Re: [ssax-sxml] // (S)Xpath query [was: duplication]

[Kirill L. <lis...@ac...>]

On Sun, 1 Dec 2002 br...@ar... wrote:

>   Kirill> (sxpath `((element                              ; node-test 
>   Kirill> 	    (@ (equal? (name "sym_overname")))) ; predicate @name="sym_overname"
>   Kirill> 	  @ (or@ value id addr) ))              ;  OR
> 
>                   ^
> Is the @  misplaced ?
>
No, it is not. 
There are three (BTW: it's 'length' of location-path) SXPath location steps 
in this location path

Probably better indentation is:
(sxpath `((element 
	    (@ (equal? (name "sym_overname")))) 
	  @ 
	  (or@ value id addr) ))


In accordance with W3C XPath rec. a location step may be cosidered as:
  Step ::= AxisSpecifier NodeTest Predicate*
Let's consider this SXPath expression from this point of view:

   1. (element 
	(@ (equal? (name "sym_overname")))) 
  Selects all the child elements (axis - child)
   of type 'element' (node-test - element)  
   and filter them using predicate 
	(@ (equal? (name "sym_overname")))
      
   2. @
  Selects a list of attributes
  axis - child, 
  node-test - @ 
  A list of SXML (and thus SXPath) attributes is just a nested element
with "special" name - '@'
  The symbol '@' is allowed by R5RS but is forbidden by XML Rec.,
so there is no possibility for a name clash here. Besides, its used
for attributes in XPath
  That is, SXML/SXPath considers elements/attributes uniformely, so 
an attribute node is not distinct from an element node, it's just an 
element node nested in an attribute-list.
  In XPath, where is a special type of attribute nodes: they are attached
to an element node directly, but are not considered as "children" for such 
a node. 
 While in SXPath expression '@' is a node-test, in XPath it's an axis 
specifier. 
  
  So, here we have selected lists of attributes for all the elements
selected by the location step one.

  3. (or@ value id addr)
  axis - child
  node-test - value | id | addr
'or@' is abbreviated syntax for logical 'or' which is used for 
union of node-sets. 

Best regards,
         Kirill.