XML::Validator::Schema(3) validate XML against a subset of W3C XML Schema

SYNOPSIS


use XML::SAX::ParserFactory;
use XML::Validator::Schema;
#
# create a new validator object, using foo.xsd
#
$validator = XML::Validator::Schema->new(file => 'foo.xsd');
#
# create a SAX parser and assign the validator as a Handler
#
$parser = XML::SAX::ParserFactory->parser(Handler => $validator);
#
# validate foo.xml against foo.xsd
#
eval { $parser->parse_uri('foo.xml') };
die "File failed validation: $@" if $@;

DESCRIPTION

This module allows you to validate XML documents against a W3C XML Schema. This module does not implement the full W3C XML Schema recommendation (http://www.w3.org/XML/Schema), but a useful subset. See the SCHEMA SUPPORT section below.

IMPORTANT NOTE: To get line and column numbers in the error messages generated by this module you must install XML::Filter::ExceptionLocator and use XML::SAX::ExpatXS as your SAX parser. This module is much more useful if you can tell where your errors are, so using these modules is highly recommeded!

INTERFACE

  • "XML::Validator::Schema->new(file => 'file.xsd', cache => 1)"

    Call this method to create a new XML::Validator:Schema object. The only required option is "file" which must provide a path to an XML Schema document.

    Setting the optional "cache" parameter to 1 causes XML::Validator::Schema to keep a copy of the schema parse tree in memory. The tree will be reused on subsequent calls with the same "file" parameter, as long as the mtime on the schema file hasn't changed. This can save a lot of time if you're validating many documents against a single schema.

    Since XML::Validator::Schema is a SAX filter you will normally pass this object to a SAX parser: 

      $validator = XML::Validator::Schema->new(file => 'foo.xsd');
  $parser = XML::SAX::ParserFactory->parser(Handler => $validator);

    Then you can proceed to validate files using the parser: 

      eval { $parser->parse_uri('foo.xml') };
  die "File failed validation: $@" if $@;

    Setting the optional "debug" parameter to 1 causes XML::Validator::Schema to output elements and associated attributes while parsing and validating the XML document. This provides useful information on the position where the validation failed (although not at useful as the line and column numbers included when XML::Filter::ExceptiionLocator and XML::SAX::ExpatXS are used).

RATIONALE

I'm writing a piece of software which uses Xerces/C++ ( http://xml.apache.org/xerces-c/ ) to validate documents against XML Schema schemas. This works very well, but I'd like to release my project to the world. Requiring users to install Xerces is simply too onerous a requirement; few will have it already and the Xerces installation system leaves much to be desired.

On CPAN, the only available XML Schema validator is XML::Schema. Unfortunately, this module isn't ready for use as it lacks the ability to actually parse the XML Schema document format! I looked into enhancing XML::Schema but I must admit that I'm not smart enough to understand the code... One day, when XML::Schema is completed I will replace this module with a wrapper around it.

This module represents my attempt to support enough XML Schema syntax to be useful without attempting to tackle the full standard. I'm sure this will mean that it can't be used in all situations, but hopefully that won't prevent it from being used at all.

SCHEMA SUPPORT

Supported Elements

The following elements are supported by the XML Schema parser. If you don't see an element or an attribute here then you definitely can't use it in a schema document.

You can expect that the schema document parser will produce an error if you include elements which are not supported. However, unsupported attributes may be silently ignored. This should not be misconstrued as a feature and will eventually be fixed.

All of these elements must be in the http://www.w3.org/2001/XMLSchema namespace, either using a default namespace or a prefix. 

  <schema>
     Supported attributes: targetNamespace, elementFormDefault,
     attributeFormDefault
     Notes: the only supported values for elementFormDefault and
     attributeFormDefault are "unqualified."  As such, targetNamespace
     is essentially ignored.
        
  <element name="foo">
     Supported attributes: name, type, minOccurs, maxOccurs, ref
  <attribute>
     Supported attributes: name, type, use, ref
  <sequence>
     Supported attributes: minOccurs, maxOccurs
  <choice>
     Supported attributes: minOccurs, maxOccurs
  <all>
     Supported attributes: minOccurs, maxOccurs
  <complexType>
    Supported attributes: name
  <simpleContent>
    The only supported sub-element is <extension>.
  <extension>
    Supported attributes: base
    Notes: only allowed inside <simpleContent>
  <simpleType>
    Supported attributes: name
  <restriction>
    Supported attributes: base
    Notes: only allowed inside <simpleType>
  <whiteSpace>
    Supported attributes: value
  <pattern>
    Supported attributes: value
  <enumeration>
    Supported attributes: value
  <length>
    Supported attributes: value
  <minLength>
    Supported attributes: value
  <maxLength>
    Supported attributes: value
  <minInclusive>
    Supported attributes: value
  <minExclusive>
    Supported attributes: value
  <maxInclusive>
    Supported attributes: value
  <maxExclusive>
    Supported attributes: value
  <totalDigits>
    Supported attributes: value
  <fractionDigits>
    Supported attributes: value
  <annotation>
  <documentation>
    Supported attributes: name
  <union>
    Supported attributes: MemberTypes

Simple Type Support

Supported built-in types are: 

  string
  normalizedString
  token
  NMTOKEN
   Notes: the spec says NMTOKEN should only be used for attributes,
   but this rule is not enforced.
  boolean
  decimal
   Notes: the enumeration facet is not supported on decimal or any
   types derived from decimal.
  integer
  int
  short
  byte
  unsignedInt
  unsignedShort
  unsignedByte
  positiveInteger
  negativeInteger
  nonPositiveInteger
  nonNegativeInteger
  dateTime
    Notes: Although dateTime correctly validates the lexical format it does not
    offer comparison facets (min*, max*, enumeration).
  double
    Notes: Although double correctly validates the lexical format it
    does not offer comparison facets (min*, max*, enumeration).  Also,
    minimum and maximum constraints as described in the spec are not
    checked.
  float
    Notes: The restrictions on double support apply to float as well.
  duration
  time
  date
  gYearMonth
  gYear
  gMonthDay
  gDay
  gMonth
  hexBinary
  base64Binary
  anyURI
  QName
  NOTATION

Miscellaneous Details

Other known devations from the specification:
  • Patterns specified in pattern simpleType restrictions are Perl regexes with none of the XML Schema extensions available.
  • No effort is made to prevent the declaration of facets which ``loosen'' the restrictions on a type. This is a bug and will be fixed in a future release. Until then types which attempt to loosen restrictions on their base class will behave unpredictably.
  • No attempt has been made to exclude content models which are ambiguous, as the spec demands. In fact, I don't see any compelling reason to do so, aside from strict compliance to the spec. The content model implementaton uses regular expressions which should be able to handle loads of ambiguity without significant performance problems.
  • Marking a facet ``fixed'' has no effect.
  • SimpleTypes must come after their base types in the schema body. For example, this is ok: 

        <xs:simpleType name="foo">
        <xs:restriction base="xs:string">
            <xs:minLength value="10"/>
        </xs:restriction>
    </xs:simpleType>
    <xs:simpleType name="foo_bar">
        <xs:restriction base="foo">
            <xs:length value="10"/>
        </xs:restriction>
    </xs:simpleType>

    But this is not: 

        <xs:simpleType name="foo_bar">
        <xs:restriction base="foo">
            <xs:length value="10"/>
        </xs:restriction>
    </xs:simpleType>
    <xs:simpleType name="foo">
        <xs:restriction base="xs:string">
            <xs:minLength value="10"/>
        </xs:restriction>
    </xs:simpleType>

CAVEATS

Here are a few gotchas that you should know about:
  • No Unicode testing has been performed, although it seems possible that the module will handle Unicode data correctly.
  • Namespace processing is almost entirely missing from the module.
  • Little work has been done to ensure that invalid schemas fail gracefully. Until that is done you may want to develop your schemas using a more mature validator (like Xerces or XML Spy) before using them with this module.

BUGS

Please use "rt.cpan.org" to report bugs in this module: 

  http://rt.cpan.org

Please note that I will delete bugs which merely point out the lack of support for a particular feature of XML Schema. Those are feature requests, and believe me, I know we've got a long way to go.

SUPPORT

This module is supported on the perl-xml mailing-list. Please join the list if you have questions, suggestions or patches: 

  http://listserv.activestate.com/mailman/listinfo/perl-xml

CVS

If you'd like to help develop XML::Validator::Schema you'll want to check out a copy of the CVS tree: 

  http://sourceforge.net/cvs/?group_id=89764

CREDITS

The following people have contributed bug reports, test cases and/or code: 

  Russell B Cecala (aka Plankton)
  David Wheeler
  Toby Long-Leather
  Mathieu
  [email protected]
  [email protected]
  [email protected]
  [email protected]
  Jean Flouret

AUTHOR

Sam Tregar <[email protected]>

COPYRIGHT AND LICENSE

Copyright (C) 2002-2003 Sam Tregar

This program is free software; you can redistribute it and/or modify it under the same terms as Perl 5 itself.

A NOTE ON DEVELOPMENT METHODOLOGY

This module isn't just an XML Schema validator, it's also a test of the Test Driven Development methodology. I've been writing tests while I develop code for a while now, but TDD goes further by requiring tests to be written before code. One consequence of this is that the module code may seem naive; it really is just enough code to pass the current test suite. If I'm doing it right then there shouldn't be a single line of code that isn't directly related to passing a test. As I add functionality (by way of writing tests) I'll refactor the code a great deal, but I won't add code only to support future development.

For more information I recommend ``Test Driven Development: By Example'' by Kent Beck.

LAST SEARCHED