Linux Bash XMLLINT with XPATH

Question

Today I get to learn how to use xmllint properly. It does not seem to be well covered or explained. I plan to use a single language resource file to run my entire system. I have a mixture of bash scripts and php pages that must read from this language file.

Currently I am using the following format in my xml file en.xml:

<?xml version="1.0" encoding="utf-8"?>
<resources>

   <item id="index.php">
        <label>LABEL</label>
        <value>VALUE</value>
        <description>DESCRIPTION</description>
   </item>
   <item id="config.php">
        <label>LABEL</label>
        <value>VALUE</value>
        <description>DESCRIPTION</description>
   </item>

</resources>

Now I need to start with a bash script line that should pull the data values from the xml file. For example I want to get the value of DESCRIPTION from the index.php item.

I was using

xmllint --xpath 'string(//description)' /path/en.xml

for a different layout which worked, but now that I am changing the layout of my xml file, I am lost as to how best to target a specific <item> and then drill down to its child element in the bash script.

Can someone help with a xmllint --xpath line to get this value please?

Answer 1

My favorite is xmlstarlet because it seems to be more powerful than xmllint:

xmlstarlet sel -t -v '/resources/item[@id="index.php"]/description/text()' en.xml

Answer 2

I had the same problem a few minutes ago and saw this post.

After hacking a bit I found the following solution to extract the city:

(wget 'http://maps.googleapis.com/maps/api/geocode/xml?latlng=53.244921,-2.479539&sensor=true' -O dummy.xml -o /dev/null;xmllint --format --xpath '/GeocodeResponse/result[type = "postal_town"]/address_component[type = "postal_town"]/short_name/node()'  dummy.xml)

You nee to specify the correct X-Path to get the desired XML-Tag and then return only the node value.

Answer 3

how best to target a specific and then drill down to its child element

The correct XPath expression to do this is:

/resources/item[@id="index.php"]/description/text()

In plain English: Start from the document node, to the document element resources, on to its child item, but only if the value of the id attribute is "index.php", on to its child description and retrieve its textual value.

I use xmllint to validate XML documents, but never for path expressions. In a bash shell (at least with Mac OS) there is an even simpler tool for evaluating XPath expressions, called "xpath":

$ xpath en.xml '/resources/item[@id="index.php"]/description/text()'

Then, the following result is obtained:

Found 1 nodes:
-- NODE --
DESCRIPTION

---

If you still prefer xmllint, use it in the following way:

$ xmllint --xpath '/resources/item[@id="index.php"]/description/text()' en.xml > result.txt

By default, --xpath implies --noout, which prevents xmllint from outputting anything. Redirect the output to a file.

$ cat result.txt 
DESCRIPTION