What's%20New%20in%20XSLT%202.0

1
What's New in XSLT 2.0

• Jeni Tennison
• http//www.jenitennison.com

2
Overview

• Grouping
• Function Definitions
• Result Documents
• Multiple Result Documents
• Output Serialisation
• Temporary Trees
• Sequences
• Text Parsing
• Typing

3
Grouping

• Perennial requirement
• usually use Muenchian Method (keys) in XSLT 1.0
• XSLT 2.0 has ltxslfor-each-groupgt
• select attribute identifies items to group
• grouping by value calculates grouping key
• group-by groups all items
• group-adjacent groups adjacent items
• grouping in sequence identifies start/end of
  group
• group-starting-with identifies start of group
• group-ending-with identifies end of group
• Use current-group() to get members of current
  group, current-grouping-key() to get value for
  current group

4
Grouping by Value
ltpapergt lttitlegtXML and PDF in Publishing
Workflowslt/titlegt ltauthorgtMyers,
Charleslt/authorgt lt/papergt ltpapergt lttitlegtOn the
Way to XMLlt/titlegt ltauthorgtParsons,
Jonathanlt/authorgt ltauthorgtCaisley,
Phillt/authorgt lt/papergt
ltauthorgt ltnamegtCaisley, Phillt/namegt ltpapergtOn
the Way to XMLlt/papergt lt/authorgt ltauthorgt
ltnamegtMyers, Charleslt/namegt ltpapergtXML and PDF
in Publishing Workflowslt/papergt lt/authorgt ltauthorgt
ltnamegtParsons, Jonathanlt/namegt ltpapergtOn the
Way to XMLlt/papergt lt/authorgt
5
Grouping by Value
ltxslfor-each-group select"paper"
group-by"author"gt ltxslsort
select"current-grouping-key()" /gt ltauthorgt
ltnamegt ltxslvalue-of select"current-groupin
g-key()" /gt lt/namegt ltxslfor-each
select"current-group()"gt ltpapergt
ltxslvalue-of select"title" /gt lt/papergt
lt/xslfor-eachgt lt/authorgt lt/xslfor-each-groupgt
6
Grouping in Sequence

• Can use to do grouping by position
• Or to "levitate" structure from flat documents
• group the content of paragraphs into text or
  block-level elements

ltxslfor-each-group select"paper"
group-starting-with"paperposition() mod 10
1"gt ltxslresult-document href"papersposition(
).html"gt ltxslapply-templates
select"current-group()" /gt
lt/xslresult-documentgt lt/xslfor-each-groupgt
7
Implications for XSLT 2.0 Use

• No more Muenchian Grouping!
• easier to create indexes
• easier to create summaries/roll-ups
• easier to create paginated documents
• Much easier to convert from flat to hierarchical
  structures
• processing XHTML to DocBook (or XHTML 2.0)

8
Function Definitions

• Use XSLT code to create new functions
• no facility to use scripting languages such as
  JavaScript
• similar to ltfuncfunctiongt from EXSLT
• Function must be in a namespace
• All parameters are required
• but can have multiple definitions with different
  numbers of arguments
• supports optional arguments, not polymorphic
  functions
• Need parameters for context item, position
• can't default to using context node as argument
• Body of function is result of function
• similar to named templates
• use ltxslsequencegt instead of ltfuncresultgt

9
Function Definition
ltxslfunction name"stralign"gt ltxslparam
name"string" /gt ltxslparam name"padding" /gt
ltxslparam name"alignment" /gt
lt/xslfunctiongt ltxslfunction
name"stralign"gt ltxslparam name"string" /gt
ltxslparam name"padding" /gt ltxslsequence
select"stralign(string, padding, 'left')"
/gt lt/xslfunctiongt
10
Implications for XSLT 2.0 Use

• General replacement for named templates
• Particular use where XSLT code can't be used
• creating a value to sort by using ltxslsortgt
• creating a value to index by using ltxslkeygt
• creating a value to group by using
  ltxslforeach-groupgt
• carrying out complex tests on nodes, for use in
  match patterns in templates or keys

ltxsltemplate match"htmlis-heading(.)"gt
... lt/xsltemplategt
11
Multiple Result Documents

• Many XSLT 1.0 processors have extension elements
  to create multiple output documents
• XSLT 2.0 has ltxslresult-documentgt

ltxslfor-each select"section"gt
ltxslresult-document href"_at_id.html"gt
ltxslapply-templates select"."
mode"html" /gt lt/xslresult-documentgt lt/
xslfor-eachgt
12
Multiple Result Documents

• Document is associated with href URI
• accessible via API
• should enable client-side support
• Make it easier to create
• paginated output
• page per chapter
• page per 20 records
• pages using HTML frames
• supplementary files
• CSS stylesheets
• SVG graphics

13
Output Serialisation

• Several changes to ltxsloutputgt
• output definitions are named
• referenced from ltxslresult-documentgt elements
• additional xhtml output method
• extra attributes to control HTML/XHTML
  serialisation
• escape-uri-attributes controls URI-escaping of
  attributes
• include-content-type controls addition of ltmetagt
  element
• normalize-unicode attribute provides Unicode
  normalisation
• character substitution provides an alternative
  for disable-output-escaping

14
Character Substitution

• Map of characters-to-strings
• On serialisation, each character in a text node
  or attribute is substituted for the relevant
  string
• Simple use is to force use of entities

ltxslcharacter-map name"html"gt
ltxsloutput-character character"160"
string"ampnbsp" /gt
lt/xslcharacter-mapgt ltxsloutput
use-character-maps"html" /gt
lteggtblah160blahlt/eggt
lteggtblahnbspblahlt/eggt
15
Character Substitution

• Complex use is to create non-well-formed output
• use characters from private use areas to
  represent illegal sequences of characters

ltxslcharacter-map name"jsp"gt lt!-- JSP start
--gt ltxsloutput-character character"xE001"
string"lt" /gt lt!--
JSP end --gt ltxsloutput-character
character"xE002"
string"gt" /gt lt/xslcharacter-mapgt
xE001_at_ page language"java" xE002
lt_at_ page language"java" gt
16
Implications for XSLT 2.0 Use

• Much more control over output
• control over automatic serialisation
• addition of ltmetagt element
• Unicode normalisation
• control over character serialisation
• which entities get used
• what form of character references
• No more d-o-e?
• character substitution is better
• works in attribute values
• supported by all (serialising) processors
• persists through variables
• d-o-e is still easier to use

17
Temporary Trees and RTFs

• XSLT 1.0 had Result Tree Fragments
• created when use content of ltxslvariablegt
• tree that couldn't be accessed with location path
• most processors have xxxnode-set() extension
  function
• convert result tree fragment to node tree
• In XSLT 2.0, have temporary trees
• can copy in same way as RTFs
• can access without using extension function

18
Example Temporary Tree
ltxslvariable name"menus"gt ltmenu name"File"gt
ltmenuItem name"New..." shortcut"Ctrl-N" /gt
ltmenuItem name"Open..." shortcut"Ctrl-O" /gt
ltmenuItem name"Save..." shortcut"Ctrl-S" /gt
... lt/menugt ... lt/xslvariablegt
1.0
ltxslvalue-of select"exslnode-set(menus)/menu
/menuItem_at_shortcut shortcut/_at_name" /gt
2.0
ltxslvalue-of select"menus/menu
/menuItem_at_shortcut shortcut/_at_name" /gt
19
Implications for XSLT 2.0 Use

• Break up complex transformations into several
  steps
• filter, sort, annotate nodes in early steps
• later steps are easier to write
• Create lookup tables
• translate from codes or numbers to labels
• similar to arrays or matrices
• Iteratively process a document until it fulfils
  some test
• add content until a document is valid

20
Sequences in XPath 2.0

• New fundamental type in XPath 2.0
• everything is a sequence
• similar to node sets, but
• ordered
• allow duplicates
• can contain atomic values as well as nodes
• Sequences are flat
• for structured data, use XML
• Singleton sequences are the same as the single
  value they contain

21
Using Sequences in XSLT 2.0

• Iterate over a sequence
• Create a text node from a sequence

ltxslfor-each select"1 to 3"gt lttrgtlttd
colspan"4" /gtlt/trgt lt/xslfor-eachgt
lttrgtlttd colspan"4" /gtlt/trgt lttrgtlttd colspan"4"
/gtlt/trgt lttrgtlttd colspan"4" /gtlt/trgt
ltxslvalue-of select"author/surname"
separator", " /gt
Thompson, Tobin
22
Creating Sequences in XSLT 2.0

• Every sequence of instructions creates a sequence
  of items
• When a sequence is added to a node
• atomic values are converted to text nodes
• spaces added between atomic values
• nodes are copied to create children

sequence of any items
sequence of new nodes
children of node
23
Temporary Trees

• Variables can be set using select attribute or
  using content
• When setting value using content
• if as attribute not present, create temporary
  tree
• if as attribute present, create sequence

ltxslvariable name"tree"gt 42 lt/xslvariablegt
ltxslvariable name"sequence"
asxsinteger"gt 42 lt/xslvariablegt
24
Creating Sequences in XSLT 2.0

• New instruction ltxslsequencegt
• adds existing nodes or new atomic values to a
  sequence
• Select the line item with highest subtotal

ltxslvariable name"max-expenditure"
as"element(lineitem)"gt ltxslfor-each
select"lineitem"gt ltxslsort select"_at_price
_at_quantity" order"descending" /gt
ltxslif test"position() 1"gt
ltxslsequence select"." /gt lt/xslifgt
lt/xslfor-eachgt lt/xslvariablegt
25
Implications for XSLT 2.0 Use

• Less need for recursive templates
• use integer sequences to iterate a number of
  times
• use ltxslsequencegt to build node sequences by
  iteration rather than recursion
• Less need for temporary elements
• atomic values don't need to be wrapped in an
  element in order to be passed around in a list

26
Text Parsing

• XPath 2.0 has regular expression support
• match(string, regex, flags?) returns true if a
  regular expression matches a substring
• replace(string, regex, replacement, flags?)
  returns the string with all occurrences of the
  regular expression replaced using the replacement
  string
• tokenize(string, regex, flags?) returns a
  sequence of strings created by splitting the
  string on every occurrence of the regular
  expression
• Can do more complex regular expression processing
  using XSLT 2.0

27
Analysing Strings

• XSLT 2.0 has ltxslanalyze-stringgt instruction
• select attribute selects string
• regex attribute holds regular expression
• string split into a sequence of matching and
  non-matching substrings
• processed in turn by
• ltxslmatching-substringgt for matching
• ltxslnon-matching-substringgt for non-matching

28
Example String Analysis
ltpoemgt Mary had a little lamb, Its fleece was
white as snow And everywhere that Mary went
The lamb was sure to go. lt/poemgt
ltxsltemplate match"poem"gt ltpoemgt
ltxslanalyze-string select"." regex"\S."
flags"m"gt ltxslmatching-substringgt
ltlinegtltxslvalue-of select"." /gtlt/linegt
lt/xslmatching-substringgt lt/xslanalyze-string
gt lt/poemgt lt/xsltemplategt
ltpoemgt ltlinegtMary had a little lamb,lt/linegt
ltlinegtIts fleece was white as snowlt/linegt
ltlinegtAnd everywhere that Mary wentlt/linegt
ltlinegtThe lamb was sure to go.lt/linegt lt/poemgt
29
More Text Parsing

• Within ltxslanalyze-stringgt, use regex-group()
  function to get value of matched subexpression

ltxsltemplate match"_at_date"gt ltxslattribute
name"date"gt ltxslvariable name"UK-date-regex
" select"(\d2)\\(\d2)\\(\d
2)" /gt ltxslanalyze-string select"."
regex"UK-date-regex"gt
ltxslmatching-substringgt ltxslsequence
select"concat('20', regex-group(3),
'-', regex-group(2),
'-',
regex-group(1))" /gt lt/xslmatching-substring
gt lt/xslanalyze-stringgt lt/xslattributegt lt/x
sltemplategt
30
Implications for XSLT 2.0 Use

• XSLT 2.0 also allows access to text files with
  unparsed-text() function
• works in a similar way to document()
• Potential to process any text format
• comma-delimited and fixed-format files
• CSS files
• HTML? Java code?
• these are hard because of matching tags/braces
• XSLT could be used for up-conversion to XML

31
Strong Typing in XSLT 2.0

• XPath 2.0 is strongly typed
• the type of a value determines how it is treated
  by some operators (e.g. , )
• if the wrong type of value is passed to a
  function, you will get an error
• Similarly, in XSLT 2.0
• the type of a sort key determines how values are
  sorted
• if the wrong type of value is passed as a
  parameter to a template, you will get an error

32
Declaring Types of Variables

• Declare the type of variables and parameters with
  as attribute
• holds a SequenceType
• item test plus occurrence indicator
• xsinteger means "one or more integers"
• element()? means "an optional element"
• Error if value doesn't comply with type

ltxslfunction name"mathpower"gt ltxslparam
name"base" as"xsdecimal" /gt ltxslparam
name"power" as"xsinteger" /gt
lt/xslfunctiongt
33
Declaring Type of Functions

• Declare the return type of functions and
  templates with as attribute
• holds a SequenceType
• Generated sequence will be converted to atomic
  sequence type if possible

ltxslfunction name"mathpower" as"xsdecimal"gt
ltxslparam name"base" as"xsdecimal" /gt
ltxslparam name"power" as"xsinteger" /gt
lt/xslfunctiongt
ltxsltemplate match"_at_date" as"attribute(_at_date,
)"gt ltxslattribute name"date"gtlt/xslattribute
gt lt/xsltemplategt
34
Node Typing in XSLT 2.0

• In XPath 2.0, every element and attribute has a
  type
• can select nodes based on their type
• //attribute(_at_, xsdate) selects all the
  attributes in the document that hold dates
• Similarly, in XSLT 2.0
• can match nodes based on their type
• attribute(_at_, xsdate) matches all attributes
  that hold dates
• can create elements and attributes of particular
  types

35
Creating Elements and Attributes

• Use xsltype attribute to indicate type of
  element/attribute
• can use this to type-annotate documents without
  schema validation

ltxsltemplate match"_at_date" as"attribute(_at_date,
xsdate)"gt ltxslattribute name"date"
type"xsdate"gt lt/xslattributegt lt/xsltem
plategt
ltxsltemplate match"employee" as"element(person,
xstoken)"gt ltperson xsltype"xstoken"gt
ltxslvalue-of select"name" /gt
lt/persongt lt/xsltemplategt
36
Importing Schemas

• Need to import schemas to use
• user-defined types
• substitution groups
• Import with ltxslimport-schemagt
• namespace identifies target namespace
• schema-location locates schema
• Enables validation of result tree

ltxslimport-schema namespace"http//www.w3.org/19
99/xhtml" schema-location"xhtm
l.xsd" /gt ltxsltemplate match"element(_inline)"gt
lt/xsltemplategt
37
Implications for XSLT 2.0 Use

• Easy to get errors from a stylesheet unless
  you're rigorous in keeping track of types
• declare types of variables and parameters
• cast elements/attributes to particular types
• Well-designed schemas become a useful tool
• substitution groups and appropriate types help
  reduce number of templates
• check whether the result conforms to the schema
  while generating it

38
Conclusions

• XSLT 2.0 introduces a lot of new features
• Many stylesheets can be simpler
• multi-step processing with temporary trees
• grouping using ltxslfor-each-groupgt
• user-defined functions for repetitive code
• Stylesheet applications can be simpler
• multiple result documents should reduce need for
  client-side scripting
• XSLT 2.0 expands into text parsing
• Using schemas/types well will make things easier
  not using them will make things harder

39
For Details

• XPath 2.0 Working Draft
• http//www.w3.org/TR/xpath20
• XSLT 2.0 Working Draft
• http//www.w3.org/TR/xslt20
• Saxon implementation
• http//saxon.sourceforge.net
• Comments please!
• public-qt-comments_at_w3.org