source: greenstone3/trunk/src/java/org/greenstone/gsdl3/LibraryServlet.java@ 16688

package org.greenstone.gsdl3;

import org.greenstone.gsdl3.comms.*;
import org.greenstone.gsdl3.core.*;
import org.greenstone.gsdl3.util.*;
import org.greenstone.gsdl3.action.PageAction; // used to get the default action
import org.w3c.dom.Document;
import org.w3c.dom.Element;
import org.w3c.dom.Node;
import org.w3c.dom.NodeList;
import java.io.*;
import javax.servlet.*;
import javax.servlet.http.*;
import java.util.Enumeration;
import java.util.ArrayList;
import java.util.HashMap;
import java.io.File;
import java.util.Hashtable;
import org.apache.log4j.*;


/** a servlet to serve the greenstone library - we are using servlets instead
 * of cgi
 * the init method is called only once - the first time the servlet classes
 * are loaded. Each time a request comes in to the servlet, the session()
 * method is called in a new thread (calls doGet/doPut etc)
 * takes the a=p&p=home type args and builds a simple request to send to
 * its receptionist, which returns a result in html, cos output=html
 * is set in the request
 * 
 * 18/Jul/07 xiao
 * modify to make the cached parameters collection-specific.
 * Most of the work is done in doGet(), except adding an inner class UserSessionCache.
 * 
 * @see Receptionist
 */
public class LibraryServlet extends HttpServlet {
  
  /** the receptionist to send messages to */
  protected Receptionist recept=null;
  /** the default language - is specified by setting a servlet param,
   * otherwise DEFAULT_LANG is used*/
  protected String default_lang= null;
  /** The default default - used if a default lang is not specified
   * in the servlet params */
  protected final String DEFAULT_LANG = "en";
  /** container Document to create XML Nodes */
  protected Document doc=null;
  /** a converter class to parse XML and create Docs */
  protected XMLConverter converter=null;
  /** the cgi stuff - the Receptionist can add new args to this
   *
   * its used by the servlet to determine what args to save */
  protected GSParams params = null;
  
  /** user id - new one per session. This doesn't work if session state is saved between restarts - this requires this value to be saved too. */
  protected int next_user_id = 0;

  /** a hash that contains all the active session IDs mapped to the cached items
   *  It is updated whenever the whole site or a particular collection is reconfigured
   *  using the command a=s&sa=c or a=s&sa=c&c=xxx
   *  It is in the form: sid -> (UserSessionCache object)
   */
  protected Hashtable session_ids_table = new Hashtable();
  //this name is combined with the collection name and used for caching 
  protected String valid_site_name = "";
  
  /** the maximum interval that the cached info remains in session_ids_table (in seconds) 
   *  This is set in web.xml
   */
  protected int session_expiration = 1800;
  
  static Logger logger = Logger.getLogger(org.greenstone.gsdl3.LibraryServlet.class.getName());
  
  /** initialise the servlet
   */
  public void init(ServletConfig config) throws ServletException {
    // always call super.init;
    super.init(config);
    // disable preferences - does this work anyway??
    //System.setProperty("java.util.prefs.PreferencesFactory", "org.greenstone.gsdl3.util.DisabledPreferencesFactory");
    
    
    String library_name = config.getInitParameter(GSConstants.LIBRARY_NAME);
    String gsdl3_home = config.getInitParameter(GSConstants.GSDL3_HOME);
    String interface_name = config.getInitParameter(GSConstants.INTERFACE_NAME);
    this.default_lang = config.getInitParameter(GSConstants.DEFAULT_LANG);
    String sess_expire = config.getInitParameter(GSXML.SESSION_EXPIRATION);
    if (sess_expire != null && !sess_expire.equals("")) {
        this.session_expiration = Integer.parseInt(sess_expire);
    }
    
    if (library_name == null || interface_name ==null) {
      // must have this
      System.err.println("initialisation parameters not all set!");
      System.err.println(" you must have libraryname and interfacename");
      System.exit(1);
    }
    
    String site_name = config.getInitParameter(GSConstants.SITE_NAME);
    String remote_site_name = null;
    String remote_site_type = null;
    String remote_site_address = null;
    
    if (site_name == null) {
      // no site, try for communicator
      remote_site_name = config.getInitParameter("remote_site_name");
      remote_site_type = config.getInitParameter("remote_site_type");
      remote_site_address = config.getInitParameter("remote_site_address");
      if (remote_site_name == null || remote_site_type == null || remote_site_address == null) {
        System.err.println("initialisation paramters not all set!");
        System.err.println("if site_name is not set, then you must have remote_site_name, remote_site_type and remote_site_address set");
        System.exit(1);
      }
    }
    valid_site_name = (site_name != null)? site_name : remote_site_name;
    
    if (this.default_lang == null) {
      // choose english
      this.default_lang = DEFAULT_LANG;
    }
    
    HashMap config_params = new HashMap();
    
    config_params.put(GSConstants.LIBRARY_NAME, library_name);
    config_params.put(GSConstants.INTERFACE_NAME, interface_name);
    if (site_name != null) {
      config_params.put(GSConstants.SITE_NAME, site_name);
    }
    this.converter = new XMLConverter();
    this.doc = this.converter.newDOM();
    
    // the receptionist -the servlet will talk to this
    String recept_name = (String)config.getInitParameter("receptionist_class");
    if (recept_name == null) {
      this.recept = new DefaultReceptionist();
    } else {
      try {
        this.recept = (Receptionist)Class.forName("org.greenstone.gsdl3.core."+recept_name).newInstance();
      } catch (Exception e) { // cant use this new one, so use normal one
        System.err.println("LibraryServlet configure exception when trying to use a new Receptionist "+recept_name+": "+e.getMessage());
        e.printStackTrace();
        this.recept = new DefaultReceptionist();
      }
    }
    this.recept.setConfigParams(config_params);
    
    // the receptionist uses a MessageRouter or Communicator to send its requests to. We either create a MessageRouter here for the designated site (if site_name set), or we create a Communicator for a remote site. The is given to teh Receptionist, and the servlet never talks to it again.directly.
    if (site_name != null) {
      String mr_name = (String)config.getInitParameter("messagerouter_class");
      MessageRouter message_router = null;
      if (mr_name == null) { // just use the normal MR
        message_router = new MessageRouter();
      } else { // try the specified one
        try {
          message_router = (MessageRouter)Class.forName("org.greenstone.gsdl3.core."+mr_name).newInstance();
        } catch (Exception e) { // cant use this new one, so use normal one
          System.err.println("LibraryServlet configure exception when trying to use a new MessageRouter "+mr_name+": "+e.getMessage());
          e.printStackTrace();
          message_router = new MessageRouter();
        }
      }
      
      message_router.setSiteName(site_name);
      message_router.setLibraryName(library_name);
      message_router.configure();
      this.recept.setMessageRouter(message_router);
    } else {
      // talking to a remote site, create a communicator
      Communicator communicator = null;
      // we need to create the XML to configure the communicator
      Element site_elem = this.doc.createElement(GSXML.SITE_ELEM);
      site_elem.setAttribute(GSXML.TYPE_ATT, remote_site_type);
      site_elem.setAttribute(GSXML.NAME_ATT, remote_site_name);
      site_elem.setAttribute(GSXML.ADDRESS_ATT, remote_site_address);
      
      if (remote_site_type.equals(GSXML.COMM_TYPE_SOAP_JAVA)) {
        communicator = new SOAPCommunicator();
      } else {
        System.err.println("LibraryServlet.init Error: invalid Communicator type: "+remote_site_type);
        System.exit(1);
      }
      
      if (!communicator.configure(site_elem)) {
        System.err.println("LibraryServlet.init Error: Couldn't configure communicator");
        System.exit(1);
      }
      this.recept.setMessageRouter(communicator);
    }
    
    // the params arg thingy
    
    String params_name = (String)config.getInitParameter("params_class");
    if (params_name == null) {
      this.params = new GSParams();
    } else {
      try {
        this.params = (GSParams)Class.forName("org.greenstone.gsdl3.util."+params_name).newInstance();
      } catch (Exception e) {
        System.err.println("LibraryServlet configure exception when trying to use a new params thing "+params_name+": "+e.getMessage());
        e.printStackTrace();
        this.params = new GSParams();
      }
    }
    // pass it to the receptionist
    this.recept.setParams(this.params);
    this.recept.configure();
    
  }
  
  
  private void logUsageInfo(HttpServletRequest request){
    String usageInfo = "";
    
    //session-info: get params stored in the session
    HttpSession session = request.getSession(true);
    Enumeration attributeNames = session.getAttributeNames();
    while(attributeNames.hasMoreElements()) {
      String name = (String)attributeNames.nextElement();
      usageInfo +=name+"="+session.getAttribute(name)+" ";
    }
    
    //logged info = general-info + session-info
    usageInfo =
      request.getServletPath()+" "+ //serlvet
      "["+request.getQueryString()+"]" +" "+ //the query string
      "["+usageInfo.trim()+"]" +" "+ // params stored in a session
      request.getRemoteAddr()+" "+   //remote address
      request.getRequestedSessionId()+" "+ //session id
      request.getHeader("user-agent")+" "; //the remote brower info
    
    logger.info(usageInfo);
    
  }
  
  public class UserSessionCache implements HttpSessionBindingListener {

    String session_id = "";
    
    /** a hash that maps the session ID to a hashtable that maps the coll_name to its parameters 
     *  coll_name -> Hashtable (param_name -> param_value)
     */
    protected Hashtable coll_name_params_table = null;
    
    public UserSessionCache(String id, Hashtable table) {
      session_id = id;      
      coll_name_params_table = (table == null)? new Hashtable() : table;      
    }

    protected void cleanupCache(String coll_name) {
      if (coll_name_params_table.containsKey(coll_name)) {
          coll_name_params_table.remove(coll_name);
      }
    }
    protected Hashtable getParamsTable() {
        return coll_name_params_table;
    }
    public void valueBound(HttpSessionBindingEvent event) {
    // Do nothing
    }

    public void valueUnbound(HttpSessionBindingEvent event) {
      if(session_ids_table.containsKey(session_id)) {
        session_ids_table.remove(session_id);
      }
    }
    public int tableSize() {
        return (coll_name_params_table == null)? 0 : coll_name_params_table.size();
    }
  }
  
  public void doGet (HttpServletRequest request,
    HttpServletResponse response)
    throws ServletException, IOException {
    logUsageInfo (request);

    String query_string = request.getQueryString();
    if (query_string!=null){
    String[] query_arr = query_string.split("&");
    boolean redirect = false;
    String href = null;
    String rl = null;
    for (int i=0;i<query_arr.length;i++){
        if (query_arr[i].startsWith("el")){
        if (query_arr[i].substring(query_arr[i].indexOf("=")+1,query_arr[i].length()).equals("direct")){
            redirect = true;
        }
        }else if(query_arr[i].startsWith("href")){
        href = query_arr[i].substring(query_arr[i].indexOf("=")+1,query_arr[i].length());
        href = href.replaceAll("%2f", "/");
        href = href.replaceAll("%7e", "~");
        href = href.replaceAll("%3f", "?");
        href = href.replaceAll("%3A", "\\:");
        }else if(query_arr[i].startsWith("rl")){
        rl = query_arr[i].substring(query_arr[i].indexOf("=")+1,query_arr[i].length());
        }
    }
    //if query_string contains "el=", the web page will be redirected to the external URl, otherwise a greenstone page with an external URL will be displayed
    //"rl=0" this is an external link
    //"rl=1" this is an internal link
    if ((redirect) && (href != null) && (rl.equals("0"))){// This is an external link, the web page is re-directed to the external URL (&el=&rl=0&href="http://...")
        response.setContentType("text/xml");
        response.sendRedirect(href);
    }
    }
    // Nested Diagnostic Configurator to identify the client for
    
    HttpSession session = request.getSession (true);
    session.setMaxInactiveInterval(session_expiration);
    String uid = (String)session.getAttribute (GSXML.USER_ID_ATT);
    if (uid ==null) {
      uid = ""+getNextUserId ();
      session.setAttribute (GSXML.USER_ID_ATT, uid);
    }
    request.setCharacterEncoding ("UTF-8");
    response.setContentType ("text/html;charset=UTF-8");
    PrintWriter out = response.getWriter ();
    
    String lang = request.getParameter (GSParams.LANGUAGE);
    if (lang==null || lang.equals ("")) {
      // try the session cached lang
      lang = (String)session.getAttribute (GSParams.LANGUAGE);
      if (lang==null || lang.equals ("")) {
        // still not set, use the default
        lang = this.default_lang;
      }
    }
    
    // set the lang in the session
    session.setAttribute (GSParams.LANGUAGE, lang);
    
    String output = request.getParameter (GSParams.OUTPUT);
    if (output==null || output.equals ("")) {
      output = "html"; // uses html by default
    }
    
    // the request to the receptionist
    Element xml_message = this.doc.createElement (GSXML.MESSAGE_ELEM);
    Element xml_request = GSXML.createBasicRequest (this.doc, GSXML.REQUEST_TYPE_PAGE, "", lang, uid);
    xml_request.setAttribute (GSXML.OUTPUT_ATT, output);
    xml_message.appendChild (xml_request);
    
    String action = request.getParameter (GSParams.ACTION);
    String subaction = request.getParameter (GSParams.SUBACTION);
    String collection = request.getParameter(GSParams.COLLECTION);

    //specifically we clean up the cache session_ids_table if the two reconfigure command
    //are issued: a=s&sa=c and a=s&sa=c&c=coll_name, in which case there is no caching action to be taken
    boolean should_cache = true;
    if(action != null && action.equals(GSParams.SYSTEM) 
            && subaction != null && subaction.equals(GSParams.CONFIGURE)) {
        if (collection == null || collection.equals("")) {
            //user reconfiugred the whole site, clean up all cached info
            //logger.info("clear cache for the whole site.");
            session_ids_table = new Hashtable(); 
            session.removeAttribute(GSXML.USER_SESSION_CACHE_ATT);
        } else {
            //clean up all cache info related to the collection
            //logger.info("clear cache for collection: " + collection);
            ArrayList cache_list = new ArrayList(session_ids_table.values());
            for (int i=0; i<cache_list.size(); i++) {
                UserSessionCache cache = (UserSessionCache)cache_list.get(i);
                cache.cleanupCache(collection);
            }
        }
        should_cache = false;
    }

    //  logger.info("should_cache= " + should_cache);
    //clear the collection-specific cache in the session, since we have no way to know whether this session is 
    //about the same collection as the last session or not.
    Enumeration attributeNames = session.getAttributeNames();
    while(attributeNames.hasMoreElements()) {
        String name = (String)attributeNames.nextElement();
        if (!name.equals (GSXML.USER_SESSION_CACHE_ATT)
                && !name.equals (GSParams.LANGUAGE) 
                && !name.equals (GSXML.USER_ID_ATT)) {
            session.removeAttribute(name);
        }
    }

    // create a dumy collection name for Authentication 
    if (subaction != null && subaction.equals("authen")){
    collection = "0000000000";
    }
    
    UserSessionCache session_cache = null;
    Hashtable param_table = null;    
    Hashtable table = null;
    String sid = session.getId();
    if (should_cache == true && collection != null && !collection.equals("")) {
        String key_str = valid_site_name + collection;
        if (session_ids_table.containsKey(sid)) {
            session_cache = (UserSessionCache)session_ids_table.get(sid);
            param_table = session_cache.getParamsTable();
            logger.info("collections in table: " + tableToString(param_table));
                if (param_table.containsKey(key_str)) {
                    //logger.info("existing table: " + collection);
                    table = (Hashtable)param_table.get(key_str);
                } else {
                    table = new Hashtable();
                    param_table.put(key_str, table);
                    //logger.info("new table: " + collection);
                }
        } else {
            param_table = new Hashtable();
            table = new Hashtable();
            param_table.put(key_str, table);
            session_cache = new UserSessionCache(sid, param_table);
            session_ids_table.put(sid, session_cache);
        session.setAttribute(GSXML.USER_SESSION_CACHE_ATT, session_cache);
            //logger.info("new session id");
        }
    }

    if (action==null || action.equals ("")) {
      // should we do all the following stuff if using default page?
      // display the home page  - the default page
      action = "p";
      subaction = PageAction.HOME_PAGE;
      
      xml_request.setAttribute (GSXML.ACTION_ATT, action);
      xml_request.setAttribute (GSXML.SUBACTION_ATT, subaction);
      
    } else {
      
      xml_request.setAttribute (GSXML.ACTION_ATT, action);
      if (subaction != null) {
        xml_request.setAttribute (GSXML.SUBACTION_ATT, subaction);
      }
      
      //  create the param list for the greenstone request - includes
      // the params from the current request and any others from the saved session
      Element xml_param_list = this.doc.createElement (GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
      xml_request.appendChild (xml_param_list);
      
      Enumeration params = request.getParameterNames ();
      while(params.hasMoreElements ()) {
        String name = (String)params.nextElement ();
        if (!name.equals (GSParams.ACTION) 
                && !name.equals (GSParams.SUBACTION) 
                && !name.equals (GSParams.LANGUAGE) 
                && !name.equals (GSParams.OUTPUT)) {// we have already dealt with these
          String value="";
          String [] values = request.getParameterValues (name);
          value = values[0];
          if (values.length > 1) {
            for (int i=1; i< values.length; i++) {
              value += ","+values[i];
            }
          }
          // either add it to the param list straight away, or save it to the session and add it later
          if (this.params.shouldSave (name)) {
              if (table != null) {              
                  table.put(name, value);                    
              }
          } else {
            Element param = this.doc.createElement (GSXML.PARAM_ELEM);
            param.setAttribute (GSXML.NAME_ATT, name);
            param.setAttribute (GSXML.VALUE_ATT, GSXML.xmlSafe (value));
            xml_param_list.appendChild (param);
          }
        }
      }
      //put everything in the table into the session
      if (table != null) {
          Enumeration keys = table.keys ();
          while(keys.hasMoreElements ()) {
              String name = (String)keys.nextElement();
              session.setAttribute(name, (String)table.get(name));
          }
      }
      
      // put in all the params from the session cache
      params = session.getAttributeNames ();
      while(params.hasMoreElements ()) {
        String name = (String)params.nextElement ();
        
        if ( !name.equals (GSXML.USER_SESSION_CACHE_ATT) 
                && !name.equals (GSParams.LANGUAGE) 
                && !name.equals (GSXML.USER_ID_ATT)) { 
        // lang and uid are stored but we dont want it in the param list cos its already in the request
          Element param = this.doc.createElement (GSXML.PARAM_ELEM);
          param.setAttribute (GSXML.NAME_ATT, name);
          String value =  GSXML.xmlSafe ((String)session.getAttribute (name));
          // ugly hack to undo : escaping
          value = value.replaceAll ("%3A", "\\:");
          param.setAttribute (GSXML.VALUE_ATT,value);
          xml_param_list.appendChild (param);
        }
      }
    }
    
    if (!output.equals ("html")) {
      response.setContentType ("text/xml"); // for now use text
    }
    
    //GSXML.printXMLNode(xml_message);
    
    Node xml_result = this.recept.process(xml_message);
    encodeURLs (xml_result, response);
    out.println (this.converter.getPrettyString (xml_result));
    
    displaySize(session_ids_table);
  }
  //a debugging method
  private void displaySize(Hashtable table) {
      if(table == null) {
          logger.info("cached table is null");
          return;
      }
      if (table.size() == 0) {
          logger.info("cached table size is zero");
          return;
      }
      int num_cached_coll = 0;
        ArrayList cache_list = new ArrayList(table.values());
        for (int i=0; i<cache_list.size(); i++) {
            num_cached_coll += ((UserSessionCache)cache_list.get(i)).tableSize();           
        }
      logger.info("Number of sessions : total number of cached collection info = " + table.size() + " : " + num_cached_coll);
  }  
  /** merely a debugging method! */
  private String tableToString(Hashtable table) {
     String str = "";
     Enumeration keys = table.keys ();
     while(keys.hasMoreElements ()) {
      String name = (String)keys.nextElement();
      str += name + ", ";
     }
     return str;
  }

  /** this goes through each URL and adds in a session id if needed--
   * its needed if the browser doesn't accept cookies
   * also escapes things if needed
   */
  protected void encodeURLs(Node dataNode, HttpServletResponse response) {
    
    if (dataNode == null) {
      return;
    }

    Element data =null;

    short nodeType = dataNode.getNodeType();
    if (nodeType == Node.DOCUMENT_NODE) {
    Document docNode = (Document)dataNode;
    data = docNode.getDocumentElement() ;
    }
    else {
    data = (Element)dataNode;
    }

    // get all the <a> elements
    NodeList hrefs = data.getElementsByTagName("a");
    for (int i=0; i<hrefs.getLength(); i++) {
      Element a = (Element)hrefs.item(i);
      // ugly hack to get rid of : in the args - interferes with session handling
      String href = a.getAttribute("href");
      if (!href.equals("")) {
        if (href.indexOf("?")!=-1) {
          String[] parts = href.split("\\?", -1);
          parts[1]=parts[1].replaceAll(":", "%3A");
          href = parts[0]+"?"+parts[1];
        }
        a.setAttribute("href", response.encodeURL(href));
      }
    }
    
    // now find any submit bits - get all the <form> elements
    NodeList forms = data.getElementsByTagName("form");
    for (int i=0; i<forms.getLength(); i++) {
      Element form = (Element)forms.item(i);
      form.setAttribute("action", response.encodeURL(form.getAttribute("action")));
    }
    // are these the only cases where URLs occur??
    // we should only do this for greenstone urls?
    
  }
  
  synchronized protected int getNextUserId() {
    next_user_id++;
    return next_user_id;
  }
  
  public void doPost(HttpServletRequest request,
    HttpServletResponse response)
    throws ServletException, IOException {
    doGet(request,response);
    
  }
}