List of usage examples for org.xml.sax XMLReader setEntityResolver
public void setEntityResolver(EntityResolver resolver);
From source file:org.infoscoop.web.MultiRssServlet.java
private void mergeRssAnd2JSON(HttpServletRequest request, HttpServletResponse response, String uid, String widgetId, int pageSize, NodeList urlList) throws Exception { try {// w w w.j a va 2s . c o m RssJsonResultBuilder resultBuilder = new SortedRssJsonResultBuilder(pageSize); String dateTimeFormat = request.getHeader("X-IS-DATETIMEFORMAT"); if (dateTimeFormat != null) { dateTimeFormat = URLDecoder.decode(dateTimeFormat, "UTF-8"); } String freshTime = request.getHeader("X-IS-FRESHTIME"); String maxCountString = request.getHeader("X-IS-RSSMAXCOUNT"); int maxCount = 1000; if (maxCountString != null) { try { int paramMaxCount = Integer.parseInt(maxCountString); if (paramMaxCount >= 0) { maxCount = paramMaxCount; } } catch (NumberFormatException e) { log.warn("rssmaxcount \"" + maxCountString + "\" isn't integer value."); } } // Norrowing String titleFilter = request.getHeader("X-IS-TITLEFILTER"); if (titleFilter != null) titleFilter = URLDecoder.decode(titleFilter, "UTF-8"); String creatorFilter = request.getHeader("X-IS-CREATORFILTER"); if (creatorFilter != null) creatorFilter = URLDecoder.decode(creatorFilter, "UTF-8"); String categoryFilter = request.getHeader("X-IS-CATEGORYFILTER"); if (categoryFilter != null) categoryFilter = URLDecoder.decode(categoryFilter, "UTF-8"); int DEFAULT_TIMEOUT = 15 * 1000; boolean modified = false; Map cacheHeaders = new HashMap(); Map errorMap = new HashMap(); List siteCacheHeaders = new ArrayList(); for (int i = 0; i < urlList.getLength(); i++) { Element rssEl = (Element) urlList.item(i); String url = rssEl.getAttribute("url"); ProxyRequest proxyRequest = new ProxyRequest(url, "NoOperation"); proxyRequest.setLocales(request.getLocales()); proxyRequest.setPortalUid(uid); int timeout = request.getIntHeader("MSDPortal-Timeout") - 1000; proxyRequest.setTimeout((timeout > 0) ? timeout : DEFAULT_TIMEOUT); //proxyRequest.setTimeout(timeout); proxyRequest.addIgnoreHeader("user-agent"); proxyRequest.addIgnoreHeader("X-IS-DATETIMEFORMAT"); proxyRequest.addIgnoreHeader("X-IS-FRESHTIME"); proxyRequest.addIgnoreHeader("X-IS-REFRESH"); proxyRequest.addIgnoreHeader("X-IS-RSSMAXCOUNT"); proxyRequest.addIgnoreHeader("X-IS-PAGESIZE"); Enumeration headers = request.getHeaderNames(); while (headers.hasMoreElements()) { String headerName = (String) headers.nextElement(); proxyRequest.putRequestHeader(headerName, request.getHeader(headerName)); } NodeList rssChildNodes = rssEl.getElementsByTagName("header"); for (int j = 0; j < rssChildNodes.getLength(); j++) { Element header = (Element) rssChildNodes.item(j); if (header.getFirstChild() != null) { String name = header.getAttribute("name"); String value = header.getFirstChild().getNodeValue(); if (name == null || name.trim().length() == 0 || value == null || value.trim().length() == 0) continue; proxyRequest.putRequestHeader(name, value); } } int statusCode = 0; String methodType = rssEl.getAttribute("method"); try { if ("post".equals(methodType)) { statusCode = proxyRequest.executePost(); } else { statusCode = proxyRequest.executeGet(); } } catch (SocketTimeoutException ex) { log.error("url: [" + url + "] socket timeout.", ex); errorMap.put(url, new Integer(HttpStatusCode.MSD_SC_TIMEOUT)); } catch (ConnectTimeoutException ex) { log.error("url: [" + url + "] connection timeout.", ex); errorMap.put(url, new Integer(500)); } catch (SocketException ex) { log.error("url: [" + url + "] socket error.", ex); errorMap.put(url, new Integer(HttpStatus.SC_NOT_FOUND)); } catch (IOException ex) { log.error("url: [" + url + "] I/O error.", ex); errorMap.put(url, new Integer(HttpStatus.SC_NOT_FOUND)); } catch (Exception ex) { log.error("url: [" + url + "]" + ex.getMessage(), ex); errorMap.put(url, new Integer(500)); } BufferedInputStream bis = null; if (errorMap.containsKey(url)) { // nothing } else if (statusCode == 204) { log.warn("url:[" + url + "] is no content #" + statusCode); modified = true; } else if (statusCode == 304) { log.warn("url:[" + url + "] is not modified #" + statusCode); } else if (statusCode != 200) { log.error("url:[" + url + "] had error status code #" + statusCode); errorMap.put(url, new Integer(statusCode)); } else { log.info("url:[" + url + "] is succed #" + statusCode); try { modified = true; bis = new BufferedInputStream(proxyRequest.getResponseBody()); ByteArrayOutputStream baos = new ByteArrayOutputStream(); byte[] buf = new byte[10240]; int c; while ((c = bis.read(buf)) != -1) { baos.write(buf, 0, c); baos.flush(); } bis.close(); byte[] data = baos.toByteArray(); baos.close(); //RssCacheDAO.newInstance().insertCache(uid, widgetId+url, -1,data ); Map responseHeaders = proxyRequest.getResponseHeaders(); String lastModifiedName = "Last-Modified".toLowerCase(); if (responseHeaders.containsKey(lastModifiedName)) { siteCacheHeaders.add(lastModifiedName); siteCacheHeaders.add(responseHeaders.get(lastModifiedName)); } String etagName = "ETag".toLowerCase(); if (responseHeaders.containsKey(etagName)) { siteCacheHeaders.add(etagName); siteCacheHeaders.add(responseHeaders.get(etagName)); } if (siteCacheHeaders.size() > 0) { cacheHeaders.put(url, siteCacheHeaders.toArray()); siteCacheHeaders.clear(); } bis = new BufferedInputStream(new ByteArrayInputStream(data)); } catch (IOException ex) { log.error("rss reading " + url + " is failed.", ex); cacheHeaders.remove(url); errorMap.put(url, new Integer(500)); bis.close(); bis = null; } } if (bis == null) continue; RssHandler handler; boolean isAtom = RssFilter.isAtom(bis); XMLFilter.skipEmptyLine(bis); if (isAtom) { handler = new AtomHandler(resultBuilder, dateTimeFormat, freshTime, maxCount, titleFilter, creatorFilter, categoryFilter, i); } else { handler = new RssHandler(resultBuilder, dateTimeFormat, freshTime, maxCount, titleFilter, creatorFilter, categoryFilter, i); } try { XMLReader reader = factory.newSAXParser().getXMLReader(); reader.setEntityResolver(NoOpEntityResolver.getInstance()); reader.setContentHandler(handler); reader.parse(new InputSource(bis)); } catch (SAXException e) { log.info("Parsing rss " + url + " is failed.", e); cacheHeaders.remove(url); errorMap.put(url, new Integer(HttpStatusCode.MSD_SC_CONTENT_PARSE_ERROR)); } } if (!modified && errorMap.isEmpty()) { log.warn("multi rss is not modified."); response.setStatus(304); return; } else { try { long freshTimeLong = new Date().getTime(); if (freshTime != null) freshTimeLong = Long.parseLong(freshTime.trim()); setOldData(resultBuilder, uid, widgetId, freshTimeLong, titleFilter, creatorFilter, categoryFilter); } catch (NumberFormatException e) { log.error("", e); } //} int pageCount = resultBuilder.getPageCount(); // We create the result cash by all means. //if( pageCount > 1 ) { for (int pageNum = 0; pageNum < pageCount; pageNum++) { RssCacheDAO.newInstance().insertCache(uid, widgetId, pageNum, resultBuilder.getResult(pageNum)); } //} } response.addHeader("Content-Type", "text/plain; charset=UTF-8"); String result = resultBuilder.getResult(); if (!errorMap.isEmpty()) { JSONObject errors = new JSONObject(errorMap); result = "{errors:" + errors.toString() + "," + result.substring(result.indexOf("{") + 1); } if (!cacheHeaders.isEmpty()) { StringBuffer cacheHeadersBuf = new StringBuffer(); cacheHeadersBuf.append("cacheHeaders : {"); for (Iterator keys = cacheHeaders.keySet().iterator(); keys.hasNext();) { String url = (String) keys.next(); Object[] headers = (Object[]) cacheHeaders.get(url); cacheHeadersBuf.append("\"").append(url).append("\" : {"); for (int i = 0; i < headers.length; i += 2) { cacheHeadersBuf.append("\"").append(headers[i]).append("\""); cacheHeadersBuf.append(" : '").append(headers[i + 1]).append("'"); if (i + 2 < headers.length) cacheHeadersBuf.append(","); } cacheHeadersBuf.append("}"); if (keys.hasNext()) cacheHeadersBuf.append(","); } cacheHeadersBuf.append("}"); result = "{" + cacheHeadersBuf.toString() + "," + result.substring(result.indexOf("{") + 1); } response.setContentLength(result.getBytes("UTF-8").length); OutputStreamWriter out = new OutputStreamWriter(response.getOutputStream(), "UTF-8"); try { out.write(result); out.flush(); } catch (SocketException ex) { // ignore client abort exception } finally { if (out != null) { try { out.close(); } catch (IOException ex) { // ignore } } } } catch (Exception e) { log.error("unexpected error occurred.", e); response.sendError(500, e.getMessage()); } }
From source file:org.infoscoop.web.ValidationServlet.java
private static boolean checkXml(String text) { try {/*from ww w .ja v a 2s .c om*/ XMLReader reader = factory.newSAXParser().getXMLReader(); reader.setEntityResolver(NoOpEntityResolver.getInstance()); reader.parse(text); return true; } catch (SAXException e) { return false; } catch (IOException e) { return false; } catch (ParserConfigurationException e) { return false; } }
From source file:org.jboss.confluence.plugin.docbook_tools.docbookimport.DocbookImporter.java
/** * Process XSLT transformation./* w ww . j a va2s.c o m*/ * * @param xsltTemplate input stream with XSLT template file used to transform (closed inside this method) * @param xmlToTransform input stream with XML file to transform (closed inside this method) * @param xmlToTransformURL URL of <code>xmlToTransform</code> file (may be <code>file://</code> too). We need it to * correctly evaluate relative paths. * @param output stream to write transformed output to * @throws javax.xml.transform.TransformerException */ protected void processXslt(final InputStream xsltTemplate, final InputStream xmlToTransform, final String xmlToTransformURL, final OutputStream output) throws Exception { final XSLTErrorListener errorListener = new XSLTErrorListener(); final SAXErrorHandler eh = new SAXErrorHandler(); Thread th = new Thread(new Runnable() { public void run() { try { org.xml.sax.InputSource xmlSource = new org.xml.sax.InputSource(xmlToTransform); xmlSource.setSystemId(xmlToTransformURL); javax.xml.transform.Source xsltSource = new javax.xml.transform.stream.StreamSource( xsltTemplate); javax.xml.transform.Result result = new javax.xml.transform.stream.StreamResult(output); // prepare XInclude aware parser which resolves necessary entities correctly XMLReader reader = new ParserAdapter(saxParserFactory.newSAXParser().getParser()); reader.setEntityResolver(new JDGEntityResolver(reader.getEntityResolver())); reader.setErrorHandler(eh); SAXSource xmlSAXSource = new SAXSource(reader, xmlSource); javax.xml.transform.Transformer trans = transformerFact.newTransformer(xsltSource); trans.setErrorListener(errorListener); trans.transform(xmlSAXSource, result); } catch (Exception e) { if (e instanceof TransformerException) { errorListener.setException((TransformerException) e); } else { errorListener.setException(new TransformerException(e)); } } finally { FileUtils.closeInputStream(xmlToTransform); FileUtils.closeInputStream(xsltTemplate); } } }); th.setName("DocbookImporter XSLT transformation thread"); th.setDaemon(true); th.setContextClassLoader(DocbookImporter.class.getClassLoader()); th.start(); th.join(); if (eh.getException() != null) { throw eh.getException(); } if (errorListener.getException() != null) { throw errorListener.getException(); } }
From source file:org.jbuiltDemo.managed.view.Xhtml2Jbuilt.java
public Xhtml2Jbuilt() throws Exception { factory = SAXParserFactory.newInstance(); factory.setNamespaceAware(true);// w w w .ja va 2s. c o m factory.setFeature("http://xml.org/sax/features/namespace-prefixes", true); factory.setFeature("http://xml.org/sax/features/validation", false); factory.setValidating(false); parser = factory.newSAXParser(); XMLReader reader = parser.getXMLReader(); reader.setDTDHandler(this); reader.setContentHandler(this); reader.setProperty("http://xml.org/sax/properties/lexical-handler", this); reader.setErrorHandler(this); reader.setEntityResolver(this); }
From source file:org.mycore.common.content.transformer.MCRXSLTransformer.java
protected XMLReader getXMLReader(LinkedList<TransformerHandler> transformHandlerList) throws SAXException { XMLReader reader = XMLReaderFactory.createXMLReader(); reader.setEntityResolver(ENTITY_RESOLVER); reader.setContentHandler(transformHandlerList.getFirst()); return reader; }
From source file:org.olat.ims.qti.qpool.ItemFileResourceValidator.java
private boolean validateDocument(Document in) { try {/* w w w . j av a 2 s.c o m*/ SAXParserFactory factory = SAXParserFactory.newInstance(); factory.setValidating(true); factory.setNamespaceAware(true); SimpleErrorHandler errorHandler = new SimpleErrorHandler(); ItemContentHandler contentHandler = new ItemContentHandler(); SAXParser parser = factory.newSAXParser(); XMLReader reader = parser.getXMLReader(); reader.setEntityResolver(new IMSEntityResolver()); reader.setErrorHandler(errorHandler); reader.setContentHandler(contentHandler); SAXValidator validator = new SAXValidator(reader); validator.validate(in); return errorHandler.isValid() && contentHandler.isItem(); } catch (ParserConfigurationException e) { return false; } catch (SAXException e) { return false; } catch (Exception e) { return false; } }
From source file:org.opencms.xml.CmsXmlUtils.java
/** * Validates the structure of a XML document contained in a byte array * with the DTD or XML schema used by the document.<p> * /*from w ww . jav a 2 s . c o m*/ * @param xmlStream a source providing a XML document that should be validated * @param resolver the XML entity resolver to use * * @throws CmsXmlException if the validation fails */ public static void validateXmlStructure(InputStream xmlStream, EntityResolver resolver) throws CmsXmlException { XMLReader reader; try { reader = XMLReaderFactory.createXMLReader("org.apache.xerces.parsers.SAXParser"); } catch (SAXException e) { // xerces parser not available - no schema validation possible if (LOG.isWarnEnabled()) { LOG.warn(Messages.get().getBundle().key(Messages.LOG_VALIDATION_INIT_XERXES_SAX_READER_FAILED_0), e); } // no validation of the content is possible return; } // turn on validation try { reader.setFeature("http://xml.org/sax/features/validation", true); // turn on schema validation reader.setFeature("http://apache.org/xml/features/validation/schema", true); // configure namespace support reader.setFeature("http://xml.org/sax/features/namespaces", true); reader.setFeature("http://xml.org/sax/features/namespace-prefixes", false); } catch (SAXNotRecognizedException e) { // should not happen as Xerces 2 support this feature if (LOG.isWarnEnabled()) { LOG.warn(Messages.get().getBundle().key(Messages.LOG_SAX_READER_FEATURE_NOT_RECOGNIZED_0), e); } // no validation of the content is possible return; } catch (SAXNotSupportedException e) { // should not happen as Xerces 2 support this feature if (LOG.isWarnEnabled()) { LOG.warn(Messages.get().getBundle().key(Messages.LOG_SAX_READER_FEATURE_NOT_SUPPORTED_0), e); } // no validation of the content is possible return; } // add an error handler which turns any errors into XML CmsXmlValidationErrorHandler errorHandler = new CmsXmlValidationErrorHandler(); reader.setErrorHandler(errorHandler); if (resolver != null) { // set the resolver for the "opencms://" URIs reader.setEntityResolver(resolver); } try { reader.parse(new InputSource(xmlStream)); } catch (IOException e) { // should not happen since we read form a byte array if (LOG.isErrorEnabled()) { LOG.error(Messages.get().getBundle().key(Messages.LOG_READ_XML_FROM_BYTE_ARR_FAILED_0), e); } return; } catch (SAXException e) { // should not happen since all errors are handled in the XML error handler if (LOG.isErrorEnabled()) { LOG.error(Messages.get().getBundle().key(Messages.LOG_PARSE_SAX_EXC_0), e); } return; } if (errorHandler.getErrors().elements().size() > 0) { // there was at last one validation error, so throw an exception StringWriter out = new StringWriter(256); OutputFormat format = OutputFormat.createPrettyPrint(); XMLWriter writer = new XMLWriter(out, format); try { writer.write(errorHandler.getErrors()); writer.write(errorHandler.getWarnings()); writer.close(); } catch (IOException e) { // should not happen since we write to a StringWriter if (LOG.isErrorEnabled()) { LOG.error(Messages.get().getBundle().key(Messages.LOG_STRINGWRITER_IO_EXC_0), e); } } // generate String from XML for display of document in error message throw new CmsXmlException(Messages.get().container(Messages.ERR_XML_VALIDATION_1, out.toString())); } }
From source file:org.openstreetmap.josm.tools.ImageProvider.java
/** * Reads the wiki page on a certain file in html format in order to find the real image URL. *//*from ww w.jav a 2 s . c om*/ private static String getImgUrlFromWikiInfoPage(final String base, final String fn) { /** Quit parsing, when a certain condition is met */ class SAXReturnException extends SAXException { private String result; public SAXReturnException(String result) { this.result = result; } public String getResult() { return result; } } try { final XMLReader parser = XMLReaderFactory.createXMLReader(); parser.setContentHandler(new DefaultHandler() { @Override public void startElement(String uri, String localName, String qName, Attributes atts) throws SAXException { System.out.println(); if (localName.equalsIgnoreCase("img")) { String val = atts.getValue("src"); if (val.endsWith(fn)) throw new SAXReturnException(val); // parsing done, quit early } } }); parser.setEntityResolver(new EntityResolver() { public InputSource resolveEntity(String publicId, String systemId) { return new InputSource(new ByteArrayInputStream(new byte[0])); } }); parser.parse(new InputSource(new MirroredInputStream(base + fn, new File(Main.pref.getPreferencesDir(), "images").toString()))); } catch (SAXReturnException r) { return r.getResult(); } catch (Exception e) { System.out.println("INFO: parsing " + base + fn + " failed:\n" + e); return null; } System.out.println("INFO: parsing " + base + fn + " failed: Unexpected content."); return null; }
From source file:org.pentaho.reporting.libraries.xmlns.parser.AbstractXmlResourceFactory.java
/** * Creates a resource by interpreting the data given in the resource-data object. If additional datastreams need to be * parsed, the provided resource manager should be used. This method parses the given resource-data as XML stream. * * @param manager the resource manager used for all resource loading. * @param data the resource-data from where the binary data is read. * @param context the resource context used to resolve relative resource paths. * @return the parsed result, never null. * @throws ResourceCreationException if the resource could not be parsed due to syntaxctial or logical errors in the * data. * @throws ResourceLoadingException if the resource could not be accessed from the physical storage. *//*from ww w .j a v a2 s. c o m*/ public Resource create(final ResourceManager manager, final ResourceData data, final ResourceKey context) throws ResourceCreationException, ResourceLoadingException { try { final SAXParser parser = getParser(); final XMLReader reader = parser.getXMLReader(); final XmlFactoryModule[] rootHandlers = getModules(); if (rootHandlers.length == 0) { throw new ResourceCreationException( "There are no root-handlers registered for the factory for type " + getFactoryType()); } final ResourceDataInputSource input = new ResourceDataInputSource(data, manager); final ResourceKey contextKey; final long version; final ResourceKey targetKey = data.getKey(); if (context == null) { contextKey = targetKey; version = data.getVersion(manager); } else { contextKey = context; version = -1; } final RootXmlReadHandler handler = createRootHandler(manager, targetKey, rootHandlers, contextKey, version); final DefaultConfiguration parserConfiguration = handler.getParserConfiguration(); final URL value = manager.toURL(contextKey); if (value != null) { parserConfiguration.setConfigProperty(CONTENTBASE_KEY, value.toExternalForm()); } configureReader(reader, handler); reader.setContentHandler(handler); reader.setDTDHandler(handler); reader.setEntityResolver(handler.getEntityResolver()); reader.setErrorHandler(getErrorHandler()); final Map parameters = targetKey.getFactoryParameters(); final Iterator it = parameters.keySet().iterator(); while (it.hasNext()) { final Object o = it.next(); if (o instanceof FactoryParameterKey) { final FactoryParameterKey fpk = (FactoryParameterKey) o; handler.setHelperObject(fpk.getName(), parameters.get(fpk)); } } reader.parse(input); final Object createdProduct = finishResult(handler.getResult(), manager, data, contextKey); handler.getDependencyCollector().add(targetKey, data.getVersion(manager)); return createResource(targetKey, handler, createdProduct, getFactoryType()); } catch (ParserConfigurationException e) { throw new ResourceCreationException("Unable to initialize the XML-Parser", e); } catch (SAXException e) { throw new ResourceCreationException("Unable to parse the document: " + data.getKey(), e); } catch (IOException e) { throw new ResourceLoadingException("Unable to read the stream from document: " + data.getKey(), e); } }
From source file:org.pentaho.reporting.libraries.xmlns.parser.AbstractXmlResourceFactory.java
/** * A method to allow to invoke the parsing without accessing the LibLoader layer. The data to be parsed is held in the * given InputSource object.//w w w . jav a 2 s . co m * * @param manager the resource manager used for all resource loading. * @param input the raw-data given as SAX-InputSource. * @param context the resource context used to resolve relative resource paths. * @param parameters the parse parameters. * @return the parsed result, never null. * @throws ResourceCreationException if the resource could not be parsed due to syntaxctial or logical errors in * the data. * @throws ResourceLoadingException if the resource could not be accessed from the physical storage. * @throws ResourceKeyCreationException if creating the context key failed. */ public Object parseDirectly(final ResourceManager manager, final InputSource input, final ResourceKey context, final Map parameters) throws ResourceKeyCreationException, ResourceCreationException, ResourceLoadingException { try { final SAXParser parser = getParser(); final XMLReader reader = parser.getXMLReader(); final ResourceKey targetKey = manager.createKey(EMPTY_DATA); final ResourceKey contextKey; if (context == null) { contextKey = targetKey; } else { contextKey = context; } final XmlFactoryModule[] rootHandlers = getModules(); final RootXmlReadHandler handler = createRootHandler(manager, targetKey, rootHandlers, contextKey, -1); final DefaultConfiguration parserConfiguration = handler.getParserConfiguration(); final URL value = manager.toURL(contextKey); if (value != null) { parserConfiguration.setConfigProperty(CONTENTBASE_KEY, value.toExternalForm()); } configureReader(reader, handler); reader.setContentHandler(handler); reader.setDTDHandler(handler); reader.setEntityResolver(handler.getEntityResolver()); reader.setErrorHandler(getErrorHandler()); final Iterator it = parameters.keySet().iterator(); while (it.hasNext()) { final Object o = it.next(); if (o instanceof FactoryParameterKey) { final FactoryParameterKey fpk = (FactoryParameterKey) o; handler.setHelperObject(fpk.getName(), parameters.get(fpk)); } } reader.parse(input); return finishResult(handler.getResult(), manager, new RawResourceData(targetKey), contextKey); } catch (ParserConfigurationException e) { throw new ResourceCreationException("Unable to initialize the XML-Parser", e); } catch (SAXException e) { throw new ResourceCreationException("Unable to parse the document", e); } catch (IOException e) { throw new ResourceLoadingException("Unable to read the stream", e); } }