xsl:key element is used at the top level of the stylesheet to declare an attribute, or other value,
that may be used as a key to identify nodes using the
key() function within an expression. Each xsl:key
definition declares a named key, which must match the name of the key used in the
The set of nodes to which the key applies is defined by a pattern in the match attribute: for example,
match="ACT|SCENE" then every
ACT element and every
SCENE element is indexed by this key.
The value of the key, for each of these matched elements, is determined by the
This is an expression, which is evaluated for each matched element. If the expression returns a node-set,
the typed value of each node in this node-set acts as a key value. For example, if
use="AUTHOR", then each
AUTHOR child of the matched element supplies one key value.
Keys are not unique: the same value may identify many different nodes
Keys are multi-valued: each matched node may have several (zero or more) values of the key, any one of which may be used to locate that node
Keys can only be used to identify nodes within a single XML document: the key() function will return nodes that are in the same document as the current node.
All three attributes,
use, are mandatory.
collation attribute can be used when comparing strings.
A key declaration does not apply to any particular source document; in principle, it applies to every
document. Saxon builds an index for a particular document the first time the
is used to search that document. The Saxon implementation goes to some lengths to ensure that when the
same stylesheet is applied repeatedly to the same source document (perhaps a lookup document used in many
transformations), the key is only built once. The memory used by a key is freed as soon as either the
source document or the stylesheet are garbage-collected.
The Saxon-EE optimizer will automatically create additional keys where these appear to be useful.