解析一個(gè)XML文件有很多方法,最常用的就是Dom4j,和JDOM.這個(gè)我們就是講使用Xalan來(lái)解析一個(gè)XML文件.
先有個(gè)xml文件如下:
<?xml version="1.0" encoding="UTF-8"?>
<a:main xmlns="test0" xmlns:a="testt" xmlns:b="test2">
<a:node>
<a:nodeA1>A1</a:nodeA1>
<a:nodeA2>
<b:nodeB>B1</b:nodeB>
<b:nodeB>B2</b:nodeB>
</a:nodeA2>
</a:node>
<node>
<nodeD/>
</node>
</a:main>
這個(gè)一個(gè)簡(jiǎn)單的xml文件.但是我們現(xiàn)在要用xpath來(lái)查找
"/a:main/node"
一般你是一個(gè)很復(fù)雜的表達(dá)式,這個(gè)xpath是經(jīng)過(guò)簡(jiǎn)化的.簡(jiǎn)化的代碼你可以使用如下:
1 private static String simplifyXPathExpression(String xpathExpression) {
2
3 Perl5Matcher matcher = new Perl5Matcher();
4
5 Perl5Compiler compiler = new Perl5Compiler();
6
7 Pattern pattern = null;
8 try {
9 pattern = compiler.compile("(.*)/\\s*\\w+\\s*(/(\\.\\.|parent))(.*)");
10 } catch (MalformedPatternException e) {
11 ExceptionHandler.process(e);
12 }
13
14 Perl5Substitution substitution = new Perl5Substitution("$1$4", Perl5Substitution.INTERPOLATE_ALL);
15
16 int lengthOfPreviousXPath = 0;
17
18 do {
19 lengthOfPreviousXPath = xpathExpression.length();
20 if (matcher.matches(xpathExpression, pattern)) {
21 xpathExpression = Util.substitute(matcher, pattern, substitution, xpathExpression, Util.SUBSTITUTE_ALL);
22 }
23 } while (xpathExpression.length() != lengthOfPreviousXPath);
24
25 return xpathExpression;
26 }
這里使用的org.apache.oro.text.regex包里面的類(lèi).
現(xiàn)在就開(kāi)始解析一個(gè)XML文件:在Dengues 項(xiàng)目中的org.dengues.commons.jdk.xpath.ComplexXPathUsedXalan類(lèi).
1 public ComplexXPathUsedXalan(String xmlFilename) {
2 xmlInput = new File(xmlFilename);
3 if (!xmlInput.exists()) {
4 throw new RuntimeException("Specified file does not exist!");
5 }
6 try {
7 DocumentBuilderFactory docFactory = DocumentBuilderFactory.newInstance();
8 docFactory.setNamespaceAware(true);
9 DocumentBuilder builder = docFactory.newDocumentBuilder();
10 document = builder.parse(xmlInput);
11 initLastNodes(document.getDocumentElement());
12 prefixToNamespace.put(XMLConstants.XML_NS_PREFIX, XMLConstants.XML_NS_URI);
13 } catch (ParserConfigurationException e) {
14 e.printStackTrace();
15 } catch (IOException e) {
16 e.printStackTrace();
17 } catch (SAXException exception) {
18 exception.printStackTrace();
19 }
20 XPathFactory factory = XPathFactory.newInstance();
21 xPath = factory.newXPath();
22 xPath.setNamespaceContext(getNamespaceContext());
23 }
這是類(lèi)的構(gòu)造函數(shù).這里要注意的幾個(gè)地方就是:1.第8行設(shè)置這個(gè)DocFactory支持Namespace.這里一定要設(shè)置否則你將不能找所需要的節(jié)點(diǎn).
2.在第12行,此前我新建了一個(gè)Map用來(lái)存儲(chǔ)Prefix和Namespace的對(duì)應(yīng)關(guān)系:
1 private final Map<String, String> prefixToNamespace = new HashMap<String, String>();
在12行就是講XML對(duì)應(yīng)Namespace設(shè)置進(jìn)去.
3.第11行,它的作用的就是取出所有的Prefix和Namespace.代碼如下:
1 private void initLastNodes(Node node) {
2 NodeList childNodes = node.getChildNodes();
3 int length = childNodes.getLength();
4 int type = node.getNodeType();
5 if (type == Node.ELEMENT_NODE) {
6 setPrefixToNamespace(node);
7 }
8 for (int i = 0; i < length; i++) {
9 Node item = childNodes.item(i);
10 if (item.getChildNodes().getLength() > 0) {
11 initLastNodes(item);
12 }
13 }
14 }
15
16 /**
17 * DOC qzhang Comment method "setPrefixToNamespace".
18 *
19 * @param node
20 */
21 private void setPrefixToNamespace(Node node) {
22 NamedNodeMap nnm = node.getAttributes();
23 for (int i = 0; i < nnm.getLength(); i++) {
24 Node attr = nnm.item(i);
25 String aname = attr.getNodeName();
26 boolean isPrefix = aname.startsWith(XMLConstants.XMLNS_ATTRIBUTE + ":");
27 if (isPrefix || aname.equals(XMLConstants.XMLNS_ATTRIBUTE)) {
28 int index = aname.indexOf(':');
29 String p = isPrefix ? aname.substring(index + 1) : XMLConstants.NULL_NS_URI;
30 prefixToNamespace.put(p, attr.getNodeValue());
31 }
32 }
33 }
這樣的就可以去到當(dāng)前XML的所有Prefix的所有節(jié)點(diǎn).
4.其實(shí)去Prefix和Namepace還由另外的API:com.sun.org.apache.xml.interal.utils.PrefixResloverDefault,你可以使用它:
1 PrefixResolverDefault resolverDefault = new PrefixResolverDefault(document.getDocumentElement());
2 String namespace = resolverDefault.getNamespaceForPrefix(prefix);
你可以使用這種方法.但是我覺(jué)得這種方法在大部分情況下解決不了問(wèn)題.原因:(1)當(dāng)前你只有document的根結(jié)點(diǎn).也就是說(shuō)它只能得到根結(jié)點(diǎn)定義的Namespace.(2)當(dāng)根結(jié)點(diǎn)也使用了Namespace的話(huà)(a:main),解析就有問(wèn)題.
5.構(gòu)造函數(shù)的22行設(shè)置一個(gè)NamespaceContext,代碼如下:
1 private NamespaceContext getNamespaceContext() {
2 return new NamespaceContext() {
3
4 public String getNamespaceURI(String prefix) {
5 String namespaceForPrefix = getNamespaceForPrefix(prefix);
6 return namespaceForPrefix;
7
8 }
9
10 public java.util.Iterator getPrefixes(String val) {
11 return null;
12 }
13
14 public String getPrefix(String uri) {
15 return null;
16 }
17 };
18 }
19
20 /**
21 * DOC qzhang Comment method "getNamespaceForPrefix".
22 *
23 * @param prefix
24 * @return
25 */
26 protected String getNamespaceForPrefix(String prefix) {
27 String namespace = prefixToNamespace.get(prefix);
28 if (namespace != null) {
29 return namespace;
30 }
31 return getDefaultNamespace();
32 }
33
34 private String getDefaultNamespace() {
35 Node parent = document.getDocumentElement();
36 int type = parent.getNodeType();
37 if (type == Node.ELEMENT_NODE) {
38 NamedNodeMap nnm = parent.getAttributes();
39 for (int i = 0; i < nnm.getLength(); i++) {
40 Node attr = nnm.item(i);
41 String aname = attr.getNodeName();
42 if (aname.equals(XMLConstants.XMLNS_ATTRIBUTE)) {
43 return attr.getNodeValue();
44 }
45 }
46 }
47 return XMLConstants.NULL_NS_URI;
48 }
這個(gè)總要就是取Namespace值,但是當(dāng)你輸入Prefix不在Map里面的時(shí)候,它將取默認(rèn)的Namespace,也就是在getDefaultNamespace().
現(xiàn)在就是最后要提供一個(gè)API:
1 public Object parseXPath(String expression, QName name) {
2 try {
3 expression = addDefaultPrefix(expression);
4 XPathExpression xexpr = xPath.compile(expression);
5 Object cd = xexpr.evaluate(getDocumnent(), name);
6 return cd;
7 } catch (XPathExpressionException e) {
8 e.printStackTrace();
9 }
10 return null;
11 }
12
13 private String addDefaultPrefix(String xPathExpression) {
14 if (XMLConstants.NULL_NS_URI.equals(getDefaultNamespace())) {
15 return xPathExpression;
16 } else {
17 StringBuilder expr = new StringBuilder();
18 String[] split = xPathExpression.split("/");
19 for (String string : split) {
20 if (!string.equals("") && string.indexOf(':') == -1 && string.indexOf('.') == -1) {
21 expr.append(XMLConstants.DEFAULT_NS_PREFIX + ":");
22 }
23 expr.append(string + "/");
24 }
25 if (split.length > 0) {
26 expr.deleteCharAt(expr.length() - 1);
27 }
28 return expr.toString();
29 }
30 }
這個(gè)就是提供的解析xpath表達(dá)是的接口.大家會(huì)看到了要執(zhí)行一個(gè)添加addDefaultPrefix()的函數(shù),它的作用就是在
當(dāng)存在默認(rèn)Namespace的時(shí)候,添加一個(gè)默認(rèn)的Prefix.這樣的問(wèn)題解決了.
再看看測(cè)試的代碼:
1 private static void testXPath() {
2 ComplexXPathUsedXalan complexXPath = new ComplexXPathUsedXalan("c:/SimpleNamespace.xml");
3 String expression = "/a:main/node";
4 try {
5 NodeList nodes = (NodeList) complexXPath.parseXPath(expression, XPathConstants.NODESET);
6 System.out.println("length: " + nodes.getLength());
7 for (int i = 0; i < nodes.getLength(); i++) {
8 Node item = nodes.item(i);
9 System.out.println("Name:" + item.getNodeName() + " value:" + item.getNodeValue());
10 }
11 } catch (Exception e) {
12 e.printStackTrace();
13 }
14 }
這樣就可以得到打印結(jié)果:
1 length: 1
2 Name:node value:null
就是這樣了.歡迎大家指正.
Dengues論壇(http://groups.google.com/group/dengues/),一個(gè)很好的Eclipse開(kāi)發(fā)者樂(lè)園.