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 19 years ago
parent 2a18e6486b
commit 5c4db6968e
1 changed files with 218 additions and 152 deletions
Show Stats Download Patch File Download Diff File

370
htroot/IndexControl_p.java
Unescape View File

@ -1,10 +1,13 @@
// IndexControl_p.java // IndexControl_p.java
// ----------------------- // -----------------------
// part of the AnomicHTTPD caching proxy // part of the AnomicHTTPD caching proxy
// (C) by Michael Peter Christen; mc@anomic.de // (C) by Michael Peter Christen; mc@anomic.de
// first published on http://www.anomic.de // first published on http://www.anomic.de
// Frankfurt, Germany, 2004 // Frankfurt, Germany, 2004
// last change: 02.05.2004 //
// $LastChangedDate$
// $LastChangedRevision$
// $LastChangedBy$
// //
// This program is free software; you can redistribute it and/or modify // 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 // 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. // Contributions and changes to the program code must be marked as such.
// You must compile this file with // 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 // if the shell's current path is HTROOT
import java.io.IOException; import java.io.IOException;
@ -50,7 +53,7 @@ import java.util.Enumeration;
import java.util.HashSet; import java.util.HashSet;
import java.util.HashMap; import java.util.HashMap;
import java.util.Iterator; import java.util.Iterator;
import java.util.TreeMap;
import de.anomic.htmlFilter.htmlFilterContentScraper; import de.anomic.htmlFilter.htmlFilterContentScraper;
import de.anomic.http.httpHeader; import de.anomic.http.httpHeader;
import de.anomic.plasma.plasmaCrawlLURL; import de.anomic.plasma.plasmaCrawlLURL;
@ -67,11 +70,11 @@ import de.anomic.yacy.yacySeed;
public class IndexControl_p { public class IndexControl_p {
public static serverObjects respond(httpHeader header, serverObjects post, serverSwitch env) { public static serverObjects respond(httpHeader header, serverObjects post, serverSwitch env) {
// return variable that accumulates replacements // return variable that accumulates replacements
plasmaSwitchboard switchboard = (plasmaSwitchboard) env; plasmaSwitchboard switchboard = (plasmaSwitchboard) env;
serverObjects prop = new serverObjects(); serverObjects prop = new serverObjects();
if ((post == null) || (env == null)) { if (post == null || env == null) {
prop.put("keystring", ""); prop.put("keystring", "");
prop.put("keyhash", ""); prop.put("keyhash", "");
prop.put("urlstring", ""); prop.put("urlstring", "");
@ -81,51 +84,69 @@ public class IndexControl_p {
prop.put("ucount", Integer.toString(switchboard.urlPool.loadedURL.size())); prop.put("ucount", Integer.toString(switchboard.urlPool.loadedURL.size()));
prop.put("otherHosts", ""); prop.put("otherHosts", "");
prop.put("indexDistributeChecked", (switchboard.getConfig("allowDistributeIndex", "true").equals("true")) ? "checked" : ""); prop.put("indexDistributeChecked", (switchboard.getConfig("allowDistributeIndex", "true").equals("true")) ? "checked" : "");
prop.put("indexDistributeWhileCrawling", (switchboard.getConfig("allowDistributeIndexWhileCrawling", "true").equals("true")) ? "checked" : ""); prop.put("indexDistributeWhileCrawling", (switchboard.getConfig("allowDistributeIndexWhileCrawling", "true").equals("true")) ? "checked" : "");
prop.put("indexReceiveChecked", (switchboard.getConfig("allowReceiveIndex", "true").equals("true")) ? "checked" : ""); prop.put("indexReceiveChecked", (switchboard.getConfig("allowReceiveIndex", "true").equals("true")) ? "checked" : "");
prop.put("indexReceiveBlockBlacklistChecked", (switchboard.getConfig("indexReceiveBlockBlacklist", "true").equals("true")) ? "checked" : ""); prop.put("indexReceiveBlockBlacklistChecked", (switchboard.getConfig("indexReceiveBlockBlacklist", "true").equals("true")) ? "checked" : "");
return prop; // be save return prop; // be save
} }
// default values // default values
String keystring = ((String) post.get("keystring")).trim(); String keystring = ((String) post.get("keystring")).trim();
String keyhash = ((String) post.get("keyhash")).trim(); String keyhash = ((String) post.get("keyhash")).trim();
String urlstring = ((String) post.get("urlstring")).trim(); String urlstring = ((String) post.get("urlstring")).trim();
String urlhash = ((String) post.get("urlhash")).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("keystring", keystring);
prop.put("keyhash", keyhash); prop.put("keyhash", keyhash);
prop.put("urlstring", urlstring); prop.put("urlstring", urlstring);
prop.put("urlhash", urlhash); prop.put("urlhash", urlhash);
prop.put("result", ""); prop.put("result", "");
// read values from checkboxes // read values from checkboxes
String[] urlx = post.getAll("urlhx.*"); String[] urlx = post.getAll("urlhx.*");
boolean delurl = post.containsKey("delurl"); boolean delurl = post.containsKey("delurl");
boolean delurlref = post.containsKey("delurlref"); 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")) { if (post.containsKey("setIndexTransmission")) {
boolean allowDistributeIndex = ((String) post.get("indexDistribute", "")).equals("on"); if (post.get("indexDistribute", "").equals("on")) {
switchboard.setConfig("allowDistributeIndex", (allowDistributeIndex) ? "true" : "false"); switchboard.setConfig("allowDistributeIndex", "true");
if (allowDistributeIndex) switchboard.indexDistribution.enable(); else switchboard.indexDistribution.disable(); switchboard.indexDistribution.enable();
} else {
boolean allowDistributeIndexWhileCrawling = post.containsKey("indexDistributeWhileCrawling"); switchboard.setConfig("allowDistributeIndex", "false");
switchboard.setConfig("allowDistributeIndexWhileCrawling", (allowDistributeIndexWhileCrawling) ? "true" : "false"); switchboard.indexDistribution.disable();
if (allowDistributeIndexWhileCrawling) switchboard.indexDistribution.enableWhileCrawling(); else switchboard.indexDistribution.disableWhileCrawling(); }
boolean allowReceiveIndex = ((String) post.get("indexReceive", "")).equals("on"); if (post.containsKey("indexDistributeWhileCrawling")) {
switchboard.setConfig("allowReceiveIndex", (allowReceiveIndex) ? "true" : "false"); switchboard.setConfig("allowDistributeIndexWhileCrawling", "true");
yacyCore.seedDB.mySeed.setFlagAcceptRemoteIndex(allowReceiveIndex); switchboard.indexDistribution.enableWhileCrawling();
} else {
boolean indexReceiveBlockBlacklist = ((String) post.get("indexReceiveBlockBlacklist", "")).equals("on"); switchboard.setConfig("allowDistributeIndexWhileCrawling", "false");
switchboard.setConfig("indexReceiveBlockBlacklist", (indexReceiveBlockBlacklist) ? "true" : "false"); switchboard.indexDistribution.disableWhileCrawling();
}
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);
}
if (post.get("indexReceiveBlockBlacklist", "").equals("on")) {
switchboard.setConfig("indexReceiveBlockBlacklist", "true");
} else {
switchboard.setConfig("indexReceiveBlockBlacklist", "false");
}
} }
// delete word
if (post.containsKey("keyhashdeleteall")) { if (post.containsKey("keyhashdeleteall")) {
if ((delurl) || (delurlref)) { if (delurl || delurlref) {
// generate an urlx array // generate an urlx array
plasmaWordIndexEntity index = null; plasmaWordIndexEntity index = null;
try { try {
@ -135,43 +156,67 @@ public class IndexControl_p {
Enumeration en = index.elements(true); Enumeration en = index.elements(true);
int i = 0; int i = 0;
urlx = new String[index.size()]; urlx = new String[index.size()];
while (en.hasMoreElements()) urlx[i++] = ((plasmaWordIndexEntry) en.nextElement()).getUrlHash(); while (en.hasMoreElements()) {
index.close(); index = null; urlx[i++] = ((plasmaWordIndexEntry) en.nextElement()).getUrlHash();
}
index.close();
index = null;
} catch (IOException e) { } catch (IOException e) {
urlx = new String[0]; urlx = new String[0];
} finally { } finally {
if (index != null) try { index.close(); } catch (Exception e) {} 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 (delurlref) {
if ((delurl) || (delurlref)) for (int i = 0; i < urlx.length; i++) switchboard.urlPool.loadedURL.remove(urlx[i]); 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); switchboard.wordIndex.deleteIndex(keyhash);
post.remove("keyhashdeleteall"); 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"); post.put("keystringsearch", "generated");
else } else {
post.put("keyhashsearch", "generated"); post.put("keyhashsearch", "generated");
}
} }
// delete selected URLs
if (post.containsKey("keyhashdelete")) { if (post.containsKey("keyhashdelete")) {
if (delurlref) for (int i = 0; i < urlx.length; i++) switchboard.removeAllUrlReferences(urlx[i], true); if (delurlref) {
if ((delurl) || (delurlref)) for (int i = 0; i < urlx.length; i++) switchboard.urlPool.loadedURL.remove(urlx[i]); 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); switchboard.wordIndex.removeEntries(keyhash, urlx, true);
// this shall lead to a presentation of the list; so handle that the remaining program // this shall lead to a presentation of the list; so handle that the remaining program
// thinks that it was called for a list presentation // thinks that it was called for a list presentation
post.remove("keyhashdelete"); 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"); post.put("keystringsearch", "generated");
else } else {
post.put("keyhashsearch", "generated"); 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")) { if (post.containsKey("urlhashdeleteall")) {
int i = switchboard.removeAllUrlReferences(urlhash, true); int i = switchboard.removeAllUrlReferences(urlhash, true);
prop.put("result", "Deleted URL and " + i + " references from " + i + " word indexes."); prop.put("result", "Deleted URL and " + i + " references from " + i + " word indexes.");
} }
if (post.containsKey("urlhashdelete")) { if (post.containsKey("urlhashdelete")) {
plasmaCrawlLURL.Entry entry = switchboard.urlPool.loadedURL.getEntry(urlhash); plasmaCrawlLURL.Entry entry = switchboard.urlPool.loadedURL.getEntry(urlhash);
URL url = entry.url(); URL url = entry.url();
@ -192,18 +237,23 @@ public class IndexControl_p {
prop.put("urlhash", ""); prop.put("urlhash", "");
prop.put("result", genUrlList(switchboard, keyhash, keystring)); prop.put("result", genUrlList(switchboard, keyhash, keystring));
} }
if (post.containsKey("keyhashsearch")) { 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("keystring", "<not possible to compute word from hash>");
}
prop.put("urlstring", ""); prop.put("urlstring", "");
prop.put("urlhash", ""); prop.put("urlhash", "");
prop.put("result", genUrlList(switchboard, keyhash, "")); prop.put("result", genUrlList(switchboard, keyhash, ""));
} }
// transfer to other peer
if (post.containsKey("keyhashtransfer")) { 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("keystring", "<not possible to compute word from hash>");
}
prop.put("urlstring", ""); prop.put("urlstring", "");
prop.put("urlhash", ""); prop.put("urlhash", "");
plasmaWordIndexEntity[] indexes = new plasmaWordIndexEntity[1]; plasmaWordIndexEntity[] indexes = new plasmaWordIndexEntity[1];
@ -232,43 +282,46 @@ public class IndexControl_p {
} }
// now delete all entries that have no url entry // now delete all entries that have no url entry
Iterator hashIter = unknownURLEntries.iterator(); Iterator hashIter = unknownURLEntries.iterator();
while (hashIter.hasNext()) try { while (hashIter.hasNext()) {
indexes[0].removeEntry((String) hashIter.next(), false); try {
} catch (IOException e) {} indexes[0].removeEntry((String) hashIter.next(), false);
// use whats remaining } catch (IOException e) {}
}
// use whats remaining
String gzipBody = switchboard.getConfig("indexControl.gzipBody","false"); String gzipBody = switchboard.getConfig("indexControl.gzipBody","false");
int timeout = (int) switchboard.getConfigLong("indexControl.timeout",60000); int timeout = (int) switchboard.getConfigLong("indexControl.timeout",60000);
result = yacyClient.transferIndex( result = yacyClient.transferIndex (
yacyCore.seedDB.getConnected(post.get("hostHash", "")), yacyCore.seedDB.getConnected(post.get("hostHash", "")),
indexes, indexes,
knownURLs, knownURLs,
"true".equalsIgnoreCase(gzipBody), "true".equalsIgnoreCase(gzipBody),
timeout); timeout);
prop.put("result", (result == null) ? ("Successfully transferred " + indexes[0].size() + " words in " + ((System.currentTimeMillis() - starttime) / 1000) + " seconds") : result); prop.put("result", (result == null) ? ("Successfully transferred " + indexes[0].size() + " words in " + ((System.currentTimeMillis() - starttime) / 1000) + " seconds") : result);
try {indexes[0].close();} catch (IOException e) {} try {indexes[0].close();} catch (IOException e) {}
} }
// generate list
if (post.containsKey("keyhashsimilar")) { if (post.containsKey("keyhashsimilar")) {
Iterator hashIt = switchboard.wordIndex.wordHashes(keyhash, true, true); final Iterator hashIt = switchboard.wordIndex.wordHashes(keyhash, true, true);
String result = "Sequential List of Word-Hashes:<br>"; StringBuffer result = new StringBuffer("Sequential List of Word-Hashes:<br>");
String hash; String hash;
int i = 0; int i = 0;
while ((hashIt.hasNext()) && (i < 256)) { while (hashIt.hasNext() && i < 256) {
hash = (String) hashIt.next(); hash = (String) hashIt.next();
result += "<a href=\"/IndexControl_p.html?" + result.append("<a href=\"/IndexControl_p.html?")
"keystring=" + .append("keystring=")
"&keyhash=" + hash + .append("&keyhash=").append(hash)
"&urlhash=" + .append("&urlhash=")
"&urlstring=" + .append("&urlstring=")
"&keyhashsearch=" + .append("&urlhashsearch=")
"\" class=\"tt\">" + hash + "</a> " + (((i + 1) % 8 == 0) ? "<br>" : ""); .append("\" class=\"tt\">").append(hash).append("</a> ")
i++; .append(((i + 1) % 8 == 0) ? "<br>" : "");
} i++;
prop.put("result", result); }
prop.put("result", result);
} }
if (post.containsKey("urlstringsearch")) { if (post.containsKey("urlstringsearch")) {
try { try {
URL url = new URL(urlstring); URL url = new URL(urlstring);
urlhash = plasmaURL.urlHash(url); urlhash = plasmaURL.urlHash(url);
@ -280,7 +333,7 @@ public class IndexControl_p {
prop.put("urlhash", ""); prop.put("urlhash", "");
} }
} }
if (post.containsKey("urlhashsearch")) { if (post.containsKey("urlhashsearch")) {
plasmaCrawlLURL.Entry entry = switchboard.urlPool.loadedURL.getEntry(urlhash); plasmaCrawlLURL.Entry entry = switchboard.urlPool.loadedURL.getEntry(urlhash);
URL url = entry.url(); URL url = entry.url();
@ -293,62 +346,64 @@ public class IndexControl_p {
} }
} }
if (post.containsKey("urlhashsimilar")) { // generate list
if (post.containsKey("urlhashsimilar")) {
try { try {
Iterator hashIt = switchboard.urlPool.loadedURL.urlHashes(urlhash, true); final Iterator hashIt = switchboard.urlPool.loadedURL.urlHashes(urlhash, true);
String result = "Sequential List of URL-Hashes:<br>"; StringBuffer result = new StringBuffer("Sequential List of URL-Hashes:<br>");
String hash; String hash;
int i = 0; int i = 0;
while ((hashIt.hasNext()) && (i < 256)) { while (hashIt.hasNext() && i < 256) {
hash = (String) hashIt.next(); hash = (String) hashIt.next();
result += "<a href=\"/IndexControl_p.html?" + result.append("<a href=\"/IndexControl_p.html?")
"keystring=" + .append("keystring=")
"&keyhash=" + .append("&keyhash=")
"&urlhash=" + hash + .append("&urlhash=").append(hash)
"&urlstring=" + .append("&urlstring=")
"&urlhashsearch=" + .append("&urlhashsearch=")
"\" class=\"tt\">" + hash + "</a> " + (((i + 1) % 8 == 0) ? "<br>" : ""); .append("\" class=\"tt\">").append(hash).append("</a> ")
.append(((i + 1) % 8 == 0) ? "<br>" : "");
i++; i++;
} }
prop.put("result", result); prop.put("result", result.toString());
} catch (IOException e) { } catch (IOException e) {
prop.put("result", "Error: " + e.getMessage()); prop.put("result", "Error: " + e.getMessage());
} }
} }
//List known hosts // list known hosts
yacySeed seed; yacySeed seed;
int hc = 0; 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); Enumeration e = yacyCore.dhtAgent.getAcceptRemoteIndexSeeds(keyhash);
while (e.hasMoreElements()) { while (e.hasMoreElements()) {
seed = (yacySeed) e.nextElement(); seed = (yacySeed) e.nextElement();
if (seed != null) { if (seed != null) {
prop.put("hosts_" + hc + "_hosthash", seed.hash); prop.put("hosts_" + hc + "_hosthash", seed.hash);
prop.put("hosts_" + hc + "_hostname", /*seed.hash + " " +*/ seed.get("Name", "nameless")); prop.put("hosts_" + hc + "_hostname", /*seed.hash + " " +*/ seed.get("Name", "nameless"));
hc++; hc++;
} }
} }
prop.put("hosts", Integer.toString(hc)); prop.put("hosts", Integer.toString(hc));
} else { } else {
prop.put("hosts", "0"); prop.put("hosts", "0");
} }
// insert constants // insert constants
prop.put("wcount", Integer.toString(switchboard.wordIndex.size())); prop.put("wcount", Integer.toString(switchboard.wordIndex.size()));
prop.put("ucount", Integer.toString(switchboard.urlPool.loadedURL.size())); prop.put("ucount", Integer.toString(switchboard.urlPool.loadedURL.size()));
prop.put("indexDistributeChecked", (switchboard.getConfig("allowDistributeIndex", "true").equals("true")) ? "checked" : ""); prop.put("indexDistributeChecked", (switchboard.getConfig("allowDistributeIndex", "true").equals("true")) ? "checked" : "");
prop.put("indexDistributeWhileCrawling", (switchboard.getConfig("allowDistributeIndexWhileCrawling", "true").equals("true")) ? "checked" : ""); prop.put("indexDistributeWhileCrawling", (switchboard.getConfig("allowDistributeIndexWhileCrawling", "true").equals("true")) ? "checked" : "");
prop.put("indexReceiveChecked", (switchboard.getConfig("allowReceiveIndex", "true").equals("true")) ? "checked" : ""); prop.put("indexReceiveChecked", (switchboard.getConfig("allowReceiveIndex", "true").equals("true")) ? "checked" : "");
prop.put("indexReceiveBlockBlacklistChecked", (switchboard.getConfig("indexReceiveBlockBlacklist", "true").equals("true")) ? "checked" : ""); prop.put("indexReceiveBlockBlacklistChecked", (switchboard.getConfig("indexReceiveBlockBlacklist", "true").equals("true")) ? "checked" : "");
// return rewrite properties // return rewrite properties
return prop; return prop;
} }
public static String genUrlProfile(plasmaSwitchboard switchboard, plasmaCrawlLURL.Entry entry, String urlhash) { 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(); 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>" + String result = "<table>" +
"<tr><td class=\"small\">URL String</td><td class=\"tt\">" + htmlFilterContentScraper.urlNormalform(url) + "</td></tr>" + "<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>" + "<tr><td class=\"small\">Hash</td><td class=\"tt\">" + urlhash + "</td></tr>" +
@ -374,66 +429,77 @@ public class IndexControl_p {
"<span class=\"small\">&nbsp;this may produce unresolved references at other word indexes but they do not harm</span><br><br>" + "<span class=\"small\">&nbsp;this may produce unresolved references at other word indexes but they do not harm</span><br><br>" +
"<input type=\"submit\" value=\"Delete URL and remove all references from words\" name=\"urlhashdeleteall\"><br>" + "<input type=\"submit\" value=\"Delete URL and remove all references from words\" name=\"urlhashdeleteall\"><br>" +
"<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>" + "<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>"; "</form>";
return result; return result;
} }
public static String genUrlList(plasmaSwitchboard switchboard, String keyhash, String keystring) { public static String genUrlList(plasmaSwitchboard switchboard, String keyhash, String keystring) {
// search for a word hash and generate a list of url links // search for a word hash and generate a list of url links
plasmaWordIndexEntity index = null; plasmaWordIndexEntity index = null;
try { try {
HashSet keyhashes = new HashSet(); final HashSet keyhashes = new HashSet();
keyhashes.add(keyhash); keyhashes.add(keyhash);
index = switchboard.searchManager.searchHashes(keyhashes, 10000); index = switchboard.searchManager.searchHashes(keyhashes, 10000);
String result = "";
final StringBuffer result = new StringBuffer(1024);
if (index.size() == 0) { 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 { } else {
Enumeration en = index.elements(true); final Enumeration en = index.elements(true);
plasmaWordIndexEntry ie; result.append("URL entries related to this word hash <span class=\"tt\">").append(keyhash).append("</span><br>");
result = "URL entries related to this word hash <span class=\"tt\">" + keyhash + "</span>:<br>"; result.append("<form action=\"IndexControl_p.html\" method=\"post\" enctype=\"multipart/form-data\">");
result += "<form action=\"IndexControl_p.html\" method=\"post\" enctype=\"multipart/form-data\">";
String us, uh; String us, uh;
int i = 0; int i = 0;
final TreeMap tm = new TreeMap();
while (en.hasMoreElements()) { while (en.hasMoreElements()) {
ie = (plasmaWordIndexEntry) en.nextElement(); uh = ((plasmaWordIndexEntry)en.nextElement()).getUrlHash();
uh = ie.getUrlHash();
result +=
"<input type=\"checkbox\" name=\"urlhx" + i++ + "\" value=\"" + uh + "\" align=\"top\">";
if (switchboard.urlPool.loadedURL.exists(uh)) { if (switchboard.urlPool.loadedURL.exists(uh)) {
us = switchboard.urlPool.loadedURL.getEntry(uh).url().toString(); us = switchboard.urlPool.loadedURL.getEntry(uh).url().toString();
result += tm.put(us, uh);
"<a href=\"/IndexControl_p.html?" + "keystring=" + keystring + } else {
"&keyhash=" + keyhash + "&urlhash=" + uh + "&urlstringsearch=" + "&urlstring=" + us + tm.put("", uh);
"\" class=\"tt\">" + uh + "</a><span class=\"tt\">&nbsp;" + us + "</span><br>"; }
}
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 { } else {
result += result.append("<span class=\"tt\">").append(uh).append("&nbsp;&lt;unresolved URL Hash&gt;</span><br>");
"<span class=\"tt\">" + uh + "&nbsp;&lt;unresolved URL Hash&gt;</span><br>";
} }
} }
result += result.append("<input type=\"hidden\" name=\"keystring\" value=\"").append(keystring).append("\">")
"<input type=\"hidden\" name=\"keystring\" value=\"" + keystring + "\">" + .append("<input type=\"hidden\" name=\"keyhash\" value=\"").append(keyhash).append("\">")
"<input type=\"hidden\" name=\"keyhash\" value=\"" + keyhash + "\">" + .append("<input type=\"hidden\" name=\"urlstring\" value=\"\">")
"<input type=\"hidden\" name=\"urlstring\" value=\"\">" + .append("<input type=\"hidden\" name=\"urlhash\" value=\"\">")
"<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>")
"<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>")
"<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>")
"<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\">")
"</td><td width=\"150\">" + .append("<center><input type=\"checkbox\" name=\"delurl\" value=\"\" align=\"top\" checked></center><br>")
"<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>")
"<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\">")
"</td><td width=\"150\">" + .append("<center><input type=\"checkbox\" name=\"delurlref\" value=\"\" align=\"top\"></center><br>")
"<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>")
"<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>");
"</td></tr></table></fieldset></form>";
} }
index.close(); index = null; index.close();
return result; index = null;
return result.toString();
} catch (IOException e) { } catch (IOException e) {
return ""; return "";
} finally { } 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