xpath.md

XPath

Uses

1. find elements within an XML document
2. test whether particular element(s) exist in an XML document

Background

• A DSL for selecting nodes from an XML document

• Spec:

  • Versions (1.0 is defacto default)
    • As of 2020-04-09, current version is 3.1 https://www.w3.org/TR/xpath-3/
    • 1.0 spec: https://www.w3.org/TR/1999/REC-xpath-19991116/
    • Many/most implementations only support 1.0
      • e.g. Nokogiri depends on libxml2 which supports 1.0 and has no plans for 2.0 support because the spec changed a lot from 1.0 to 2.0
      • e.g. Firefox Gecko implements 1.0
  • defines path expressions

• can be used in any XML like document: XML, HTML, SVG

• implemented in all language stacks

  • not all browsers/stacks implement all features

• xpath has functions e.g.

  • position()
    • lets you choose an element by position
  • number()
  • string()
  • string-length()
  • ...

• syntax

  • starting points for the search
    • //
      • recursive descent operator
      • starting the path with // means starting anywhere within the document
    • /
      • selects the document root
      • ignores whatever context node you might have passed in to begin with
  • attribute names being with @
  • conditions go within []
    • logical AND conditions by specifying them concurrently e.g. //things[@type="good"][@size="big"] will return all <thing> elements which have an attribute type="good" AND size="big"
    • OR ?
    • NOT ?

• axes

  • there are 13 axes
  • an axis represents a relationship to the context node and is used to locate nodes relative to that node in the tree
  • not all implementations support all axes

• models the document as a tree of nodes

• types of nodes

  1. element nodes
  2. attribute node
  3. text node

• defines a way to compute a string value for each type of node

• some types of nodes have "names"

  • a "name" is a tuple of (local-part, namespace-uri)
    • the namespace-uri can be null

• xpath expressions return one of 4 basic types when evaluated

  1. node-set
     • unordered collection of nodes without duplicates
  2. boolean (true or false)
  3. number ( floating point number)
  4. string (a sequence of UCS characters)

• note that "a single node" is not a basic type - nodes are always in a node-set

• expressions are evaluated with respect to a context (think of it like "context" in the sense of evaluating a function in a lang with lexical scope

• context consists of

  1. a node (the context node)
  2. a pair of non-zero positive integers (context-position, cotext-size)
     • context-position is always less than or equal to context size
  3. a set of variable bindings
     • a mapping from variable names to variable values
     • variable values can have any type valid for the return value of an expression and other types (???)
  4. a function library
     • a mapping of function names to functions
     • each function takes 0 or more args
     • each function returns one of the 4 basic types (see above)
     • the spec defines a set of core functions
     • XSLT and XPointer define additional functions and data types
  5. a set of namespace declarations in scope for the expression
     • a mapping from prefixes to namespace URIs

• the expression is a set of location steps each separated by /

• the set of nodes selected by each step becomes the context node for the location step to it's right e.g.

  • example:

    a/b/c
    
  • the nodes selected by a become the context node for b
  • the nodes selecte by b become the context node for c and so on

• each location step has 3 parts:

  1. an axis
     • specifies the tree relationship between the nodes selected by this location step and the context node
  2. a node test
     • specifies node type and expanded name of the nodes selected by this location step
  3. 0-many predicates
     • use arbitrary expressions to further refine the set of nodes selected