Introduction to XML Databases

1
Introduction to XML Databases

• Sylvia Osborn

2
What is XML?

• XML (eXtensible Markup Language) is a language
  that describes text and tags.
• It is defined by a number of documents available
  on the w3c (world wide web consortium) web site
  (http//www.w3.org).
• Traditionally, tagged text has been used for
  publishing (e.g. SGML) documents.
• HTML also consists of tags and text, but HTML has
  a pre-defined set of tags to markup (or show some
  formatting) for text that is to be displayed in a
  web browser.
• there was a lot of interest in the database
  community in semi-structured data, i.e. data
  with some structure, and with some parts that
  were just very long. E.g. a book would have a
  title, author, publisher and then just the book
  contents.
• When we have any collection of similar objects on
  a persistent storage medium, we want to query
  them. We will see a couple of query language
  ideas for XML.
• Like most standards, things keep evolving.

3
An Sample XML Document

• lt?xml version"1.0" encoding"UTF-8"?gt
• ltstudentList xmlnsxsi"http//www.w3.org/2001/XML
  Schema-instance" xsinoNamespaceSchemaLocation"Z
  \853\exampleschema.xsd"gt
• ltstudentgt
• ltNamegtJohn Doelt/Namegt
• ltStudentIDgt123456lt/StudentIDgt
• ltAddressgt
• ltStNumbergt414lt/StNumbergt
• ltStNamegtPine Streetlt/StNamegt
• ltCitygtLondonlt/Citygt
• lt/Addressgt
• lt/studentgt
• ltstudentgt
• ltNamegtBob Daylt/Namegt
• ltStudentIDgt9876543212lt/StudentIDgt
• ltAddressgt
• ltPOBoxgtBox 255lt/POBoxgt
• ltCitygtLondonlt/Citygt
• lt/Addressgt
• lt/studentgt

4
Some parts of XML

• an XML element is any text properly nested
  between two matching tags
• ltsampleTaggt ... lt/sampleTaggt.
• the name of the element is the tag text (here
  sampleTag)
• content is the text between the tags
• ltsampleTaggtexample contentlt/sampleTaggt
• ltsampleTaggt example content lt/sampleTaggt
• has different content i.e. the blanks are part
  of the tagged text.
• parent-child relationships between elements are
  given by the nesting of the tags. In the previous
  example, Name is a child of student, and
  studentList is a parent of student.

5
XML parts, contd

• attributes can be attached to the opening tag. In
  the above example, there are two attributes for
  studentList. Their values are enclosed in .
• attribute values must be enclosed in quotes.
• Element text is not quoted.
• empty elements can be given by ltPOBox/gt.
• An element can be empty but have attribute
  values, e.g. ltsampleTag attr1value
  attr2value2/gt
• comments are any text enclosed in lt! ... gt. In
  Galax, a comment is enclosed in ( )
• processing instructions are enclosed in lt? ...
  ?gt and may be used by the XML processor receiving
  the document.

6
How does XML differ from HTML?

• XML documents can contain any number of different
  tags chosen by the document author, whereas HTML
  tags are predefined in the official specification
  of HTML
• In XML, all opening tags must have a matching
  closing tag. Some HTML tags (like ltpgt, which
  starts a new paragraph) do not need closing tags.
  Others may be omitted and the browsers won't
  complain.

7
A well-formed XML Document

• is one in which
• there is a root element
• the tags are properly nested, and every opening
  tag is followed by the corresponding closing tag
  
• attributes occur at most once in an opening tag,
  and their values must be given and must be in
  quotes.
• At first glance, they look like aggregation
  hierarchies, expressed a different way. There
  are some differences...

8
Differences between XML documents and database
objects

• XML is a markup language, so the following is a
  legal ltAddressgt
• ltAddressgt John Doe lives on
• at house number ltStNumbergt 123 lt/StNumbergt
• on ltStNamegt Main St. lt/StNamegt in
• ltCitygt London lt/Citygt
• which is on the north side, at the corner.
  
• lt/Addressgt
• The data existing outside of the structure would
  not be allowed in a database system.

9
Differences, contd

• 2. XML documents are ORDERED. These 2 objects for
  an OODB are identical (the same)
• OID 123 StNumber "123", StName "Main
  Street"
• OID 123 StName "Main Street", StNumber
  "123"
• whereas these 2 XML documents are different
• ltAddressgt ltStNamegt Main St. lt/StNamegt
• ltStNumbergt 123 lt/StNumbergt
• lt/Addressgt
• ltAddressgt ltStNumbergt 123 lt/StNumbergt
• ltStNamegt Main St. lt/StNamegt
• lt/Addressgt

10
Schema definitions for XMLdocuments

• There are two ways of specifying document type
  definitions DTDs and XML Schema
• DTDs are older XML Schema is fairly new
• DTD syntax is not in XML, XML Schema is
• DTDs allow you to specify the elements, their
  nesting, the attributes, and to give a data type
  to the element contents.
• With DTDs, the choice of data types is extremely
  limited - basically just PCDATA which means
  parsed character data
• attributes in a DTD can be of types string, ID,
  IDREF, IDREFS
• ID attributes must have a value which is unique
  throughout the document (can act like an object
  ID)
• IDREF refers to an ID somewhere else, IDREFS is
  one or more IDREF
• they also did not know about namespaces

11
Sample DTD

• lt!DOCTYPE PersonList
• lt!ELEMENT PersonList (Title, Contents)gt
• lt!ELEMENT Title EMPTY gt
• lt!ELEMENT Contents (Person)gt
• lt!ELEMENT Person (Name, Id, Address))gt
• lt!ELEMENT Name (PCDATA)gt
• lt!ELEMENT Id (PCDATA)gt
• lt!ELEMENT Address (Number, Street)gt
• lt!ELEMENT Number (PCDATA)gt
• lt!ELEMENT Street (PCDATA)gt
• lt!ATTLIST PersonList Type CDATA IMPLIED
• Date CDATA IMPLIEDgt
• lt!ATTLIST Title Value CDATA REQUIREDgt
• gt

12
Namespaces

• a namespace is a way of specifying where the
  definition of a tag should come from, so that
  multiple definitions of the same tag can be used
  in the same document.
• Namespaces are defined by using the special
  attribute xmlns
• ltitem xmlns"http//www.somesite.com/mypagepartd
  efs" xmlnsmice"http//www.somesite.com/m
  ypagemicedefs"gt
• ltnamegtcpult/namegt ltmicenamegtwirelesslt/micenam
  egtlt/itemgt
• the definitions of item and name come from the
  default namespace defined by the xmlns
  attribute. Anything prefixed by mice is defined
  in the alternate micedefs namespace.
• you can have many alternate namespaces as long as
  their names are distinct.
• you can even nest namespace definitions in the
  document. A scoping rule is used.
• The idea is that within certain areas of
  discourse, e.g. e-commerce, some well accepted
  namespaces will evolve and be used by everyone
  doing this type of business.

13
XML Schema

• XML Schema is a data definition language for XML
  documents (i.e. it gives the schema definition)
• XML Schema uses XML syntax
• it is integrated with the namespace mechanism
• it provides a number of built-in types
• it allows us to define complex types
• it supports key and referential integrity
  constraints
• it provides a better way than DTDs of specifying
  documents where the order does not matter

14
Structures in XML Schema

• Primitive types are decimal, integer, float,
  boolean, date, string, ID and IDREF
• Simple types can also include lists and unions
  (either this structure or that)
• Simple types can be defined by patterns
• There are constraints (called restrictions) like
  min and max value
• Can specify enumerated types (just list all the
  valid values)

15
Examples of Simple Type Definitions in XML Schema

• ltxssimpleType name"uwoStudentNumberType"gt
• ltxsrestriction base"xsstring"gt
• ltxspattern value"1-90-97"/gt
• lt/xsrestrictiongt
• lt/xssimpleTypegt
• ltxssimpleType name"workerAgeType"gt
• ltxsrestriction base"xsinteger"gt
• ltxsminInclusive value"14"/gt
• ltxsmaxInclusive value"65"/gt
• lt/xsrestrictiongt
• lt/xssimpleTypegt

16
More Simple Type Examples

• ltxssimpleType name"eyeColourType"gt
• ltxsrestriction base"xsstring"gt
• ltxsenumeration value"blue"/gt
• ltxsenumeration value"brown"/gt
• ltxsenumeration value"hazel"/gt
• lt/xsrestrictiongt
• lt/xssimpleTypegt
• ltxssimpleType name"studentIDsType"gt
• ltxslist itemType"uwoStudentNumberType"/gt
• lt/xssimpleTypegt

17
Element Types and Complex Types

• Simple elements are declared as follows
• ltelement namePersonName typeString/gt
• Attributes are declared as follows
• ltattribute nameage typeworkerAgeType/gt
• Complex types can consist of a sequence of
  elements (which are ordered), an unordered set
  (specified by the tag all), or a choice
• None of them is default, so a complex type must
  always say what it is

18
Example Complex Type

• ltxscomplexType name"addressType"gt
• ltxssequencegt
• ltxschoicegt
• ltxselement name"POBox" type"xsstring"/gt
• ltxssequencegt
• ltxselement name"StNumber"
  type"xsstring"/gt
• ltxselement name"StName"
  type"xsstring"/gt
• lt/xssequencegt
• lt/xschoicegt
• ltxselement name"City" type"xsstring"/gt
• lt/xssequencegt
• lt/xscomplexTypegt

19
A type with some subelements

• ltxscomplexType name"StudentType"gt
• ltxssequencegt
• ltxselement name"Name" type"xsstring"/gt
• ltxselement name"StudentID"
  type"uwoStudentNumberType"/gt
• ltxselement name"Address" type"addressType"/gt
• lt/xssequencegt
• lt/xscomplexTypegt

20
To have lists of them

• ltxscomplexType name"studentListType"gt
• ltxssequence maxOccurs"unbounded"gt
• ltxselement name"student" type"StudentType"/gt
• lt/xssequencegt
• lt/xscomplexTypegt
• ltxselement name"studentList" type"studentListTy
  pe"/gt

21
Some data using this schema

• lt?xml version"1.0" encoding"UTF-8"?gt
• ltstudentList xmlnsxsi"http//www.w3.org/2001/XML
  Schema-instance" xsinoNamespaceSchemaLocation"Z
  \853\exampleschema.xsd"gt
• ltstudentgt
• ltNamegtJohn Doelt/Namegt
• ltStudentIDgt12345678lt/StudentIDgt
• ltAddressgt
• ltStNumbergt414lt/StNumbergt
• ltStNamegtPine Streetlt/StNamegt
• ltCitygtLondonlt/Citygt
• lt/Addressgt
• lt/studentgt
• ltstudentgt
• ltNamegtBob Daylt/Namegt
• ltStudentIDgt98765432lt/StudentIDgt
• ltAddressgt
• ltPOBoxgtBox 255lt/POBoxgt
• ltCitygtLondonlt/Citygt
• lt/Addressgt
• lt/studentgt

22
ID and IDref

• An attribute can be declared to be an ID
• Values of type ID are what XML calls Names
• These Names must be unique values throughout the
  whole documents (i.e. provide a unique ID, given
  with the data)
• Elements must have at most one ID attribute
• Values of type IDREF must match the value of an
  ID attribute for some element in the document
• Names start with a letter _ or and can contain
  letters, digits, and other characters (see XML
  documents)
• IDs are like user-defined object ids with a
  specific syntax, and an IDREF is a pointer to them