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

sorted URL list, StringBuffers, finals, cleaned, Properties

Browse Source 
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@860 6c8d7289-2bf4-0310-a012-ef5d649a1542
pull/1/head
borg-0300 20 years ago
parent 2a18e6486b
commit 5c4db6968e
1 changed files with 218 additions and 152 deletions
Show Stats Download Patch File Download Diff File

256
htroot/IndexControl_p.java
Unescape View File

@ -4,7 +4,10 @@
// (C) by Michael Peter Christen; mc@anomic.de
// first published on http://www.anomic.de
// Frankfurt, Germany, 2004
// last change: 02.05.2004
//
// $LastChangedDate$
// $LastChangedRevision$
// $LastChangedBy$
//
// This program is free software; you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
@ -40,7 +43,7 @@
// Contributions and changes to the program code must be marked as such.
// You must compile this file with
// javac -classpath .:../Classes IndexControl_p.java
// javac -classpath .:../classes IndexControl_p.java
// if the shell's current path is HTROOT
import java.io.IOException;
@ -50,7 +53,7 @@ import java.util.Enumeration;
import java.util.HashSet;
import java.util.HashMap;
import java.util.Iterator;
import java.util.TreeMap;
import de.anomic.htmlFilter.htmlFilterContentScraper;
import de.anomic.http.httpHeader;
import de.anomic.plasma.plasmaCrawlLURL;
@ -71,7 +74,7 @@ public class IndexControl_p {
plasmaSwitchboard switchboard = (plasmaSwitchboard) env;
serverObjects prop = new serverObjects();
if ((post == null) || (env == null)) {
if (post == null || env == null) {
prop.put("keystring", "");
prop.put("keyhash", "");
prop.put("urlstring", "");
@ -93,7 +96,8 @@ public class IndexControl_p {
String urlstring = ((String) post.get("urlstring")).trim();
String urlhash = ((String) post.get("urlhash")).trim();
if ((!(urlstring.startsWith("http://"))) && (!(urlstring.startsWith("https://")))) urlstring = "http://" + urlstring;
if (!urlstring.startsWith("http://") &&
!urlstring.startsWith("https://")) { urlstring = "http://" + urlstring; }
prop.put("keystring", keystring);
prop.put("keyhash", keyhash);
@ -105,27 +109,44 @@ public class IndexControl_p {
String[] urlx = post.getAll("urlhx.*");
boolean delurl = post.containsKey("delurl");
boolean delurlref = post.containsKey("delurlref");
//System.out.println("DEBUG CHECK: " + ((delurl) ? "delurl" : "") + " " + ((delurlref) ? "delurlref" : ""));
// System.out.println("DEBUG CHECK: " + ((delurl) ? "delurl" : "") + " " + ((delurlref) ? "delurlref" : ""));
// DHT control
if (post.containsKey("setIndexTransmission")) {
boolean allowDistributeIndex = ((String) post.get("indexDistribute", "")).equals("on");
switchboard.setConfig("allowDistributeIndex", (allowDistributeIndex) ? "true" : "false");
if (allowDistributeIndex) switchboard.indexDistribution.enable(); else switchboard.indexDistribution.disable();
if (post.get("indexDistribute", "").equals("on")) {
switchboard.setConfig("allowDistributeIndex", "true");
switchboard.indexDistribution.enable();
} else {
switchboard.setConfig("allowDistributeIndex", "false");
switchboard.indexDistribution.disable();
}
boolean allowDistributeIndexWhileCrawling = post.containsKey("indexDistributeWhileCrawling");
switchboard.setConfig("allowDistributeIndexWhileCrawling", (allowDistributeIndexWhileCrawling) ? "true" : "false");
if (allowDistributeIndexWhileCrawling) switchboard.indexDistribution.enableWhileCrawling(); else switchboard.indexDistribution.disableWhileCrawling();
if (post.containsKey("indexDistributeWhileCrawling")) {
switchboard.setConfig("allowDistributeIndexWhileCrawling", "true");
switchboard.indexDistribution.enableWhileCrawling();
} else {
switchboard.setConfig("allowDistributeIndexWhileCrawling", "false");
switchboard.indexDistribution.disableWhileCrawling();
}
boolean allowReceiveIndex = ((String) post.get("indexReceive", "")).equals("on");
switchboard.setConfig("allowReceiveIndex", (allowReceiveIndex) ? "true" : "false");
yacyCore.seedDB.mySeed.setFlagAcceptRemoteIndex(allowReceiveIndex);
if (post.get("indexReceive", "").equals("on")) {
switchboard.setConfig("allowReceiveIndex", "true");
yacyCore.seedDB.mySeed.setFlagAcceptRemoteIndex(true);
} else {
switchboard.setConfig("allowReceiveIndex", "false");
yacyCore.seedDB.mySeed.setFlagAcceptRemoteIndex(false);
}
boolean indexReceiveBlockBlacklist = ((String) post.get("indexReceiveBlockBlacklist", "")).equals("on");
switchboard.setConfig("indexReceiveBlockBlacklist", (indexReceiveBlockBlacklist) ? "true" : "false");
if (post.get("indexReceiveBlockBlacklist", "").equals("on")) {
switchboard.setConfig("indexReceiveBlockBlacklist", "true");
} else {
switchboard.setConfig("indexReceiveBlockBlacklist", "false");
}
}
// delete word
if (post.containsKey("keyhashdeleteall")) {
if ((delurl) || (delurlref)) {
if (delurl || delurlref) {
// generate an urlx array
plasmaWordIndexEntity index = null;
try {
@ -135,36 +156,60 @@ public class IndexControl_p {
Enumeration en = index.elements(true);
int i = 0;
urlx = new String[index.size()];
while (en.hasMoreElements()) urlx[i++] = ((plasmaWordIndexEntry) en.nextElement()).getUrlHash();
index.close(); index = null;
while (en.hasMoreElements()) {
urlx[i++] = ((plasmaWordIndexEntry) en.nextElement()).getUrlHash();
}
index.close();
index = null;
} catch (IOException e) {
urlx = new String[0];
} finally {
if (index != null) try { index.close(); } catch (Exception e) {}
}
}
if (delurlref) for (int i = 0; i < urlx.length; i++) switchboard.removeAllUrlReferences(urlx[i], true);
if ((delurl) || (delurlref)) for (int i = 0; i < urlx.length; i++) switchboard.urlPool.loadedURL.remove(urlx[i]);
if (delurlref) {
for (int i = 0; i < urlx.length; i++) {
switchboard.removeAllUrlReferences(urlx[i], true);
}
}
if (delurl || delurlref) {
for (int i = 0; i < urlx.length; i++) {
switchboard.urlPool.loadedURL.remove(urlx[i]);
}
}
switchboard.wordIndex.deleteIndex(keyhash);
post.remove("keyhashdeleteall");
if ((keystring.length() > 0) && (plasmaWordIndexEntry.word2hash(keystring).equals(keyhash)))
if (keystring.length() > 0 &&
plasmaWordIndexEntry.word2hash(keystring).equals(keyhash)) {
post.put("keystringsearch", "generated");
else
} else {
post.put("keyhashsearch", "generated");
}
}
// delete selected URLs
if (post.containsKey("keyhashdelete")) {
if (delurlref) for (int i = 0; i < urlx.length; i++) switchboard.removeAllUrlReferences(urlx[i], true);
if ((delurl) || (delurlref)) for (int i = 0; i < urlx.length; i++) switchboard.urlPool.loadedURL.remove(urlx[i]);
if (delurlref) {
for (int i = 0; i < urlx.length; i++) {
switchboard.removeAllUrlReferences(urlx[i], true);
}
}
if (delurl || delurlref) {
for (int i = 0; i < urlx.length; i++) {
switchboard.urlPool.loadedURL.remove(urlx[i]);
}
}
switchboard.wordIndex.removeEntries(keyhash, urlx, true);
// this shall lead to a presentation of the list; so handle that the remaining program
// thinks that it was called for a list presentation
post.remove("keyhashdelete");
if ((keystring.length() > 0) && (plasmaWordIndexEntry.word2hash(keystring).equals(keyhash)))
if (keystring.length() > 0 &&
plasmaWordIndexEntry.word2hash(keystring).equals(keyhash)) {
post.put("keystringsearch", "generated");
else
} else {
post.put("keyhashsearch", "generated");
//prop.put("result", "Delete of relation of url hashes " + result + " to key hash " + keyhash);
// prop.put("result", "Delete of relation of url hashes " + result + " to key hash " + keyhash);
}
}
if (post.containsKey("urlhashdeleteall")) {
@ -194,16 +239,21 @@ public class IndexControl_p {
}
if (post.containsKey("keyhashsearch")) {
if ((keystring.length() == 0) || (!(plasmaWordIndexEntry.word2hash(keystring).equals(keyhash))))
if (keystring.length() == 0 ||
!plasmaWordIndexEntry.word2hash(keystring).equals(keyhash)) {
prop.put("keystring", "<not possible to compute word from hash>");
}
prop.put("urlstring", "");
prop.put("urlhash", "");
prop.put("result", genUrlList(switchboard, keyhash, ""));
}
// transfer to other peer
if (post.containsKey("keyhashtransfer")) {
if ((keystring.length() == 0) || (!(plasmaWordIndexEntry.word2hash(keystring).equals(keyhash))))
if (keystring.length() == 0 ||
!plasmaWordIndexEntry.word2hash(keystring).equals(keyhash)) {
prop.put("keystring", "<not possible to compute word from hash>");
}
prop.put("urlstring", "");
prop.put("urlhash", "");
plasmaWordIndexEntity[] indexes = new plasmaWordIndexEntity[1];
@ -232,13 +282,15 @@ public class IndexControl_p {
}
// now delete all entries that have no url entry
Iterator hashIter = unknownURLEntries.iterator();
while (hashIter.hasNext()) try {
while (hashIter.hasNext()) {
try {
indexes[0].removeEntry((String) hashIter.next(), false);
} catch (IOException e) {}
}
// use whats remaining
String gzipBody = switchboard.getConfig("indexControl.gzipBody","false");
int timeout = (int) switchboard.getConfigLong("indexControl.timeout",60000);
result = yacyClient.transferIndex(
result = yacyClient.transferIndex (
yacyCore.seedDB.getConnected(post.get("hostHash", "")),
indexes,
knownURLs,
@ -248,21 +300,22 @@ public class IndexControl_p {
try {indexes[0].close();} catch (IOException e) {}
}
// generate list
if (post.containsKey("keyhashsimilar")) {
Iterator hashIt = switchboard.wordIndex.wordHashes(keyhash, true, true);
String result = "Sequential List of Word-Hashes:<br>";
final Iterator hashIt = switchboard.wordIndex.wordHashes(keyhash, true, true);
StringBuffer result = new StringBuffer("Sequential List of Word-Hashes:<br>");
String hash;
int i = 0;
while ((hashIt.hasNext()) && (i < 256)) {
while (hashIt.hasNext() && i < 256) {
hash = (String) hashIt.next();
result += "<a href=\"/IndexControl_p.html?" +
"keystring=" +
"&keyhash=" + hash +
"&urlhash=" +
"&urlstring=" +
"&keyhashsearch=" +
"\" class=\"tt\">" + hash + "</a> " + (((i + 1) % 8 == 0) ? "<br>" : "");
result.append("<a href=\"/IndexControl_p.html?")
.append("keystring=")
.append("&keyhash=").append(hash)
.append("&urlhash=")
.append("&urlstring=")
.append("&urlhashsearch=")
.append("\" class=\"tt\">").append(hash).append("</a> ")
.append(((i + 1) % 8 == 0) ? "<br>" : "");
i++;
}
prop.put("result", result);
@ -293,33 +346,35 @@ public class IndexControl_p {
}
}
// generate list
if (post.containsKey("urlhashsimilar")) {
try {
Iterator hashIt = switchboard.urlPool.loadedURL.urlHashes(urlhash, true);
String result = "Sequential List of URL-Hashes:<br>";
final Iterator hashIt = switchboard.urlPool.loadedURL.urlHashes(urlhash, true);
StringBuffer result = new StringBuffer("Sequential List of URL-Hashes:<br>");
String hash;
int i = 0;
while ((hashIt.hasNext()) && (i < 256)) {
while (hashIt.hasNext() && i < 256) {
hash = (String) hashIt.next();
result += "<a href=\"/IndexControl_p.html?" +
"keystring=" +
"&keyhash=" +
"&urlhash=" + hash +
"&urlstring=" +
"&urlhashsearch=" +
"\" class=\"tt\">" + hash + "</a> " + (((i + 1) % 8 == 0) ? "<br>" : "");
result.append("<a href=\"/IndexControl_p.html?")
.append("keystring=")
.append("&keyhash=")
.append("&urlhash=").append(hash)
.append("&urlstring=")
.append("&urlhashsearch=")
.append("\" class=\"tt\">").append(hash).append("</a> ")
.append(((i + 1) % 8 == 0) ? "<br>" : "");
i++;
}
prop.put("result", result);
prop.put("result", result.toString());
} catch (IOException e) {
prop.put("result", "Error: " + e.getMessage());
}
}
//List known hosts
// list known hosts
yacySeed seed;
int hc = 0;
if ((yacyCore.seedDB != null) && (yacyCore.seedDB.sizeConnected() > 0)) {
if (yacyCore.seedDB != null && yacyCore.seedDB.sizeConnected() > 0) {
Enumeration e = yacyCore.dhtAgent.getAcceptRemoteIndexSeeds(keyhash);
while (e.hasMoreElements()) {
seed = (yacySeed) e.nextElement();
@ -346,9 +401,9 @@ public class IndexControl_p {
}
public static String genUrlProfile(plasmaSwitchboard switchboard, plasmaCrawlLURL.Entry entry, String urlhash) {
if (entry == null) return "No entry found for URL-hash " + urlhash;
if (entry == null) { return "No entry found for URL-hash " + urlhash; }
URL url = entry.url();
if (url == null) return "No entry found for URL-hash " + urlhash;
if (url == null) { return "No entry found for URL-hash " + urlhash; }
String result = "<table>" +
"<tr><td class=\"small\">URL String</td><td class=\"tt\">" + htmlFilterContentScraper.urlNormalform(url) + "</td></tr>" +
"<tr><td class=\"small\">Hash</td><td class=\"tt\">" + urlhash + "</td></tr>" +
@ -382,57 +437,68 @@ public class IndexControl_p {
// search for a word hash and generate a list of url links
plasmaWordIndexEntity index = null;
try {
HashSet keyhashes = new HashSet();
final HashSet keyhashes = new HashSet();
keyhashes.add(keyhash);
index = switchboard.searchManager.searchHashes(keyhashes, 10000);
String result = "";
final StringBuffer result = new StringBuffer(1024);
if (index.size() == 0) {
result = "No URL entries related to this word hash <span class=\"tt\">" + keyhash + "</span>.";
result.append("No URL entries related to this word hash <span class=\"tt\">").append(keyhash).append("</span>.");
} else {
Enumeration en = index.elements(true);
plasmaWordIndexEntry ie;
result = "URL entries related to this word hash <span class=\"tt\">" + keyhash + "</span>:<br>";
result += "<form action=\"IndexControl_p.html\" method=\"post\" enctype=\"multipart/form-data\">";
final Enumeration en = index.elements(true);
result.append("URL entries related to this word hash <span class=\"tt\">").append(keyhash).append("</span><br>");
result.append("<form action=\"IndexControl_p.html\" method=\"post\" enctype=\"multipart/form-data\">");
String us, uh;
int i = 0;
final TreeMap tm = new TreeMap();
while (en.hasMoreElements()) {
ie = (plasmaWordIndexEntry) en.nextElement();
uh = ie.getUrlHash();
result +=
"<input type=\"checkbox\" name=\"urlhx" + i++ + "\" value=\"" + uh + "\" align=\"top\">";
uh = ((plasmaWordIndexEntry)en.nextElement()).getUrlHash();
if (switchboard.urlPool.loadedURL.exists(uh)) {
us = switchboard.urlPool.loadedURL.getEntry(uh).url().toString();
result +=
"<a href=\"/IndexControl_p.html?" + "keystring=" + keystring +
"&keyhash=" + keyhash + "&urlhash=" + uh + "&urlstringsearch=" + "&urlstring=" + us +
"\" class=\"tt\">" + uh + "</a><span class=\"tt\">&nbsp;" + us + "</span><br>";
tm.put(us, uh);
} else {
result +=
"<span class=\"tt\">" + uh + "&nbsp;&lt;unresolved URL Hash&gt;</span><br>";
tm.put("", uh);
}
}
result +=
"<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>";
}
index.close(); index = null;
return result;
final Iterator iter = tm.keySet().iterator();
result.ensureCapacity((tm.size() + 2) * 384);
while (iter.hasNext()) {
us = iter.next().toString();
uh = (String)tm.get(us);
result.append("<input type=\"checkbox\" name=\"urlhx").append(i++).append("\" value=\"").append(uh).append("\" align=\"top\">");
if (us.length() > 0) {
result.append("<a href=\"/IndexControl_p.html?").append("keystring=").append(keystring)
.append("&keyhash=").append(keyhash).append("&urlhash=").append(uh)
.append("&urlstringsearch=").append("&urlstring=").append(us).append("\" class=\"tt\">")
.append(uh).append("</a><span class=\"tt\">&nbsp;").append(us).append("</span><br>");
} else {
result.append("<span class=\"tt\">").append(uh).append("&nbsp;&lt;unresolved URL Hash&gt;</span><br>");
}
}
result.append("<input type=\"hidden\" name=\"keystring\" value=\"").append(keystring).append("\">")
.append("<input type=\"hidden\" name=\"keyhash\" value=\"").append(keyhash).append("\">")
.append("<input type=\"hidden\" name=\"urlstring\" value=\"\">")
.append("<input type=\"hidden\" name=\"urlhash\" value=\"\">")
.append("<br><fieldset><legend>Reference Deletion</legend><table border=\"0\" cellspacing=\"5\" cellpadding=\"5\"><tr valign=\"top\"><td><br><br>")
.append("<input type=\"submit\" value=\"Delete reference to selected URLs\" name=\"keyhashdelete\"><br><br>")
.append("<input type=\"submit\" value=\"Delete reference to ALL URLs\" name=\"keyhashdeleteall\"><span class=\"small\"><br>&nbsp;&nbsp;(= delete Word)</span>")
.append("</td><td width=\"150\">")
.append("<center><input type=\"checkbox\" name=\"delurl\" value=\"\" align=\"top\" checked></center><br>")
.append("<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>")
.append("</td><td width=\"150\">")
.append("<center><input type=\"checkbox\" name=\"delurlref\" value=\"\" align=\"top\"></center><br>")
.append("<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>")
.append("</td></tr></table></fieldset></form><br>");
}
index.close();
index = null;
return result.toString();
} catch (IOException e) {
return "";
} finally {
if (index != null) try { index.close(); } catch (Exception e) {};
if (index != null) try { index.close(); index = null; } catch (Exception e) {};
}
}

Write Preview
Loading…
Cancel
Save