(no commit message)

[utils] / crawler / kissweb / src / org / wamblee / crawler / kiss / scheduling / CrawlerSchedule.java

diff --git a/crawler/kissweb/src/org/wamblee/crawler/kiss/scheduling/CrawlerSchedule.java b/crawler/kissweb/src/org/wamblee/crawler/kiss/scheduling/CrawlerSchedule.java
index 09ada84da8d4f9bf0411ef1b6cb228a252ce4f27..e341169a82e76e5cad7da3f6cacda4b7cc84097b 100644 (file)
--- a/crawler/kissweb/src/org/wamblee/crawler/kiss/scheduling/CrawlerSchedule.java
+++ b/crawler/kissweb/src/org/wamblee/crawler/kiss/scheduling/CrawlerSchedule.java
@@ -20,6 +20,9 @@ import java.io.Serializable;
 import java.util.Calendar;
 import java.util.Date;
 
+import org.apache.commons.logging.Log;
+import org.apache.commons.logging.LogFactory;
+
 /**
  * This class encapsulates the logic for deciding whether to
  * run the crawler. This provides the mechanism to keep the 
@@ -29,9 +32,12 @@ import java.util.Date;
  */
 public class CrawlerSchedule implements Serializable {
     
+    private static final Log LOG = LogFactory.getLog(CrawlerSchedule.class);
+    
     private CrawlerExecutor _crawler; 
     private Date _lastExecuted;
-    private Exception _lastResult;
+    private boolean _lastResult;
+    private Exception _lastException;
     private int _hourMin; 
     private int _hourMax;
     
@@ -44,7 +50,8 @@ public class CrawlerSchedule implements Serializable {
     public CrawlerSchedule(CrawlerExecutor aCrawler, int aHourMin, int aHourMax) { 
         _crawler = aCrawler; 
         _lastExecuted = null; 
-        _lastResult = null; 
+        _lastResult = false;
+        _lastException = null; 
         _hourMin = aHourMin;
         _hourMax = aHourMax;
     }
@@ -56,15 +63,18 @@ public class CrawlerSchedule implements Serializable {
      */
     public void execute(Date aDate) { 
         if (mustExecute(aDate)) { 
-            try { 
-                _lastResult = null; 
+            LOG.info("Executing crawler at " + aDate);
+            try {  
                 _crawler.execute(aDate);
+                _lastResult = true; 
+                _lastException = null; 
             } catch (Exception e) {
-                _lastResult = e; 
+                _lastResult = false; 
+                _lastException = e; 
             } finally { 
                 _lastExecuted = aDate;
             }
-        }
+        } 
     }
     
     /**
@@ -77,11 +87,19 @@ public class CrawlerSchedule implements Serializable {
     
     /**
      * Gets the result of the last execution. 
+     * @return True iff last execution was a success. 
+     */
+    public boolean getLastResult() { 
+        return _lastResult; 
+    }
+    
+    /**
+     * Gets the exception thrown by the last execution. 
      * @return null if the last execution was successful or an exception 
      *   otherwise.
      */
-    public Exception getLastResult() { 
-        return _lastResult; 
+    public Exception getLastException() { 
+        return _lastException; 
     }
     
     /**
@@ -99,13 +117,32 @@ public class CrawlerSchedule implements Serializable {
         if (hour > _hourMax ) { 
             return false; 
         }
-        if ( hour == _hourMin ) { 
+     
+        if ( !lastExecutionWasOnSameDay(aDate)) { 
             return true; // First execution of today. 
-        }
-        if ( _lastResult != null ) { 
+        } 
+        // last execution was on the same day.
+        if ( !_lastResult ) { 
             return true; // last execution of today was unsuccessful, retry. 
         }
-        
         return false; // already run successfully today.
     }
+    
+    private boolean lastExecutionWasOnSameDay(Date aDate) {
+        if ( _lastExecuted == null ) { 
+            return false;
+        }
+        int curDay = getDayOfYear(aDate);
+        int lastDay = getDayOfYear(_lastExecuted);
+        return curDay == lastDay; 
+    }
+
+    /**
+     * @param aDate
+     */
+    private int getDayOfYear(Date aDate) {
+        Calendar calendar = Calendar.getInstance();
+        calendar.setTime(aDate);
+        return calendar.get(Calendar.DAY_OF_YEAR);
+    }
 }