Static Validation of Dynamically Generated XML Documents

1
Static Validation of Dynamically Generated XML
Documents
A survey on a series of papers by the BRICS
research group at the University of Aarhus,
Denmark. BRICS Basic Research in Computer
Science www.brics.dk
2
Dynamic Construction of Web Pages

• Interactive Web Services

request
response
Client with a browser
Web server

• Application-to-application Web Services

Clients are not humans but general programs
XML widely used as data interchange format for
Web Services
Typical solutions embedded server-side scripting
languages (PHP,ASP,JSP,etc)
Problems 1.Linear document construction
(printing string fragments to output stream)
2.No compile-time validation of the
generated XML (according to a schema)
3
Approaches to Statically Validate Dynamic XML

• Main ideas 1.Construction based on XML templates
  (sequences of XML trees
• containing named gaps)
• 2.Provide compile-time validation of
  transformed XML documents
• An XML schema S describes a set L(S) of XML
  documents
• An XML document X is valid with respect to an XML
  schema S iff X is in the set L(S)
• For a program, verify at compile-time that
  given that
• for each i.

• Evolution
• bigwig - high-level C-based domain-specific
  programming language for developing
• interactive Web services HTML templates are
  first-class values that may be computed
• and stored in variables
• JWIG - Java-based framework for Web service
  development a flexible mechanism for
• dynamic construction and validation of XML
  documents, in particular XHTML
• Xact an extension of JWIG applied on general
  XML, supports also deconstruction of
• XML documents XML templates are first-class JAVA
  values

4
XML templates
An XML template is a well-formed XML fragment
with named gaps
ltrecipe numbernumbergt lttitlegtlttitlegtlt/tit
legt ltingredientsgtltlistgtlt/ingredientsgt
ltpreparationgt ltstepsgtlt/preparationgt
ltnutritiongt lt/recipegt

• There are template gaps and attribute gaps.
• Both templates and strings (or arrays of these)
  can be plugged into template gaps.
• Only strings can be plugged into attribute gaps.

5
Construction
Deconstructionof XML Templates
The plug operation
The select and gapify operations
6
Compile-time Output Validation
Given an XML transformation T and an XML schema
S, is it the case that whenever T outputs an XML
document X, then X is valid with respect to S?

• This is a hard problem
• Practical solution give approximate but safe
  (conservative) answers
• Perform dataflow analysis using Summary Graphs
  (a set of XML template values
• as nodes, and edges corresponding to possible
  plugging of gaps)

The Program Analyzer

• Construct summary graphs for each XML variable
  at each program point. The
• set of summary graphs constitutes a finite
  lattice (suitable for dataflow analysis).
• 2. Construct summary graphs from DTD schemas.
• 3. Validate summary graphs (of output XML
  docs) against DTD schemas.

?
1.
2.
Set of possible XML docs
Set of valid XML docs
3.Validation!
7
Example no errors
import dk.brics.xact. public class Hello
XML person ltpersongtltnamegtJohn
Doelt/namegtlt/persongt XML html
lthtmlgt
ltheadgtlttitlegtHellolt/titlegtlt/headgt
ltbody bgcolor"red"gt
lth1gtHello ltnamegtlt/h1gt
lt/bodygt lt/htmlgt
public XML sayHello()
XML name person.select("/person/name/text()")0
return html.plug("name", name)
public static void main(String
args) Hello hello new Hello()
XML greeting hello.sayHello()
greeting.analyze("xhtml1-transitional.dtd")
System.out.println(greeting)
Output Code
lthtmlgt ltheadgt lttitlegtHellolt/titlegt
lt/headgt ltbody bgcolor"red"gt lth1gtHello
John Doelt/h1gt lt/bodygt lt/htmlgt
Browser Output
Hello John Doe 
8
Example invalid output
import dk.brics.xact. public class Hello
XML person ltpersongtltnamegtJohn
Doelt/namegtlt/persongt XML html
lthtmlgt
ltheadgtlttitlegtHellolt/titlegtlt/headgt
ltbody gbcolor"red"gt
lth1gtHello ltnamegtlt/h1gt
lt/bodygt lt/htmlgt
public XML sayHello()
XML name person.select("/person/name/text()")0
return html.plug("name", name)
public static void main(String
args) Hello hello new Hello()
XML greeting hello.sayHello()
greeting.analyze("xhtml1-transitional.dtd")
System.out.println(greeting)
Output Code
lthtmlgt ltheadgt lttitlegtHellolt/titlegt
lt/headgt ltbody gbcolor"red"gt lth1gtHello
John Doelt/h1gt lt/bodygt lt/htmlgt
Browser Output
Hello John Doe 
The XACT analyzer reports the error! Invalid
XML at line 26 XML template at line 8 attribute
gap 'gbcolor' in element 'body' not declared
9
Example plug error
import dk.brics.xact. public class Hello
XML person ltpersongtltnamegtJohn
Doelt/namegtlt/persongt XML html
lthtmlgt
ltheadgtlttitlegtHellolt/titlegtlt/headgt
ltbody bgcolor"red"gt
lth1gtHello ltnamegtlt/h1gt
lt/bodygt lt/htmlgt
public XML sayHello()
XML name person.select("/person/name/text()")0
return html.plug(nema", name)
public static void main(String
args) Hello hello new Hello()
XML greeting hello.sayHello()
greeting.analyze("xhtml1-transitional.dtd")
System.out.println(greeting)
Output Code
lthtmlgt ltheadgt lttitlegtHellolt/titlegt
lt/headgt ltbody bgcolor"red"gt lth1gtHello
lt/h1gt lt/bodygt lt/htmlgt
Browser Output
Hello
The XACT analyzer reports the error!
Inconsistent plug operation at line 20 Template
gap 'nema' does not exist