]> git.argeo.org Git - lgpl/argeo-commons.git/blobdiff - org.argeo.cms/src/org/argeo/cms/internal/kernel/NodeHttp.java
Rather take the first found image than the last one.
[lgpl/argeo-commons.git] / org.argeo.cms / src / org / argeo / cms / internal / kernel / NodeHttp.java
index 5e1f39acb1f7a57d0651a0976a5c0cc786a781a1..27083aeaba11812fd219ac60ab7e950ee3c01114 100644 (file)
@@ -1,11 +1,26 @@
 package org.argeo.cms.internal.kernel;
 
+import static javax.jcr.Property.JCR_DESCRIPTION;
+import static javax.jcr.Property.JCR_LAST_MODIFIED;
+import static javax.jcr.Property.JCR_TITLE;
+import static org.argeo.cms.CmsTypes.CMS_IMAGE;
+
 import java.io.IOException;
+import java.io.PrintWriter;
+import java.security.PrivilegedExceptionAction;
 import java.security.cert.X509Certificate;
+import java.util.Calendar;
 import java.util.Enumeration;
 
+import javax.jcr.Node;
+import javax.jcr.NodeIterator;
+import javax.jcr.Repository;
+import javax.jcr.RepositoryException;
+import javax.jcr.Session;
+import javax.security.auth.Subject;
 import javax.servlet.FilterChain;
 import javax.servlet.ServletException;
+import javax.servlet.http.HttpServlet;
 import javax.servlet.http.HttpServletRequest;
 import javax.servlet.http.HttpServletResponse;
 import javax.servlet.http.HttpSession;
@@ -13,30 +28,37 @@ import javax.servlet.http.HttpSession;
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
 import org.argeo.cms.CmsException;
+import org.argeo.cms.util.CmsUtils;
 import org.argeo.jcr.ArgeoJcrConstants;
+import org.argeo.jcr.JcrUtils;
 import org.eclipse.equinox.http.servlet.ExtendedHttpService;
 
 /**
  * Intercepts and enriches http access, mainly focusing on security and
  * transactionality.
  */
-@Deprecated
 class NodeHttp implements KernelConstants, ArgeoJcrConstants {
        private final static Log log = LogFactory.getLog(NodeHttp.class);
 
        // Filters
-       private final RootFilter rootFilter;
+       // private final RootFilter rootFilter;
 
        // private final DoSFilter dosFilter;
        // private final QoSFilter qosFilter;
 
-       NodeHttp(ExtendedHttpService httpService) {
-               rootFilter = new RootFilter();
+       private Repository repository;
+
+       NodeHttp(ExtendedHttpService httpService, NodeRepository node) {
+               this.repository = node;
+               // rootFilter = new RootFilter();
                // dosFilter = new CustomDosFilter();
                // qosFilter = new QoSFilter();
 
                try {
-                       httpService.registerFilter("/", rootFilter, null, null);
+                       httpService.registerServlet("/!", new LinkServlet(repository),
+                                       null, null);
+                       httpService.registerServlet("/robots.txt", new RobotServlet(),
+                                       null, null);
                } catch (Exception e) {
                        throw new CmsException("Cannot register filters", e);
                }
@@ -45,6 +67,165 @@ class NodeHttp implements KernelConstants, ArgeoJcrConstants {
        public void destroy() {
        }
 
+       static class LinkServlet extends HttpServlet {
+               private static final long serialVersionUID = 3749990143146845708L;
+               private final Repository repository;
+
+               public LinkServlet(Repository repository) {
+                       this.repository = repository;
+               }
+
+               @Override
+               protected void service(HttpServletRequest request,
+                               HttpServletResponse response) throws ServletException,
+                               IOException {
+                       String path = request.getPathInfo();
+                       String userAgent = request.getHeader("User-Agent").toLowerCase();
+                       boolean isBot = false;
+                       boolean isCompatibleBrowser = false;
+                       if (userAgent.contains("bot") || userAgent.contains("facebook")
+                                       || userAgent.contains("twitter")) {
+                               isBot = true;
+                       } else if (userAgent.contains("webkit")
+                                       || userAgent.contains("gecko")
+                                       || userAgent.contains("firefox")
+                                       || userAgent.contains("msie")
+                                       || userAgent.contains("chrome")
+                                       || userAgent.contains("chromium")
+                                       || userAgent.contains("opera")
+                                       || userAgent.contains("browser")) {
+                               isCompatibleBrowser = true;
+                       }
+
+                       if (isBot) {
+                               log.warn("# BOT " + request.getHeader("User-Agent"));
+                               canonicalAnswer(request, response, path);
+                               return;
+                       }
+
+                       if (isCompatibleBrowser && log.isTraceEnabled())
+                               log.trace("# BWS " + request.getHeader("User-Agent"));
+                       redirectTo(response, "/#" + path);
+               }
+
+               private void redirectTo(HttpServletResponse response, String location) {
+                       response.setHeader("Location", location);
+                       response.setStatus(HttpServletResponse.SC_FOUND);
+               }
+
+               // private boolean canonicalAnswerNeededBy(HttpServletRequest request) {
+               // String userAgent = request.getHeader("User-Agent").toLowerCase();
+               // return userAgent.startsWith("facebookexternalhit/");
+               // }
+
+               /** For bots which don't understand RWT. */
+               private void canonicalAnswer(HttpServletRequest request,
+                               HttpServletResponse response, String path) {
+                       Session session = null;
+                       try {
+                               PrintWriter writer = response.getWriter();
+                               session = Subject.doAs(KernelUtils.anonymousLogin(),
+                                               new PrivilegedExceptionAction<Session>() {
+
+                                                       @Override
+                                                       public Session run() throws Exception {
+                                                               return repository.login();
+                                                       }
+
+                                               });
+                               Node node = session.getNode(path);
+                               String title = node.hasProperty(JCR_TITLE) ? node.getProperty(
+                                               JCR_TITLE).getString() : node.getName();
+                               String desc = node.hasProperty(JCR_DESCRIPTION) ? node
+                                               .getProperty(JCR_DESCRIPTION).getString() : null;
+                               Calendar lastUpdate = node.hasProperty(JCR_LAST_MODIFIED) ? node
+                                               .getProperty(JCR_LAST_MODIFIED).getDate() : null;
+                               String url = CmsUtils.getCanonicalUrl(node, request);
+                               String imgUrl = null;
+                               loop: for (NodeIterator it = node.getNodes(); it.hasNext();) {
+                                       // Takes the first found cms:image
+                                       Node child = it.nextNode();
+                                       if (child.isNodeType(CMS_IMAGE)){
+                                               imgUrl = CmsUtils.getDataUrl(child, request);
+                                               break loop;
+                                       }
+                               }
+                               StringBuilder buf = new StringBuilder();
+                               buf.append("<html>");
+                               buf.append("<head>");
+                               writeMeta(buf, "og:title", title);
+                               writeMeta(buf, "og:type", "website");
+                               buf.append("<meta name='twitter:card' content='summary' />");
+                               buf.append("<meta name='twitter:site' content='@argeo_org' />");
+                               writeMeta(buf, "og:url", url);
+                               if (desc != null)
+                                       writeMeta(buf, "og:description", desc);
+                               if (imgUrl != null)
+                                       writeMeta(buf, "og:image", imgUrl);
+                               if (lastUpdate != null)
+                                       writeMeta(buf, "og:updated_time",
+                                                       Long.toString(lastUpdate.getTime().getTime()));
+                               buf.append("</head>");
+                               buf.append("<body>");
+                               buf.append(
+                                               "<p><b>!! This page is meant for indexing robots, not for real people,"
+                                                               + " visit <a href='/#").append(path)
+                                               .append("'>").append(title)
+                                               .append("</a> instead.</b></p>");
+                               writeCanonical(buf, node);
+                               buf.append("</body>");
+                               buf.append("</html>");
+                               writer.print(buf.toString());
+
+                               response.setHeader("Content-Type", "text/html");
+                               writer.flush();
+                       } catch (Exception e) {
+                               throw new CmsException("Cannot write canonical answer", e);
+                       } finally {
+                               JcrUtils.logoutQuietly(session);
+                       }
+               }
+
+               private void writeMeta(StringBuilder buf, String tag, String value) {
+                       buf.append("<meta property='").append(tag).append("' content='")
+                                       .append(value).append("'/>");
+               }
+
+               private void writeCanonical(StringBuilder buf, Node node)
+                               throws RepositoryException {
+                       buf.append("<div>");
+                       if (node.hasProperty(JCR_TITLE))
+                               buf.append("<p>")
+                                               .append(node.getProperty(JCR_TITLE).getString())
+                                               .append("</p>");
+                       if (node.hasProperty(JCR_DESCRIPTION))
+                               buf.append("<p>")
+                                               .append(node.getProperty(JCR_DESCRIPTION).getString())
+                                               .append("</p>");
+                       NodeIterator children = node.getNodes();
+                       while (children.hasNext()) {
+                               writeCanonical(buf, children.nextNode());
+                       }
+                       buf.append("</div>");
+               }
+       }
+
+       class RobotServlet extends HttpServlet {
+               private static final long serialVersionUID = 7935661175336419089L;
+
+               @Override
+               protected void service(HttpServletRequest request,
+                               HttpServletResponse response) throws ServletException,
+                               IOException {
+                       PrintWriter writer = response.getWriter();
+                       writer.append("User-agent: *\n");
+                       writer.append("Disallow:\n");
+                       response.setHeader("Content-Type", "text/plain");
+                       writer.flush();
+               }
+
+       }
+
        /** Intercepts all requests. Authenticates. */
        class RootFilter extends HttpFilter {