Re: [xml-dev] XPath 3.1 introduction ?

Hi Christophe,

I'm glad it helps.

If you need something similar for XPath 3.0, here is a very quick intro as well to the main new features (of course, it's much more superficial than a complete course such as Dimitre's).

Higher-order functions:

===============

Create an inline function:

let $function := function($x as xs:integer) as xs:boolean { $x eq 1 }

return $function(2)

Reference a named function with its name and arity:

let $function as function(xs:string, xs:string) as xs:string := fn:concat#2

return $function("foo", "bar")

Partial application (currying):

let $partial-function := fn:concat#2("foo", ?)

return $partial-function("bar")

String concatenation inline

===============

|| can be used to concatenates strings

"foo" || "bar" || "foobar"

Map operator

========

The ! operator is identical to a for-return that binds the context item in turn to each item in the input sequence.

(1 to 10) ! (. * .)

It's also convenient for function call chains:

$node ! local:function1(.) ! local:function2(.) ! local:function3(.)

Extended QNames

============

You can build QNames for XPath navigation without binding a prefix (similar to the Clark notation):

doc("http://www.example.com/schema.xsd")/Q{http://www.w3.org/2001/XMLSchema}schema/Q{http://www.w3.org/2001/XMLSchema}element

Kind regards,

Ghislain

On Wed, Jun 22, 2016 at 11:43 AM, <cmarchand@oxiane.com> wrote:

Thanks Ghislain ! This is very clear an understandable !

In fact, I need also an introduction to XPath 3.0 ...

Best regards,

Christophe

Le 2016-06-22 11:35, Ghislain Fourny a écrit :

Hi Christophe

The most prominent add to XPath 3.1 is maps and arrays.

Here's a very abstract overview (which of course misses plenty of smaller details). It also applies to XQuery 3.1, which supports them as well.

In short:

========

- Items in sequences can now also be maps or arrays.

- Maps are associative arrays. Keys are any atomic value. Values are any sequences of items (including arrays and maps, so they can nest).

- Arrays are ordered lists of values. A values is any sequence of items (including arrays and maps, so they can nest).

- Values put into maps and arrays are *not* copied, so you can use maps, say, for indexing. This is different from XML node constructors.

To build them:

========

- One syntax for objects, looking a bit like JSON but more generic (since you can of course nest expressions) and with a keyword "map", like node constructors:

map { "foo" : "bar", "bar" : (1, 2, 3) }

- Two syntaxes for arrays:

Syntax 1 similar to that of maps, with an "array" keyword. It simply takes the sequence and each item in it becomes a value of the array (So you can't construct arrays with sequences of more than one item with this one):

array { 1, (2, 3, 4), (), 5 }

creates an array with 5 values, 1, 2, 3, 4 and 5.

Syntax 2 similar to JSON, but it's comma-sensitive in the sense that they delimit the values (so it's not a comma expression inside: the commas are part of the constructor)

[ 1, 2, 3, 4, (), (5, 6) ]

creates an array with 5 values : 1, 2, 3, 4, the empty sequence, and the sequence (5,6)

Navigation:

========

Alternative 1: Use maps and arrays as function items:

let $map := map { "foo" : "bar" }

return $map("foo")

let $array := array { 1, 2, 3 }

return $array(2)

Alternative 2: Use the ? operator (for NCNames and integers):

let $map := map { "foo" : "bar" }

return $map?foo

let $array := array { 1, 2, 3 }

return $array?2

The ? operator works with wildcards to "unbox" arrays or maps to a sequence:

let $array := array { 1, 2, 3 }

return $array?*

(it also exists as a unary operator, which implicitly assumes the context item on the left-hand-side)

Functions:

========

There are plenty of new functions for maps and arrays documented here: https://www.w3.org/TR/xpath-functions-31/#maps-and-arrays

Types:

=======

There are item type notations such as map(*), map(xs:integer, node()), array(xs:integer), array(*), etc.

I hope it helps you get started!

Feel free to correct me if I got anything wrong in the above.

Kind regards,

Ghislain