Michael Peter Christen
13 years ago
parent
00c1c777fa
commit
fa7f6f0be8
@ -0,0 +1,118 @@
|
||||
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "DTD/xhtml1-transitional.dtd">
|
||||
<!-- This page is only XHTML 1.0 Transitional because target is being used in a links -->
|
||||
<html xmlns="http://www.w3.org/1999/xhtml">
|
||||
<head>
|
||||
<title>YaCy '#[clientname]#': URL References Administration</title>
|
||||
#%env/templates/metas.template%#
|
||||
<script type="text/javascript">
|
||||
//<![CDATA[
|
||||
function xmlhttpPost() {
|
||||
var searchform = document.getElementById('searchform');
|
||||
search(searchform.urlstring.value);
|
||||
}
|
||||
|
||||
function search(query) {
|
||||
var xmlHttpReq = false;
|
||||
var self = this;
|
||||
if (window.XMLHttpRequest) { // Mozilla/Safari
|
||||
self.xmlHttpReq = new XMLHttpRequest();
|
||||
}
|
||||
else if (window.ActiveXObject) { // IE
|
||||
self.xmlHttpReq = new ActiveXObject("Microsoft.XMLHTTP");
|
||||
}
|
||||
self.xmlHttpReq.open('GET', "/solr/select?q=sku:\"" + query + "\" OR host_s:\"" + query + "\" OR host_dnc_s:\"" + query + "\" OR host_organization_s:\"" + query + "\" OR host_organizationdnc_s:\"" + query + "\" OR host_subdomain_s:\"" + query + "\"&start=0&rows=100&wt=json", true);
|
||||
self.xmlHttpReq.setRequestHeader('Content-Type', 'application/x-www-form-urlencoded');
|
||||
self.xmlHttpReq.onreadystatechange = function() {
|
||||
if (self.xmlHttpReq.readyState == 4) {
|
||||
updatepage(self.xmlHttpReq.responseText);
|
||||
}
|
||||
}
|
||||
self.xmlHttpReq.send(null);
|
||||
}
|
||||
|
||||
function updatepage(str) {
|
||||
var raw = document.getElementById("raw");
|
||||
if (raw != null) raw.innerHTML = str;
|
||||
var rsp = eval("("+str+")");
|
||||
var firstChannel = rsp.channels[0];
|
||||
var totalResults = firstChannel.totalResults.replace(/[,.]/,"");
|
||||
var startIndex = firstChannel.startIndex;
|
||||
var itemsPerPage = firstChannel.itemsPerPage;
|
||||
var navigation = firstChannel.navigation;
|
||||
|
||||
var html = "";
|
||||
|
||||
if (totalResults > 0 && firstChannel.items.length > 0) {
|
||||
var item;
|
||||
html += "<table class=\"networkTable\" border=\"0\" cellpadding=\"2\" cellspacing=\"1\" width=\"99%\">";
|
||||
html += "<tr class=\"TableHeader\" valign=\"bottom\">";
|
||||
html += "<td>URL from index (total results = " + totalResults + ")<\/td>";
|
||||
for (var i = 0; i < firstChannel.items.length; i++) {
|
||||
item = firstChannel.items[i];
|
||||
html += "<tr class=\"TableCellLight\"><td align=\"left\"><a href=\"HostBrowser.html?urlstringsearch=&urlstring=" + item.link + "\">" + item.link + "<\/a><\/td>";
|
||||
}
|
||||
html += "<\/table>";
|
||||
}
|
||||
document.getElementById("searchresults").innerHTML = html;
|
||||
}
|
||||
//]]>
|
||||
</script>
|
||||
</head>
|
||||
<body id="IndexControl">
|
||||
#%env/templates/header.template%#
|
||||
#%env/templates/submenuIndexControl.template%#
|
||||
<h2>URL References Administration</h2>
|
||||
<p>The local index currently contains #[ucount]# URL references</p>
|
||||
<form action="HostBrowser.html" id="searchform" method="post" enctype="multipart/form-data" accept-charset="UTF-8" onkeyup="xmlhttpPost(); return false;">
|
||||
<fieldset><legend>URL Retrieval</legend>
|
||||
<dl>
|
||||
|
||||
<dt class="TableCellDark">Retrieve by URL:</dt>
|
||||
<dd><input type="text" name="urlstring" value="#[urlstring]#" size="40" maxlength="250" />
|
||||
<input type="submit" name="urlstringsearch" value="Show Details for URL" class="submitready" style="width:240px;"/><br />
|
||||
<div id="searchresults"></div>
|
||||
</dd>
|
||||
|
||||
</dl>
|
||||
</fieldset>
|
||||
</form>
|
||||
|
||||
<div id="raw"></div>
|
||||
|
||||
#(statisticslines)#::
|
||||
<p><em>Statistics about the top-#[domains]# domains in the database:</em></p>
|
||||
<table cellpadding="2" cellspacing="1" >
|
||||
<tr class="TableHeader">
|
||||
<td><strong>Domain</strong></td>
|
||||
<td><strong>URLs</strong></td>
|
||||
</tr>
|
||||
#{domains}#
|
||||
<tr class="TableCell#(dark)#Light::Dark#(/dark)#">
|
||||
<td><a href="http://#[domain]#/" target="_blank">#[domain]#</a></td>
|
||||
<td>#[count]#</td>
|
||||
</tr>
|
||||
#{/domains}#
|
||||
</table>
|
||||
#(/statisticslines)#
|
||||
|
||||
|
||||
#(genUrlProfile)#
|
||||
::No entry found for URL-hash #[urlhash]#
|
||||
::<iframe src="/api/yacydoc.html?urlhash=#[urlhash]#" width="100%" height="420" frameborder="0" scrolling="no"></iframe><br />
|
||||
<div id="api">
|
||||
<a href="/api/yacydoc.html?urlhash=#[urlhash]#">
|
||||
<img src="env/grafics/api.png" width="60" height="40" alt="API" /></a>
|
||||
<span>These document details can be retrieved as <a href="http://www.w3.org/TR/xhtml-rdfa-primer/">XHTML+RDFa</a>
|
||||
document containg <a href="http://www.w3.org/RDF/">RDF</a> annotations in <a href="http://dublincore.org/">Dublin Core</a> vocabulary.
|
||||
The XHTML+RDFa data format is both a XML content format and a HTML display format and is considered as an important <a href="http://www.w3.org/2001/sw/">Semantic Web</a> content format.
|
||||
The same content can also be retrieved as pure <a href="/api/yacydoc.xml?urlhash=#[urlhash]#">XML metadata</a> with DC tag name vocabulary.
|
||||
Click the API icon to see an example call to the search rss API.
|
||||
To see a list of all APIs, please visit the <a href="http://www.yacy-websuche.de/wiki/index.php/Dev:API">API wiki page</a>.</span>
|
||||
</div>
|
||||
#(/genUrlProfile)#
|
||||
|
||||
#[result]#
|
||||
|
||||
#%env/templates/footer.template%#
|
||||
</body>
|
||||
</html>
|
||||
@ -0,0 +1,107 @@
|
||||
|
||||
import java.net.MalformedURLException;
|
||||
|
||||
import net.yacy.cora.document.ASCII;
|
||||
import net.yacy.cora.protocol.RequestHeader;
|
||||
import net.yacy.kelondro.data.meta.DigestURI;
|
||||
import net.yacy.kelondro.data.meta.URIMetadata;
|
||||
import net.yacy.kelondro.data.word.Word;
|
||||
import net.yacy.search.Switchboard;
|
||||
import net.yacy.search.index.Segment;
|
||||
import net.yacy.server.serverObjects;
|
||||
import net.yacy.server.serverSwitch;
|
||||
|
||||
public class HostBrowser {
|
||||
|
||||
public static serverObjects respond(@SuppressWarnings("unused") final RequestHeader header, final serverObjects post, final serverSwitch env) {
|
||||
// return variable that accumulates replacements
|
||||
final Switchboard sb = (Switchboard) env;
|
||||
|
||||
final serverObjects prop = new serverObjects();
|
||||
|
||||
Segment segment = sb.index;
|
||||
|
||||
// set default values
|
||||
prop.put("urlstring", "");
|
||||
prop.put("urlhash", "");
|
||||
prop.put("result", "");
|
||||
prop.putNum("ucount", segment.fulltext().size());
|
||||
prop.put("otherHosts", "");
|
||||
prop.put("genUrlProfile", 0);
|
||||
prop.put("statistics", 1);
|
||||
prop.put("statistics_lines", 100);
|
||||
prop.put("statisticslines", 0);
|
||||
|
||||
if (post == null || env == null) {
|
||||
return prop; // nothing to do
|
||||
}
|
||||
|
||||
// post values that are set on numerous input fields with same name
|
||||
String urlstring = post.get("urlstring", "").trim();
|
||||
|
||||
if (!urlstring.startsWith("http://") &&
|
||||
!urlstring.startsWith("https://") &&
|
||||
!urlstring.startsWith("ftp://") &&
|
||||
!urlstring.startsWith("smb://") &&
|
||||
!urlstring.startsWith("file://")) { urlstring = "http://" + urlstring; }
|
||||
|
||||
prop.putHTML("urlstring", urlstring);
|
||||
prop.put("result", " ");
|
||||
|
||||
if (post.containsKey("urlstringsearch")) {
|
||||
try {
|
||||
final DigestURI url = new DigestURI(urlstring);
|
||||
String urlhash = ASCII.String(url.hash());
|
||||
prop.put("urlhash", urlhash);
|
||||
final URIMetadata entry = segment.fulltext().getMetadata(ASCII.getBytes(urlhash));
|
||||
if (entry == null) {
|
||||
prop.putHTML("result", "No Entry for URL " + url.toNormalform(true, true));
|
||||
prop.putHTML("urlstring", urlstring);
|
||||
prop.put("urlhash", "");
|
||||
} else {
|
||||
prop.putAll(genUrlProfile(segment, entry, urlhash));
|
||||
prop.put("statistics", 0);
|
||||
}
|
||||
} catch (final MalformedURLException e) {
|
||||
prop.putHTML("result", "bad url: " + urlstring);
|
||||
prop.put("urlhash", "");
|
||||
}
|
||||
prop.put("lurlexport", 0);
|
||||
}
|
||||
|
||||
// insert constants
|
||||
prop.putNum("ucount", segment.fulltext().size());
|
||||
// return rewrite properties
|
||||
return prop;
|
||||
}
|
||||
|
||||
private static serverObjects genUrlProfile(final Segment segment, final URIMetadata entry, final String urlhash) {
|
||||
final serverObjects prop = new serverObjects();
|
||||
if (entry == null) {
|
||||
prop.put("genUrlProfile", "1");
|
||||
prop.put("genUrlProfile_urlhash", urlhash);
|
||||
return prop;
|
||||
}
|
||||
final URIMetadata le = (entry.referrerHash() == null || entry.referrerHash().length != Word.commonHashLength) ? null : segment.fulltext().getMetadata(entry.referrerHash());
|
||||
if (entry.url() == null) {
|
||||
prop.put("genUrlProfile", "1");
|
||||
prop.put("genUrlProfile_urlhash", urlhash);
|
||||
return prop;
|
||||
}
|
||||
prop.put("genUrlProfile", "2");
|
||||
prop.putHTML("genUrlProfile_urlNormalform", entry.url().toNormalform(false, true));
|
||||
prop.put("genUrlProfile_urlhash", urlhash);
|
||||
prop.put("genUrlProfile_urlDescr", entry.dc_title());
|
||||
prop.put("genUrlProfile_moddate", entry.moddate().toString());
|
||||
prop.put("genUrlProfile_loaddate", entry.loaddate().toString());
|
||||
prop.put("genUrlProfile_referrer", (le == null) ? 0 : 1);
|
||||
prop.putHTML("genUrlProfile_referrer_url", (le == null) ? "<unknown>" : le.url().toNormalform(false, true));
|
||||
prop.put("genUrlProfile_referrer_hash", (le == null) ? "" : ASCII.String(le.hash()));
|
||||
prop.put("genUrlProfile_doctype", String.valueOf(entry.doctype()));
|
||||
prop.put("genUrlProfile_language", entry.language());
|
||||
prop.put("genUrlProfile_size", entry.size());
|
||||
prop.put("genUrlProfile_wordCount", entry.wordCount());
|
||||
return prop;
|
||||
}
|
||||
|
||||
}
|
||||
Loading…
Reference in new issue