boxtec
/
yacy_search_server
mirror of https://github.com/yacy/yacy_search_server
4
1
Fork 0
Code Issues Projects Releases Wiki Activity

- added buttons to add complete domain or single URL to blacklist to IndexControl_p

Browse Source 
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3400 6c8d7289-2bf4-0310-a012-ef5d649a1542
pull/1/head
karlchenofhell 18 years ago
parent 8668ac5d91
commit 31ad42535a
2 changed files with 119 additions and 32 deletions
Show Stats Download Patch File Download Diff File

84
htroot/IndexControl_p.html
Unescape View File

@ -32,7 +32,7 @@
</tr>
<tr valign="top" class="TableCellLight">
<td colspan="2"></td>
<td><input type="submit" name="keyhashtransfer" value="Transfer to other peer" />
<td><input type="submit" name="keyhashtransfer" value="Transfer to other peer" /><br />
<select name="hostHash">
#{hosts}#
<option value="#[hosthash]#">#[hostname]#</option>
@ -117,44 +117,66 @@
</table>
</form>
#(keyhashsimilar)#
::Sequential List of Word-Hashes:<br />#{rows}# #{cols}#<a href="/IndexControl_p.html?keyhash=#[wordHash]#&amp;keyhashsearch=1" class="tt">#[wordHash]#</a> #{/cols}#<br />#{/rows}#
::Sequential List of Word-Hashes:<br />#{rows}#
#{cols}#
<a href="/IndexControl_p.html?keyhash=#[wordHash]#&amp;keyhashsearch=1" class="tt">#[wordHash]#</a>#{/cols}#<br />
#{/rows}#
#(/keyhashsimilar)#
#(urlhashsimilar)#
::Sequential List of URL-Hashes:<br />#{rows}# #{cols}#<a href="/IndexControl_p.html?urlhash=#[urlHash]#&amp;urlhashsearch=1" class="tt">#[urlHash]#</a> #{/cols}#<br />#{/rows}#
::Sequential List of URL-Hashes:<br />
#{rows}#
#{cols}#<a href="/IndexControl_p.html?urlhash=#[urlHash]#&amp;urlhashsearch=1" class="tt">#[urlHash]#</a> #{/cols}#<br />
#{/rows}#
#(/urlhashsimilar)#
#(genUrlList)#
::
No URL entries related to this word hash <span class="tt">#[keyHash]#</span>.
::
#[count]# URL entries related to this word hash <span class="tt">#[keyHash]#</span><br /><br />
<p>#[count]# URL entries related to this word hash <span class="tt">#[keyHash]#</span></p>
<form action="IndexControl_p.html" method="post" enctype="multipart/form-data">
#{urlList}# #(urlExists)#
<input type="checkbox" name="urlhx.#[urlhxCount]#" checked value="#[urlhxValue]#" align="top" /><span class="tt">#[urlhxValue]#&nbsp;&lt;unresolved URL Hash&gt;</span><br />
::<input type="checkbox" name="urlhx.#[urlhxCount]#" #(urlhxChecked)#::checked="checked" #(/urlhxChecked)#value="#[urlhxValue]#" align="top" />
<a href="/IndexControl_p.html?keystring=#[keyString]#&amp;keyhash=#[keyHash]#&amp;urlhash=#[urlhxValue]#&amp;urlstringsearch=&amp;urlstring=#[urlString]#" class="tt">#[urlhxValue]#</a><span class="tt">&nbsp;#[urlString]#, pos=#[pos]#</span><br />
#(/urlExists)# #{/urlList}#
<input type="hidden" name="keystring" value="#[keyString]#" />
<input type="hidden" name="keyhash" value="#[keyHash]#" />
<input type="hidden" name="urlstring" value="" />
<input type="hidden" name="urlhash" value="" /><br />
<fieldset><legend>Reference Deletion</legend>
<table border="0" cellspacing="5" cellpadding="5">
<tr valign="top">
<td>
<br /><br /><input type="submit" value="Delete reference to selected URLs" name="keyhashdelete" />
<br /><br /><input type="submit" value="Delete reference to ALL URLs" name="keyhashdeleteall" /><span class="small"><br />&nbsp;&nbsp;(= delete Word)</span>
</td>
<td width="150">
<center><input type="checkbox" name="delurl" value="" align="top" checked /></center><br />
<span class="small">delete also the referenced URL itself (reasonable and recommended, may produce unresolved references at other word indexes but they do not harm)</span>
</td>
<td width="150">
<center><input type="checkbox" name="delurlref" value="" align="top" /></center><br />
<span class="small">for every resolveable and deleted URL reference, delete the same reference at every other word where the reference exists (very extensive, but prevents further unresolved references)</span>
</td>
</tr>
</table>
</fieldset></form><br />#(/genUrlList)#
<p>#{urlList}# #(urlExists)#
<input type="checkbox" id="urlhx.#[urlhxCount]#" name="urlhx.#[urlhxCount]#" checked value="#[urlhxValue]#" align="top" /><label for="urlhx.#[urlhxCount]#" class="tt">#[urlhxValue]#&nbsp;&lt;unresolved URL Hash&gt;</span><br />
::<input type="checkbox" id="urlhx.#[urlhxCount]#" name="urlhx.#[urlhxCount]#" #(urlhxChecked)#::checked="checked" #(/urlhxChecked)#value="#[urlhxValue]#" align="top" />
<a href="/IndexControl_p.html?keystring=#[keyString]#&amp;keyhash=#[keyHash]#&amp;urlhash=#[urlhxValue]#&amp;urlstringsearch=&amp;urlstring=#[urlString]#" class="tt">#[urlhxValue]#</a><label for="urlhx.#[urlhxCount]#" class="tt">&nbsp;#[urlString]#, pos=#[pos]#</label><br />
#(/urlExists)# #{/urlList}#</p>
<input type="hidden" name="keystring" value="#[keyString]#" />
<input type="hidden" name="keyhash" value="#[keyHash]#" />
<input type="hidden" name="urlstring" value="" />
<input type="hidden" name="urlhash" value="" /><br />
<fieldset><legend>Reference Deletion</legend>
<dl>
<dt><input type="checkbox" id="delurl" name="delurl" value="" align="top" checked /></dt>
<dd>
<label for="delurl" class="small">
delete also the referenced URL itself (reasonable and recommended, may produce unresolved references<br />
at other word indexes but they do not harm)
</label>
</dt>
<dt><input type="checkbox" id="delurlref" name="delurlref" value="" align="top" /></dt>
<dd>
<label for="delurlref" class="small">
for every resolveable and deleted URL reference, delete the same reference at every other word where<br />
the reference exists (very extensive, but prevents further unresolved references)
</label>
</dt>
<dt>&nbsp;</dt>
<dd>
<input type="submit" value="Delete reference to selected URLs" name="keyhashdelete" /><br />
<input type="submit" value="Delete reference to ALL URLs" name="keyhashdeleteall" /><span class="small">&nbsp;&nbsp;(= delete Word)</span>
</dd>
<dt>
<select name="blacklist" size="1">#{blacklists}#
<option value="#[name]#">#[name]#</option>#{/blacklists}#
</select>
</dt>
<dd>
<input type="submit" value="Add selected URLs to blacklist" name="blacklisturls" /><br />
<input type="submit" value="Add selected domains to blacklist" name="blacklistdomains" />
</dd>
</dl>
</fieldset>
</form>
<br />#(/genUrlList)#
#(genUrlProfile)#
::No entry found for URL-hash #[urlhash]#
::<table>

67
htroot/IndexControl_p.java
Unescape View File

@ -46,7 +46,10 @@
// javac -classpath .:../classes IndexControl_p.java
// if the shell's current path is HTROOT
import java.io.File;
import java.io.FileWriter;
import java.io.IOException;
import java.io.PrintWriter;
import java.net.MalformedURLException;
import java.util.Enumeration;
import java.util.HashMap;
@ -55,6 +58,7 @@ import java.util.Iterator;
import java.util.Set;
import java.util.TreeMap;
import de.anomic.data.listManager;
import de.anomic.http.httpHeader;
import de.anomic.index.indexContainer;
import de.anomic.index.indexRWIEntry;
@ -383,6 +387,61 @@ public class IndexControl_p {
}
}
if (post.containsKey("blacklist")) {
String blacklist = post.get("blacklist", "");
if (post.containsKey("blacklisturls")) {
PrintWriter pw;
try {
String[] supportedBlacklistTypes = env.getConfig("BlackLists.types", "").split(",");
pw = new PrintWriter(new FileWriter(new File(listManager.listsPath, blacklist), true));
URL url;
for (int i=0; i<urlx.length; i++) {
indexURLEntry e = switchboard.wordIndex.loadedURL.load(urlx[i], null);
if (e != null) {
url = e.comp().url();
pw.println(url.getHost() + "/" + url.getFile());
for (int blTypes=0; blTypes < supportedBlacklistTypes.length; blTypes++) {
if (listManager.ListInListslist(supportedBlacklistTypes[blTypes] + ".BlackLists", blacklist)) {
plasmaSwitchboard.urlBlacklist.add(
supportedBlacklistTypes[blTypes],
url.getHost(),
url.getFile());
}
}
}
}
pw.close();
} catch (IOException e) {
}
}
if (post.containsKey("blacklistdomains")) {
PrintWriter pw;
try {
String[] supportedBlacklistTypes = env.getConfig("BlackLists.types", "").split(",");
pw = new PrintWriter(new FileWriter(new File(listManager.listsPath, blacklist), true));
URL url;
for (int i=0; i<urlx.length; i++) {
indexURLEntry e = switchboard.wordIndex.loadedURL.load(urlx[i], null);
if (e != null) {
url = e.comp().url();
pw.println(url.getHost() + "/.*");
for (int blTypes=0; blTypes < supportedBlacklistTypes.length; blTypes++) {
if (listManager.ListInListslist(supportedBlacklistTypes[blTypes] + ".BlackLists", blacklist)) {
plasmaSwitchboard.urlBlacklist.add(
supportedBlacklistTypes[blTypes],
url.getHost(),
".*");
}
}
}
}
pw.close();
} catch (IOException e) {
}
}
}
listHosts(prop, keyhash);
// insert constants
@ -512,6 +571,7 @@ public class IndexControl_p {
prop.put("genUrlList_urlList", i);
prop.put("genUrlList_keyString", keystring);
prop.put("genUrlList_count", i);
putBlacklists(prop, listManager.getDirListing(listManager.listsPath));
}
index = null;
return prop;
@ -521,5 +581,10 @@ public class IndexControl_p {
if (index != null) index = null;
}
}
private static void putBlacklists(serverObjects prop, String[] lists) {
prop.put("genUrlList_blacklists", lists.length);
for (int i=0; i<lists.length; i++)
prop.put("genUrlList_blacklists_" + i + "_name", lists[i]);
}
}

Write Preview
Loading…
Cancel
Save