org.argeo.cms.jcr/src/org/argeo/cms/jcr/internal/servlet/LinkServlet.java

   1 package org.argeo.cms.jcr.internal.servlet;
   2
   3 import static javax.jcr.Property.JCR_DESCRIPTION;
   4 import static javax.jcr.Property.JCR_LAST_MODIFIED;
   5 import static javax.jcr.Property.JCR_TITLE;
   6
   7 import java.io.IOException;
   8 import java.io.PrintWriter;
   9 import java.net.MalformedURLException;
  10 import java.net.URL;
  11 import java.security.PrivilegedExceptionAction;
  12 import java.util.Calendar;
  13 import java.util.Collection;
  14
  15 import javax.jcr.Node;
  16 import javax.jcr.NodeIterator;
  17 import javax.jcr.Repository;
  18 import javax.jcr.RepositoryException;
  19 import javax.jcr.Session;
  20 import javax.security.auth.Subject;
  21 import javax.security.auth.login.LoginContext;
  22 import javax.security.auth.login.LoginException;
  23 import javax.servlet.ServletException;
  24 import javax.servlet.http.HttpServlet;
  25 import javax.servlet.http.HttpServletRequest;
  26 import javax.servlet.http.HttpServletResponse;
  27
  28 import org.argeo.api.cms.CmsAuth;
  29 import org.argeo.api.cms.CmsConstants;
  30 import org.argeo.cms.jcr.CmsJcrUtils;
  31 import org.argeo.jcr.JcrUtils;
  32 import org.osgi.framework.BundleContext;
  33 import org.osgi.framework.FrameworkUtil;
  34 import org.osgi.framework.ServiceReference;
  35
  36 public class LinkServlet extends HttpServlet {
  37         private final BundleContext bc = FrameworkUtil.getBundle(getClass()).getBundleContext();
  38
  39         private static final long serialVersionUID = 3749990143146845708L;
  40
  41         @Override
  42         protected void service(HttpServletRequest request, HttpServletResponse response)
  43                         throws ServletException, IOException {
  44                 String path = request.getPathInfo();
  45                 String userAgent = request.getHeader("User-Agent").toLowerCase();
  46                 boolean isBot = false;
  47                 // boolean isCompatibleBrowser = false;
  48                 if (userAgent.contains("bot") || userAgent.contains("facebook") || userAgent.contains("twitter")) {
  49                         isBot = true;
  50                 }
  51                 // else if (userAgent.contains("webkit") ||
  52                 // userAgent.contains("gecko") || userAgent.contains("firefox")
  53                 // || userAgent.contains("msie") || userAgent.contains("chrome") ||
  54                 // userAgent.contains("chromium")
  55                 // || userAgent.contains("opera") || userAgent.contains("browser"))
  56                 // {
  57                 // isCompatibleBrowser = true;
  58                 // }
  59
  60                 if (isBot) {
  61                         // log.warn("# BOT " + request.getHeader("User-Agent"));
  62                         canonicalAnswer(request, response, path);
  63                         return;
  64                 }
  65
  66                 // if (isCompatibleBrowser && log.isTraceEnabled())
  67                 // log.trace("# BWS " + request.getHeader("User-Agent"));
  68                 redirectTo(response, "/#" + path);
  69         }
  70
  71         private void redirectTo(HttpServletResponse response, String location) {
  72                 response.setHeader("Location", location);
  73                 response.setStatus(HttpServletResponse.SC_FOUND);
  74         }
  75
  76         // private boolean canonicalAnswerNeededBy(HttpServletRequest request) {
  77         // String userAgent = request.getHeader("User-Agent").toLowerCase();
  78         // return userAgent.startsWith("facebookexternalhit/");
  79         // }
  80
  81         /** For bots which don't understand RWT. */
  82         private void canonicalAnswer(HttpServletRequest request, HttpServletResponse response, String path) {
  83                 Session session = null;
  84                 try {
  85                         PrintWriter writer = response.getWriter();
  86                         session = Subject.doAs(anonymousLogin(), new PrivilegedExceptionAction<Session>() {
  87
  88                                 @Override
  89                                 public Session run() throws Exception {
  90                                         Collection<ServiceReference<Repository>> srs = bc.getServiceReferences(Repository.class,
  91                                                         "(" + CmsConstants.CN + "=" + CmsConstants.EGO_REPOSITORY + ")");
  92                                         Repository repository = bc.getService(srs.iterator().next());
  93                                         return repository.login();
  94                                 }
  95
  96                         });
  97                         Node node = session.getNode(path);
  98                         String title = node.hasProperty(JCR_TITLE) ? node.getProperty(JCR_TITLE).getString() : node.getName();
  99                         String desc = node.hasProperty(JCR_DESCRIPTION) ? node.getProperty(JCR_DESCRIPTION).getString() : null;
 100                         Calendar lastUpdate = node.hasProperty(JCR_LAST_MODIFIED) ? node.getProperty(JCR_LAST_MODIFIED).getDate()
 101                                         : null;
 102                         String url = getCanonicalUrl(node, request);
 103                         String imgUrl = null;
 104                         // TODO support images
 105 //                      loop: for (NodeIterator it = node.getNodes(); it.hasNext();) {
 106 //                              // Takes the first found cms:image
 107 //                              Node child = it.nextNode();
 108 //                              if (child.isNodeType(CMS_IMAGE)) {
 109 //                                      imgUrl = getDataUrl(child, request);
 110 //                                      break loop;
 111 //                              }
 112 //                      }
 113                         StringBuilder buf = new StringBuilder();
 114                         buf.append("<html>");
 115                         buf.append("<head>");
 116                         writeMeta(buf, "og:title", escapeHTML(title));
 117                         writeMeta(buf, "og:type", "website");
 118                         buf.append("<meta name='twitter:card' content='summary' />");
 119                         buf.append("<meta name='twitter:site' content='@argeo_org' />");
 120                         writeMeta(buf, "og:url", url);
 121                         if (desc != null)
 122                                 writeMeta(buf, "og:description", escapeHTML(desc));
 123                         if (imgUrl != null)
 124                                 writeMeta(buf, "og:image", imgUrl);
 125                         if (lastUpdate != null)
 126                                 writeMeta(buf, "og:updated_time", Long.toString(lastUpdate.getTime().getTime()));
 127                         buf.append("</head>");
 128                         buf.append("<body>");
 129                         buf.append("<p><b>!! This page is meant for indexing robots, not for real people," + " visit <a href='/#")
 130                                         .append(path).append("'>").append(escapeHTML(title)).append("</a> instead.</b></p>");
 131                         writeCanonical(buf, node);
 132                         buf.append("</body>");
 133                         buf.append("</html>");
 134                         writer.print(buf.toString());
 135
 136                         response.setHeader("Content-Type", "text/html");
 137                         writer.flush();
 138                 } catch (Exception e) {
 139                         throw new IllegalStateException("Cannot write canonical answer", e);
 140                 } finally {
 141                         JcrUtils.logoutQuietly(session);
 142                 }
 143         }
 144
 145         /**
 146          * From http://stackoverflow.com/questions/1265282/recommended-method-for-
 147          * escaping-html-in-java (+ escaping '). TODO Use
 148          * org.apache.commons.lang.StringEscapeUtils
 149          */
 150         private String escapeHTML(String s) {
 151                 StringBuilder out = new StringBuilder(Math.max(16, s.length()));
 152                 for (int i = 0; i < s.length(); i++) {
 153                         char c = s.charAt(i);
 154                         if (c > 127 || c == '\'' || c == '"' || c == '<' || c == '>' || c == '&') {
 155                                 out.append("&#");
 156                                 out.append((int) c);
 157                                 out.append(';');
 158                         } else {
 159                                 out.append(c);
 160                         }
 161                 }
 162                 return out.toString();
 163         }
 164
 165         private void writeMeta(StringBuilder buf, String tag, String value) {
 166                 buf.append("<meta property='").append(tag).append("' content='").append(value).append("'/>");
 167         }
 168
 169         private void writeCanonical(StringBuilder buf, Node node) throws RepositoryException {
 170                 buf.append("<div>");
 171                 if (node.hasProperty(JCR_TITLE))
 172                         buf.append("<p>").append(node.getProperty(JCR_TITLE).getString()).append("</p>");
 173                 if (node.hasProperty(JCR_DESCRIPTION))
 174                         buf.append("<p>").append(node.getProperty(JCR_DESCRIPTION).getString()).append("</p>");
 175                 NodeIterator children = node.getNodes();
 176                 while (children.hasNext()) {
 177                         writeCanonical(buf, children.nextNode());
 178                 }
 179                 buf.append("</div>");
 180         }
 181
 182         // DATA
 183         private StringBuilder getServerBaseUrl(HttpServletRequest request) {
 184                 try {
 185                         URL url = new URL(request.getRequestURL().toString());
 186                         StringBuilder buf = new StringBuilder();
 187                         buf.append(url.getProtocol()).append("://").append(url.getHost());
 188                         if (url.getPort() != -1)
 189                                 buf.append(':').append(url.getPort());
 190                         return buf;
 191                 } catch (MalformedURLException e) {
 192                         throw new IllegalArgumentException("Cannot extract server base URL from " + request.getRequestURL(), e);
 193                 }
 194         }
 195
 196         private String getDataUrl(Node node, HttpServletRequest request) throws RepositoryException {
 197                 try {
 198                         StringBuilder buf = getServerBaseUrl(request);
 199                         buf.append(CmsJcrUtils.getDataPath(CmsConstants.EGO_REPOSITORY, node));
 200                         return new URL(buf.toString()).toString();
 201                 } catch (MalformedURLException e) {
 202                         throw new IllegalArgumentException("Cannot build data URL for " + node, e);
 203                 }
 204         }
 205
 206         // public static String getDataPath(Node node) throws
 207         // RepositoryException {
 208         // assert node != null;
 209         // String userId = node.getSession().getUserID();
 210         //// if (log.isTraceEnabled())
 211         //// log.trace(userId + " : " + node.getPath());
 212         // StringBuilder buf = new StringBuilder();
 213         // boolean isAnonymous =
 214         // userId.equalsIgnoreCase(NodeConstants.ROLE_ANONYMOUS);
 215         // if (isAnonymous)
 216         // buf.append(WEBDAV_PUBLIC);
 217         // else
 218         // buf.append(WEBDAV_PRIVATE);
 219         // Session session = node.getSession();
 220         // Repository repository = session.getRepository();
 221         // String cn;
 222         // if (repository.isSingleValueDescriptor(NodeConstants.CN)) {
 223         // cn = repository.getDescriptor(NodeConstants.CN);
 224         // } else {
 225         //// log.warn("No cn defined in repository, using " +
 226         // NodeConstants.NODE);
 227         // cn = NodeConstants.NODE;
 228         // }
 229         // return
 230         // buf.append('/').append(cn).append('/').append(session.getWorkspace().getName()).append(node.getPath())
 231         // .toString();
 232         // }
 233
 234         private String getCanonicalUrl(Node node, HttpServletRequest request) throws RepositoryException {
 235                 try {
 236                         StringBuilder buf = getServerBaseUrl(request);
 237                         buf.append('/').append('!').append(node.getPath());
 238                         return new URL(buf.toString()).toString();
 239                 } catch (MalformedURLException e) {
 240                         throw new IllegalArgumentException("Cannot build data URL for " + node, e);
 241                 }
 242                 // return request.getRequestURL().append('!').append(node.getPath())
 243                 // .toString();
 244         }
 245
 246         private Subject anonymousLogin() {
 247                 Subject subject = new Subject();
 248                 LoginContext lc;
 249                 try {
 250                         lc = new LoginContext(CmsAuth.LOGIN_CONTEXT_ANONYMOUS, subject);
 251                         lc.login();
 252                         return subject;
 253                 } catch (LoginException e) {
 254                         throw new IllegalStateException("Cannot login as anonymous", e);
 255                 }
 256         }
 257
 258 }