enhanced re-crawl settings

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1960 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/IndexCreate_p.html

@@ -45,9 +45,18 @@ You can define URLs as start points for Web page crawling and start crawling her
   </tr>
   <tr valign="top" class="TableCellLight">
     <td class=small>Re-Crawl Option:</td>
-    <td class=small><input name="crawlingIfOlder" type="text" size="7" maxlength="7" value="#[crawlingIfOlder]#"></td>
     <td class=small>
-    
+    <input type="checkbox" name="crawlingIfOlderCheck" align="top" #(crawlingIfOlderCheck)#::checked#(/crawlingIfOlderCheck)#>
+    <input name="crawlingIfOlderNumber" type="text" size="7" maxlength="7" value="#[crawlingIfOlderNumber]#"></td>
+	Year(s)&nbsp;<input type="radio" name="crawlingIfOlderUnit" value="year"  #(crawlingIfOlderUnitYearCheck)#::checked#(/crawlingIfOlderYearUnitCheck)#>
+	Month(s)&nbsp;<input type="radio" name="crawlingIfOlderUnit" value="month" #(crawlingIfOlderUnitMonthCheck)#::checked#(/crawlingIfOlderMonthUnitCheck)#>
+    Day(s)&nbsp;<input type="radio" name="crawlingIfOlderUnit" value="day" #(crawlingIfOlderUnitDayCheck)#::checked#(/crawlingIfOlderDayUnitCheck)#>
+    Hour(s)&nbsp;<input type="radio" name="crawlingIfOlderUnit" value="hour" #(crawlingIfOlderUnitHourCheck)#::checked#(/crawlingIfOlderHourUnitCheck)#>
+    Minute(s)&nbsp;<input type="radio" name="crawlingIfOlderUnit" value="minute" #(crawlingIfOlderUnitMinuteCheck)#::checked#(/crawlingIfOlderMinuteUnitCheck)#>
+    <td class=small>
+    If you use this option, web pages that are already existent in your database are crawled and indexed again.
+    It depends on the age of the last crawl if this is done or not: if the last crawl is older than the given
+    date, the page is crawled again, othervise it is treaded as 'double' and not loaded or indexed again.
     </td>
   </tr>
   <tr valign="top" class="TableCellDark">

htroot/IndexCreate_p.java

@@ -95,8 +95,11 @@ public class IndexCreate_p {
                     env.setConfig("crawlingFilter", newcrawlingfilter);
                     int newcrawlingdepth = Integer.parseInt(post.get("crawlingDepth", "0"));
                     env.setConfig("crawlingDepth", Integer.toString(newcrawlingdepth));
-                    int recrawlIfOlder = Integer.parseInt(post.get("crawlingIfOlder", "-1"));
+                    boolean crawlingIfOlderCheck = post.get("crawlingIfOlderCheck", "").equals("on");
-                    env.setConfig("crawlingIfOlder", recrawlIfOlder);
+                    int crawlingIfOlderNumber = Integer.parseInt(post.get("crawlingIfOlderNumber", "-1"));
+                    String crawlingIfOlderUnit = post.get("crawlingIfOlderUnit","year");
+                    int crawlingIfOlder = recrawlIfOlderC(crawlingIfOlderCheck, crawlingIfOlderNumber, crawlingIfOlderUnit);                    
+                    env.setConfig("crawlingIfOlder", crawlingIfOlder);
                     int domFilterDepth = Integer.parseInt(post.get("crawlingDomFilterDepth", "-1"));
                     env.setConfig("crawlingDomFilterDepth", Integer.toString(domFilterDepth));
                     int domMaxPages = Integer.parseInt(post.get("crawlingDomMaxPages", "-1"));
@@ -151,7 +154,7 @@ public class IndexCreate_p {
                             switchboard.urlPool.errorURL.remove(urlhash);
                             
                             // stack url
-                            plasmaCrawlProfile.entry pe = switchboard.profiles.newEntry(crawlingStartURL.getHost(), crawlingStart, newcrawlingfilter, newcrawlingfilter, newcrawlingdepth, newcrawlingdepth, recrawlIfOlder, domFilterDepth, domMaxPages, crawlingQ, storeHTCache, true, localIndexing, crawlOrder, xsstopw, xdstopw, xpstopw);
+                            plasmaCrawlProfile.entry pe = switchboard.profiles.newEntry(crawlingStartURL.getHost(), crawlingStart, newcrawlingfilter, newcrawlingfilter, newcrawlingdepth, newcrawlingdepth, crawlingIfOlder, domFilterDepth, domMaxPages, crawlingQ, storeHTCache, true, localIndexing, crawlOrder, xsstopw, xdstopw, xpstopw);
                             String reasonString = switchboard.sbStackCrawlThread.stackCrawl(crawlingStart, null, yacyCore.seedDB.mySeed.hash, "CRAWLING-ROOT", new Date(), 0, pe);
                             
                             if (reasonString == null) {
@@ -212,7 +215,7 @@ public class IndexCreate_p {
                                 HashMap hyperlinks = (HashMap) scraper.getAnchors();
                                 
                                 // creating a crawler profile
-                                plasmaCrawlProfile.entry profile = switchboard.profiles.newEntry(fileName, file.toURL().toString(), newcrawlingfilter, newcrawlingfilter, newcrawlingdepth, newcrawlingdepth, recrawlIfOlder, domFilterDepth, domMaxPages, crawlingQ, storeHTCache, true, localIndexing, crawlOrder, xsstopw, xdstopw, xpstopw);                                
+                                plasmaCrawlProfile.entry profile = switchboard.profiles.newEntry(fileName, file.toURL().toString(), newcrawlingfilter, newcrawlingfilter, newcrawlingdepth, newcrawlingdepth, crawlingIfOlder, domFilterDepth, domMaxPages, crawlingQ, storeHTCache, true, localIndexing, crawlOrder, xsstopw, xdstopw, xpstopw);                                
                                 
                                 // loop through the contained links
                                 Iterator interator = hyperlinks.entrySet().iterator();
@@ -301,7 +304,32 @@ public class IndexCreate_p {
         prop.put("proxyPrefetchDepth", env.getConfig("proxyPrefetchDepth", "0"));
         prop.put("crawlingDepth", env.getConfig("crawlingDepth", "0"));
         prop.put("crawlingFilter", env.getConfig("crawlingFilter", "0"));
-        prop.put("crawlingIfOlder", env.getConfig("crawlingIfOlder", "-1"));
+        
+        int crawlingIfOlder = (int) env.getConfigLong("crawlingIfOlder", -1);
+        prop.put("crawlingIfOlderCheck", (crawlingIfOlder == Integer.MAX_VALUE) ? 0 : 1);
+        prop.put("crawlingIfOlderUnitYearCheck", 0);
+        prop.put("crawlingIfOlderUnitMonthCheck", 0);
+        prop.put("crawlingIfOlderUnitDayCheck", 0);
+        prop.put("crawlingIfOlderUnitHourCheck", 0);
+        prop.put("crawlingIfOlderUnitMinuteCheck", 0);
+        if (crawlingIfOlder == Integer.MAX_VALUE) {
+        } else if (crawlingIfOlder >= 60*24*365) {
+            prop.put("crawlingIfOlderNumber", crawlingIfOlder / 60*24*365);
+            prop.put("crawlingIfOlderUnitYearCheck", 1);
+        } else if (crawlingIfOlder >= 60*24*30) {
+            prop.put("crawlingIfOlderNumber", crawlingIfOlder / 60*24*30);
+            prop.put("crawlingIfOlderUnitMonthCheck", 1);
+        } else if (crawlingIfOlder >= 60*24) {
+            prop.put("crawlingIfOlderNumber", crawlingIfOlder / 60*24);
+            prop.put("crawlingIfOlderUnitDayCheck", 1);
+        } else if (crawlingIfOlder >= 60) {
+            prop.put("crawlingIfOlderNumber", crawlingIfOlder / 60);
+            prop.put("crawlingIfOlderUnitHourCheck", 1);
+        } else {
+            prop.put("crawlingIfOlderNumber", crawlingIfOlder);
+            prop.put("crawlingIfOlderUnitMinuteCheck", 1);
+        }
+        //prop.put("crawlingIfOlder", crawlingIfOlder);
         prop.put("crawlingDomFilterDepth", env.getConfig("crawlingDomFilterDepth", "-1"));
         prop.put("crawlingDomMaxPages", env.getConfig("crawlingDomMaxPages", "-1"));
         prop.put("crawlingQChecked", env.getConfig("crawlingQ", "").equals("true") ? 1 : 0);
@@ -477,6 +505,15 @@ public class IndexCreate_p {
         return prop;
     }
 
+    private static int recrawlIfOlderC(boolean recrawlIfOlderCheck, int recrawlIfOlderNumber, String crawlingIfOlderUnit) {
+        if (!recrawlIfOlderCheck) return -1;
+        if (crawlingIfOlderUnit.equals("year")) return recrawlIfOlderNumber * 60 * 24 * 356;
+        if (crawlingIfOlderUnit.equals("month")) return recrawlIfOlderNumber * 60 * 24 * 30;
+        if (crawlingIfOlderUnit.equals("day")) return recrawlIfOlderNumber * 60 * 24;
+        if (crawlingIfOlderUnit.equals("hour")) return recrawlIfOlderNumber * 60;
+        if (crawlingIfOlderUnit.equals("minute")) return recrawlIfOlderNumber;
+        return -1;
+    }
 }
 
 

htroot/env/templates/header.template

@@ -37,13 +37,7 @@
   <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/IndexCreate_p.html" class="MenuItemLink">Index Create</a></td></tr>
   <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/IndexControl_p.html" class="MenuItemLink">Index Control</a></td></tr>
   <tr><td class="MenuItem">&nbsp;<a href="/IndexMonitor.html" class="MenuItemLink">Index Monitor</a></td></tr>
-  <tr><td class="MenuSpacer"></td></tr>
-
-  <tr><td class="MenuHeader">&nbsp;Local Proxy</td></tr>
   <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/Blacklist_p.html" class="MenuItemLink">Blacklist</a></td></tr>
-  <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/ProxyIndexingMonitor_p.html" class="MenuItemLink">Proxy Indexing</a></td></tr>
-  <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/CacheAdmin_p.html" class="MenuItemLink">Cache Monitor</a></td></tr>
-  <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/CookieMonitorIncoming_p.html" class="MenuItemLink">Cookie Monitor</a></td></tr>
   <tr><td class="MenuSpacer"></td></tr>
 
   <tr><td class="MenuHeader">&nbsp;Communication / Publication</td></tr>
@@ -65,6 +59,12 @@
   <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/Connections_p.html" class="MenuItemLink">Connections</a></td></tr>
   <tr><td class="MenuSpacer"></td></tr>
 
+  <tr><td class="MenuHeader">&nbsp;Local Proxy</td></tr>
+  <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/ProxyIndexingMonitor_p.html" class="MenuItemLink">Proxy Indexing</a></td></tr>
+  <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/CacheAdmin_p.html" class="MenuItemLink">Cache Monitor</a></td></tr>
+  <tr><td class="MenuItem">&nbsp;<img border="0" src="/env/grafics/lock.gif" align="top">&nbsp;<a href="/CookieMonitorIncoming_p.html" class="MenuItemLink">Cookie Monitor</a></td></tr>
+  <tr><td class="MenuSpacer"></td></tr>
+
   <tr><td class="MenuHeader">&nbsp;The Project</td></tr>
   <tr><td class="MenuItem">&nbsp;<a href="http://www.yacy.net/yacy/" class="MenuItemLink">Project Home</a></td></tr>
   <tr><td class="MenuItem">&nbsp;<a href="http://www.yacy.net/yacy/News.html" class="MenuItemLink">Project News</a></td></tr>

source/yacy.java

@@ -138,7 +138,7 @@ public final class yacy {
     private static float version = (float) 0.1;
 
     private static final String vDATE   = "@REPL_DATE@";
-    private static final String copyright = "[ YACY Proxy v" + vString + ", build " + vDATE + " by Michael Christen / www.yacy.net ]";
+    private static final String copyright = "[ YaCy v" + vString + ", build " + vDATE + " by Michael Christen / www.yacy.net ]";
     private static final String hline = "-------------------------------------------------------------------------------";
    
     /**
@@ -163,10 +163,10 @@ public final class yacy {
     }
 
     /**
-    * Combines the version of the proxy with the versionnumber from SVN to a
+    * Combines the version of YaCy with the versionnumber from SVN to a
     * combined version
     *
-    * @param version Current given version for this proxy.
+    * @param version Current given version.
     * @param svn Current version given from svn.
     * @return String with the combined version
     */