List of usage examples for java.net HttpURLConnection getAllowUserInteraction
public boolean getAllowUserInteraction()
From source file:org.oclc.oai.harvester.verb.HarvesterVerb.java
/** * Performs the OAI request, recovering from typical XML error * /*from www. jav a 2 s . c o m*/ * @author nfreire Nuno Freire / Gilberto Pedrosa * @param requestURL * @throws IOException * @throws ParserConfigurationException * @throws SAXException * @throws TransformerException */ private void harvest(String requestURL) throws IOException, ParserConfigurationException, SAXException, TransformerException { this.requestURL = requestURL; logger.debug("requestURL=" + requestURL); InputStream in; URL url = new URL(requestURL); HttpURLConnection con; int responseCode; do { con = (HttpURLConnection) url.openConnection(); con.setConnectTimeout(30000); con.setReadTimeout(600000); if (con.getAllowUserInteraction()) { con.setRequestProperty("User-Agent", "OAIHarvester/2.0"); con.setRequestProperty("Accept-Encoding", "compress, gzip, identify"); } try { responseCode = con.getResponseCode(); logger.debug("responseCode=" + responseCode); } catch (FileNotFoundException e) { // assume it's a 503 response logger.error(requestURL, e); responseCode = HttpURLConnection.HTTP_UNAVAILABLE; } if (responseCode == HttpURLConnection.HTTP_UNAVAILABLE) { long retrySeconds = con.getHeaderFieldInt("Retry-After", -1); if (retrySeconds == -1) { long now = (new Date()).getTime(); long retryDate = con.getHeaderFieldDate("Retry-After", now); retrySeconds = retryDate - now; } if (retrySeconds == 0) { // Apparently, it's a bad URL throw new FileNotFoundException("Bad URL?"); } logger.warn("Server response: Retry-After=" + retrySeconds); if (retrySeconds > 0) { try { Thread.sleep(retrySeconds * 1000); } catch (InterruptedException ex) { ex.printStackTrace(); } } } } while (responseCode == HttpURLConnection.HTTP_UNAVAILABLE); String contentEncoding = con.getHeaderField("Content-Encoding"); logger.debug("contentEncoding=" + contentEncoding); if ("compress".equals(contentEncoding)) { ZipInputStream zis = new ZipInputStream(con.getInputStream()); zis.getNextEntry(); in = zis; } else if ("gzip".equals(contentEncoding)) { in = new GZIPInputStream(con.getInputStream()); } else if ("deflate".equals(contentEncoding)) { in = new InflaterInputStream(con.getInputStream()); } else { in = con.getInputStream(); } byte[] inputBytes = IOUtils.toByteArray(in); InputSource data = new InputSource(new ByteArrayInputStream(inputBytes)); Thread t = Thread.currentThread(); DocumentBuilder builder = builderMap.get(t); if (builder == null) { builder = factory.newDocumentBuilder(); builderMap.put(t, builder); } try { doc = builder.parse(data); } catch (SAXException e) { try { //Here we can try to recover the xml from known typical problems //Recover from invalid characters //we assume this is UTF-8... String xmlString = new String(inputBytes, "UTF-8"); xmlString = XmlUtil.removeInvalidXMLCharacters(xmlString); data = new InputSource(new ByteArrayInputStream(xmlString.getBytes("UTF-8"))); doc = builder.parse(data); } catch (Exception e2) { //the recovered version did not work either. Throw the original exception throw e; } } catch (IOException e3) { System.out.println("e = " + e3.getMessage()); } catch (Exception e4) { System.out.println("e = " + e4.getMessage()); } StringTokenizer tokenizer = new StringTokenizer(getSingleString("/*/@xsi:schemaLocation"), " "); StringBuffer sb = new StringBuffer(); while (tokenizer.hasMoreTokens()) { if (sb.length() > 0) sb.append(" "); sb.append(tokenizer.nextToken()); } this.schemaLocation = sb.toString(); this.defaultNamespace = getDocument().getDocumentElement().getNamespaceURI(); }
From source file:org.oclc.oai.harvester.verb.HarvesterVerb.java
/** * @param args/*from w w w .j a v a2 s. co m*/ * @throws IOException * @throws ParserConfigurationException * @throws SAXException */ public static void main(String[] args) throws IOException, ParserConfigurationException, SAXException { String requestURL = "https://databank.ora.ox.ac.uk/oaipmh?verb=ListRecords&resumptionToken=20121206_TKMGW4A_SWT3NHV"; //String requestURL = "http://bd2.inesc-id.pt:8080/repox2Eudml/OAIHandler?verb=ListRecords&resumptionToken=1354116062009:ELibM_external:eudml-article2:33753:37054::"; //String requestURL = "http://bd2.inesc-id.pt:8080/repox2Eudml/OAIHandler?verb=GetRecord&identifier=urn:eudml.eu:ELibM_external:05152756&metadataPrefix=eudml-article2"; //String requestURL = "C:/Users/Gilberto Pedrosa/Desktop/OAIHandler.xml"; //FileInputStream fis = new FileInputStream(requestURL); //InputStream in = fis; logger.debug("requestURL=" + requestURL); DocumentBuilderFactory factory; factory = DocumentBuilderFactory.newInstance(); factory.setNamespaceAware(true); Thread t = Thread.currentThread(); DocumentBuilder builder = factory.newDocumentBuilder(); HashMap<Thread, DocumentBuilder> builderMap = new HashMap<Thread, DocumentBuilder>(); builderMap.put(t, builder); InputStream in; URL url = new URL(requestURL); HttpURLConnection con; int responseCode; do { con = (HttpURLConnection) url.openConnection(); con.setConnectTimeout(30000); con.setReadTimeout(600000); if (con.getAllowUserInteraction()) { con.setRequestProperty("User-Agent", "OAIHarvester/2.0"); con.setRequestProperty("Accept-Encoding", "compress, gzip, identify"); } try { responseCode = con.getResponseCode(); logger.debug("responseCode=" + responseCode); } catch (FileNotFoundException e) { // assume it's a 503 response logger.error(requestURL, e); responseCode = HttpURLConnection.HTTP_UNAVAILABLE; } if (responseCode == HttpURLConnection.HTTP_UNAVAILABLE) { long retrySeconds = con.getHeaderFieldInt("Retry-After", -1); if (retrySeconds == -1) { long now = (new Date()).getTime(); long retryDate = con.getHeaderFieldDate("Retry-After", now); retrySeconds = retryDate - now; } if (retrySeconds == 0) { // Apparently, it's a bad URL throw new FileNotFoundException("Bad URL?"); } logger.warn("Server response: Retry-After=" + retrySeconds); if (retrySeconds > 0) { try { Thread.sleep(retrySeconds * 1000); } catch (InterruptedException ex) { ex.printStackTrace(); } } } } while (responseCode == HttpURLConnection.HTTP_UNAVAILABLE); String contentEncoding = con.getHeaderField("Content-Encoding"); logger.debug("contentEncoding=" + contentEncoding); if ("compress".equals(contentEncoding)) { ZipInputStream zis = new ZipInputStream(con.getInputStream()); zis.getNextEntry(); in = zis; } else if ("gzip".equals(contentEncoding)) { in = new GZIPInputStream(con.getInputStream()); } else if ("deflate".equals(contentEncoding)) { in = new InflaterInputStream(con.getInputStream()); } else { in = con.getInputStream(); } byte[] inputBytes = IOUtils.toByteArray(in); InputSource data = new InputSource(new ByteArrayInputStream(inputBytes)); String xmlString = new String(inputBytes, "UTF-8"); xmlString = XmlUtil.removeInvalidXMLCharacters(xmlString); builder.parse(data); System.out.println("data = " + data); }
From source file:org.oclc.oai.harvester2.verb.HarvesterVerb.java
/** * Preforms the OAI request, recovering from typical XML error * * @author nfreire Nuno Freire / Gilberto Pedrosa * @param requestURL/*from w ww.jav a2s .c om*/ * @throws IOException * @throws ParserConfigurationException * @throws SAXException * @throws TransformerException */ private void harvest(String requestURL) throws IOException, ParserConfigurationException, SAXException, TransformerException { this.requestURL = requestURL; logger.debug("requestURL=" + requestURL); InputStream in; URL url = new URL(requestURL); HttpURLConnection con; int responseCode; do { con = (HttpURLConnection) url.openConnection(); con.setConnectTimeout(30000); con.setReadTimeout(600000); if (con.getAllowUserInteraction()) { con.setRequestProperty("User-Agent", "OAIHarvester/2.0"); con.setRequestProperty("Accept-Encoding", "compress, gzip, identify"); } try { responseCode = con.getResponseCode(); logger.debug("responseCode=" + responseCode); } catch (FileNotFoundException e) { // assume it's a 503 response logger.error(requestURL, e); responseCode = HttpURLConnection.HTTP_UNAVAILABLE; } if (responseCode == HttpURLConnection.HTTP_UNAVAILABLE) { long retrySeconds = con.getHeaderFieldInt("Retry-After", -1); if (retrySeconds == -1) { long now = (new Date()).getTime(); long retryDate = con.getHeaderFieldDate("Retry-After", now); retrySeconds = retryDate - now; } if (retrySeconds == 0) { // Apparently, it's a bad URL throw new FileNotFoundException("Bad URL?"); } logger.warn("Server response: Retry-After=" + retrySeconds); if (retrySeconds > 0) { try { Thread.sleep(retrySeconds * 1000); } catch (InterruptedException ex) { ex.printStackTrace(); } } } } while (responseCode == HttpURLConnection.HTTP_UNAVAILABLE); String contentEncoding = con.getHeaderField("Content-Encoding"); logger.debug("contentEncoding=" + contentEncoding); if ("compress".equals(contentEncoding)) { ZipInputStream zis = new ZipInputStream(con.getInputStream()); zis.getNextEntry(); in = zis; } else if ("gzip".equals(contentEncoding)) { in = new GZIPInputStream(con.getInputStream()); } else if ("deflate".equals(contentEncoding)) { in = new InflaterInputStream(con.getInputStream()); } else { in = con.getInputStream(); } byte[] inputBytes = IOUtils.toByteArray(in); InputSource data = new InputSource(new ByteArrayInputStream(inputBytes)); Thread t = Thread.currentThread(); DocumentBuilder builder = (DocumentBuilder) builderMap.get(t); if (builder == null) { builder = factory.newDocumentBuilder(); builderMap.put(t, builder); } try { doc = builder.parse(data); } catch (SAXException e) { try { //Here we can try to recover the xml from known typical problems //Recover from invalid characters //we assume this is UTF-8... String xmlString = new String(inputBytes, "UTF-8"); xmlString = XmlUtil.removeInvalidXMLCharacters(xmlString); data = new InputSource(new ByteArrayInputStream(xmlString.getBytes("UTF-8"))); doc = builder.parse(data); } catch (Exception e2) { //the recovered version did not work either. Throw the original exception throw e; } } catch (IOException e3) { System.out.println("e = " + e3.getMessage()); } catch (Exception e4) { System.out.println("e = " + e4.getMessage()); } StringTokenizer tokenizer = new StringTokenizer(getSingleString("/*/@xsi:schemaLocation"), " "); StringBuffer sb = new StringBuffer(); while (tokenizer.hasMoreTokens()) { if (sb.length() > 0) sb.append(" "); sb.append(tokenizer.nextToken()); } this.schemaLocation = sb.toString(); this.defaultNamespace = getDocument().getDocumentElement().getNamespaceURI(); }
From source file:org.oclc.oai.harvester2.verb.HarvesterVerb.java
public static void main(String[] args) throws IOException, ParserConfigurationException, SAXException { String requestURL = "https://databank.ora.ox.ac.uk/oaipmh?verb=ListRecords&resumptionToken=20121206_TKMGW4A_SWT3NHV"; //String requestURL = "http://bd2.inesc-id.pt:8080/repox2Eudml/OAIHandler?verb=ListRecords&resumptionToken=1354116062009:ELibM_external:eudml-article2:33753:37054::"; //String requestURL = "http://bd2.inesc-id.pt:8080/repox2Eudml/OAIHandler?verb=GetRecord&identifier=urn:eudml.eu:ELibM_external:05152756&metadataPrefix=eudml-article2"; //String requestURL = "C:/Users/Gilberto Pedrosa/Desktop/OAIHandler.xml"; //FileInputStream fis = new FileInputStream(requestURL); //InputStream in = fis; logger.debug("requestURL=" + requestURL); DocumentBuilderFactory factory; factory = DocumentBuilderFactory.newInstance(); factory.setNamespaceAware(true);/* w ww .j a v a2 s . c o m*/ Thread t = Thread.currentThread(); DocumentBuilder builder = factory.newDocumentBuilder(); HashMap builderMap = new HashMap(); builderMap.put(t, builder); InputStream in; URL url = new URL(requestURL); HttpURLConnection con; int responseCode; do { con = (HttpURLConnection) url.openConnection(); con.setConnectTimeout(30000); con.setReadTimeout(600000); if (con.getAllowUserInteraction()) { con.setRequestProperty("User-Agent", "OAIHarvester/2.0"); con.setRequestProperty("Accept-Encoding", "compress, gzip, identify"); } try { responseCode = con.getResponseCode(); logger.debug("responseCode=" + responseCode); } catch (FileNotFoundException e) { // assume it's a 503 response logger.error(requestURL, e); responseCode = HttpURLConnection.HTTP_UNAVAILABLE; } if (responseCode == HttpURLConnection.HTTP_UNAVAILABLE) { long retrySeconds = con.getHeaderFieldInt("Retry-After", -1); if (retrySeconds == -1) { long now = (new Date()).getTime(); long retryDate = con.getHeaderFieldDate("Retry-After", now); retrySeconds = retryDate - now; } if (retrySeconds == 0) { // Apparently, it's a bad URL throw new FileNotFoundException("Bad URL?"); } logger.warn("Server response: Retry-After=" + retrySeconds); if (retrySeconds > 0) { try { Thread.sleep(retrySeconds * 1000); } catch (InterruptedException ex) { ex.printStackTrace(); } } } } while (responseCode == HttpURLConnection.HTTP_UNAVAILABLE); String contentEncoding = con.getHeaderField("Content-Encoding"); logger.debug("contentEncoding=" + contentEncoding); if ("compress".equals(contentEncoding)) { ZipInputStream zis = new ZipInputStream(con.getInputStream()); zis.getNextEntry(); in = zis; } else if ("gzip".equals(contentEncoding)) { in = new GZIPInputStream(con.getInputStream()); } else if ("deflate".equals(contentEncoding)) { in = new InflaterInputStream(con.getInputStream()); } else { in = con.getInputStream(); } byte[] inputBytes = IOUtils.toByteArray(in); InputSource data = new InputSource(new ByteArrayInputStream(inputBytes)); String xmlString = new String(inputBytes, "UTF-8"); xmlString = XmlUtil.removeInvalidXMLCharacters(xmlString); builder.parse(data); System.out.println("data = " + data); }