Programming

Objectives
How XML may be manipulated from general-

An Introduction to XML and Web Technologies purpose programming languages
How streaming may be useful for handling large
XML Programming documents
Anders Møller & Michael I. Schwartzbach

© 2006 Addison-Wesley
An Introduction to XML and Web Technologies 2
General Purpose XML Programming The JDOM Framework
Needed for: An implementation of generic XML trees in Java

• domain-specific applications Nodes are represented as classes and interfaces
• implementing new generic tools
DOM is a language-independent alternative

Important constituents:
• parsing XML documents into XML trees
• navigating through XML trees
• manipulating XML trees
• serializing XML trees as XML documents
An Introduction to XML and Web Technologies 3 An Introduction to XML and Web Technologies 4
1
JDOM Classes and Interfaces A Simple Example
The abstract class Content has subclasses: int xmlHeight(Element e) {

java.util.List contents = e.getContent();
• Comment java.util.Iterator i = contents.iterator();
int max = 0;
• DocType while (i.hasNext()) {
• Element Object c = i.next();
int h;
• EntityRef if (c instanceof Element)
• ProcessingInstruction h = xmlHeight((Element)c);
else
• Text h = 1;
Other classes are Attribute and Document if (h > max)
max = h;
The Parent interface describes Document and }
Element return max+1;

}
Another Example A Final Example (1/3)
static void doubleSugar(Document d)

throws DataConversionException {
Modify all elements like
Namespace rcp = <ingredient name="butter" amount="0.25" unit="cup"/>
Namespace.getNamespace("http://www.brics.dk/ixwt/recipes");
Filter f = new ElementFilter("ingredient",rcp); into a more elaborate version:
java.util.Iterator i = d.getDescendants(f); <ingredient name="butter">
while (i.hasNext()) { <ingredient name="cream" unit="cup" amount="0.5" />
Element e = (Element)i.next();
<preparation>
if (e.getAttributeValue("name").equals("sugar")) {
Churn until the cream turns to butter.
double amount = e.getAttribute("amount").getDoubleValue();
e.setAttribute("amount",new Double(2*amount).toString()); </preparation>
} </ingredient>
}
}
2
A Final Example (2/3) A Final Example (3/3)
void makeButter(Element e) throws DataConversionException { Element cream = new Element("ingredient",rcp);

Namespace rcp = cream.setAttribute("name","cream");
Namespace.getNamespace("http://www.brics.dk/ixwt/recipes"); cream.setAttribute("unit",c.getAttributeValue("unit"));
java.util.ListIterator i = e.getChildren().listIterator(); double amount = c.getAttribute("amount").getDoubleValue();
while (i.hasNext()) { cream.setAttribute("amount",new Double(2*amount).toString());
Element c = (Element)i.next(); butter.addContent(cream);
if (c.getName().equals("ingredient") && Element churn = new Element("preparation",rcp);
c.getAttributeValue("name").equals("butter")) { churn.addContent("Churn until the cream turns to butter.");
Element butter = new Element("ingredient",rcp); butter.addContent(churn);
butter.setAttribute("name","butter"); i.set((Element)butter);
} else {
makeButter(c);
}
}
}
Parsing and Serializing Validation (DTD)
public class ValidateDTD {

public class ChangeDescription {
public static void main(String[] args) {
try {
try {
SAXBuilder b = new SAXBuilder();
SAXBuilder b = new SAXBuilder();
b.setValidation(true);
Document d = b.build(new File("recipes.xml"));
String msg = "No errors!";
Namespace rcp =
try {
Namespace.getNamespace("http://www.brics.dk/ixwt/recipes");
Document d = b.build(new File(args[0]));
d.getRootElement().getChild("description",rcp)
} catch (JDOMParseException e ) {
.setText("Cool recipes!");
msg = e.getMessage();
XMLOutputter outputter = new XMLOutputter();
}
outputter.output(d,System.out);
System.out.println(msg);
} catch (Exception e) { e.printStackTrace(); }
}
}
}
}
3
Validation (XML Schema) XPath Evaluation
public class ValidateXMLSchema {
public static void main(String[] args) { void doubleSugar(Document d) throws JDOMException {
try { XPath p = XPath.newInstance("//rcp:ingredient[@name='sugar']");
SAXBuilder b = new SAXBuilder(); p.addNamespace("rcp","http://www.brics.dk/ixwt/recipes");
b.setValidation(true); java.util.Iterator i = p.selectNodes(d).iterator();
b.setProperty( while (i.hasNext()) {
"http://java.sun.com/xml/jaxp/properties/schemaLanguage", Element e = (Element)i.next();
"http://www.w3.org/2001/XMLSchema"); double amount = e.getAttribute("amount").getDoubleValue();
String msg = "No errors!"; e.setAttribute("amount",new Double(2*amount).toString());
try { }
Document d = b.build(new File(args[0])); }
} catch (JDOMParseException e ) {
msg = e.getMessage();
}
System.out.println(msg);
}
}
XSLT Transformation Business Cards

<cardlist xmlns="http://businesscard.org"
public class ApplyXSLT { xmlns:xhtml="http://www.w3.org/1999/xhtml">
<title>
<xhtml:h1>My Collection of Business Cards</xhtml:h1>
try {
containing people from <xhtml:em>Widget Inc.</xhtml:em>
SAXBuilder b = new SAXBuilder(); </title>
Document d = b.build(new File(args[0])); <card>
XSLTransformer t = new XSLTransformer(args[1]); <name>John Doe</name>
Document h = t.transform(d); <title>CEO, Widget Inc.</title>
XMLOutputter outputter = new XMLOutputter(); <email>john.doe@widget.com</email>
outputter.output(h,System.out); <phone>(202) 555-1414</phone>
</card>
<card>
}
<name>Joe Smith</name>
} <title>Assistant</title>
<email>thrall@widget.com</email>
</card>
</cardlist>
4
Business Card Editor Class Representation
class Card {
public String name,title,email,phone,logo;
public Card(String name, String title, String email,

String phone, String logo) {
this.name=name;
this.title=title;
this.email=email;
this.phone=phone;
this.logo=logo;
}
}
From JDOM to Classes From Classes to JDOM (1/2)
Vector doc2vector(Document d) {
Vector v = new Vector(); Document vector2doc() {
Iterator i = d.getRootElement().getChildren().iterator(); Element cardlist = new Element("cardlist");
while (i.hasNext()) { for (int i=0; i<cardvector.size(); i++) {
Element e = (Element)i.next(); Card c = (Card)cardvector.elementAt(i);
String phone = e.getChildText("phone",b);
if (c!=null) {
if (phone==null) phone="";
Element card = new Element("card",b);
Element logo = e.getChild("logo",b);
String uri; Element name = new Element("name",b);
if (logo==null) uri=""; name.addContent(c.name); card.addContent(name);
else uri=logo.getAttributeValue("uri"); Element title = new Element("title",b);
Card c = new Card(e.getChildText("name",b), title.addContent(c.title); card.addContent(title);
e.getChildText("title",b), Element email = new Element("email",b);
e.getChildText("email",b),
email.addContent(c.email); card.addContent(email);
phone, uri);
v.add(c);
}
return v;
}
5
From Classes to JDOM (2/2) A Little Bit of Code
if (!c.phone.equals("")) { void addCards() {

Element phone = new Element("phone",b); cardpanel.removeAll();
phone.addContent(c.phone); for (int i=0; i<cardvector.size(); i++) {
card.addContent(phone);
Card c = (Card)cardvector.elementAt(i);
if (c!=null) {
}
Button b = new Button(c.name);
if (!c.logo.equals("")) {
b.setActionCommand(String.valueOf(i));
Element logo = new Element("logo",b);
b.addActionListener(this);
logo.setAttribute("uri",c.logo); cardpanel.add(b);
card.addContent(logo); }
} }
cardlist.addContent(card); this.pack();
} }
}
return new Document(cardlist);
}
The Main Application XML Data Binding
public BCedit(String cardfile) {

The methods doc2vector and vector2doc are
super("BCedit");
this.cardfile=cardfile; tedious to write
try {
cardvector = doc2vector(
new SAXBuilder().build(new File(cardfile))); XML data binding provides tools to:
} catch (Exception e) { e.printStackTrace(); } • map schemas to class declarations
// initialize the user interface
• automatically generate unmarshalling code
...
} • automatically generate marshalling code
• automatically generate validation code
6
Binding Compilers The JAXB Framework
Which schemas are supported? It supports most of XML Schema

Fixed or customizable binding? The binding is customizable (annotations)
Does roundtripping preserve information? Roundtripping is almost complete
What is the support for validation? Validation is supported during unmarshalling or
Are the generated classes implemented by some on demand
generic framework? JAXB only specifies the interfaces to the
generated classes
Business Card Schema (1/3) Business Card Schema (2/3)
<schema xmlns="http://www.w3.org/2001/XMLSchema" <complexType name="cardlist_type">

xmlns:b="http://businesscard.org"
<sequence>
targetNamespace="http://businesscard.org"
<element name="title" type="b:cardlist_title_type"/>
elementFormDefault="qualified">
<element ref="b:card" minOccurs="0" maxOccurs="unbounded"/>
</sequence>
<element name="cardlist" type="b:cardlist_type"/>
</complexType>
<element name="card" type="b:card_type"/>
<element name="name" type="string"/>
<complexType name="cardlist_title_type" mixed="true">
<element name="email" type="string"/>
<sequence>
<element name="phone" type="string"/>
<any namespace="http://www.w3.org/1999/xhtml"
<element name="logo" type="b:logo_type"/>
minOccurs="0" maxOccurs="unbounded"
processContents="lax"/>
<attribute name="uri" type="anyURI"/>
</sequence>
</complexType>
7
Business Card Schema (3/3) The org.businesscard Package
<complexType name="card_type"> The binding compiler generates :

<sequence> • Cardlist, CardlistType
<element ref="b:name"/>
<element name="title" type="string"/> • CardlistImpl, CardlistTypeImpl
<element ref="b:email"/> • ...
<element ref="b:phone" minOccurs="0"/>
<element ref="b:logo" minOccurs="0"/> • Logo, LogoType
</sequence> • LogoImpl, LogoTypeImpl
</complexType>
• ObjectFactory
<complexType name="logo_type">
<attribute ref="b:uri" use="required"/>
</complexType> The Title element is not a class, since it is
</schema>
declared as a local element.
The CardType Interface A Little Bit of Code
public interface CardType { void addCards() {

java.lang.String getEmail(); cardpanel.removeAll();
void setEmail(java.lang.String value); Iterator i = cardlist.iterator();
org.businesscard.LogoType getLogo(); int j = 0;
void setLogo(org.businesscard.LogoType value); while (i.hasNext()) {
java.lang.String getTitle(); Card c = (Card)i.next();
void setTitle(java.lang.String value); Button b = new Button(c.getName());
java.lang.String getName(); b.setActionCommand(String.valueOf(j++));
void setName(java.lang.String value); b.addActionListener(this);
java.lang.String getPhone(); cardpanel.add(b);
void setPhone(java.lang.String value); }
} this.pack();
}
8
The Main Application Streaming XML
public BCedit(String cardfile) {

JDOM and JAXB keeps the entire XML tree in
super("BCedit");
this.cardfile=cardfile; memory
try { Huge documents can only be streamed:
jc = JAXBContext.newInstance("org.businesscard");
Unmarshaller u = jc.createUnmarshaller();
• movies on the Internet
cl = (Cardlist)u.unmarshal( • Unix file commands using pipes
new FileInputStream(cardfile)
What is streaming for XML documents?
);
cardlist = cl.getCard();
// initialize the user interface
The SAX framework has the answer...
...
}
Parsing Events Tracing All Events (1/4)
View the XML document as a stream of events: public class Trace extends DefaultHandler {
• the document starts int indent = 0;
• a start tag is encountered void printIndent() {

• an end tag is encountered for (int i=0; i<indent; i++) System.out.print("-");
}
• a namespace declaration is seen
• some whitespace is seen public void startDocument() {
System.out.println("start document");
• character data is encountered }
• the document ends
public void endDocument() {
The SAX tool observes these events System.out.println("end document");
It reacts by calling corresponding methods }
specified by the programmer

9
Tracing All Events (2/4) Tracing All Events (3/4)
public void startElement(String uri, String localName,

public void ignorableWhitespace(char[] ch, int start, int length) {
String qName, Attributes atts) {
printIndent();
printIndent();
System.out.println("whitespace, length " + length);
System.out.println("start element: " + qName);
}
indent++;
}
public void processingInstruction(String target, String data) {
printIndent();
public void endElement(String uri, String localName,
System.out.println("processing instruction: " + target);
String qName) {
}
indent--;
printIndent();
public void characters(char[] ch, int start, int length){
System.out.println("end element: " + qName);
printIndent();
}
System.out.println("character data, length " + length);
}
Tracing All Events (4/4) Output for the Recipe Collection

start document
public static void main(String[] args) { start element: rcp:collection
-character data, length 3
try {
-start element: rcp:description
Trace tracer = new Trace();
--character data, length 44
XMLReader reader = XMLReaderFactory.createXMLReader(); --character data, length 3
reader.setContentHandler(tracer); -end element: rcp:description
reader.parse(args[0]); -character data, length 3
} catch (Exception e) { e.printStackTrace(); } -start element: rcp:recipe
} --character data, length 5
--start element: rcp:title
}
---character data, length 42
...
--start element: rcp:nutrition
--end element: rcp:nutrition
--character data, length 3
-end element: rcp:recipe
-character data, length 1
end element: rcp:collection
end document
10
A Simple Streaming Example (1/2) A Simple Streaming Example (2/2)
public class Height extends DefaultHandler { public static void main(String[] args) {
int h = -1; try {
int max = 0; Height handler = new Height();
XMLReader reader = XMLReaderFactory.createXMLReader();
public void startElement(String uri, String localName, reader.setContentHandler(handler);
String qName, Attributes atts) { reader.parse(args[0]);
h++; if (h > max) max = h; System.out.println(handler.max);
} } catch (Exception e) { e.printStackTrace(); }
}
public void endElement(String uri, String localName, }
String qName) {
h--;
}
public void characters(char[] ch, int start, int length){

if (h+1 > max) max = h+1;
}
Comments on The Example SAX May Emulate JDOM (1/2)
public void startElement(String uri, String localName,

This version is less intuitive (stack-like style) String qName, Attributes atts) {
if (localName.equals("card")) card = new Element("card",b);
The JDOM version: else if (localName.equals("name"))
field = new Element("name",b);
java.lang.OutOfMemoryError else if (localName.equals("title"))
on 18MB document field = new Element("title",b);
else if (localName.equals("email"))
The SAX version handles 1.2GB in 51 seconds field = new Element("email",b);
else if (localName.equals("phone"))
field = new Element("phone",b);
else if (localName.equals("logo")) {
field = new Element("logo",b);
field.setAttribute("uri",atts.getValue("","uri"));
}
}
11
SAX May Emulate JDOM (2/2) Using Contextual Information

String qName) { Check forms beyond W3C validator:
if (localName.equals("card")) contents.add(card);
else if (localName.equals("cardlist")) {
• that all form input tags are inside form tags
Element cardlist = new Element("cardlist",b); • that all form tags have distinct name attributes
cardlist.setContent(contents);
doc = new Document(cardlist);
• that form tags are not nested
} else { This requires us to keep information about the
card.addContent(field);
field = null; context of the current parsing event
}
}
public void characters(char[] ch, int start, int length) {

if (field!=null)
field.addContent(new String(ch,start,length));
}
Contextual Information in SAX (1/3) Contextual Information in SAX (2/3)

public class CheckForms extends DefaultHandler { public void startElement(String uri, String localName,
int formheight = 0; String qName, Attributes atts) {
HashSet formnames = new HashSet(); if (uri.equals("http://www.w3.org/1999/xhtml")) {
if (localName.equals("form")) {
Locator locator; if (formheight > 0) report("nested forms");
public void setDocumentLocator(Locator locator) { String name = atts.getValue("","name");
this.locator = locator; if (formnames.contains(name))
} report("duplicate form name");
else
void report(String s) { formnames.add(name);
System.out.print(locator.getLineNumber()); formheight++;
System.out.print(":"); } else
System.out.print(locator.getColumnNumber()); if (localName.equals("input") ||
System.out.println(" ---"+s); localName.equals("select") ||
} localName.equals("textarea"))
if (formheight==0) report("form field outside form");
}
}
12
Contextual Information in SAX (3/3) SAX Filters
String qName) {
if (uri.equals("http://www.w3.org/1999/xhtml")) A SAX application may be turned into a filter
if (localName.equals("form"))
formheight--;
Filters may be composed (as with pipes)
} A filter is an event handler that may pass events
public static void main(String[] args) { along in the chain
try {
CheckForms handler = new CheckForms();
reader.setContentHandler(handler);
reader.parse(args[0]);
}
}
A SAX Filter Example (1/4) A SAX Filter Example (2/4)
A filter to remove processing instructions: A filter to create unique id attributes:

class PIFilter extends XMLFilterImpl { class IDFilter extends XMLFilterImpl {
public void processingInstruction(String target, String data) int id = 0;
throws SAXException {} public void startElement(String uri, String localName,
} String qName, Attributes atts)
throws SAXException {
AttributesImpl idatts = new AttributesImpl(atts);
idatts.addAttribute("","id","id","ID",
new Integer(id++).toString());
super.startElement(uri,localName,qName,idatts);
}
}
13
A SAX Filter Example (3/4) A SAX Filter Example (4/4)
A filter to count characters: public class FilterTest {

try {
class CountFilter extends XMLFilterImpl {
FilterTest handler = new FilterTest();
public int count = 0;
public void characters(char[] ch, int start, int length)
PIFilter pi = new PIFilter();
throws SAXException {
pi.setParent(reader);
count = count+length;
IDFilter id = new IDFilter();
super.characters(ch,start,length);
id.setParent(pi);
}
CountFilter count = new CountFilter();
}
count.setParent(id);
count.parse(args[0]);
System.out.println(count.count);
}
}
Pull vs. Push Contextual Information in XMLPull (1/3)
SAX is known as a push framework

public class CheckForms2 {
static void report(XmlPullParser xpp, String s) {
• the parser has the initivative System.out.print(xpp.getLineNumber());

System.out.print(":");
• the programmer must react to events System.out.print(xpp.getColumnNumber());
System.out.println(" ---"+s);
An alternative is a pull framework }
• the programmer has the initiative public static void main (String args[])
throws XmlPullParserException, IOException {
• the parser must react to requests XmlPullParserFactory factory = XmlPullParserFactory.newInstance();
XML Pull is an example of a pull framework factory.setNamespaceAware(true);

factory.setFeature(XmlPullParser.FEATURE_PROCESS_NAMESPACES, true);
XmlPullParser xpp = factory.newPullParser();
int formheight = 0;
HashSet formnames = new HashSet();
14
Contextual Information in XMLPull (2/3) Contextual Information in XMLPull (3/3)
xpp.setInput(new FileReader(args[0])); else if (eventType==XmlPullParser.END_TAG) {
int eventType = xpp.getEventType(); if (xpp.getNamespace().equals("http://www.w3.org/1999/xhtml")
while (eventType!=XmlPullParser.END_DOCUMENT) { && xpp.getName().equals("form"))
if (eventType==XmlPullParser.START_TAG) { formheight--;
if (xpp.getNamespace().equals("http://www.w3.org/1999/xhtml") }
&& xpp.getName().equals("form")) { eventType = xpp.next();
if (formheight>0) }
report(xpp,"nested forms"); }
String name = xpp.getAttributeValue("","name"); }
if (formnames.contains(name))
report(xpp,"duplicate form name");
else
formnames.add(name);
formheight++;
} else if (xpp.getName().equals("input") ||
xpp.getName().equals("select") ||
xpp.getName().equals("textarea"))
if (formheight==0)
report(xpp,"form field outside form");
} }
Using a Pull Parser Streaming Transformations
Not that different from the push version SAX allows the programming of streaming
More direct programming style applications "by hand"
Smaller memory footprint XSLT allows high-level programming of
Pipelining with filter chains is not available applications
(but may be simulated in languages with higher- A broad spectrum of these could be streamed
order functions) But XSLT does not allow streaming...
Solution: use a domain-specific language for

streaming transformations
15
STX Similarities with XSLT
STX is a variation of XSLT suitable for streaming template text

• some features are not allowed copy element
• but every STX application can be streamed value-of attribute
if variable
The differences reflect necessary limitations in the
else param
control flow
choose with-param
when
otherwise Most XSLT functions
Differences with XSLT STXPath
apply-templates is the main problem: A subset of XPath 2.0 used by STX

• allows processing to continue anywhere in the tree
• requires moving back and forth in the input file STXPath expressions:
• or storing the whole document • look like restricted XPath 2.0 expressions
• evaluate to sequences of nodes and atomic values
mutable variables to accumulate information • but they have a different semantics
16
STXPath Syntax STXPath Semantics
Must use abbreviated XPath 2.0 syntax Evaluate the corresponding XPath 2.0 expression
The axes following and preceding are not Restrict the result to those nodes that are on the
available ancestor axis
Extra node tests: cdata() and doctype() <A>
<B/>
<C><D/></C>
</A>
Evaluate count(//B) with D as the context node
With XPath the result is 1
With STXPath the result is 0
Transformation Sheets A Simple STX Example
STX use transform instead of stylesheet Extract comments from recipes:

apply-templates is not allowed <stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns"
Processing is defined by: version="1.0"

xmlns:rcp="http://www.brics.dk/ixwt/recipes">
• process-children
<stx:template match="rcp:collection">
• process-siblings
<comments>
• process-self <stx:process-children/>
Only a single occurrence of process-children </comments>

</stx:template>
is allowed in each template (to enable streaming)
<stx:template match="rcp:comment">
<comment><stx:value-of select="."/></comment>
</stx:template>
</stx:transform>
17
SAX Version (1/2) SAX Version (2/2)
public void characters(char[] ch, int start, int length) {

public class ExtractComments extends if (chars)
DefaultHandler { System.out.print(new String(ch, start, length));
}
bool chars = true;
String qName) {
public void startElement(String uri, if (uri.equals("http://www.brics.dk/ixwt/recipes")) {
if (localName.equals("collection"))
String localName, System.out.print("</comments>");
if (localName.equals("comment")) {
String
System.out.print("</comment>");
qName, Attributes atts) { chars = false;
}
if }
(uri.equals("http://www.brics.dk/ixwt }
}
/recipes")) {
if
The Ancestor Stack Using process-

process-siblings
<stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns" <stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns"

version="1.0"> version="1.0">
<stx:template match="*"> <stx:template match="*">
<stx:message select="concat(count(//*),' ',local-name())"/> <stx:copy>
<stx:process-children/> <stx:process-children/>
</stx:template> <stx:process-siblings/>
</stx:transform> </stx:copy>
</stx:template>
</stx:transform>
<A> 1 A
<B/> 2 B
<a> <a>
<B><C/></B> 2 B
<b><c/></b> <b>
<A/>
3 C
<d><e/></d> <c/>
2 A
<B><A><C/></A></B> <d><e/></d>
2 B </a>
</A> </b>
3 A
</a>
4 C
18
Mutable Variables STX Version of CheckForms (1/2)
<stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns" <stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns"

version="1.0" version="1.0"
xmlns:rcp="http://www.brics.dk/ixwt/recipes"> xmlns:xhtml="http://www.w3.org/1999/xhtml">
<stx:variable name="depth" select="0"/> <stx:variable name="formheight" select="0"/>
<stx:variable name="maxdepth" select="0"/> <stx:variable name="formnames" select="'#'"/>
<stx:template match="rcp:collection"> <stx:template match="xhtml:form">

<stx:process-children/> <stx:if test="$formheight>0">
<maxdepth><stx:value-of select="$maxdepth"/></maxdepth> <stx:message select="'nested forms'"/>
</stx:template> </stx:if>
<stx:if test="contains($formnames,concat('#',@name,'#'))">
<stx:template match="rcp:ingredient"> <stx:message select="'duplicate form name'"/>
<stx:assign name="depth" select="$depth + 1"/> </stx:if>
<stx:if test="$depth > $maxdepth"> <stx:assign name="formheight" select="$formheight + 1"/>
<stx:assign name="maxdepth" select="$depth"/> <stx:assign name="formnames"
</stx:if> select="concat($formnames,@name,'#')"/>
<stx:process-children/> <stx:process-children/>
<stx:assign name="depth" select="$depth - 1"/> <stx:assign name="formheight" select="$formheight - 1"/>
</stx:template> </stx:template>
</stx:transform>
STX Version of CheckForms (2/2) Groups (1/2)

<stx:template match="xhtml:input|xhtml:select|xhtml:textarea"> <stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns"
<stx:if test="$formheight=0"> version="1.0"
<stx:message select="'form field outside form'"/> strip-space="yes">
</stx:if> <stx:template match="person">
<stx:process-children/> <person><stx:process-children/></person>
</stx:transform> <stx:template match="email">

<emails><stx:process-self group="foo"/></emails>
</stx:template>
<person>
<person> <emails>
<email/><email/><email/> <email/><email/><email/>
<phone/><phone/> </emails>
</person> <phone/><phone/>
</person>
19
Groups (2/2) Limitations of Streaming
Something we will never write with STX:

<stx:group name="foo">
<stx:template match="email">
<email/>
<stx:process-siblings while="email" group="foo"/>
</stx:template> <xsl:stylesheet version="2.0"
</stx:group>
xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
<stx:template match="phone">
<phone/> <xsl:template name="mirror" match="/|@*|node()">
</stx:template> <xsl:copy>
</stx:transform> <xsl:apply-templates select="@*"/>
<xsl:apply-templates select="reverse(node())"/>
<person> </xsl:copy>
<person> <emails>
<email/><email/><email/>
</xsl:template>
<email/><email/><email/>
<phone/><phone/> </emails> </xsl:stylesheet>
</person> <phone/><phone/>
</person>
STX for Recipes (1/7) STX for Recipes (2/7)

<stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns" <stx:template match="rcp:recipe">
version="1.0" <body>
xmlns:rcp="http://www.brics.dk/ixwt/recipes" <table border="1">
xmlns="http://www.w3.org/1999/xhtml" <stx:process-self group="outer"/>
strip-space="yes"> </table>
</body>
<stx:template match="rcp:collection"> </stx:template>
<html>
<stx:process-children/> <stx:group name="outer">
</html> <stx:template match="rcp:description">
</stx:template> <tr>
<td><stx:value-of select="."/></td>
<stx:template match="rcp:description"> </tr>
<head> </stx:template>
<title><stx:value-of select="."/></title>
<link href="style.css" rel="stylesheet" type="text/css"/>
</head>
</stx:template>
20
<stx:template match="rcp:recipe"> <stx:template match="rcp:ingredient" >
<tr> <ul><stx:process-self group="inner"/></ul>
<td> </stx:template>
<stx:process-children/>
</td> <stx:template match="rcp:preparation">
</tr> <ol><stx:process-children/></ol>
<stx:template match="rcp:title"> <stx:template match="rcp:step">

<h1><stx:value-of select="."/></h1> <li><stx:value-of select="."/></li>
<stx:template match="rcp:date"> <stx:template match="rcp:comment">

<i><stx:value-of select="."/></i> <ul>
</stx:template> <li type="square"><stx:value-of select="."/></li>
</ul>
</stx:template>

<stx:template match="rcp:nutrition"> <stx:group name="inner">
<table border="2"> <stx:template match="rcp:ingredient">
<tr> <stx:choose>
<th>Calories</th><th>Fat</th> <stx:when test="@amount">
<th>Carbohydrates</th><th>Protein</th> <li>
<stx:if test="@alcohol"><th>Alcohol</th></stx:if> <stx:if test="@amount!='*'">
</tr> <stx:value-of select="@amount"/>
<tr> <stx:text> </stx:text>
<td align="right"><stx:value-of select="@calories"/></td> <stx:if test="@unit">
<td align="right"><stx:value-of select="@fat"/></td> <stx:value-of select="@unit"/>
<td align="right"><stx:value-of select="@carbohydrates"/></td> <stx:if test="number(@amount)>number(1)">
<td align="right"><stx:value-of select="@protein"/></td> <stx:text>s</stx:text>
<stx:if test="@alcohol"> </stx:if>
<td align="right"><stx:value-of select="@alcohol"/></td> <stx:text> of </stx:text>
</stx:if> </stx:if>
</tr> </stx:if>
</table> <stx:text> </stx:text>
</stx:template> <stx:value-of select="@name"/>
</stx:group> </li>
</stx:when>
21
STX for Recipes (7/7) XML in Programming Languages
SAX: programmers react to parsing events

<stx:otherwise>
<li><stx:value-of select="@name"/></li>
JDOM: a general data structure for XML trees

<stx:process-children group="outer"/>
</stx:otherwise>
JAXB: a specific data structure for XML trees

</stx:choose>
<stx:process-siblings while="rcp:ingredient" group="inner"/>
</stx:template>
</stx:group>
These approaches are convenient

</stx:transform>
But no compile-time guarantees:

• about validity of the constructed XML (JDOM, JAXB)
• well-formedness of the constructed XML (SAX)
Type-
Type-Safe XML Programming Languages XDuce
With XML schemas as types A first-order functional language

Type-checking now guarantees validity XML trees are native values
Regular expression types (generalized DTDs)
An active research area
Arguments and results are explicitly typed
Type inference for pattern variables
Compile-time type checking guarantees:
• XML navigation is safe
• generated XML is valid
22
XDuce Types for Recipes (1/2) XDuce Types for Recipes (2/2)
namespace rcp = "http://www.brics.dk/ixwt/recipes" type Ingredient = rcp:ingredient[@name[String],
@amount[String]?,
type Collection = rcp:collection[Description,Recipe*] @unit[String]?,
type Description = rcp:description[String] (Ingredient*,Preparation)?]
type Recipe = rcp:recipe[@id[String]?, type Preparation = rcp:preparation[Step*]
Title, type Step = rcp:step[String]
Date, type Comment = rcp:comment[String]
Ingredient*, type Nutrition = rcp:nutrition[@calories[String],
Preparation, @carbohydrates[String],
Comment?, @fat[String],
Nutrition, @protein[String],
Related*] @alcohol[String]?]
type Title = rcp:title[String] type Related = rcp:related[@ref[String],String]
type Date = rcp:date[String]
XDuce Types of Nutrition Tables From Recipes to Tables (1/3)
type NutritionTable = nutrition[Dish*] fun extractCollection(val c as Collection) : NutritionTable =

match c with
type Dish = dish[@name[String],
rcp:collection[Description, val rs]
@calories[String], -> nutrition[extractRecipes(rs)]
@fat[String],
@carbohydrates[String], fun extractRecipes(val rs as Recipe*) : Dish* =
match rs with
@protein[String], rcp:recipe[@..,
@alcohol[String]] rcp:title[val t],
Date,
Ingredient*,
Preparation,
Comment?,
val n as Nutrition,
Related*], val rest
-> extractNutrition(t,n), extractRecipes(rest)
| () -> ()
23
From Recipes to Tables (2/3) From Recipes to Tables (3/3)
fun extractNutrition(val t as String, val n as Nutrition) : Dish = | rcp:nutrition[@calories[val calories],
match n with @carbohydrates[val carbohydrates],
rcp:nutrition[@calories[val calories], @fat[val fat],
@carbohydrates[val carbohydrates], @protein[val protein]]
@fat[val fat], -> dish[@name[t],
@protein[val protein], @calories[calories],
@alcohol[val alcohol]] @carbohydrates[carbohydrates],
-> dish[@name[t], @fat[fat],
@calories[calories], @protein[protein],
@carbohydrates[carbohydrates], @alcohol["0%"]]
@fat[fat],
@protein[protein],
@alcohol[alcohol]] let val collection = validate load_xml("recipes.xml") with Collection
let val _ = print(extractCollection(collection))
XDuce Guarantees XACT
The XDuce type checker determines that: A Java framework (like JDOM) but:
• every function returns a valid value • it is based on immutable templates, which are
• every function argument is a valid value sequences of XML trees containing named gaps
• every match has an exhaustive collection of patterns • XML trees are constructed by plugging gaps
• every pattern matches some value • it has syntactic sugar for template constants
Clearly, this will eliminate many potential errors • XML is navigated using XPath
• an analyzer can a compile-time guarantee that an XML
expression is valid according to a given DTD
24
Business Cards to Phone Lists (1/2) Business Cards to Phone Lists (2/2)
import dk.brics.xact.*; XML cardlist = XML.get("file:cards.xml",
import java.io.*; "file:businesscards.dtd",
"http://businesscard.org");
public class PhoneList { XML x = wrapper.plug("TITLE", "My Phone List")
public static void main(String[] args) throws XactException { .plug("MAIN", [[<h:ul><[CARDS]></h:ul>]]);
String[] map = {"c", "http://businesscard.org",
"h", "http://www.w3.org/1999/xhtml"};
XML.setNamespaceMap(map); XMLIterator i = cardlist.select("//c:card[c:phone]").iterator();
while (i.hasNext()) {
XML wrapper = [[<h:html> XML card = i.next();
<h:head> x = x.plug("CARDS",
<h:title><[TITLE]></h:title> [[<h:li>
</h:head> <h:b><{card.select("c:name/text()")}></h:b>,
<h:body> phone: <{card.select("c:phone/text()")}>
<h:h1><[TITLE]></h:h1> </h:li>
<[MAIN]> <[CARDS]>]]);
</h:body> }
</h:html>]]; System.out.println(x);
}
}
XML API A Highly Structured Recipe
constant(s) build a template constant from s <rcp:recipe id="117">

<rcp:title>Fried Eggs with Bacon</rcp:title>
x.plug(g,y) plugs the gap g with y <rcp:date>Fri, 10 Nov 2004</rcp:date>

<rcp:ingredient name="fried eggs">
x.select(p) returns a template containing the <rcp:ingredient name="egg" amount="2"/>

<rcp:preparation>
sequence targets of the XPath expression p <rcp:step>Break the eggs into a bowl.</rcp:step>
<rcp:step>Fry until ready.</rcp:step>
x.gapify(p,g) replaces the targets of p with </rcp:preparation>
</rcp:ingredient>
gaps named g <rcp:ingredient name="bacon" amount="3" unit="strip"/>
<rcp:preparation>
get(u,d,n) parses a template from a URL with <rcp:step>Fry the bacon until crispy.</rcp:step>
a DTD and a namespace

<rcp:step>Serve with the eggs.</rcp:step>
</rcp:preparation>
x.analyze(d,n) guarantees at compile-time <rcp:nutrition calories="517"

fat="64%" carbohydrates="0%" protein="0%"/>
that x is valid given a DTD and a namespace </rcp:recipe>
25
A Flattened Recipe A Recipe Flattener in XACT (1/2)
<rcp:recipe id="117"> public class Flatten {
<rcp:title>Fried Eggs with Bacon</rcp:title> static final String rcp = "http://www.brics.dk/ixwt/recipes";
<rcp:date>Fri, 10 Nov 2004</rcp:date> static final String[] map = { "rcp", rcp };
<rcp:ingredient name="egg" amount="2"/>
<rcp:ingredient name="bacon" amount="3" unit="strip"/> static { XML.setNamespaceMap(map); }
<rcp:preparation>
<rcp:step>Break the eggs into a bowl.</rcp:step> public static void main(String[] args) throws XactException {
<rcp:step>Fry until ready.</rcp:step> XML collection = XML.get("file:recipes.xml",
<rcp:step>Fry the bacon until crispy.</rcp:step> "file:recipes.dtd", rcp);
<rcp:step>Serve with the eggs.</rcp:step> XML recipes = collection.select("//rcp:recipe");
</rcp:preparation> XML result = [[<rcp:collection>
<rcp:nutrition calories="517" <{collection.select("rcp:description")}>
fat="64%" carbohydrates="0%" protein="36%"/> <[MORE]>
</rcp:recipe> </rcp:collection>]];
A Recipe Flattener in XACT (2/2) An Error

XMLIterator i = recipes.iterator(); <rcp:ingredient>
while (i.hasNext()) { <{r.select("rcp:title|rcp:date")}>
XML r = i.next();
<{r.select("//rcp:ingredient[@amount]")}>
result = result.plug("MORE",
<rcp:preparation>
[[<rcp:recipe>
<{r.select("rcp:title|rcp:date")}> <{r.select("//rcp:step")}>
<{r.select("//rcp:ingredient[@amount]")}> </rcp:preparation>
<rcp:preparation> <{r.select("rcp:comment|rcp:nutrition|rcp:related")}>
<{r.select("//rcp:step")}> </rcp:ingredient>
</rcp:preparation>
<{r.select("rcp:comment|rcp:nutrition|rcp:related")}>
</rcp:recipe>
<[MORE]>]]);
}
result.analyze("file:recipes.dtd", rcp);
System.out.println(result);
}
}
26
Caught at Compile-
Compile-Time Essential Online Resources
*** Invalid XML at line 31
sub-element 'rcp:ingredient' of element 'rcp:collection' not declared
required attribute 'name' missing in element 'rcp:ingredient'
http://www.jdom.org/
sub-element 'rcp:title' of element 'rcp:ingredient' not declared
sub-element 'rcp:related' of element 'rcp:ingredient' not declared
http://java.sun.com/xml/jaxp/
sub-element 'rcp:nutrition' of element 'rcp:ingredient' not declared
sub-element 'rcp:date' of element 'rcp:ingredient' not declared
http://java.sun.com/xml/jaxb/
http://www.saxproject.org/
27

Programming

Загружено:

Сведения о документе

Исходное описание:

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Programming

Загружено:

Авторское право:

Доступные форматы

Objectives

 How XML may be manipulated from general-

Anders Møller & Michael I. Schwartzbach

General Purpose XML Programming The JDOM Framework

 Needed for:  An implementation of generic XML trees in Java

 DOM is a language-independent alternative

 The abstract class Content has subclasses: int xmlHeight(Element e) {

Element return max+1;

Another Example A Final Example (1/3)

static void doubleSugar(Document d)

void makeButter(Element e) throws DataConversionException { Element cream = new Element("ingredient",rcp);

Parsing and Serializing Validation (DTD)

public class ValidateDTD {

XSLT Transformation Business Cards

public Card(String name, String title, String email,

From JDOM to Classes From Classes to JDOM (1/2)

if (!c.phone.equals("")) { void addCards() {

The Main Application XML Data Binding

public BCedit(String cardfile) {

 Which schemas are supported?  It supports most of XML Schema

Business Card Schema (1/3) Business Card Schema (2/3)

<schema xmlns="http://www.w3.org/2001/XMLSchema" <complexType name="cardlist_type">

<complexType name="card_type">  The binding compiler generates :

The CardType Interface A Little Bit of Code

public interface CardType { void addCards() {

public BCedit(String cardfile) {

Parsing Events Tracing All Events (1/4)

• a start tag is encountered void printIndent() {

 It reacts by calling corresponding methods }

specified by the programmer

public void startElement(String uri, String localName,

Tracing All Events (4/4) Output for the Recipe Collection

public void characters(char[] ch, int start, int length){

Comments on The Example SAX May Emulate JDOM (1/2)

public void startElement(String uri, String localName,

public void endElement(String uri, String localName,

public void characters(char[] ch, int start, int length) {

Contextual Information in SAX (1/3) Contextual Information in SAX (2/3)

A SAX Filter Example (1/4) A SAX Filter Example (2/4)

 A filter to remove processing instructions:  A filter to create unique id attributes:

 A filter to count characters: public class FilterTest {

Pull vs. Push Contextual Information in XMLPull (1/3)

 SAX is known as a push framework

• the parser has the initivative System.out.print(xpp.getLineNumber());

 XML Pull is an example of a pull framework factory.setNamespaceAware(true);

XmlPullParser xpp = factory.newPullParser();

Using a Pull Parser Streaming Transformations

 Solution: use a domain-specific language for

 STX is a variation of XSLT suitable for streaming  template  text

Differences with XSLT STXPath

 apply-templates is the main problem:  A subset of XPath 2.0 used by STX

Transformation Sheets A Simple STX Example

 STX use transform instead of stylesheet  Extract comments from recipes:

 Processing is defined by: version="1.0"

 Only a single occurrence of process-children </comments>

public void characters(char[] ch, int start, int length) {

The Ancestor Stack Using process-

<stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns" <stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns"

<stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns" <stx:transform xmlns:stx="http://stx.sourceforge.net/2002/ns"

<stx:template match="rcp:collection"> <stx:template match="xhtml:form">

STX Version of CheckForms (2/2) Groups (1/2)

</stx:transform> <stx:template match="email">

 Something we will never write with STX:

How XML may be manipulated from general-

Needed for: An implementation of generic XML trees in Java

DOM is a language-independent alternative

The abstract class Content has subclasses: int xmlHeight(Element e) {

Which schemas are supported? It supports most of XML Schema

<complexType name="card_type"> The binding compiler generates :

It reacts by calling corresponding methods }

A filter to remove processing instructions: A filter to create unique id attributes:

A filter to count characters: public class FilterTest {

SAX is known as a push framework

XML Pull is an example of a pull framework factory.setNamespaceAware(true);

Solution: use a domain-specific language for

STX is a variation of XSLT suitable for streaming template text

apply-templates is the main problem: A subset of XPath 2.0 used by STX

STX use transform instead of stylesheet Extract comments from recipes:

Processing is defined by: version="1.0"

Only a single occurrence of process-children </comments>

Something we will never write with STX:

SAX: programmers react to parsing events

JDOM: a general data structure for XML trees

JAXB: a specific data structure for XML trees

These approaches are convenient

But no compile-time guarantees:

With XML schemas as types A first-order functional language

constant(s) build a template constant from s <rcp:recipe id="117">

x.plug(g,y) plugs the gap g with y <rcp:date>Fri, 10 Nov 2004</rcp:date>

x.select(p) returns a template containing the <rcp:ingredient name="egg" amount="2"/>

x.analyze(d,n) guarantees at compile-time <rcp:nutrition calories="517"