')[0]
    pair = (bus_name, port_id,)
    print(pair)

# ElementTree
root1 = ET.fromstring(data)

for el in root1.findall('.//item/string[@name="BusName"]/..'):
    bus_name = el.find('./string').get('value')
    port_id=el.find("./list/item/string").get('value')
    pair=(bus_name,port_id)
    print(pair)

Output lxml:

('Kick In', 'I|Focusrite USB ASIO|Input 3')
('Live Room', 'I|Focusrite USB ASIO|Digital 10')

Output ElementTree:

('Kick In', 'I|Focusrite USB ASIO|Input 3')
('Live Room', 'I|Focusrite USB ASIO|Digital 10')