Make Seemingly Difficult Tasks Trivial

By Kevin Quillen, on October 31, 2017

Earlier today there was a ping in Slack for assistance. A coworker needed to update 140 XML files with a new element, and the element value will be populated from an existing element in the node.

For most people, they hear the number “140” and think wow, this will take a while. But the number of documents here isn’t the issue, we just need to solve the problem once and apply it everywhere.

My mind quickly jumped to a shell script and some sort of sed magic. Before I opened up Vim, I jotted down what path I needed:

  1. loop files in dir
  2. find xml node
  3. get text value attribute of node
  4. append new node with copied value as its value

Once I took a look at 2-4, I figured sed is probably not the best tool here. One small Google search brought me to xmlstarlet.

It doesn’t say this on the xmlstarlet site, but for macOS users you can use Homebrew to install it:


brew install xmlstarlet

The syntax and arguments for xmlstarlet are a little vague, but with my tasks outlined I knew what I was looking to do in general.

I fetched a dummy XML file online with made up data, and put it into a new directory. I then generated 140 files with tee:


tee xmltest-{1..140}.xml < xmltest-original.xml >/dev/null

With our test data ready, I could start tackling the problem at hand. The actual script part was not so bad, I had to look up how to get xmlstarlet to do what I wanted. But it made traversing, selecting, and appending xml so easy. I wound up with a working script in about 10 minutes:


#!/bin/bash

set -e
shopt -s nullglob

for file in *.xml
do
  SETTING_VALUE=`xmlstarlet sel -t -m "/dataset/record" -v ip_address $file`;
  echo "Appending $SETTING_VALUE as new_ip_address in $file";
  xmlstarlet ed -L -s "/dataset/record" -t elem -n new_ip_address -v "$SETTING_VALUE" $file;
done

This looks for an existing node called <ip_address>, gets the text value, and inserts a new element called <new_ip_address> with that value. The result in all 140 test files looked like this:


<dataset>
  <record>
    <id>1</id>
    <first_name>Ambrosius</first_name>
    <last_name>Alwin</last_name>
    <email>aalwin0@vk.com</email>
    <gender>Male</gender>
    <ip_address>166.77.108.157</ip_address>
    <new_ip_address>166.77.108.157</new_ip_address>
  </record>
</dataset>

Exciting stuff. I think I have some XML file tasks coming up for a new project we are working on, so I will keep xmlstarlet in mind as a useful tool depending on what we need to do.

Tagged with: shell, zsh
comments powered by Disqus