3

I need to grab attributes from the XML file with pure bash script.

So I have the following XML file with a root element Group and lots of Person elements, every of them has id and username attributes. id is unique value for each element:

<?xml version="1.0" encoding="UTF-8"?>
<Group id="D_8"
       main="false">

    <Person id="P_0001"
            email="[email protected]"
            username="person_0001"
            password="pass_0001"
            active="true"/>

    <Person id="P_0002"
            email="[email protected]"
            username="person_0002"
            password="pass_0002"
            active="true"/>

    <!--  ...and hundreds of other Person elements ...  -->
</Group>

And I need to use bash script to extract the id and username attributes into some key-value structure:

P_0001=person_0001
P_0002=person_0002

Checked other related answers, but most of them suggest to use some XML parsers like xmllint. But unfortunately I do not have them on the target machine.

Please suggest how I can achieve this.

Improve this question
6
  • Perl/PHP are out? awk? Commented Apr 5, 2020 at 12:10
  • 3
    Choose the right tool first. I suggest to use an XML/HTML parser (xmlstarlet, e.g.). Commented Apr 5, 2020 at 12:22
  • Is not it possible to use the pure bash without any additional libs? Commented Apr 5, 2020 at 12:48
  • 3
    Yes, but an XML file with the same content can also be formatted completely differently and this then causes problems. This problem does not exist if you use tools that can interpret XML files correctly. Commented Apr 5, 2020 at 13:09
  • Does this answer your question? How to parse XML in Bash? Commented Feb 28, 2024 at 14:41

2 Answers 2

Reset to default
2

As long as the username attribute does not come before id attribute, this is a bash script to give the result:

#/usr/bin/env bash

id='\bid="([^"]+)"'
username='\busername="([^"]+)"'
while IFS= read -r line; do
    [[ $line =~ $id       ]] && idv="${BASH_REMATCH[1]}"
    [[ $line =~ $username ]] && echo "$idv=${BASH_REMATCH[1]}"
done < data.xml
exit 0

It works even when username attribute and id attribute are on the same line.

Improve this answer
Sign up to request clarification or add additional context in comments.

5 Comments

Thank you very much Philippe, this is very elegant solution taking into account I cannot use xmlstarlet or any other lib as @Cyrus mentioned.
Apologies, I might have missed something, but when I run this script against my XML file data.xml I do not get any output.
How did you you run the script ? what's your bash version : echo $BASH_VERSION
I have created script.sh file, pasted the content there, and then run it as usual. data.xml file is in the same location. I am currently running it on my macos and the bash version here is 3.2.57(1)-release. I will on RHEL machine once I have access to it.
Note that it's legal XML for there to be spaces around the = in an attribute assignment.
2

Assumptions:

  • the xml file is 'nicely' formatted as the presented example (hence no need for a XML parser)
  • id is the first attribute in a Person block (ie, we'll always be able to match on the literal string Person id), otherwise additional parsing will be necessary
  • Person id and username only show up in the Group block

One awk solution:

awk -F'"' '
/<Person id=/          { pid=$2 ; next }
/[[:space:]]username=/ { printf "%s=%s\n", pid, $2 ; next }
' test.xml

Running this awk solution against the sample data file generates:

P_0001=person_0001
P_0002=person_0002

All bets are off if my assumptions are invalid.

Improve this answer

1 Comment

Writing /Person id/ is probably not a good representation as it make you believe that could also do /Person email/ which would fail. Can you separate out the id parts and add another field, like email, for example?

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.