- added domain statistic generation to IndexControlURLs_p.html servlet

- added 'delete all' button to all results of such a domain statistic output which causes that all urls to this domain are deleted
- extended stack cleaner to clean also the statistics: they are not completely destroyed, only the smallest counting domains are removed


git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@5117 6c8d7289-2bf4-0310-a012-ef5d649a1542

1 files changed, 53 insertions, 12 deletions

diff --git a/htroot/IndexControlURLs_p.html b/htroot/IndexControlURLs_p.html
index dded77a71..5eb87f8ab 100644
--- a/htroot/IndexControlURLs_p.html
+++ b/htroot/IndexControlURLs_p.html
@@ -26,19 +26,44 @@
     </fieldset>

     </form>

     

-	#(urlhashsimilar)#::Sequential List of URL-Hashes:<br />

-    #{rows}# 

-    #{cols}#<a href="/IndexControlURLs_p.html?urlhash=#[urlHash]#&amp;urlhashsearch=1" class="tt">#[urlHash]#</a> #{/cols}#<br />

-    #{/rows}# 

-    #(/urlhashsimilar)#

-    

-    #(lurlexportfinished)#::

-    <div class="commit">Finished export of #[urlcount]# URLs to file #[exportfile]#</div>::

-    #(/lurlexportfinished)#

+    #(statistics)#::

+    <form action="IndexControlURLs_p.html" method="post" enctype="multipart/form-data">

+    <fieldset><legend>Statistics about top-domains in URL Database</legend>

+      <dl>

+        <dt class="TableCellLight"></dt>

+        <dd>Show top <input type="text" name="lines" value="#[lines]#" size="6" maxlength="6" /> domains from all URLs.

+            <input type="submit" name="statistics" value="Generate Statistics" />

+        </dd>

+      </dl>

+    </fieldset>

+    </form>

+    #(/statistics)#

     

-    #(lurlexporterror)#::

-    <div class="error">Export to file #[exportfile]# failed: #[exportfailmsg]#</div>::

-    #(/lurlexporterror)#

+    #(statisticslines)#::

+    <p><em>Statistics about the top-#[domains]# domains in the database:</em>

+    <table cellpadding="2" cellspacing="1" >

+      <tr class="TableHeader">

+   		<td align="center"></td>

+        <td><strong>Domain</strong></td>

+        <td><strong>URLs</strong></td>

+      </tr>

+      #{domains}#

+      <tr class="TableCell#(dark)#Light::Dark#(/dark)#">

+        <td>

+          <form action="IndexControlURLs_p.html" method="post" enctype="multipart/form-data">

+            <div>

+              <input type="hidden" name="hashpart" value="#[hashpart]#" />

+              <input type="hidden" name="lines" value="#[lines]#" />

+              <input type="submit" name="deletedomain" value="delete all" />

+            </div>

+          </form>

+        </td>

+        <td><a href="http://#[domain]#/" target="_">#[domain]#</a></td>

+        <td>#[count]#</td>

+      </tr>

+      #{/domains}#

+    </table></p>

+    #(/statisticslines)#

     

     #(lurlexport)#::

     <form action="IndexControlURLs_p.html" method="post" enctype="multipart/form-data">

@@ -69,6 +94,21 @@
     <div class="commit" style="text-decoration:blink">Export to file #[exportfile]# is running ..  #[urlcount]# URLs so far</div>::

     #(/lurlexport)#

     

+	#(lurlexportfinished)#::

+    <div class="commit">Finished export of #[urlcount]# URLs to file #[exportfile]#</div>::

+    #(/lurlexportfinished)#

+    

+    #(lurlexporterror)#::

+    <div class="error">Export to file #[exportfile]# failed: #[exportfailmsg]#</div>::

+    #(/lurlexporterror)#

+    

+    #(urlhashsimilar)#::<p>Sequential List of URL-Hashes:<br />

+    #{rows}# 

+    #{cols}#<a href="/IndexControlURLs_p.html?urlhash=#[urlHash]#&amp;urlhashsearch=1" class="tt">#[urlHash]#</a> #{/cols}#<br />

+    #{/rows}#

+    </p>

+    #(/urlhashsimilar)#

+    

     #(genUrlProfile)#

     ::No entry found for URL-hash #[urlhash]#

     ::<table>

@@ -98,6 +138,7 @@
 		<span class="small">&nbsp;delete the reference to this url at every other word where the reference exists (very extensive, but prevents unresolved references)</span><br />

 	</form>

     #(/genUrlProfile)#

+    

     #[result]#

     

     #%env/templates/footer.template%#