Type Alias XPath

Type Declaration

• • (path: string): readonly Target[]

  • Selects values.

    A single selection reaches across the whole target set: the values retrieved from each target are merged into one list, sparing the caller a loop of its own. Every axis, node test, predicate, operator and core function XPath 1.0 defines is available, each target taken in its turn as the context node.

    An expression computing a string, a number or a boolean, count(//item) among them, reports the value it computed for each target, so that whatever the language can express is reached the same way. A value an expression computed holds no tree of its own, and so selects nothing in its turn.

    Names are matched as the tree holds them, case sensitively, prefixes compared as written and no xmlns declaration read, so that a single set of expressions serves both the XML trees and the HTML ones, whose names are folded to lower case as they are parsed:

    • an unprefixed name test matches an unprefixed name alone, so item addresses <item> however the document declares a default namespace, and leaves <d:item> out
    • a prefixed name test matches the prefix as written, so d:b addresses <d:b> whatever URI the document binds d to, and nothing at all where the same element is written under another prefix
    • local-name() reports a name without its prefix and name() reports it whole, so <d:b> answers to local-name()='b' and to name()='d:b' alike
    • namespace-uri() reports the prefix a name carries rather than the URI a declaration binds it to, as no declaration is read; the xml prefix is the exception, bound by definition, so @xml:base and lang() read what they are meant to
    • the xmlns declarations themselves are ordinary attributes, reported by an attribute step like any other

    The XML declaration and the document type declaration are not nodes, as the language prescribes, and a processing instruction is not one either, so processing-instruction() selects nothing; namespace:: selects nothing in its turn, as no namespace node is held. A CDATA section is a text node holding what it wraps, rather than a node of its own.

    Parameters

    • path: string

      The selection expression

    Returns readonly Target[]

    An immutable list of the values selected by path, ordered by target and, within each target, in document order, each node reported once; empty if path selects no value

    Throws

    SyntaxError If path is malformed