xml path language (xpath)

Resin can use XPath, the XML Path Language, to select nodes from an XML tree. Scripts can select all 'table' children, or even all 'table' elements in an entire HTML file. The XPath language is exceptionally rich. It can describe an incredible number of node selections.

Java applications can use the com.caucho.xpath package to use XPath with an XML tree.

The XML Path Language describes nodes in an XML tree. It's a mini-language for specifying nodes patterns, like regular expressions are a mini-language for specifying text patterns.

The language selects sets of nodes. Each operator in XPath selects a new set based on the old set of nodes. For example, given a set of chapters, XPath can select all sections in those chapters, chapters with 'advanced' attributes, or grandchildren with 'color' attributes of 'blue'.

Basic Patterns

The basic XPath patterns cover 90% of the cases that most stylesheets will need. Because the pattern language is based on familiar filesystem paths, the most useful patterns should be easy.

node

Selects all child elements with nodeName of node.

xml = caucho.xml.Xml.parseString(@<<END);
<top>
  <a id='1'/>
  <b>
    <a id='2'/>
  </b>
  <a id='3'/>
</top>
END

top = xml.documentElement

for (var node in top.select('a'))
  writeln(node.nodeName, ': ', node.attribute.id);

a: 1
a: 3

*

Selects all child elements.

@attr

Selects the attribute attr.

chapter/@title	All title attributes of chapters.
//var[@name='keywords']/@content	The contents of all var keywords. A web spider might use this.

@*

Selects all attributes.

ns:*

Selects elements in the given namespace. Namespace patterns only make sense in the context of XSL, where the namespace declarations have been made.

node()

Matches an org.w3c.dom.Node.

text()

Matches a org.w3c.dom.Text node.

comment()

Matches a comment.

processing-instruction()

Matches a org.w3c.dom.ProcessingInstruction node.

.

Selects the current node.

The current node is primarily useful for descendant patterns. for some filter patterns.

.//image	Any image descendant.
.//image	Any image descendant.

..

Selects the parent of current node.

../brother	All brothers.
../../aunt	All aunts.
//*[../@color='blue']	All elements with blue parents.

/

Selects the document node.

Useful for finding constants in a document.

/html	The single root element.
/html/head	The HTML head section.
/html/head/var	All var tags.

a[expr]

Select only those nodes matching a which also satisfy the expression expr.

The expression b is a combination of and, or, not, comparisons and XPath patterns. An XPath expression, e.g. chapter/verse, is true if at least one node matches the pattern.

chapter[verse]	Chapters with a verse.
chapter[not(verse)]	Chapters with no verses.
chapter[not(verse) and not(section)]	Chapters with neither verses nor sections.
*[@color='blue']	Blue children.
*[@color='blue'][position()=last() - 1]	second to last blue child

a[n]

Selects the nth matching node matching a When a filter's expression is a number, XPath selects based on position. This special case of the filter pattern treats selections as ordered lists.

The position filter is equivalent to a[position()=n]

child[3]	third child
child/grandchild[1]	first grandchild from every child (not only the first grandchild).
child[last()]	last child

a/b

For each node matching a, add the nodes matching b to the result.

The following is almost a definition of a/b.

for (var a in node.select('a')) {
  for (var b in a.select('b')) {
    // possible duplicates if a or b
    // are weird patterns.
  }
}

Some example interpretations,

chapter/verse	grandchildren verse with parent chapter.
../span	sibling span elements.
./span	children span elements.
/	All grandchildren.
*[color='blue']/verse	All grandchildren verse elements with blue colored parents.
a/b/c	Great grandchildren c, with parent b and grandparent a.

a//b

For each node matching a, add the descendant nodes matching b to the result. The '//' operator selects all descendants matching b. The '/' operator selects all children matching b.

chapter//a	All links contained in a chapter.
.//image	Any image descendant.

//b

Returns elements in the entire document matching b.

This is equivalent to /.//b, but less weird.

//image	All images in the document
//a[@href='http://www.caucho.com']	All links to caucho

a|b

All nodes matching a or b.

Some example interpretations,

lion|tiger|bear

Lions and tigers and bears.

expr/path

Uses an expression as the start of a path. Usually, expr will be a special function call, like the id() function. expr[expr] is also allowed.

id('314')/@color	The color attribute of the element with id 314.
key('id', 'a')[@color='red']	Selects the node returned by key if they have a red color.

Node Axes

child::a

The child axis selects children of the current node. It is entirely equivalent to the usual slash notation. So child::a/child::b is the same as a/b.

child::chapter	Select all chapter elements of the current node.
child::text()	All text children.
child::comment()	All comment children.
/child::h1	Select all top-level h1 elements.

descendant::a

The descendant axis selects descendants of the current node. It is equivalent to '//'. So child::a//child::b is the same as a//b.

descendant::a	Select all hyperlinks below the current node.
/descendant::table	Select all tables in the document.
child::section/descendant::image	Select all images in a section.
.[descendant::image]	Selects the current node if it contains an image.

descendant-or-self::a

Selects descendants including the current node.

attribute::a

Selects attributes of the current element. It is equivalent to @a.

attribute::width	Selects the width of the current node.
image[attribute::width > 10]	Selects image children with a width attribute greater than 10.

following-sibling::a

Selects nodes after the current node.

following-sibling::chapter	Selects following chapters.
h3/following-sibling::text()	Following text nodes after an h3.

preceding-sibling::a

Selects nodes before the current node.

preceding-siblings:: counts positions backwards. So preceding-sibling::a[1] selects the closest preceding sibling.

preceding-sibling::chapter	Selects chapters before chapter 3.
fun/preceding-sibling::comment()	Selects the comments before a function.

following::a

Selects the first matching node following in document order, excluding descendants. In other words, the following axis will scan through every node in document order, starting with getNextSibling().

following, preceding, ancestor and self partition the document.

preceding::a

Selects the first matching node preceding in document order, excluding ancestors. In other words, the preceding axis will scan through every node in document order, starting with the root and ending in the current node, but it will skip ancestors.

following, preceding, ancestor and self partition the document.

parent::a

Selects the parent if it matches. The '..' pattern from the core is equivalent to 'parent::node()'.

parent::chapter	Selects the parent if it's a chapter.
parent::*/@color	Selects the color of the parent.

ancestor::a

Selects matching ancestors.

following, preceding, ancestor and self partition the document.

ancestor::section	Selects ancestor sections.
ancestor::*/@color	Selects ancestor color attributes.

ancestor-or-self::a

Selects ancestors including the current node.

self::a

Selects the current node. '.' is equivalent to 'self::node()'.

following, preceding, ancestor and self partition the document.

self::chapter

Selects the current node if it's a chapter.

Expressions

path

Selects nodes based on the path.

The path interpretation depends on the context. When filtering nodes, the path expression is a boolean. Any matching node will return true. When the value is a string, as in the xsl:value-of, then the textValue is used.

boolean	True if path matches any nodes.
string	The text value of the first matching node.
number	The string value converted to a number.

number

Numbers have the same syntax as Java doubles.

"string"

String literals can use single or double quotes.

a = b

Standard comparisons.

XPath converts the arguments to a common type before comparison. The conversion priority is boolean, number, string. In other words, if either arg is a boolean, a boolean comparison is used.

boolean
number
string

Node-sets, e.g. chapter/@color, are compared differently. Each node in the node set is compared. If any matches, then the comparison is true.

a = b	True if a equals b.
a != b	True if a is not equal to b.
a < b	True if a is less than b.
a <= b	True if a is less than or equal to b.
a > b	True if a is greater than b.
a >= b	True if a is greater than or equal to b.

a + b

Arithmetic expressions.

- a	Unary minus
a + b	Add
a - b	Substract
a * b	Multiply
a div b	Divide
a mod b	Floating point mod, like Java.

(expr)

Parenthesized expressions.

fun(arg1, ..., argn)

Function calls.

The function library is in the XPath function section.

not(expr)

Boolean not.

a or b

Boolean or.

a and b

Boolean and.

$var

The value of a variable. Variables, in general, only make sense in a context like XSL. Normal use of XPath outside of XSL will not use variables.

Copyright © 1998-2015 Caucho Technology, Inc. All rights reserved. Resin ^® is a registered trademark. Quercus^tm, and Hessian^tm are trademarks of Caucho Technology.

Cloud-optimized Resin Server is a Java EE certified Java Application Server, and Web Server, and Distributed Cache Server (Memcached).
Leading companies worldwide with demand for reliability and high performance web applications including SalesForce.com, CNET, DZone and many more are powered by Resin.

home company docs
app server

Resin Documentation

xml path language (xpath)

Basic Patterns

node

*

@attr

@*

ns:*

node()

text()

comment()

processing-instruction()

.

..

/

a[expr]

a[n]

a/b

a//b

//b

a|b

expr/path

Node Axes

child::a

descendant::a

descendant-or-self::a

attribute::a

following-sibling::a

preceding-sibling::a

following::a

preceding::a

parent::a

ancestor::a

ancestor-or-self::a

self::a

Expressions

path

number

"string"

a = b

a + b

(expr)

fun(arg1, ..., argn)

not(expr)

a or b

a and b

$var