Changeset 7734

Timestamp:

2004-07-13T10:57:05+12:00 (20 years ago)

Author:

kjdon

Message:

commiting John T's changes (c/- DL Consulting) to stop phind hanging the browser if the gsdl server goes boom

Location:

trunk/gsdl

Files:

• bin/java/Phind.jar (modified) ( previous)
• src/java/org/nzdl/gsdl/Phind/Phind.java (modified) (6 diffs)
• src/recpt/phindaction.cpp (modified) (25 diffs)
• src/recpt/phindaction.h (modified) (6 diffs)

trunk/gsdl/src/java/org/nzdl/gsdl/Phind/Phind.java

@@ -90 +90 @@
 
 import java.net.Socket;
+import java.net.SocketTimeoutException;
 import java.net.InetAddress;
 import java.net.UnknownHostException;
@@ -187 +188 @@
     //    ignored.
 
-  
+    /** We're adding a timeout to the loop used to read the result from the
+     *  server - partially because of the 'fails to return anything' death
+     *  spiral - but also because its a good idea anyway. Why DataInputStream
+     *  doesn't have a TimeOutException, given its most often used to get data
+     *  from a URL, I don't know. Lets default it to 60 seconds.
+     *  DLConsulting 12-07-2004
+     */
+    private int timeout = 60;
+
     public String getAppletInfo() {
     return "Phind by Gordon Paynter ([email protected]). Copyright 1997-2000.";
@@ -457 +466 @@
     // Send the query to the phindcgi program 
     System.out.println("sending query: " + query);
+    /** Start timeout counter.
+     *  DLConsulting 12-07-2004
+     */
+    int time = 0;
     try {
         URL phindcgi = new URL(query);
@@ -467 +480 @@
         if (availableBytes == 0) {
             // if i had threads i'd do a wait here for 1 second
+            /** Ah-ha but we do have threads here - at least one thread 
+             *  anyway! Note that we want to prevent interrupted 
+             *  exceptions flinging us out of the read loop.
+             *  We also check that the time count hasn't reached out
+             *  timeout threshold yet. If so, make use of the nifty
+             *  execption handling.
+             *  DLConsulting 12-07-2004
+             *  Start...
+             */
+            if(time >= timeout) {
+            throw new SocketTimeoutException(
+                "No data recieved in " + time + 
+                " seconds. Connection timed-out.");
+            }
+            try {
+            Thread.sleep(1000);
+            }
+            catch(InterruptedException exception) {
+            System.err.println("Unexpected InterruptedException. Non-fatal.");
+            exception.printStackTrace();
+            }
+            time++;
+            /** ...End */
         } else {
             buffer = new byte[availableBytes];
@@ -685 +721 @@
     // expansion_bg = parameterValue("expansion_bg", Color.white);
     // document_bg = parameterValue("document_bg", Color.white);
+
+    /** How long should the read loop wait before timing out?
+     *  DLConsulting 12-07-2004
+     */
+    timeout = parameterValue("timeout", 60);
     }
 
@@ -917 +958 @@
 
 }
+

trunk/gsdl/src/recpt/phindaction.cpp

@@ -181 +181 @@
 
     DocNumArray result;
-    find_phrase_number_from_word(basepath, word, result);
+    /** In order to prevent browser crashing problems, any method which
+     *  previously suffered a silent fatal error, now instead returns false
+     *  to indicate a fatal error has occured. We can then dispatch an
+     *  appropriate error tag to the Phind applet (rather than leave it
+     *  whiling away the milliseconds until the end of existence - or at
+     *  least your browser - in an infinite loop!)
+     *  DLConsulting 12-07-2004
+     */
+    if(!find_phrase_number_from_word(basepath, word, result)) {
+    output_error("phindaction: Fatal Error! Couldn't load index information in find_phrase_number_from_word()",
+             textout, outconvert, disp, logout, XMLmode);
+    return true;
+    }
     
     if (result.empty()) {
@@ -205 +217 @@
   if (!textdata.LoadData (base, fullpathc)) {
     //    FatalError (1, "Couldn't load text information for \"%s\"", fullpathc);
-    exit (0);
+    //exit(0);
+    /** We must return something to the client, whether this error is fatal or 
+     *  no, otherwise we risk sending their browser into an infinite loop!
+     *  DLConsulting 12-07-2004
+     */
+    output_error("phindaction: Fatal Error! Couldn't load text information for collection",
+         textout, outconvert, disp, logout, XMLmode);
+    return true;
   }
 
   delete []fullpathc;
 
-  get_phrase_all_data(textdata, phrase, word, tf, ef, lf, df, el, 
-              linkdest, linktype, docNums, docfreq);
+  /** Another previously silent method can now cry out.
+   *  DLConsulting 12-07-2004
+   */
+  if(!get_phrase_all_data(textdata, phrase, word, tf, ef, lf, df, el, 
+              linkdest, linktype, docNums, docfreq)) {
+    output_error(
+      "phindaction: Fatal Error! Couldn't parse phrase in get_phrase_all_data()",
+      textout, outconvert, disp, logout, XMLmode);
+      return true;    
+  }
 
   // Output the header
@@ -243 +270 @@
           << "\" start=\"" << first_l
           << "\" end=\"" << last_l << "\">\n";
-      print_thesaurus_links(args["c"], XMLmode, textdata, linkdest, linktype,
-                first_l, last_l, disp, outconvert, textout);
+      /** DLConsulting 12-07-2004 */
+      if(!print_thesaurus_links(args["c"], XMLmode, textdata, linkdest, linktype,
+                   first_l, last_l, disp, outconvert, textout)) {
+    output_error(
+             "phindaction: Fatal Error! Couldn't get phrase in get_phrase_freq_data()",
+             textout, outconvert, disp, logout, XMLmode);
+    return true;    
+      }
       textout << "</thesauruslist>\n";
     }
@@ -257 +290 @@
 
       textout << "<p><table border=1><tr><th>type</th><th>topic</th><th>freq</th><th>docs</th></tr>\n";
-      print_thesaurus_links(args["c"], XMLmode, textdata, linkdest, linktype, 
-                first_l, last_l, disp, outconvert, textout);
+      /** DLConsulting 12-07-2004 */
+      if(!print_thesaurus_links(args["c"], XMLmode, textdata, linkdest, linktype, 
+                   first_l, last_l, disp, outconvert, textout)) {
+    output_error(
+             "phindaction: Fatal Error! Couldn't get phrase in get_phrase_freq_data()",
+             textout, outconvert, disp, logout, XMLmode);
+    return true;
+      }
       textout << "</table>\n";
 
@@ -368 +407 @@
           << "\" end=\"" << last_d << "\">\n";
       
-      print_documents(XMLmode, basepath, args["c"], docNums, docfreq,
-              first_d, last_d, disp, outconvert, textout);
+      if(!print_documents(XMLmode, basepath, args["c"], docNums, docfreq,
+              first_d, last_d, disp, outconvert, textout)) {
+    output_error(
+             "phindaction: Fatal Error! Couldn't load text information in print_documents() or get_document_all_data()",
+             textout, outconvert, disp, logout, XMLmode);
+    return true;    
+      }
 
       textout << "</documentlist>\n";
@@ -384 +428 @@
 
       textout << "<p><table border=1><tr><th align=left>document</th><th>freq</th></tr>\n";
-      print_documents(XMLmode, basepath, args["c"], docNums, docfreq, 
-              first_d, last_d, disp, outconvert, textout);
+      if(!print_documents(XMLmode, basepath, args["c"], docNums, docfreq, 
+              first_d, last_d, disp, outconvert, textout)) {
+    output_error(
+             "phindaction: Fatal Error! Couldn't load text information in print_documents()",
+             textout, outconvert, disp, logout, XMLmode);
+    return true;
+      }
       textout << "</table>\n";
       
@@ -430 +479 @@
 
 // Find the phrase number of a word in the index file
-void phindaction::find_phrase_number_from_word(const text_t &basepath, 
+bool phindaction::find_phrase_number_from_word(const text_t &basepath, 
                            const text_t &query, 
                            DocNumArray &result) {
@@ -447 +496 @@
   if (!indexData.LoadData (base, fullpathc)) {
     //    FatalError (1, "Couldn't load index information for \"%s\"", fullpathc);
-    exit (0);
+    //exit(0);
+    /** Don't handle fatal errors here anymore.
+     *  DLConsulting 12-07-2004
+     */
+    return false; // Indicates something very bad has happened
   }
 
@@ -478 +531 @@
 
   indexData.UnloadData();
+
+  /** This method now returns a boolean, so...
+   *  DLConsulting 12-07-2004
+   */
+  return true; // Indicates that what happened is all good, baby.
 }
 
@@ -492 +550 @@
 //   ll - the list of phrases that are thesaurus links 
 //   dl - the list of documents that contain phrase
-void phindaction::get_phrase_all_data(TextData &textdata, unsigned long phrase,
+bool phindaction::get_phrase_all_data(TextData &textdata, unsigned long phrase,
                       text_t &word, unsigned long &tf, unsigned long &ef,
                       unsigned long &lf, unsigned long &df,
@@ -507 +565 @@
   if (!GetDocText (textdata, docLevel, phrase, text)) {
     //    FatalError (1, "Error while trying to get phrase %u", phrase);
-    exit (0);
+    //exit(0);
+    return false; // Something very bad has happened.
   }
 
@@ -630 +689 @@
     }
   }
-}
-
-void phindaction::print_thesaurus_links(const text_t &collection, bool XMLmode, 
+
+  return true; // Indicates that what happened is all good, baby.
+}
+
+bool phindaction::print_thesaurus_links(const text_t &collection, bool XMLmode, 
                     TextData &textdata, vector <unsigned long> &linkdest,
                     vector <UCArray> &linktype, unsigned long first,
@@ -647 +708 @@
     phrase = linkdest[l];
     type = linktype[l];
-    get_phrase_freq_data(textdata, phrase, text, tf, ef, df);
+
+    /** DLConsulting 12-07-2004 */
+    if(!get_phrase_freq_data(textdata, phrase, text, tf, ef, df)) {
+      return false;
+    }
     
     if (XMLmode) {
@@ -665 +730 @@
     }
   }
+
+  /** DLConsulting 12-07-2004 */
+  return true;
 }
 
@@ -675 +743 @@
 //   ef - the expansion frequency of the phrase
 //   df - the document frequency of the phrase
-void phindaction::get_phrase_freq_data(TextData &textdata, unsigned long phrase,
+/**
+ *   Returns:
+ *     false if the method suffered a fatal error, true otherwise
+ */
+bool phindaction::get_phrase_freq_data(TextData &textdata, unsigned long phrase,
                        UCArray &word, unsigned long &tf, 
                        unsigned long &ef, unsigned long &df) {
@@ -686 +758 @@
   if (!GetDocText (textdata, docLevel, phrase, text)) {
     //    FatalError (1, "Error while trying to get phrase %u", phrase);
-    exit (0);
+    //exit(0);
+    /** DLConsulting 12-07-2004 */
+    return false;
   }
 
@@ -719 +793 @@
     df += (*next - '0');
   }
+
+  /** DLConsulting 12-07-2004 */
+  return true;
 }
 
@@ -844 +921 @@
 }
 
-void phindaction::print_documents(bool XMLmode, const text_t &basepath, 
+bool phindaction::print_documents(bool XMLmode, const text_t &basepath, 
                   const text_t &collection,
                   const vector <unsigned long> &docNums, 
@@ -865 +942 @@
   if (!docdata.LoadData (base, fullpathc)) {
     //    FatalError (1, "Couldn't load text information for \"%s\"", fullpathc);
-    exit (0);
+    //exit(0);
+    /** DLConsulting 12-07-2004 */
+    return false;
   }
 
@@ -877 +956 @@
     freq = docFreq[d];
 
-    get_document_all_data(docdata, doc, title, hash);
+    /** DLConsulting 13-07-2004 */
+    if(!get_document_all_data(docdata, doc, title, hash)) {
+      return false;
+    }
 
     if (XMLmode) {
@@ -894 +976 @@
 
   docdata.UnloadData();
+
+  /** DLConsulting 12-07-2004 */
+  return true;
 }
 
@@ -902 +987 @@
 //   title - the document's title
 //   hash - the document's unique OID
-void phindaction::get_document_all_data(TextData &docdata, unsigned long docNum,
+/** Returns:
+ *   false if a fatal error occured, true otherwise
+ *  DLConsulting 12-07-2004
+ */
+bool phindaction::get_document_all_data(TextData &docdata, unsigned long docNum,
                     UCArray &title, UCArray &hash) {
 
@@ -912 +1001 @@
   if (!GetDocText (docdata, docLevel, docNum, text)) {
     //    FatalError (1, "Error while trying to get document %u", docNum);
-    exit (0);
+    //exit(0);
+    /** DLConsulting 13-07-2004 */
+    return false;
   }
 
@@ -931 +1022 @@
     title.push_back(*next);
   }
+
+  /** DLConsulting 13-07-2004 */
+  return true;
 }
 
@@ -966 +1060 @@
 
 #endif //GSDL_USE_PHIND_ACTION
+

trunk/gsdl/src/recpt/phindaction.h

@@ -45 +45 @@
 protected:
 
-  void find_phrase_number_from_word(const text_t &basepath, 
+  bool find_phrase_number_from_word(const text_t &basepath, 
                     const text_t &query, 
                     DocNumArray &result);
 
-  void get_phrase_all_data(TextData &textdata, unsigned long phrase,
+  bool get_phrase_all_data(TextData &textdata, unsigned long phrase,
                text_t &word, unsigned long &tf, unsigned long &ef,
                unsigned long &lf, unsigned long &df,
@@ -58 +58 @@
                vector <unsigned long> &docfrq);
 
-  void print_thesaurus_links(const text_t &collection, bool XMLmode, 
+  bool print_thesaurus_links(const text_t &collection, bool XMLmode, 
                  TextData &textdata, vector <unsigned long> &linkdest,
                  vector <UCArray> &linktype, unsigned long first,
@@ -64 +64 @@
                  outconvertclass &outconvert, ostream &textout);
 
-  void get_phrase_freq_data(TextData &textdata, unsigned long phrase,
+  bool get_phrase_freq_data(TextData &textdata, unsigned long phrase,
                 UCArray &word, unsigned long &tf, 
                 unsigned long &ef, unsigned long &df);
@@ -81 +81 @@
             UCArray::const_iterator end);
 
-  void print_documents(bool XMLmode, const text_t &basepath, 
+  bool print_documents(bool XMLmode, const text_t &basepath, 
                const text_t &collection,
                const vector <unsigned long> &docNums, 
@@ -89 +89 @@
                ostream &textout);
 
-  void get_document_all_data(TextData &docdata, unsigned long docNum,
+  bool get_document_all_data(TextData &docdata, unsigned long docNum,
                  UCArray &title, UCArray &hash);
 
@@ -120 +120 @@
 #endif //!PHINDACTION_H
 #endif //GSDL_USE_PHIND_ACTION
+