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

- re-implemented image search

Browse Source 
- generalized search result status bar, is now also visible during text search


git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4077 6c8d7289-2bf4-0310-a012-ef5d649a1542
pull/1/head
orbiter 18 years ago
parent 6c819a6fd9
commit 6c3bcadc1c
7 changed files with 221 additions and 537 deletions
Show Stats Download Patch File Download Diff File

36
htroot/ViewFile.soap
Unescape View File

@ -1,36 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<yacyUrlInfo>
<!-- Resource metadata -->
<meta>
<url>#(error)##[url]#::#(/error)#</url>
<hash>#(error)##[hash]#::#(/error)#</hash>
<wordCount>#(error)##[wordCount]#::#(/error)#</wordCount>
<desc>#(error)##[desc]#::#(/error)#</desc>
<size>#(error)##[size]#::#(/error)#</size>
<mimeType>#(error)##[mimeType]#::#(/error)#</mimeType>
<statusCode>#(error)#0::1::2::3::4::5::6#(/error)#</statusCode>
<statusText>
#(error)#OK
::No URL hash submitted
::Unable to find URL Entry in DB
::Invalid URL
::Unable to download resource content: #[errorText]#
::Unable to parse resource content: #[errorText]#
::Unsupported protocol
#(/error)#
</statusText>
</meta>
<!-- resource body -->
<body mode="#(viewMode)#::1::2::3#(/viewMode)#">
#(viewMode)#
:: <!-- 1 -->
<![CDATA[#[plainText]#]]>
:: <!-- 2 -->
<![CDATA[#[parsedText]#]]>
:: <!-- 3 -->
#{sentences}#
<sentence nr="#[nr]#"><![CDATA[#[text]#]]></sentence>
#{/sentences}#
#(/viewMode)#
</body>
</yacyUrlInfo>

159
htroot/js/yacysearch.js
Unescape View File

@ -51,165 +51,6 @@ function Progressbar(length, parent) {
parent.appendChild(this.element);
}
function handleTextState(req) {
if(req.readyState != 4){
return;
}
var response = req.responseXML;
var snippetText = response.getElementsByTagName("text")[0].firstChild.data;
var urlHash = response.getElementsByTagName("urlHash")[0].firstChild.data;
var status = response.getElementsByTagName("status")[0].firstChild.data;
var span = document.getElementById("h" + urlHash);
removeAllChildren(span);
//span.removeChild(span.firstChild);
if (status < 11) {
span.className = "snippetLoaded";
} else {
span.className = "snippetError";
span.parentNode.parentNode.className = "searchresults hidden";
document.getElementById("hidden_results").innerHTML='Some results were hidden, because they do not contain your searchwords anymore, or because they are not accessible. Click here to <a href="javascript:show_hidden_results()">show them</a>';
}
// set URL to favicon (if a link-tag was found in the document)
if (response.getElementsByTagName("favicon")[0].firstChild != null) {
var img = document.getElementById("f" + urlHash);
img.src = "ViewImage.png?width=16&height=16&url=" + response.getElementsByTagName("favicon")[0].firstChild.data;
}
// replace "<b>" text by <strong> node
var pos1=snippetText.indexOf("<b>");
var pos2=snippetText.indexOf("</b>");
while (pos1 >= 0 && pos2 > pos1) {
leftString = document.createTextNode(snippetText.substring(0, pos1)); //other text
if (leftString != "") span.appendChild(leftString);
//add the bold text
strongNode=document.createElement("strong");
middleString=document.createTextNode(snippetText.substring(pos1 + 3, pos2));
strongNode.appendChild(middleString);
span.appendChild(strongNode);
// cut out left and middle and go on with remaining text
snippetText=snippetText.substring(pos2 + 4);
pos1=snippetText.indexOf("<b>");
pos2=snippetText.indexOf("</b>");
}
// add remaining string
if (snippetText != "") {
span.appendChild(document.createTextNode(snippetText));
}
}
function handleMediaState(req, progressbar) {
if(req.readyState != 4){
return;
}
var response = req.responseXML;
// On errors, the result might not contain what we look for...
if (response.getElementsByTagName("urlHash")) {
var urlHash = response.getElementsByTagName("urlHash")[0].firstChild.data;
var links = response.getElementsByTagName("links")[0].firstChild.data;
var container = document.getElementById("results");
if (links > 0) {
for (i = 0; i < links; i++) {
var type = response.getElementsByTagName("type")[i].firstChild.data;
var href = response.getElementsByTagName("href")[i].firstChild.data;
var name = response.getElementsByTagName("name")[i].firstChild.data;
var attr = response.getElementsByTagName("attr")[i].firstChild.data;
var link = document.createElement("a");
link.setAttribute("href", href);
link.setAttribute("target", "_parent");
link.appendChild(document.createTextNode(name));
var title = document.createElement("h4");
title.className = "linktitle";
title.appendChild(link);
var urllink = document.createElement("a");
urllink.setAttribute("href", href);
urllink.setAttribute("target", "_parent");
urllink.appendChild(document.createTextNode(href));
var url = document.createElement("p");
url.className = "url";
url.appendChild(urllink);
var searchresultcontainer = document.createElement("div");
searchresultcontainer.className = "searchresults";
searchresultcontainer.appendChild(title);
searchresultcontainer.appendChild(url);
container.appendChild(searchresultcontainer);
document.getElementById("linkcount").innerHTML++;
}
}
}
progressbar.step(1);
}
function handleImageState(req, progressbar) {
if(req.readyState != 4)
return;
var response = req.responseXML;
// on errors, the result might not contain the expected elements and would throw an error, so we check for it here
if (response.getElementsByTagName("urlHash")) {
// the urlHash is not needed anymore at the moment
//var urlHash = response.getElementsByTagName("urlHash")[0].firstChild.data;
var links = response.getElementsByTagName("links")[0].firstChild.data;
var div = document.getElementById("results")
if (links > 0) {
for (i = 0; i < links; i++) {
var type = response.getElementsByTagName("type")[i].firstChild.data;
var href = response.getElementsByTagName("href")[i].firstChild.data;
var code = response.getElementsByTagName("code")[i].firstChild.data;
var name = response.getElementsByTagName("name")[i].firstChild.data;
var attr = response.getElementsByTagName("attr")[i].firstChild.data;
var img = document.createElement("img");
img.setAttribute("src", "/ViewImage.png?maxwidth=96&maxheight=96&code=" + code);
img.setAttribute("alt", name);
var imganchor = document.createElement("a");
imganchor.setAttribute("href", href);
imganchor.className = "thumblink";
imganchor.appendChild(img);
var nameanchor = document.createElement("a");
nameanchor.setAttribute("href", href);
nameanchor.appendChild(document.createTextNode(name));
var textcontainer = document.createElement("div");
textcontainer.className = "TableCellDark";
textcontainer.appendChild(nameanchor);
var thumbcontainer = document.createElement("div");
thumbcontainer.className = "thumbcontainer";
thumbcontainer.appendChild(imganchor);
thumbcontainer.appendChild(textcontainer);
div.appendChild(thumbcontainer);
//span.appendChild(imganchor);
document.getElementById("linkcount").innerHTML++;
}
}
}
progressbar.step(1);
}
function addHover() {
if (document.all&&document.getElementById) {
var divs = document.getElementsByTagName("div");

103
htroot/yacysearch.html
Unescape View File

@ -103,100 +103,27 @@ document.getElementById("Enter").value = "search again - catch up more links";
::
<p>No Results. &quot;<strong>#[wrong_regex]#</strong>&quot; is no valid regular expression. Please go back to the previous page and make sure to enter a valid regular expressions for URL mask and Prefer mask.</p>
::
<p><strong id="offset">#[offset]#</strong>-<strong id="itemscount">#[linkcount]#</strong> results from a total number of <strong id="totalcount">#[totalcount]#</strong> known#(globalresults)#.::, <strong id="globalcount">#[globalcount]#</strong> links from other YaCy peers.#(/globalresults)#</p>
<p><strong id="offset">#[offset]#</strong>-<strong id="itemscount">#[linkcount]#</strong> results from a total number of <strong id="totalcount">#[totalcount]#</strong> known#(globalresults)#.::, <strong id="globalcount">#[globalcount]#</strong> links from other YaCy peers.#(/globalresults)#<div id="results"></div></p>
<p>Search Result Pages: <span id="pagenav">#[pagenav]#</span></p>
::
<p>Searching the web with this peer is disabled for unauthorized users. Please <a href="Status.html?login=">log in</a> as administrator to use the search function</p>
#(/num-results)#
<!-- view the result list -->
#(type)#
<!-- type 0: text search -->
#(combine)#
::
<p><strong>Refine with</strong>:</p>
<p>
#{words}#
<a href="yacysearch.html?search=#[newsearch]#&amp;Enter=Search&amp;count=#[count]#&amp;offset=#[offset]#&amp;resource=#[resource]#&amp;time=#[time]#">#[word]#</a>
#{/words}#
</p>
#(/combine)#
<!-- linklist begin -->
<div id="hidden_results"></div>
#{results}#
<!--#include virtual="yacysearchitem.html?item=#[item]#&eventID=#[eventID]#&display=#[display]#" -->
#{/results}#
<!-- linklist end -->
<!-- the search results -->
#(combine)#::
<p><strong>Refine with</strong>:</p>
<p>#{words}#<a href="yacysearch.html?search=#[newsearch]#&amp;Enter=Search&amp;count=#[count]#&amp;offset=#[offset]#&amp;resource=#[resource]#&amp;time=#[time]#">#[word]#</a>#{/words}#</p>
#(/combine)#
<script type="text/javascript">
var progressbar = new Progressbar(#[results]#, document.getElementById("results"));
</script>
::<!-- type 1: media search -->
<!-- non-js variant -->
<noscript>
<p>
Sorry, at the moment you need JavaScript in order to use this search. But here are the urls where YaCy would have looked for your searchresults. Perhaps you can find them there. Sorry for the inconvenience. If you want a non-JavaScript search, beg the developers for it.
</p>
<!-- linklist begin -->
<ul>
#{results}#
<!-- link begin -->
<li><a href="#[url]#">#[urlname]#</a></li>
<!-- link end -->
#{/results}#
</ul>
<!-- linklist end -->
</noscript>
<!-- linklist start-->
<div id="results">
</div>
<!-- linklist end-->
<script type="text/javascript">
var urls = "#{results}##[urlname]# #{/results}#".split(" ");
// the last element is always empty because the split gets a space at the end of the string
urls.length--;
if(urls.length > 0) {
AllMediaSnippets(urls, "#[former]#", "#[mediatype]#");
}
</script>
::<!-- type 2: image search: presents image thumbnails -->
<noscript>
<p>
Sorry, but in order to use the image search, you need JavaScript. But here are the urls where YaCy would have looked for the images you want. Perhaps you can find them there. Sorry for the inconvenience. If you want a non-JavaScript image search, beg the developers for it.
</p>
<ul>
#{results}#
<li><a href="#[url]#">#[urlname]#</a></li>
#{/results}#
</ul>
</noscript>
<!-- linklist start -->
<div id="results">
</div>
<!-- linklist end -->
<script type="text/javascript">
var urls = "#{results}##[urlname]# #{/results}#".split(" ");
// the last element is always empty because the split gets a space at the end of the string
urls.length--;
if(urls.length > 0) {
AllImageSnippets(urls, "#[former]#");
}
</script>
<br style="clear:left;" />
::<!-- type 3: image thumbnail list for one single url -->
<table border="0" cellspacing="16" cellpadding="0">
#{results}#
<tr valign="bottom">
#{line}#
<td>
<a href="#[url]#"><img src="/ViewImage.png?maxwidth=96&amp;maxheight=96&amp;url=#[url]#" /></a><br /><a href="#[url]#">#[name]#</a>
</td>
#{/line}#
</tr>
#{/results}#
</table>
#(/type)#
<!-- linklist begin -->
<div id="hidden_results"></div>
#{results}#
<!--#include virtual="yacysearchitem.html?item=#[item]#&eventID=#[eventID]#" -->
#{/results}#
<!-- linklist end -->
</body>
</html>

324
htroot/yacysearch.java
Unescape View File

@ -52,7 +52,6 @@ import java.util.HashMap;
import java.util.Iterator;
import java.util.TreeSet;
import de.anomic.htmlFilter.htmlFilterImageEntry;
import de.anomic.http.httpHeader;
import de.anomic.index.indexURLEntry;
import de.anomic.kelondro.kelondroBitfield;
@ -61,7 +60,6 @@ import de.anomic.kelondro.kelondroNaturalOrder;
import de.anomic.plasma.plasmaCondenser;
import de.anomic.plasma.plasmaParserDocument;
import de.anomic.plasma.plasmaSearchEvent;
import de.anomic.plasma.plasmaSearchImages;
import de.anomic.plasma.plasmaSearchQuery;
import de.anomic.plasma.plasmaSearchRankingProfile;
import de.anomic.plasma.plasmaSearchProcessing;
@ -141,10 +139,9 @@ public class yacysearch {
prop.put("input_contentdomCheckVideo", 0);
prop.put("input_contentdomCheckImage", 0);
prop.put("input_contentdomCheckApp", 0);
prop.put("type", 0);
prop.put("type_excluded", 0);
prop.put("type_combine", 0);
prop.put("type_results", "");
prop.put("excluded", 0);
prop.put("combine", 0);
prop.put("results", "");
prop.put("num-results", (searchAllowed) ? 0 : 6);
return prop;
@ -193,70 +190,70 @@ public class yacysearch {
int contentdomCode = plasmaSearchQuery.contentdomParser(post.get("contentdom", "text"));
// patch until better search profiles are available
if ((contentdomCode != plasmaSearchQuery.CONTENTDOM_TEXT) && (count <= 10)) count = 30;
if ((contentdomCode != plasmaSearchQuery.CONTENTDOM_TEXT) && (count <= 30)) count = 30;
serverObjects prop = new serverObjects();
if (post.get("cat", "href").equals("href")) {
final TreeSet[] query = plasmaSearchQuery.cleanQuery(querystring); // converts also umlaute
// filter out stopwords
final TreeSet filtered = kelondroMSetTools.joinConstructive(query[0], plasmaSwitchboard.stopwords);
if (filtered.size() > 0) {
kelondroMSetTools.excludeDestructive(query[0], plasmaSwitchboard.stopwords);
}
// if a minus-button was hit, remove a special reference first
if (post.containsKey("deleteref")) {
if (!sb.verifyAuthentication(header, true)) {
prop.put("AUTHENTICATE", "admin log-in"); // force log-in
return prop;
final TreeSet[] query = plasmaSearchQuery.cleanQuery(querystring); // converts also umlaute
// filter out stopwords
final TreeSet filtered = kelondroMSetTools.joinConstructive(query[0], plasmaSwitchboard.stopwords);
if (filtered.size() > 0) {
kelondroMSetTools.excludeDestructive(query[0], plasmaSwitchboard.stopwords);
}
// delete the index entry locally
final String delHash = post.get("deleteref", ""); // urlhash
sb.wordIndex.removeWordReferences(query[0], delHash);
// make new news message with negative voting
HashMap map = new HashMap();
map.put("urlhash", delHash);
map.put("vote", "negative");
map.put("refid", "");
yacyCore.newsPool.publishMyNews(yacyNewsRecord.newRecord(yacyNewsPool.CATEGORY_SURFTIPP_VOTE_ADD, map));
}
// if a plus-button was hit, create new voting message
if (post.containsKey("recommendref")) {
if (!sb.verifyAuthentication(header, true)) {
prop.put("AUTHENTICATE", "admin log-in"); // force log-in
return prop;
// if a minus-button was hit, remove a special reference first
if (post.containsKey("deleteref")) {
if (!sb.verifyAuthentication(header, true)) {
prop.put("AUTHENTICATE", "admin log-in"); // force log-in
return prop;
}
// delete the index entry locally
final String delHash = post.get("deleteref", ""); // urlhash
sb.wordIndex.removeWordReferences(query[0], delHash);
// make new news message with negative voting
HashMap map = new HashMap();
map.put("urlhash", delHash);
map.put("vote", "negative");
map.put("refid", "");
yacyCore.newsPool.publishMyNews(yacyNewsRecord.newRecord(yacyNewsPool.CATEGORY_SURFTIPP_VOTE_ADD, map));
}
final String recommendHash = post.get("recommendref", ""); // urlhash
indexURLEntry urlentry = sb.wordIndex.loadedURL.load(recommendHash, null);
if (urlentry != null) {
indexURLEntry.Components comp = urlentry.comp();
plasmaParserDocument document;
document = plasmaSnippetCache.retrieveDocument(comp.url(), true, 5000, true);
if (document != null) {
// create a news message
HashMap map = new HashMap();
map.put("url", comp.url().toNormalform(false, true).replace(',', '|'));
map.put("title", comp.title().replace(',', ' '));
map.put("description", ((document == null) ? comp.title() : document.getTitle()).replace(',', ' '));
map.put("author", ((document == null) ? "" : document.getAuthor()));
map.put("tags", ((document == null) ? "" : document.getKeywords(' ')));
yacyCore.newsPool.publishMyNews(yacyNewsRecord.newRecord(yacyNewsPool.CATEGORY_SURFTIPP_ADD, map));
document.close();
// if a plus-button was hit, create new voting message
if (post.containsKey("recommendref")) {
if (!sb.verifyAuthentication(header, true)) {
prop.put("AUTHENTICATE", "admin log-in"); // force log-in
return prop;
}
final String recommendHash = post.get("recommendref", ""); // urlhash
indexURLEntry urlentry = sb.wordIndex.loadedURL.load(recommendHash, null);
if (urlentry != null) {
indexURLEntry.Components comp = urlentry.comp();
plasmaParserDocument document;
document = plasmaSnippetCache.retrieveDocument(comp.url(), true, 5000, true);
if (document != null) {
// create a news message
HashMap map = new HashMap();
map.put("url", comp.url().toNormalform(false, true).replace(',', '|'));
map.put("title", comp.title().replace(',', ' '));
map.put("description", ((document == null) ? comp.title() : document.getTitle()).replace(',', ' '));
map.put("author", ((document == null) ? "" : document.getAuthor()));
map.put("tags", ((document == null) ? "" : document.getKeywords(' ')));
yacyCore.newsPool.publishMyNews(yacyNewsRecord.newRecord(yacyNewsPool.CATEGORY_SURFTIPP_ADD, map));
document.close();
}
}
}
}
// prepare search properties
final boolean yacyonline = ((yacyCore.seedDB != null) && (yacyCore.seedDB.mySeed != null) && (yacyCore.seedDB.mySeed.getPublicAddress() != null));
final boolean globalsearch = (global) && (yacyonline);
// prepare search properties
final boolean yacyonline = ((yacyCore.seedDB != null) && (yacyCore.seedDB.mySeed != null) && (yacyCore.seedDB.mySeed.getPublicAddress() != null));
final boolean globalsearch = (global) && (yacyonline);
// do the search
TreeSet queryHashes = plasmaCondenser.words2hashes(query[0]);
plasmaSearchQuery theQuery = new plasmaSearchQuery(
// do the search
TreeSet queryHashes = plasmaCondenser.words2hashes(query[0]);
plasmaSearchQuery theQuery = new plasmaSearchQuery(
querystring,
queryHashes,
plasmaCondenser.words2hashes(query[1]),
@ -273,95 +270,94 @@ public class yacysearch {
"",
20,
constraint);
plasmaSearchRankingProfile ranking = (sb.getConfig("rankingProfile", "").length() == 0) ? new plasmaSearchRankingProfile(contentdomCode) : new plasmaSearchRankingProfile("", crypt.simpleDecode(sb.getConfig("rankingProfile", ""), null));
plasmaSearchProcessing localTiming = new plasmaSearchProcessing(4 * theQuery.maximumTime / 10, theQuery.displayResults());
plasmaSearchRankingProfile ranking = (sb.getConfig("rankingProfile", "").length() == 0) ? new plasmaSearchRankingProfile(contentdomCode) : new plasmaSearchRankingProfile("", crypt.simpleDecode(sb.getConfig("rankingProfile", ""), null));
plasmaSearchProcessing localTiming = new plasmaSearchProcessing(4 * theQuery.maximumTime / 10, theQuery.displayResults());
String client = (String) header.get("CLIENTIP"); // the search client who initiated the search
String client = (String) header.get("CLIENTIP"); // the search client who initiated the search
// tell all threads to do nothing for a specific time
sb.intermissionAllThreads(2 * theQuery.maximumTime);
// tell all threads to do nothing for a specific time
sb.intermissionAllThreads(2 * theQuery.maximumTime);
// filter out words that appear in bluelist
theQuery.filterOut(plasmaSwitchboard.blueList);
// filter out words that appear in bluelist
theQuery.filterOut(plasmaSwitchboard.blueList);
// log
serverLog.logInfo("LOCAL_SEARCH", "INIT WORD SEARCH: " + theQuery.queryString + ":" + theQuery.queryHashes + " - " + theQuery.neededResults() + " links to be computed, " + theQuery.displayResults() + " lines to be displayed, " + (theQuery.maximumTime / 1000) + " seconds");
long timestamp = System.currentTimeMillis();
// create a new search event
String wrongregex = null;
if (plasmaSearchEvent.getEvent(theQuery.id()) == null) {
theQuery.setOffset(0); // in case that this is a new search, always start without a offset
offset = 0;
}
plasmaSearchEvent theSearch = plasmaSearchEvent.getEvent(theQuery, ranking, localTiming, sb.wordIndex, (sb.isRobinsonMode()) ? sb.clusterhashes : null, false, null);
// log
serverLog.logInfo("LOCAL_SEARCH", "INIT WORD SEARCH: " + theQuery.queryString + ":" + theQuery.queryHashes + " - " + theQuery.neededResults() + " links to be computed, " + theQuery.displayResults() + " lines to be displayed, " + (theQuery.maximumTime / 1000) + " seconds");
long timestamp = System.currentTimeMillis();
// create a new search event
String wrongregex = null;
if (plasmaSearchEvent.getEvent(theQuery.id()) == null) {
theQuery.setOffset(0); // in case that this is a new search, always start without a offset
offset = 0;
}
plasmaSearchEvent theSearch = plasmaSearchEvent.getEvent(theQuery, ranking, localTiming, sb.wordIndex, (sb.isRobinsonMode()) ? sb.clusterhashes : null, false, null);
// generate result object
serverLog.logFine("LOCAL_SEARCH", "SEARCH TIME AFTER ORDERING OF SEARCH RESULTS: " + ((System.currentTimeMillis() - timestamp) / 1000) + " seconds");
serverLog.logFine("LOCAL_SEARCH", "SEARCH TIME AFTER RESULT PREPARATION: " + ((System.currentTimeMillis() - timestamp) / 1000) + " seconds");
// generate result object
serverLog.logFine("LOCAL_SEARCH", "SEARCH TIME AFTER ORDERING OF SEARCH RESULTS: " + ((System.currentTimeMillis() - timestamp) / 1000) + " seconds");
serverLog.logFine("LOCAL_SEARCH", "SEARCH TIME AFTER RESULT PREPARATION: " + ((System.currentTimeMillis() - timestamp) / 1000) + " seconds");
// calc some more cross-reference
long remainingTime = theQuery.maximumTime - (System.currentTimeMillis() - timestamp);
if (remainingTime < 0) remainingTime = 1000;
serverLog.logFine("LOCAL_SEARCH", "SEARCH TIME AFTER XREF PREPARATION: " + ((System.currentTimeMillis() - timestamp) / 1000) + " seconds");
// calc some more cross-reference
long remainingTime = theQuery.maximumTime - (System.currentTimeMillis() - timestamp);
if (remainingTime < 0) remainingTime = 1000;
serverLog.logFine("LOCAL_SEARCH", "SEARCH TIME AFTER XREF PREPARATION: " + ((System.currentTimeMillis() - timestamp) / 1000) + " seconds");
// log
serverLog.logInfo("LOCAL_SEARCH", "EXIT WORD SEARCH: " + theQuery.queryString + " - " +
// log
serverLog.logInfo("LOCAL_SEARCH", "EXIT WORD SEARCH: " + theQuery.queryString + " - " +
(theSearch.getLocalCount() + theSearch.getGlobalCount()) + " links found, " +
((System.currentTimeMillis() - timestamp) / 1000) + " seconds");
// prepare search statistics
Long trackerHandle = new Long(System.currentTimeMillis());
HashMap searchProfile = theQuery.resultProfile(theSearch.getLocalCount() + theSearch.getGlobalCount(), System.currentTimeMillis() - timestamp);
searchProfile.put("querystring", theQuery.queryString);
searchProfile.put("time", trackerHandle);
searchProfile.put("host", client);
searchProfile.put("offset", new Integer(0));
sb.localSearches.add(searchProfile);
TreeSet handles = (TreeSet) sb.localSearchTracker.get(client);
if (handles == null) handles = new TreeSet();
handles.add(trackerHandle);
sb.localSearchTracker.put(client, handles);
// prepare search statistics
Long trackerHandle = new Long(System.currentTimeMillis());
HashMap searchProfile = theQuery.resultProfile(theSearch.getLocalCount() + theSearch.getGlobalCount(), System.currentTimeMillis() - timestamp);
searchProfile.put("querystring", theQuery.queryString);
searchProfile.put("time", trackerHandle);
searchProfile.put("host", client);
searchProfile.put("offset", new Integer(0));
sb.localSearches.add(searchProfile);
TreeSet handles = (TreeSet) sb.localSearchTracker.get(client);
if (handles == null) handles = new TreeSet();
handles.add(trackerHandle);
sb.localSearchTracker.put(client, handles);
prop = new serverObjects();
prop.put("num-results_totalcount", theSearch.getLocalCount() + theSearch.getGlobalCount());
prop.put("num-results_globalresults", 1);
prop.put("num-results_globalresults_globalcount", theSearch.getGlobalCount());
prop.put("num-results_offset", 0);
prop.put("num-results_linkcount", 0);
// compose page navigation
StringBuffer pagenav = new StringBuffer();
int thispage = offset / theQuery.displayResults();
if (thispage == 0) pagenav.append("&lt;&nbsp;"); else {
pagenav.append(navurla(thispage - 1, display, theQuery));
pagenav.append("<strong>&lt;</strong></a>&nbsp;");
}
int numberofpages = Math.min(10, Math.min(thispage + 2, (theSearch.getGlobalCount() + theSearch.getLocalCount()) / theQuery.displayResults()));
for (int i = 0; i < numberofpages; i++) {
if (i == thispage) {
pagenav.append("<strong>");
pagenav.append(i + 1);
pagenav.append("</strong>&nbsp;");
} else {
pagenav.append(navurla(i, display, theQuery));
pagenav.append(i + 1);
pagenav.append("</a>&nbsp;");
prop = new serverObjects();
prop.put("num-results_totalcount", theSearch.getLocalCount() + theSearch.getGlobalCount());
prop.put("num-results_globalresults", 1);
prop.put("num-results_globalresults_globalcount", theSearch.getGlobalCount());
prop.put("num-results_offset", 0);
prop.put("num-results_linkcount", 0);
// compose page navigation
StringBuffer pagenav = new StringBuffer();
int thispage = offset / theQuery.displayResults();
if (thispage == 0) pagenav.append("&lt;&nbsp;"); else {
pagenav.append(navurla(thispage - 1, display, theQuery));
pagenav.append("<strong>&lt;</strong></a>&nbsp;");
}
}
if (thispage >= numberofpages) pagenav.append("&gt;"); else {
pagenav.append(navurla(thispage + 1, display, theQuery));
pagenav.append("<strong>&gt;</strong></a>");
}
prop.putASIS("num-results_pagenav", pagenav.toString());
int numberofpages = Math.min(10, Math.min(thispage + 2, (theSearch.getGlobalCount() + theSearch.getLocalCount()) / theQuery.displayResults()));
for (int i = 0; i < numberofpages; i++) {
if (i == thispage) {
pagenav.append("<strong>");
pagenav.append(i + 1);
pagenav.append("</strong>&nbsp;");
} else {
pagenav.append(navurla(i, display, theQuery));
pagenav.append(i + 1);
pagenav.append("</a>&nbsp;");
}
}
if (thispage >= numberofpages) pagenav.append("&gt;"); else {
pagenav.append(navurla(thispage + 1, display, theQuery));
pagenav.append("<strong>&gt;</strong></a>");
}
prop.putASIS("num-results_pagenav", pagenav.toString());
// generate the search result lines; they will be produced by another servlet
for (int i = 0; i < theQuery.displayResults(); i++) {
prop.put("type_results_" + i + "_item", offset + i);
prop.put("type_results_" + i + "_eventID", theQuery.id());
prop.put("type_results_" + i + "_display", display);
}
prop.put("type_results", theQuery.displayResults());
// generate the search result lines; they will be produced by another servlet
for (int i = 0; i < theQuery.displayResults(); i++) {
prop.put("results_" + i + "_item", offset + i);
prop.put("results_" + i + "_eventID", theQuery.id());
}
prop.put("results", theQuery.displayResults());
// process result of search
if (filtered.size() > 0) {
@ -384,7 +380,7 @@ public class yacysearch {
prop.put("num-results", 5);
int hintcount = references.length;
if (hintcount > 0) {
prop.put("type_combine", 1);
prop.put("combine", 1);
// get the topwords
final TreeSet topwords = new TreeSet(kelondroNaturalOrder.naturalOrder);
String tmp = "";
@ -414,14 +410,14 @@ public class yacysearch {
while (iter.hasNext()) {
word = (String) iter.next();
if (word != null) {
prop.put("type_combine_words_" + hintcount + "_word", word);
prop.put("type_combine_words_" + hintcount + "_newsearch", post.get("search", "").replace(' ', '+') + "+" + word);
prop.put("type_combine_words_" + hintcount + "_count", count);
prop.put("type_combine_words_" + hintcount + "_offset", offset);
prop.put("type_combine_words_" + hintcount + "_resource", ((global) ? "global" : "local"));
prop.put("type_combine_words_" + hintcount + "_time", (searchtime / 1000));
prop.put("combine_words_" + hintcount + "_word", word);
prop.put("combine_words_" + hintcount + "_newsearch", post.get("search", "").replace(' ', '+') + "+" + word);
prop.put("combine_words_" + hintcount + "_count", count);
prop.put("combine_words_" + hintcount + "_offset", offset);
prop.put("combine_words_" + hintcount + "_resource", ((global) ? "global" : "local"));
prop.put("combine_words_" + hintcount + "_time", (searchtime / 1000));
}
prop.put("type_combine_words", hintcount);
prop.put("combine_words", hintcount);
if (hintcount++ > MAX_TOPWORDS) {
break;
}
@ -437,55 +433,14 @@ public class yacysearch {
}
}
prop.put("type", (theQuery.contentdom == plasmaSearchQuery.CONTENTDOM_TEXT) ? 0 : ((theQuery.contentdom == plasmaSearchQuery.CONTENTDOM_IMAGE) ? 2 : 1));
if (prop.getInt("type", 0) == 1) prop.put("type_mediatype", post.get("contentdom", "text"));
prop.put("input_cat", "href");
prop.put("input_depth", "0");
prop.put("type_eventID", theQuery.id());
// adding some additional properties needed for the rss feed
String hostName = (String) header.get("Host", "localhost");
if (hostName.indexOf(":") == -1) hostName += ":" + serverCore.getPortNr(env.getConfig("port", "8080"));
prop.put("rssYacyImageURL", "http://" + hostName + "/env/grafics/yacy.gif");
}
if (post.get("cat", "href").equals("image")) {
int depth = post.getInt("depth", 0);
int columns = post.getInt("columns", 6);
yacyURL url = null;
try {url = new yacyURL(post.get("url", ""), null);} catch (MalformedURLException e) {}
plasmaSearchImages si = new plasmaSearchImages(6000, url, depth);
Iterator i = si.entries();
htmlFilterImageEntry ie;
int line = 0;
while (i.hasNext()) {
int col = 0;
for (col = 0; col < columns; col++) {
if (!i.hasNext()) break;
ie = (htmlFilterImageEntry) i.next();
String urls = ie.url().toString();
String name = "";
int p = urls.lastIndexOf('/');
if (p > 0) name = urls.substring(p + 1);
prop.put("type_results_" + line + "_line_" + col + "_url", urls);
prop.put("type_results_" + line + "_line_" + col + "_name", name);
}
prop.put("type_results_" + line + "_line", col);
line++;
}
prop.put("type_results", line);
prop.put("type", 3); // set type of result: image list
prop.put("input_cat", "href");
prop.put("input_depth", depth);
}
// if user is not authenticated, he may not vote for URLs
int linkcount = Integer.parseInt(prop.get("num-results_linkcount", "0"));
for (int i=0; i<linkcount; i++) {
prop.put("type_results_" + i + "_authorized", (authenticated) ? 1 : 0);
}
prop.put("searchagain", (global) ? 1 : 0);
prop.put("input", input);
@ -509,7 +464,6 @@ public class yacysearch {
prop.put("input_contentdomCheckVideo", (contentdomCode == plasmaSearchQuery.CONTENTDOM_VIDEO) ? 1 : 0);
prop.put("input_contentdomCheckImage", (contentdomCode == plasmaSearchQuery.CONTENTDOM_IMAGE) ? 1 : 0);
prop.put("input_contentdomCheckApp", (contentdomCode == plasmaSearchQuery.CONTENTDOM_APP) ? 1 : 0);
prop.put("type_former", post.get("search", "")); //the query-string used to get the snippets
// return rewrite properties
return prop;

37
htroot/yacysearch.soap
Unescape View File

@ -1,37 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<yacySearch>
<yacySearchResult>
<title>Search for #[former]#</title>
<description>Search for #[former]#</description>
#(type)#
<totalResultCount>#(num-results)#::::::::#[totalcount]##(/num-results)#</totalResultCount>
<currentResultCount>#(num-results)#::::::::#[linkcount]##(/num-results)#</currentResultCount>
<!-- StopWords -->
<stopWords>#(excluded)#::#[stopwords]##(/excluded)#</stopWords>
<!-- YaCy Top-Worts -->
<topWords>
#(combine)#::
#{words}#<word query="#[newsearch]#">#[word]#</word>#{/words}#
#(/combine)#
</topWords>
<!-- Search Results -->
#{results}#
<yacySearchResultItem>
<description>#[description]#</description>
<link>#[url]#</link>
<hash>#[urlhash]#</hash>
<linkName>#[urlname]#</linkName>
<pubDate>#[date]#</pubDate>
<size>#[size]#</size>
<ranking>YBR-#[ybr]#</ranking>
<snippet>#(snippet)#::<![CDATA[#[text]#]]>#(/snippet)#</snippet>
</yacySearchResultItem>
#{/results}#
::
#(/type)#
</yacySearchResult>
</yacySearch>

9
htroot/yacysearchitem.html
Unescape View File

@ -15,12 +15,19 @@
#(/recommend)#
</div>
#(/authorized)#
<p class="snippet"><span class="snippetLoaded" id="h#[urlhash]#">#[snippet]#</span></p>
<p class="url"><a href="#[url]#" id="url#[urlhash]#" target="_parent">#[urlname]#</a></p>
<p class="urlinfo">#[date]# | YBR-#[ybr]# | <a href="ViewFile.html?urlHash=#[urlhash]#&amp;words=#[words]#">Info</a> | <a href="yacysearch.html?cat=image&amp;url=#[url]#&amp;search=#[former]#">Pictures</a></p>
</div>
::
#{images}#
<div class="thumbcontainer">
<a href="#[href]#" class="thumblink"><img src="/ViewImage.png?maxwidth=96&maxheight=96&code=#[code]#" alt="#[name]#"></a>
<div class="TableCellDark"><a href="#[href]#">#[name]#</a></div>
</div>
#{/images}#
#(/content)#
<script type="text/javascript">
statistics("#[offset]#", "#[items]#", "#[global]#", "#[total]#");
progressbar.step(1);
</script>

90
htroot/yacysearchitem.java
Unescape View File

@ -27,6 +27,7 @@
import java.io.UnsupportedEncodingException;
import java.net.MalformedURLException;
import java.net.URLEncoder;
import java.util.ArrayList;
import java.util.TreeSet;
import de.anomic.http.httpHeader;
@ -34,6 +35,7 @@ import de.anomic.plasma.plasmaSearchEvent;
import de.anomic.plasma.plasmaSearchPreOrder;
import de.anomic.plasma.plasmaSearchQuery;
import de.anomic.plasma.plasmaSearchRankingProfile;
import de.anomic.plasma.plasmaSnippetCache;
import de.anomic.plasma.plasmaSwitchboard;
import de.anomic.server.serverObjects;
import de.anomic.server.serverSwitch;
@ -53,7 +55,6 @@ public class yacysearchitem {
String eventID = post.get("eventID", "");
int item = post.getInt("item", -1);
//int display = post.getInt("display", 0);
boolean authenticated = sb.adminAuthenticated(header) >= 2;
// find search event
@ -75,42 +76,69 @@ public class yacysearchitem {
return prop;
}
prop.put("content", 1); // switch on content
prop.put("content_authorized", (authenticated) ? 1 : 0);
prop.put("content_authorized_recommend", (yacyCore.newsPool.getSpecific(yacyNewsPool.OUTGOING_DB, yacyNewsPool.CATEGORY_SURFTIPP_ADD, "url", result.urlstring()) == null) ? 1 : 0);
prop.put("content_authorized_recommend_deletelink", "/yacysearch.html?search=" + theQuery.queryString + "&Enter=Search&count=" + theQuery.displayResults() + "&offset=" + (theQuery.neededResults() - theQuery.displayResults()) + "&order=" + crypt.simpleEncode(ranking.toExternalString()) + "&resource=local&time=3&deleteref=" + result.hash() + "&urlmaskfilter=.*");
prop.put("content_authorized_recommend_recommendlink", "/yacysearch.html?search=" + theQuery.queryString + "&Enter=Search&count=" + theQuery.displayResults() + "&offset=" + (theQuery.neededResults() - theQuery.displayResults()) + "&order=" + crypt.simpleEncode(ranking.toExternalString()) + "&resource=local&time=3&recommendref=" + result.hash() + "&urlmaskfilter=.*");
prop.put("content_authorized_urlhash", result.hash());
prop.put("content_description", result.title());
prop.put("content_url", result.urlstring());
prop.put("content", theQuery.contentdom + 1); // switch on specific content
int port=result.url().getPort();
yacyURL faviconURL;
try {
faviconURL = new yacyURL(result.url().getProtocol() + "://" + result.url().getHost() + ((port != -1) ? (":" + String.valueOf(port)) : "") + "/favicon.ico", null);
} catch (MalformedURLException e1) {
faviconURL = null;
}
if (theQuery.contentdom == plasmaSearchQuery.CONTENTDOM_TEXT) {
// text search
prop.put("content_authorized", (authenticated) ? 1 : 0);
prop.put("content_authorized_recommend", (yacyCore.newsPool.getSpecific(yacyNewsPool.OUTGOING_DB, yacyNewsPool.CATEGORY_SURFTIPP_ADD, "url", result.urlstring()) == null) ? 1 : 0);
prop.put("content_authorized_recommend_deletelink", "/yacysearch.html?search=" + theQuery.queryString + "&Enter=Search&count=" + theQuery.displayResults() + "&offset=" + (theQuery.neededResults() - theQuery.displayResults()) + "&order=" + crypt.simpleEncode(ranking.toExternalString()) + "&resource=local&time=3&deleteref=" + result.hash() + "&urlmaskfilter=.*");
prop.put("content_authorized_recommend_recommendlink", "/yacysearch.html?search=" + theQuery.queryString + "&Enter=Search&count=" + theQuery.displayResults() + "&offset=" + (theQuery.neededResults() - theQuery.displayResults()) + "&order=" + crypt.simpleEncode(ranking.toExternalString()) + "&resource=local&time=3&recommendref=" + result.hash() + "&urlmaskfilter=.*");
prop.put("content_authorized_urlhash", result.hash());
prop.put("content_description", result.title());
prop.put("content_url", result.urlstring());
int port=result.url().getPort();
yacyURL faviconURL;
try {
faviconURL = new yacyURL(result.url().getProtocol() + "://" + result.url().getHost() + ((port != -1) ? (":" + String.valueOf(port)) : "") + "/favicon.ico", null);
} catch (MalformedURLException e1) {
faviconURL = null;
}
prop.put("content_faviconCode", sb.licensedURLs.aquireLicense(faviconURL)); // aquire license for favicon url loading
prop.put("content_urlhash", result.hash());
prop.put("content_urlhexhash", yacySeed.b64Hash2hexHash(result.hash()));
prop.put("content_urlname", nxTools.shortenURLString(result.urlname(), 120));
prop.put("content_date", plasmaSwitchboard.dateString(result.modified()));
prop.put("content_ybr", plasmaSearchPreOrder.ybr(result.hash()));
prop.put("content_size", Long.toString(result.filesize()));
prop.put("content_faviconCode", sb.licensedURLs.aquireLicense(faviconURL)); // aquire license for favicon url loading
prop.put("content_urlhash", result.hash());
prop.put("content_urlhexhash", yacySeed.b64Hash2hexHash(result.hash()));
prop.put("content_urlname", nxTools.shortenURLString(result.urlname(), 120));
prop.put("content_date", plasmaSwitchboard.dateString(result.modified()));
prop.put("content_ybr", plasmaSearchPreOrder.ybr(result.hash()));
prop.put("content_size", Long.toString(result.filesize()));
TreeSet[] query = theQuery.queryWords();
yacyURL wordURL = null;
try {
prop.put("content_words", URLEncoder.encode(query[0].toString(),"UTF-8"));
} catch (UnsupportedEncodingException e) {}
prop.put("content_former", theQuery.queryString);
prop.put("content_rankingprops", result.word().toPropertyForm() + ", domLengthEstimated=" + yacyURL.domLengthEstimation(result.hash()) +
TreeSet[] query = theQuery.queryWords();
yacyURL wordURL = null;
try {
prop.put("content_words", URLEncoder.encode(query[0].toString(),"UTF-8"));
} catch (UnsupportedEncodingException e) {}
prop.put("content_former", theQuery.queryString);
prop.put("content_rankingprops", result.word().toPropertyForm() + ", domLengthEstimated=" + yacyURL.domLengthEstimation(result.hash()) +
((yacyURL.probablyRootURL(result.hash())) ? ", probablyRootURL" : "") +
(((wordURL = yacyURL.probablyWordURL(result.hash(), query[0])) != null) ? ", probablyWordURL=" + wordURL.toNormalform(false, true) : ""));
prop.putASIS("content_snippet", result.textSnippet().getLineMarked(theQuery.queryHashes));
prop.putASIS("content_snippet", result.textSnippet().getLineMarked(theQuery.queryHashes));
}
if (theQuery.contentdom == plasmaSearchQuery.CONTENTDOM_IMAGE) {
// image search; shows thumbnails
// iterate over all images in the result
ArrayList /* of plasmaSnippetCache.MediaSnippet */ images = result.mediaSnippets();
if (images != null) {
plasmaSnippetCache.MediaSnippet ms;
yacyURL url;
int c = 0;
for (int i = 0; i < images.size(); i++) {
ms = (plasmaSnippetCache.MediaSnippet) images.get(i);
try {url = new yacyURL(ms.href, null);} catch (MalformedURLException e) {continue;}
prop.put("content_images_" + i + "_href", ms.href);
prop.put("content_images_" + i + "_code", sb.licensedURLs.aquireLicense(url));
prop.put("content_images_" + i + "_name", ms.name);
prop.put("content_images_" + i + "_attr", ms.attr); // attributes, here: original size of image
c++;
}
prop.put("content_images", c);
} else {
prop.put("content_images", 0);
}
}
return prop;
}

Write Preview
Loading…
Cancel
Save