@Test
public void testHtmlWithTags() throws Exception {
final String htmlText =
"<html><head><title>Title</title></head>" + "<body><p>this is a test</p></body></html>";
// Create FetchedDatum using data
String url = "http://domain.com/page.html";
String contentType = "text/html; charset=utf-8";
HttpHeaders headers = new HttpHeaders();
headers.add(HttpHeaderNames.CONTENT_TYPE, contentType);
ContentBytes content = new ContentBytes(htmlText.getBytes("utf-8"));
FetchedDatum fetchedDatum =
new FetchedDatum(url, url, System.currentTimeMillis(), headers, content, contentType, 0);
// Call parser.parse
SimpleParser parser = new SimpleParser(new ParserPolicy(), true);
ParsedDatum parsedDatum = parser.parse(fetchedDatum);
// Now take the resulting HTML, process it using Dom4J
SAXReader reader = new SAXReader(new Parser());
reader.setEncoding("UTF-8");
String htmlWithMarkup = parsedDatum.getParsedText();
Document doc = reader.read(new StringInputStream(htmlWithMarkup));
// We have to do helicopter stunts since HTML has a global namespace on it, set
// at the <html> element level.
XPath xpath = DocumentHelper.createXPath("/xhtml:html/xhtml:body/xhtml:p");
Map<String, String> namespaceUris = new HashMap<String, String>();
namespaceUris.put("xhtml", "http://www.w3.org/1999/xhtml");
xpath.setNamespaceURIs(namespaceUris);
Node paragraphNode = xpath.selectSingleNode(doc);
Assert.assertNotNull(paragraphNode);
Assert.assertEquals("this is a test", paragraphNode.getText());
}
/**
* Finds the first occurrence of an element in a document.
*
* @param document the document to be searched for the element. Must not be null.
* @param childName the name of the element to be searched for. Must not be null.
* @return the found element or null if it could not be found
* @throws NullPointerException thrown if any of the arguments is null
*/
public static Element findElement(Document document, String childName) {
XPath xpath = document.createXPath("//" + childName + "[1]");
Object result = xpath.evaluate(document.getRootElement());
if (result instanceof List) {
return null;
}
return (Element) result;
}
private List<String> findValues(String xpathString, Node node) {
XPath xpath = getXpath(xpathString);
xpath.setNamespaceURIs(prefixNamespaceMap);
List<? extends Node> nodeList = xpath.selectNodes(node);
List<String> result = new ArrayList<String>(nodeList.size());
for (Node foundNode : nodeList) {
result.add(foundNode.getText());
}
return result;
}
private String findValue(String xpathString, Node node) {
XPath xpath = getXpath(xpathString);
xpath.setNamespaceURIs(prefixNamespaceMap);
Node resultNode = xpath.selectSingleNode(node);
if (resultNode != null) {
return resultNode.getText();
} else {
return null;
}
}
public StrategyQueryBuilder(String topicsFile, boolean timesKey)
throws MalformedURLException, DocumentException {
queryProcessors = new ArrayList<QueryProcessor>();
topics = Dom4jUtil.parse(new File(topicsFile));
XPath xPath = topics.createXPath("//topic");
List<Element> topicsElems = xPath.selectNodes(topics);
for (Element topic : topicsElems) {
QueryParser queryParser = new QueryParser(topic);
QueryProcessor queryProcessor = new QueryProcessor(queryParser.getQuery(), timesKey);
queryProcessors.add(queryProcessor);
}
}
public String getNamespaceNodeText(
Document document, String namespaceKey, String namespaceValue, String xPath) {
HashMap<String, String> xmlMap = new HashMap<String, String>();
// xmlMap.put("tns","http://www.99bill.com/schema/fo/settlement");
xmlMap.put(namespaceKey, namespaceValue);
// XPath xpath=document.createXPath("//tns:status"); //要获取哪个节点，改这里就可以了
XPath xpath = document.createXPath(xPath); // 要获取哪个节点，改这里就可以了
xpath.setNamespaceURIs(xmlMap);
Element element = (Element) xpath.selectSingleNode(document);
return element.getText();
}
public static Document unwrapMessage(Document soapEnvelope) throws InvalidInputFormatException {
if (_log.isDebugEnabled()) _log.debug(soapEnvelope.asXML());
Document result = DocumentHelper.createDocument();
Node node = null;
synchronized (path) {
node = path.selectSingleNode(soapEnvelope);
}
result.add((Node) node.clone());
return result;
}
private void readSignals(Element root, String antFile) throws DocumentException, IOException {
SAXReader xmlReader = new SAXReader();
Document antDoc = xmlReader.read(new File(antFile));
XPath xpath = DocumentHelper.createXPath("//hlm:signalListenerConfig");
xpath.setNamespaceURIs(map);
List signalNodes = xpath.selectNodes(antDoc);
for (Iterator iterator = signalNodes.iterator(); iterator.hasNext(); ) {
signaldoc = antDoc;
Element propertyNode = (Element) iterator.next();
String signalid = propertyNode.attributeValue("id");
String signaltarget = propertyNode.attributeValue("target");
List existinglist = globalSignalList.get(signaltarget);
String failbuild = signalType(signalid, signaldoc);
if (existinglist == null) existinglist = new ArrayList<String>();
existinglist.add(signalid + "," + failbuild);
globalSignalList.put(signaltarget, existinglist);
}
}
public List<Publication> toObject(final Document doc) throws XOMException {
final List<Publication> l = new ArrayList<Publication>();
List nodes = xpathSelector.selectNodes(doc);
for (final Iterator pubs = nodes.iterator(); pubs.hasNext(); ) {
final Element p = (Element) pubs.next();
final Publication publication = new Publication();
publication.setName(p.elementText("name"));
publication.setDescription(p.elementText("description"));
publication.setPubroot(p.elementText("pubroot"));
publication.setPreview(p.elementText("cs_preview"));
publication.setPrefix(p.elementText("cs_prefix"));
publication.setPreviewAsset(p.elementText("cs_previewasset"));
final String v = p.elementText("id");
publication.setId(Long.parseLong(v));
l.add(publication);
}
return l;
}
private String signalType(String signalid, Document antDoc) {
XPath xpath2 =
DocumentHelper.createXPath(
"//hlm:signalListenerConfig[@id='" + signalid + "']/signalNotifierInput/signalInput");
xpath2.setNamespaceURIs(map);
List signalNodes3 = xpath2.selectNodes(antDoc);
for (Iterator iterator3 = signalNodes3.iterator(); iterator3.hasNext(); ) {
Element propertyNode3 = (Element) iterator3.next();
String signalinputid = propertyNode3.attributeValue("refid");
XPath xpath3 = DocumentHelper.createXPath("//hlm:signalInput[@id='" + signalinputid + "']");
xpath3.setNamespaceURIs(map);
List signalNodes4 = xpath3.selectNodes(antDoc);
for (Iterator iterator4 = signalNodes4.iterator(); iterator4.hasNext(); ) {
Element propertyNode4 = (Element) iterator4.next();
return propertyNode4.attributeValue("failbuild");
}
}
return null;
}
private boolean accept(Object obj) {
if (obj == null) {
logger.warn("Applying JXPathFilter to null object.");
return false;
}
if (pattern == null) {
logger.warn("Expression for JXPathFilter is not set.");
return false;
}
if (expectedValue == null) {
// Handle the special case where the expected value really is null.
if (pattern.endsWith("= null") || pattern.endsWith("=null")) {
expectedValue = "null";
pattern = pattern.substring(0, pattern.lastIndexOf("="));
} else {
if (logger.isInfoEnabled()) {
logger.info("Expected value for JXPathFilter is not set, using 'true' by default");
}
expectedValue = Boolean.TRUE.toString();
}
}
Object xpathResult = null;
boolean accept = false;
Document dom4jDoc;
try {
dom4jDoc = XMLUtils.toDocument(obj, muleContext);
} catch (Exception e) {
logger.warn(
"JxPath filter rejected message because of an error while parsing XML: " + e.getMessage(),
e);
return false;
}
// Payload is XML
if (dom4jDoc != null) {
if (namespaces == null) {
// no namespace defined, let's perform a direct evaluation
xpathResult = dom4jDoc.valueOf(pattern);
} else {
// create an xpath expression with namespaces and evaluate it
XPath xpath = DocumentHelper.createXPath(pattern);
xpath.setNamespaceURIs(namespaces);
xpathResult = xpath.valueOf(dom4jDoc);
}
}
// Payload is a Java object
else {
if (logger.isDebugEnabled()) {
logger.debug("Passing object of type " + obj.getClass().getName() + " to JXPathContext");
}
JXPathContext context = JXPathContext.newContext(obj);
initialise(context);
xpathResult = context.getValue(pattern);
}
if (logger.isDebugEnabled()) {
logger.debug(
"JXPathFilter Expression result = '"
+ xpathResult
+ "' - Expected value = '"
+ expectedValue
+ "'");
}
// Compare the XPath result with the expected result.
if (xpathResult != null) {
accept = xpathResult.toString().equals(expectedValue);
} else {
// A null result was actually expected.
if (expectedValue.equals("null")) {
accept = true;
}
// A null result was not expected, something probably went wrong.
else {
logger.warn("JXPathFilter expression evaluates to null: " + pattern);
}
}
if (logger.isDebugEnabled()) {
logger.debug("JXPathFilter accept object : " + accept);
}
return accept;
}
static {
path.setNamespaceURIs(MessageConstants._nsMap);
}
public IdMap handle(Element element) throws IOException {
XPath idXPath = element.createXPath(idXpath);
if (namespaces != null) idXPath.setNamespaceURIs(namespaces);
Node id = idXPath.selectSingleNode(element);
if (id == null && idXpathSecondary != null) {
idXPath = element.createXPath(idXpathSecondary);
if (namespaces != null) idXPath.setNamespaceURIs(namespaces);
id = idXPath.selectSingleNode(element);
} else if (id == null && idXpathSecondary == null) {
logger.error("Doc without id field: " + element);
return null;
}
Map<String, String> textFields = new HashMap<String, String>();
Map<String, String> storedFields = new HashMap<String, String>();
Map<String, Field> uniqueFields = new HashMap<String, Field>();
Collection<IdMap.TextField> isolatedFields = new HashSet<IdMap.TextField>();
for (XmlFieldHandler handler : fieldHandlers) {
XPath xPath = element.createXPath(handler.getFieldXpath());
if (namespaces != null) xPath.setNamespaceURIs(namespaces);
List<Node> nodes = xPath.selectNodes(element);
for (Node node : nodes) {
FilteredFields fields = handler.getFields(node);
// TextFields
if (fields != null && fields.getTextFields() != null) {
for (Map.Entry<String, String> entry : fields.getTextFields().entrySet()) {
String old = textFields.get(entry.getKey());
String newOne = entry.getValue();
if (old == null) textFields.put(entry.getKey(), newOne);
else textFields.put(entry.getKey(), old + ' ' + newOne);
}
}
// StoredFields
if (fields != null && fields.getStoredTextFields() != null) {
for (Map.Entry<String, String> entry : fields.getStoredTextFields().entrySet()) {
String old = storedFields.get(entry.getKey());
if (old == null) storedFields.put(entry.getKey(), entry.getValue());
else storedFields.put(entry.getKey(), old + ' ' + entry.getValue());
}
}
// IsolatedFields
if (fields != null && fields.getIsolatedTextFields() != null) {
for (FilteredFields.TextField t : fields.getIsolatedTextFields()) {
isolatedFields.add(new IdMap.TextField(t.name, t.value));
}
}
// PreparedFields
if (fields != null && fields.getPreparedFields() != null) {
for (Field f : fields.getPreparedFields()) {
Field old = uniqueFields.get(f.name());
if (old == null) uniqueFields.put(f.name(), f);
else logger.error("Not UNIQUE Prepared field " + f.name() + " in doc: " + id.getText());
}
}
}
}
if (id instanceof Element)
return new IdMap(
((Element) id).getTextTrim(),
textFields,
storedFields,
new ArrayList<Field>(uniqueFields.values()),
isolatedFields);
else
return new IdMap(
id.getText().trim(),
textFields,
storedFields,
new ArrayList<Field>(uniqueFields.values()),
isolatedFields);
}
private boolean accept(Object obj) {
if (obj == null) {
logger.warn("Applying JXPathFilter to null object.");
return false;
}
if (pattern == null) {
logger.warn("Expression for JXPathFilter is not set.");
return false;
}
if (expectedValue == null) {
// Handle the special case where the expected value really is null.
if (pattern.endsWith("= null") || pattern.endsWith("=null")) {
expectedValue = "null";
pattern = pattern.substring(0, pattern.lastIndexOf("="));
} else {
if (logger.isInfoEnabled()) {
logger.info("Expected value for JXPathFilter is not set, using 'true' by default");
}
expectedValue = Boolean.TRUE.toString();
}
}
Object xpathResult = null;
boolean accept = false;
// Payload is a DOM Document
if (obj instanceof Document) {
if (namespaces == null) {
// no namespace defined, let's perform a direct evaluation
xpathResult = ((Document) obj).valueOf(pattern);
} else {
// create an xpath expression with namespaces and evaluate it
XPath xpath = DocumentHelper.createXPath(pattern);
xpath.setNamespaceURIs(namespaces);
xpathResult = xpath.valueOf(obj);
}
}
// Payload is a String of XML
else if (obj instanceof String) {
try {
return accept(DocumentHelper.parseText((String) obj));
} catch (DocumentException e) {
logger.warn("JXPathFilter unable to parse XML document: " + e.getMessage(), e);
if (logger.isDebugEnabled())
logger.debug("XML = " + StringMessageUtils.truncate((String) obj, 200, false));
return false;
}
}
// Payload is a Java object
else {
if (logger.isDebugEnabled()) {
logger.debug("Passing object of type " + obj.getClass().getName() + " to JXPathContext");
}
JXPathContext context = JXPathContext.newContext(obj);
initialise(context);
xpathResult = context.getValue(pattern);
}
if (logger.isDebugEnabled()) {
logger.debug(
"JXPathFilter Expression result = '"
+ xpathResult
+ "' - Expected value = '"
+ expectedValue
+ "'");
}
// Compare the XPath result with the expected result.
if (xpathResult != null) {
accept = xpathResult.toString().equals(expectedValue);
} else {
// A null result was actually expected.
if (expectedValue.equals("null")) {
accept = true;
}
// A null result was not expected, something probably went wrong.
else {
logger.warn("JXPathFilter expression evaluates to null: " + pattern);
}
}
if (logger.isDebugEnabled()) {
logger.debug("JXPathFilter accept object : " + accept);
}
return accept;
}