Schematron

1
Schematron

• Tim Bornholtz

2
Schema languages

• Many people turn to schema languages when they
  want to be sure that an XML instance follows
  certain rules
• DTD
• XML Schema
• Relax NG

3
What is wrong with that?

• These schema languages have very complicated
  grammars
• Not flexible enough to accurately reflect real
  business rules
• Often very difficult to map the business rules
  written in English to the technical rules defined
  by the Schema

4
What is Schematron

• Schematron is a language that allows you to
  directly express rules
• Without a whole grammatical infrastructure
• Schematron is an ISO standard
• Largest benefit is the ability to describe
  dynamic constraints
• Schematron can describe syntax constraints like
  XML Schema
• Can also describe semantic constraints which XML
  Schema cannot describe

5
Comparing Schema Languages

• White is core capability
• Light gray representscapabilities that
  arepossible butinconvenient
• Dark gray capabilitiesrequire
  specializedschemaorganization

6
Attributes Elements
7
Data Types
8
General Features
9
Why Schematron in Financial Aid?

• We trade many files with our partners and these
  are increasingly XML.
• Schematron is more expressive than other schema
  languages
• Can handle more complicated dependencies that XML
  Schema cannot.

10
Pre-requisites

• I'm going to assume that you're familiar with
• XML
• XML Namespaces
• XPath
• XSLT

11
Components of a Schematron file

• ltschemagt
• ltphasegt - top level construct
• ltpatterngt
• ltrulegt - defines the context
• ltassertgt
• ltreportgt

12
ltassertgt

• Assert is the basic rule within a Schematron file
  to test for a valid condition
• Message will be displayed whenever the test is
  false
• Syntaxltassert testexpression to
  evaluategtMessage to display if the expression is
  falselt/assertgt
• Any XSLT or XPath expression can be used
  including boolean logic and complicated formulas

13
ltreportgt

• Use the ltreportgt instruction to communicate
  information apart from the validation errors
• The message will be displayed when the test
  condition is true.
• Syntax
• ltreport testexpression to evaluategtMessage to
  display if the expression is truelt/reportgt

14
ltrulegt

• Basic building block of a Schematron file.
• Syntax
• ltrule contextLocation in the filegtlt/rulegt
• ltrule contextAwardgt
• ltrule contextStudentgt
• The context can be any valid XPattern
• XPattern is a subset of XPath
• XPattern is the way to identify templates in XSLT
• All of the ltassertgt statements for a particular
  rule context are grouped together

15
ltpatterngt

• A ltpatterngt is a collection of related rules.
• Syntax
• ltpattern nameDescriptive name of the rulesgt
• The rules within a pattern do not need to work on
  the same elements.
• The name will be displayed in the output.
• The name will help you identify which section of
  the document is failing the rules.

16
ltschemagt

• Root element of the document
• Schematron 1.5 must use the namespace
  http//www.ascc.net/xml/schematron
• ISO Schematron must use the namespacehttp//purl
  .oclc.org/dsdl/schematron

17
Very simple example

• lt?xml version"1.0" encoding"UTF-8"?gtltschema
  xmlns"http//www.ascc.net/xml/schematron"gt
  lttitlegtExample 1lt/titlegt ltpattern
  name"Document root"gt ltrule context"/"gt
  ltassert test"doc"gtRoot element should
  be
  "doc".lt/assertgt lt/rulegt
  lt/patterngtlt/schemagt

18
Basic assertions

• Schematron can validate many basic conditions
• Presence of elements
• Absence of elements
• Sequence of elements within an element
• Relative order of elements to each other
• Validate for a certain number of elements

19
Presence or absence of tags

• ltassert testLoanTypegt
• Require that the loan type be an element that is
  a direct child element of the current context
• ltassert testcount(LoanType) 0gt
• Require that the LoanType element is not present
• ltassert testcount() 5gt
• Require that there are exactly 5 child elements
  of the current context

20
Order of elements

• ltrule context"Student"gt
• ltassert test"following-sibling1/selfFFELPL
  US"gt
• A "title" must be immediately followed by a
  "subtitle".
• lt/assertgt
• lt/rulegt

21
Taking your validation further

• Schematron can do may things that XML Schemas
  cannot handle
• Cross field relationships
• Meaningful error messages

22
Cross Field Relationships

• XML Schemas can validate the contents and the
  data type of one field
• Not able to easily validate relationships between
  fields.
• If field A contains a value X then field B must
  contain a value Y

23
PLUS Student cannot be Borrower

• ltrule context"FFELPPLUS"gt ltassert
  test"Borrower/Index/SSN !
  ../Student/Index/SSN"gtBorrower for a PLUS
  loan must not be the studentlt/assertgtlt/rulegt
• FFEPGradPLUS the borrower must be the student
• ltrule context"FFELPGradPLUS"gt ltassert
  test"Borrower/Index/SSN
  ../Student/Index/SSN"gtBorrower for a Grad
  PLUS loan must be the studentlt/assertgtlt/rulegt

24
Grad PLUS Grade Level

• If a student is getting a GradPLUS loan, they
  must be a graduate student.
• ltrule context"FFELPGraduatePLUS"gt ltassert
  test"StudentLevelCode 'FirstYearGraduate' or
  StudentLevelCode'SecondYearGraduate' or
  StudentLevelCode'ThirdYearGraduate' or
  StudentLevelCode'FourthYearGraduate' or
  StudentLevelCode'ContinuingGraduate'"gt Student
  must be a Graduate level for FFELPGradPLUS
  awardlt/assertgtlt/rulegt

25
Stafford Loan Borrower

• For a Stafford loan, the Borrower is the Student
• No Borrower section is sent (different than
  GradPLUS)?
• ltrule context"FFELPStafford"gt ltassert
  test"count(Borrower) 0"gt Borrower
  section not necessary for Stafford
  loanlt/assertgtlt/rulegt

26
Date Ranges

• Checking validity of two related dates
• ltxslimportgtdate.difference.function.xsllt/xslimpo
  rtgtltassert test" datedifference(FinancialAwardB
  eginDate, ../FFELPSubsidized/Disbursement_at_Number
  1 /DisbursementDate) le 30"gt Award Begin
  Date must be within 30 days of first
  disbursementlt/assertgt

27
Meaningful messages

• To get the name of the XML element used in an
  ltassertgt or ltreportgt statement use the ltname/gt
  tag
• The ltvalue-ofgt tag will display the actual value
  of the element.
• ltrule context"FFELPStafford"gt ltassert
  testlength(Email) gt 128"gt ltname/gt
  value ltvalue-of/gt is not valid lt/assertgtlt/rulegt

28
Phases

• If we were to combine all of the business rules
  for CRC into one schema it would be a very large
  file.
• A ltphasegt is a simple collection of patterns that
  are executed together.

29
Namespaces

• Schematron can be used to validate XML that uses
  namespaces
• Declare the namespace as a child of the ltschemagt
  element
• ltns uriurnorgpescmessageCommonLineRequestv1
  .1.0 prefixreqgt
• Then use the namespace like normal
• ltassert testcount(reqLoanType) 1gt

30
Running the examples

• There are implementations available for many
  languages .Net, Java, Python, Perl, Ruby
• The 1.5 reference implementation compiles the
  Schematron file into a XSLT file that can be used
  against the XML instance document.
• Most any XSLT engine can be used with the
  reference implementation but the fastest are
  Saxon for Java and MSXSLT for .Net

31
Running the examples (cont.)?

• I'm using Saxon for Java but the process is
  similar for most all XSLT processors
• 1) java -jar saxon8.jar -o temp.xsl file.sch
  schematron-basic.xsl
• 2) java -jar saxon8.jar instance.xml temp.xsl
• 3) rm temp.xsl

32

• Schematron provides rule processing capabilities
  that other schema languages cannot provide
• Schematron can be used in conjunction with XML
  Schemas
• Continue to use the XML Schemas that are already
  defined
• Define complex rules with Schematron

33
Contact Information

• Tim Bornholtz
• President, The Bornholtz GroupEmail
  tim_at_bornholtz.comPhone [phone]Web
  http//www.bornholtz.com