yq/pkg/yqlib/doc/usage/xml.md

XML

Encode and decode to and from XML. Whitespace is not conserved for round trips - but the order of the fields are.

Consecutive xml nodes with the same name are assumed to be arrays.

XML content data, attributes processing instructions and directives are all created as plain fields.

This can be controlled by:

Flag	Default	Sample XML
--xml-attribute-prefix	+ (changing to +@ soon)	Legs in <cat legs="4"/>
--xml-content-name	+content	Meow in <cat>Meow <fur>true</true></cat>
--xml-directive-name	+directive	<!DOCTYPE config system "blah">
--xml-proc-inst-prefix	+p_	<?xml version="1"?>

{% hint style="warning" %} Default Attribute Prefix will be changing in v4.30! In order to avoid name conflicts (e.g. having an attribute named "content" will create a field that clashes with the default content name of "+content") the attribute prefix will be changing to "+@".

This will affect users that have not set their own prefix and are not roundtripping XML changes.

{% endhint %}

Encoder / Decoder flag options

In addition to the above flags, there are the following xml encoder/decoder options controlled by flags:

Flag	Default	Description
--xml-strict-mode	false	Strict mode enforces the requirements of the XML specification. When switched off the parser allows input containing common mistakes. See the Golang xml decoder for more details.
--xml-keep-namespace	true	Keeps the namespace of attributes
--xml-raw-token	true	Does not verify that start and end elements match and does not translate name space prefixes to their corresponding URLs.
--xml-skip-proc-inst	false	Skips over processing instructions, e.g. <?xml version="1"?>
--xml-skip-directives	false	Skips over directives, e.g. <!DOCTYPE config system "blah">

See below for examples

Parse xml: simple

Notice how all the values are strings, see the next example on how you can fix that.

Given a sample.xml file of:

<?xml version="1.0" encoding="UTF-8"?>
<cat>
  <says>meow</says>
  <legs>4</legs>
  <cute>true</cute>
</cat>

then

yq -p=xml '.' sample.xml

will output

+p_xml: version="1.0" encoding="UTF-8"
cat:
  says: meow
  legs: "4"
  cute: "true"

Parse xml: number

All values are assumed to be strings when parsing XML, but you can use the from_yaml operator on all the strings values to autoparse into the correct type.

Given a sample.xml file of:

<?xml version="1.0" encoding="UTF-8"?>
<cat>
  <says>meow</says>
  <legs>4</legs>
  <cute>true</cute>
</cat>

then

yq -p=xml ' (.. | select(tag == "!!str")) |= from_yaml' sample.xml

will output

+p_xml: version="1.0" encoding="UTF-8"
cat:
  says: meow
  legs: 4
  cute: true

Parse xml: array

Consecutive nodes with identical xml names are assumed to be arrays.

Given a sample.xml file of:

<?xml version="1.0" encoding="UTF-8"?>
<animal>cat</animal>
<animal>goat</animal>

then

yq -p=xml '.' sample.xml

will output

+p_xml: version="1.0" encoding="UTF-8"
animal:
  - cat
  - goat

Parse xml: attributes

Attributes are converted to fields, with the default attribute prefix '+'. Use '--xml-attribute-prefix` to set your own.

Given a sample.xml file of:

<?xml version="1.0" encoding="UTF-8"?>
<cat legs="4">
  <legs>7</legs>
</cat>

then

yq -p=xml '.' sample.xml

will output

+p_xml: version="1.0" encoding="UTF-8"
cat:
  +legs: "4"
  legs: "7"

Parse xml: attributes with content

Content is added as a field, using the default content name of +content. Use --xml-content-name to set your own.

Given a sample.xml file of:

<?xml version="1.0" encoding="UTF-8"?>
<cat legs="4">meow</cat>

then

yq -p=xml '.' sample.xml

will output

+p_xml: version="1.0" encoding="UTF-8"
cat:
  +content: meow
  +legs: "4"

Parse xml: custom dtd

DTD entities are processed as directives.

Given a sample.xml file of:

<?xml version="1.0"?>
<!DOCTYPE root [
<!ENTITY writer "Blah.">
<!ENTITY copyright "Blah">
]>
<root>
    <item>&writer;&copyright;</item>
</root>

then

yq -p=xml -o=xml '.' sample.xml

will output

<?xml version="1.0"?>
<!DOCTYPE root [
<!ENTITY writer "Blah.">
<!ENTITY copyright "Blah">
]>
<root>
  <item>&amp;writer;&amp;copyright;</item>
</root>

Parse xml: skip custom dtd

DTDs are directives, skip over directives to skip DTDs.

Given a sample.xml file of:

<?xml version="1.0"?>
<!DOCTYPE root [
<!ENTITY writer "Blah.">
<!ENTITY copyright "Blah">
]>
<root>
    <item>&writer;&copyright;</item>
</root>

then

yq -p=xml -o=xml --xml-skip-directives '.' sample.xml

will output

<?xml version="1.0"?>
<root>
  <item>&amp;writer;&amp;copyright;</item>
</root>

Parse xml: with comments

A best attempt is made to preserve comments.

Given a sample.xml file of:

<!-- before cat -->
<cat>
	<!-- in cat before -->
	<x>3<!-- multi
line comment 
for x --></x>
	<!-- before y -->
	<y>
		<!-- in y before -->
		<d><!-- in d before -->z<!-- in d after --></d>
		
		<!-- in y after -->
	</y>
	<!-- in_cat_after -->
</cat>
<!-- after cat -->

then

yq -p=xml '.' sample.xml

will output

# before cat
cat:
  # in cat before
  x: "3" # multi
  # line comment 
  # for x
  # before y

  y:
    # in y before
    # in d before
    d: z # in d after
    # in y after
  # in_cat_after
# after cat

Parse xml: keep attribute namespace

Given a sample.xml file of:

<?xml version="1.0"?>
<map xmlns="some-namespace" xmlns:xsi="some-instance" xsi:schemaLocation="some-url">
</map>

then

yq -p=xml -o=xml --xml-keep-namespace '.' sample.xml

will output

<?xml version="1.0"?>
<map xmlns="some-namespace" xmlns:xsi="some-instance" some-instance:schemaLocation="some-url"></map>

instead of

<?xml version="1.0"?>
<map xmlns="some-namespace" xmlns:xsi="some-instance" some-instance:schemaLocation="some-url"></map>

Parse xml: keep raw attribute namespace

Given a sample.xml file of:

<?xml version="1.0"?>
<map xmlns="some-namespace" xmlns:xsi="some-instance" xsi:schemaLocation="some-url">
</map>

then

yq -p=xml -o=xml --xml-keep-namespace --xml-raw-token '.' sample.xml

will output

<?xml version="1.0"?>
<map xmlns="some-namespace" xmlns:xsi="some-instance" some-instance:schemaLocation="some-url"></map>

instead of

<?xml version="1.0"?>
<map xmlns="some-namespace" xsi="some-instance" schemaLocation="some-url"></map>

Encode xml: simple

Given a sample.yml file of:

cat: purrs

then

yq -o=xml '.' sample.yml

will output

<cat>purrs</cat>

Encode xml: array

Given a sample.yml file of:

pets:
  cat:
    - purrs
    - meows

then

yq -o=xml '.' sample.yml

will output

<pets>
  <cat>purrs</cat>
  <cat>meows</cat>
</pets>

Encode xml: attributes

Fields with the matching xml-attribute-prefix are assumed to be attributes.

Given a sample.yml file of:

cat:
  +name: tiger
  meows: true

then

yq -o=xml '.' sample.yml

will output

<cat name="tiger">
  <meows>true</meows>
</cat>

Encode xml: attributes with content

Fields with the matching xml-content-name is assumed to be content.

Given a sample.yml file of:

cat:
  +name: tiger
  +content: cool

then

yq -o=xml '.' sample.yml

will output

<cat name="tiger">cool</cat>

Encode xml: comments

A best attempt is made to copy comments to xml.

Given a sample.yml file of:

# header comment
# above_cat
cat: # inline_cat
  # above_array
  array: # inline_array
    - val1 # inline_val1
    # above_val2
    - val2 # inline_val2
# below_cat

then

yq -o=xml '.' sample.yml

will output

<!--
 header comment
 above_cat
 --><!-- inline_cat --><cat><!-- above_array inline_array -->
  <array>val1<!-- inline_val1 --></array>
  <array><!-- above_val2 -->val2<!-- inline_val2 --></array>
</cat><!-- below_cat -->

Encode: doctype and xml declaration

Use the special xml names to add/modify proc instructions and directives.

Given a sample.yml file of:

+p_xml: version="1.0"
+directive: 'DOCTYPE config SYSTEM "/etc/iwatch/iwatch.dtd" '
apple:
  +p_coolioo: version="1.0"
  +directive: 'CATYPE meow purr puss '
  b: things

then

yq -o=xml '.' sample.yml

will output

<?xml version="1.0"?>
<!DOCTYPE config SYSTEM "/etc/iwatch/iwatch.dtd" >
<apple><?coolioo version="1.0"?><!CATYPE meow purr puss >
  <b>things</b>
</apple>

Round trip: with comments

A best effort is made, but comment positions and white space are not preserved perfectly.

Given a sample.xml file of:

<!-- before cat -->
<cat>
	<!-- in cat before -->
	<x>3<!-- multi
line comment 
for x --></x>
	<!-- before y -->
	<y>
		<!-- in y before -->
		<d><!-- in d before -->z<!-- in d after --></d>
		
		<!-- in y after -->
	</y>
	<!-- in_cat_after -->
</cat>
<!-- after cat -->

then

yq -p=xml -o=xml '.' sample.xml

will output

<!-- before cat --><cat><!-- in cat before -->
  <x>3<!-- multi
line comment 
for x --></x><!-- before y -->
  <y><!-- in y before
in d before -->
    <d>z<!-- in d after --></d><!-- in y after -->
  </y><!-- in_cat_after -->
</cat><!-- after cat -->

Roundtrip: with doctype and declaration

yq parses XML proc instructions and directives into nodes. Unfortunately the underlying XML parser loses whitespace information.

Given a sample.xml file of:

<?xml version="1.0"?>
<!DOCTYPE config SYSTEM "/etc/iwatch/iwatch.dtd" >
<apple>
  <?coolioo version="1.0"?>
  <!CATYPE meow purr puss >
  <b>things</b>
</apple>

then

yq -p=xml -o=xml '.' sample.xml

will output

<?xml version="1.0"?>
<!DOCTYPE config SYSTEM "/etc/iwatch/iwatch.dtd" >
<apple><?coolioo version="1.0"?><!CATYPE meow purr puss >
  <b>things</b>
</apple>