XML & XPath

Process XML with xmllint and xmlstarlet — XPath queries, validation, and in-place editing.

xmllint — Query and Format

Pretty-print XML — reformat for readability

xmllint --format response.xml

XPath query — extract a specific element value

xmllint --xpath '//hostname/text()' inventory.xml

XPath with attribute selection — find elements by attribute

xmllint --xpath '//device[@role="core"]/hostname/text()' inventory.xml

XPath returning multiple values — each on its own line

xmllint --xpath '//device/hostname/text()' inventory.xml 2>/dev/null | tr -s '\n'

Validate XML against a DTD — check structural correctness

xmllint --valid --dtdvalid schema.dtd config.xml --noout

Validate well-formedness only — no schema required

xmllint --noout config.xml && echo "Well-formed" || echo "Malformed"

Extract XML from an API response — curl + xmllint pipeline

curl -s https://api.example.com/data.xml | xmllint --format --xpath '//result' -

xmlstarlet — Select, Edit, Validate

Select element text — equivalent to XPath text() extraction

xmlstarlet sel -t -v '//device/hostname' inventory.xml

Select with template — format output with custom separators

xmlstarlet sel -t -m '//device' -v 'hostname' -o ',' -v 'ip' -n inventory.xml

Select attributes — extract attribute values from elements

xmlstarlet sel -t -v '//device/@role' inventory.xml

Count matching elements — useful for validation scripts

xmlstarlet sel -t -v 'count(//device)' inventory.xml

Edit XML — update an element value in-place

xmlstarlet ed -u '//device[@name="sw-core"]/vlan' -v '20' inventory.xml

Edit XML — add a new child element

xmlstarlet ed -s '//device[@name="sw-core"]' -t elem -n "location" -v "rack-a1" inventory.xml

Edit XML — delete an element

xmlstarlet ed -d '//device[@name="decommissioned"]' inventory.xml

In-place editing — modify the file directly (like sed -i)

xmlstarlet ed -L -u '//config/timeout' -v '60' config.xml

Validate against XSD schema

xmlstarlet val -e -s schema.xsd config.xml

Common XPath Expressions

Root element children — /root/element

xmlstarlet sel -t -v '/inventory/device/hostname' inventory.xml

Any depth search — //element finds at any nesting level

xmlstarlet sel -t -v '//hostname' inventory.xml

Predicate filtering — [condition] narrows the selection

# By position
xmlstarlet sel -t -v '//device[1]/hostname' inventory.xml

# By child element value
xmlstarlet sel -t -v '//device[vlan=10]/hostname' inventory.xml

# By attribute
xmlstarlet sel -t -v '//device[@enabled="true"]/hostname' inventory.xml

Boolean operators in predicates — and, or, not()

xmlstarlet sel -t -v '//device[@role="access" and vlan > 10]/hostname' inventory.xml

XPath functions — string-length, contains, starts-with

# Elements containing a substring
xmlstarlet sel -t -v '//device[contains(hostname, "core")]/ip' inventory.xml

# Elements starting with a prefix
xmlstarlet sel -t -v '//device[starts-with(hostname, "sw-")]/hostname' inventory.xml

Namespace Handling

Query with default namespace — must register it first

xmlstarlet sel -N ns="http://example.com/schema" \
  -t -v '//ns:device/ns:hostname' namespaced.xml

xmllint with namespace — use local-name() to ignore namespace prefix

xmllint --xpath '//*[local-name()="device"]/*[local-name()="hostname"]/text()' namespaced.xml

Strip namespaces entirely — simplify processing when namespace is noise

xmlstarlet ed -d '//@*[local-name()="schemaLocation"]' namespaced.xml \
  | sed 's/ xmlns[^"]*"[^"]*"//g' \
  | xmlstarlet sel -t -v '//device/hostname' -

XSLT Processing

Transform XML with an XSLT stylesheet — xsltproc

xsltproc transform.xsl input.xml > output.xml

Pass parameters to XSLT — inject values at transform time

xsltproc --stringparam environment "production" transform.xsl input.xml

Quick XSLT to extract CSV from XML — inline stylesheet

xsltproc - inventory.xml <<'XSLT'
<?xml version="1.0"?>
<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
  <xsl:output method="text"/>
  <xsl:template match="/">
    <xsl:text>hostname,ip,role&#10;</xsl:text>
    <xsl:for-each select="//device">
      <xsl:value-of select="hostname"/>,<xsl:value-of select="ip"/>,<xsl:value-of select="@role"/>
      <xsl:text>&#10;</xsl:text>
    </xsl:for-each>
  </xsl:template>
</xsl:stylesheet>
XSLT

API Response Processing

Cisco ISE API — extract endpoint data from XML response

curl -sk -u admin:password https://ise-01:9060/ers/config/endpoint \
  -H "Accept: application/xml" \
  | xmllint --format --xpath '//resources/resource/name/text()' -

Parse SOAP response — extract the payload from the envelope

xmlstarlet sel -N soap="http://schemas.xmlsoap.org/soap/envelope/" \
  -t -c '//soap:Body/*' response.xml | xmllint --format -

Extract error messages from XML API errors

xmlstarlet sel -t -v '//error/message' -n error_response.xml