Home > Articles > Programming

  • Print
  • + Share This

XML Schema

XML Schema is an initiative from the W3C that attempts to eliminate the problems with DTDs by replacing them with something better. XML Schemas are more flexible than DTDs, more powerful, and more familiar, as they use XML syntax.


Note - As of this writing, the XML Schema has advanced to "Candidate Recommendation" status at W3C, but Xerces version 1.2.3 uses the April 7, 2000 draft, located at

http://www.w3.org/TR/2000/WD-xmlschema-1-20000407/

http://www.w3.org/TR/2000/WD-xmlschema-2-20000407/

and

http://www.w3.org/TR/2000/WD-xmlschema-3-20000407/


To begin converting our DTD to a schema, we'll create a third text file called products.xsd (for XML Schema Document) and save it in the same directory as products.xml. The file should contain the text in Listing 3.17.

Listing 3.17  products.xsd: The Basic Schema Document

0: <?xml version="1.0" ?>
1: <schema>
2:
3: </schema>

Notice that this is just an XML document, same as all the others we've used so far. All definitions we create will be part of the schema element. This file is called our schema document. The file we're validating—our XML file—is called our instance document, because it's an instance of the data that we're defining.

Now, you may remember that an XSL style sheet was also an XML document, and that we separated the XSL commands from the output using namespaces—specifically, the xsl: namespace. We're going to do the same with our schema document, but namespaces allow us to do something else: We can exclude elements from validation, if we want to.

To do this, we'll create a namespace that points to schema validation, and then tell the parser that any elements that are not part of that namespace need to be processed according to our schema document. We'll do this by making the changes in Listing 3.18 to products.xml.

Listing 3.18  products.xml: Associating Our XML with a Schema Document

<?xml version="1.0"?>
<products xmlns:xsi="http://www.w3.org/1999/XMLSchema-instance"
      xsi:noNamespaceSchemaLocation='products.xsd'>

<vendor webvendor="full" id="conners">
...

Now we're ready to start converting our DTD to a Schema.

Datatypes

The concept behind XML Schema is that we start with simple datatypes (such as a string of text or a number) and build them into more complex structures, known as complex datatypes. This in and of itself is an improvement over DTDs, where we couldn't specify that, say, a price had to be a number. Some of the simple types built into XML Schema include the following:

  • string—This is simple text.

  • numeric types—These include float, double, decimal, integer, nonPositiveInteger, negativeInteger, long, int, short, byte, nonNegativeInteger, unsignedLong, unsignedInt, unsignedShort, unsignedByte, and positiveInteger.

  • time-related types—These include timeInstant, timeperiod, month, year, century, recurringDate, recurringDay, timeDuration, recurringDuration, date, and time.

  • XML 1.0 tokenized types—These are type names that have special meanings, such as ID, IDREF, ENTITY, and NMTOKEN.

Let's start by creating the elements that use simple datatypes in Listing 3.19.

Listing 3.19  Our Simple Elements

0: <?xml version="1.0">
1: <schema>
2: 
3: <element name="vendor_name" type="string"/>
4: <element name="b" type="string"/>
5: <element name="i" type="string"/>
6: <element name="p" type="string"/>
7: <element name="short_desc" type="string"/>
8: <element name="product_desc" type="string"/>
9: <element name="giveaway_item" type="string"/>
10:<element name="giveaway_desc" type="string"/>
11:<element name="long_desc" type="string"/>
12:<element name="product_id" type="string"/>
13:
14:</schema>

Now, it just so happens that all the simple types we had originally defined as #PCDATA really were strings, but if we had any that needed to be, say, numbers or dates, we would define them here.

Next, we'll start to build up the elements that use these simple elements into complex datatypes. The simplest way of doing this is shown in Listing 3.20.

Listing 3.20  products.xsd: Creating an Element of Elements

0: <?xml version="1.0"?>
1: <schema>
2:
3: <element name="vendor_name" type="string"/>
4: <element name="b" type="string"/>
5: <element name="i" type="string"/>
6: <element name="p" type="string"/>
7: <element name="short_desc" type="string"/>
8: <element name="product_desc" type="string"/>
9: <element name="giveaway_item" type="string"/>
10:<element name="giveaway_desc" type="string"/>
11:<element name="long_desc" type="string"/>
12:<element name="product_id" type="string"/>
13:
14:<element name="giveaway">
15:   <element name="giveaway_item" type="string"/>
16:   <element name="giveaway_desc" type="string"/>
17:</element>
18:
19:</schema>

On lines 14 through 17, we're building the definition of the giveaway element the same way that we'll build the element itself: by nesting the elements inside it. A giveaway consists of a giveaway_item and a giveaway_desc, so we include them as part of this definition on lines 15 and 16. We also add attributes to an element this way, as shown in Listing 3.21.

Listing 3.21  products.xsd: Creating Attributes

... 
16:   <element name="giveaway_desc" type="string"/>
17:</element>
18:
19:<element name="special" type="string">
20:  <attribute name="specialtype" type="string" use="fixed" value="weekly" />
21:</element>
22:
23:</schema>

Here, on lines 19 through 21 we're adding attributes much the same way we added them under XSL. We do, however, have two new values to look at on line 20. The use value (in this case "fixed") tells the processor whether this attribute is implied, required, and so on. If there is a default to be specified, it will be found as the value, as we did here with "weekly".

Next we'll construct some of the more complex elements. So far we've looked at elements that will contain text. In Listing 3.22, we will look at elements that are made up of other elements.

Listing 3.22  Specifying Element Occurrences

... 
19:<element name="special" type="string">
20: <attribute name="specialtype" type="string" use="fixed" value="weekly" />
21:</element>
22:
23:<element name="item" content="elementOnly">
24:   <element ref="product_desc" />
25:   <element ref="price" minOccurs="0" maxOccurs="unbounded"/>
26:</element>
27:
28:<element name="suite" content="elementOnly">
29:   <element ref="product_id" />
30:   <element ref="short_desc" />
31:   <element ref="long_desc" />
32:   <element ref="price" />
33:   <element ref="product" minOccurs="0" maxOccurs="unbounded" />
34:</element>
35:
36:<element name="product" content="elementOnly">
37:   <element ref="product_id" />
38:   <element ref="short_desc" />
39:   <element ref="product_desc" minOccurs="0" maxOccurs="1" />
40:   <element ref="price" minOccurs="1" maxOccurs="unbounded" />
41:   <element ref="item" minOccurs="0" maxOccurs="unbounded" />
42:   <element ref="inventory" minOccurs="0" maxOccurs="unbounded" />
43:   <element ref="giveaway" minOccurs="0" maxOccurs="1" />
44:</element>
45:
46:<element name="vendor" content="elementOnly">
47:   <element ref="vendor_name" />
48:   <element ref="advertisement" minOccurs="0" maxOccurs="1" />
49:   <element ref="suite" minOccurs="0" maxOccurs="unbounded" />
50:   <element ref="product" minOccurs="0" maxOccurs="unbounded" />
51:   <attribute name="webvendor" type="string" use="required" />
52:   <attribute name="id" type="ID" use="required" />
53:</element>
54:
55:</schema>

There's a lot of new material here, but most of it is along the same lines, so let's start with items on lines 23 through 26. First of all, an item element can contain only other elements, which is what content="elementOnly" means on line 23. The actual elements themselves are specified on lines 24 and 25, but because both product_desc and price appear elsewhere, rather than defining them here, we'll simply refer to the datatypes we've already created.

On line 25, we see the first of the attributes that limit how many times an element can occur. When we created the DTD, we specified a price element within an item element with *, meaning that it could occur 0 or more times. Now we're duplicating this requirement with minOccurs and maxOccurs. The advantage here over a DTD is that if we wanted to, we could set these values to specific integers, as opposed to just 0, 1, or unbounded.

We've used the same information to describe the suite element on lines 28 through 34, product on lines 36 through 44, and vendor on lines 46 through 53.

That leaves us with price, inventory, advertisement, and our root element, products. We'll look at advertisement and products in a moment, but price and inventory have a special complication to them: enumerated values. In Listing 3.23, we see how to specify enumerated values for an attribute.

Listing 3.23  products.xsd: Enumerated Values

... 
52:   <attribute name="id" type="ID" use="required" />
53:</element>
54:
55:<element name="price" type="decimal">
56: <attribute name="pricetype" type="NMTOKEN" use="default" value="retail">
57:    <simpleType base="string">
58:      <enumeration value="cost"/>
59:      <enumeration value="sale"/>
60:      <enumeration value="retail"/>
61:      <enumeration value="starting"/>
62:    </simpleType>
63: </attribute>
64: <attribute name="color" type="string" use="implied" />
65:</element>
66:
67:<element name="inventory" type="integer">
68: <attribute name="location" type="NMTOKEN" use="default" value="warehouse">
69:    <simpleType base="string">
70:      <enumeration value="warehouse"/>
71:      <enumeration value="showroom"/>
72:    </simpleType>
73: </attribute>
74:  <attribute name="color" type="string" use="implied" />
75:</element>
76:
77:</schema>

On lines 55 through 65, we're specifying the price element. The first thing to notice is that on line 55 we specified the type as decimal, which we certainly couldn't have done with a DTD. On lines 56 through 63, we're defining the pricetype attribute. This attribute is a new datatype, NMTOKEN. That means that we can use only certain defined values, or tokens. On lines 57 through 62, we're defining those values. The simpleType element tells the processor that ultimately, we're going to wind up with a string. The elements on lines 58 through 61 place further conditions on the value. In this case, we're specifying certain values, but other restrictions are possible.

Line 64 is just a straight attribute definition, as we've already seen, and we're repeating the process for the inventory element on lines 67 through 75.

We'll create the products element in Listing 3.24.

Listing 3.24  products.xsd: Creating the Root Element

... 
74:  <attribute name="color" type="string" use="implied" />
75:</element>
76:
77:<element name="products" content="elementOnly">
78:   <choice maxOccurs="unbounded">
79:    <element ref="vendor" />
80:    <element ref="special" />
81:   </choice>
82:</element>
83:
84:</schema>

Our DTD specified that the products element could have any number of vendors or specials, in no particular order. To achieve this same result in the schema, we can use the choice element. Alone, it allows one of those elements to be present. After we add maxOccurs="unbounded", any number of them is allowed.

Finally, we're down to our problem child, advertisement. It's our problem child because it has both text and elements. This is known as mixed content. We'll take care of it in Listing 3.25.

Listing 3.25  Specifying Mixed Content

... 
82:</element>
83:
84:<element name="advertisement" content="mixed">
85:   <element name="ad_sentence" content="mixed">
86:     <element ref="b" minOccurs="0"/>
87:     <element ref="i" minOccurs="0"/>/>
88:     <element ref="p" minOccurs="0"/>/>
89:   </element>
90:</element>
91:
92:</schema>

I've included this example here even though it's not really what we want. When we defined mixed content with the DTD, we were not able to control the order of content, or even whether items appeared. With schemas, we can, by specifying the content as it is in Listing 3.25. In this case, however, we don't care about the order, so it's better to use a choose element as we did in Listing 3.24.

Features Still to Come

Two of the most promising features of XML Schema, uniqueness and referential integrity, have not been implemented in the version of Xerces that we are using. It is, however, worth taking a few moments to discuss how they work.

Uniqueness, as the name implies, is the ability to specify that every value of a particular element or attribute must be different; there can be no duplicates. Referential integrity, as we discussed earlier, is when we specify that one value, such as a special's vendor_id attribute, must match a value elsewhere, such as the vendor's id attribute.

For both uniqueness and referential integrity, we need to create keys. These keys involve the element that the key is attached to and the field or attribute that is being checked. For instance, to specify that we want the id to be unique for a vendor, we would add the following to the schema:

<unique>
   <selector>/products/vendor</selector>
   <field>vendor[@id]</field>
</unique>

If that selector looks familiar, it should. It's an XPath expression, just like the ones that we used with XSL. This way, we can distinguish between a product_id, and a vendor id.

To create referential integrity, we first create the vendor_id as a key:

<key name="vendor_key">
<selector>/products/vendor</selector>
<field>vendor[@id]</field>
</key>

We then create a keyref type:

<keyref refer="vendor_key">
<selector>/products/special</selector>
<field>special[@vendor_id]</field>_</keyref>

This forces any vendor_id attribute to match the vendor_key.

  • + Share This
  • 🔖 Save To Your Account

Related Resources

There are currently no related titles. Please check back later.