search.jsp revision 1164
883N/A<%--
883N/ACDDL HEADER START
883N/A
883N/AThe contents of this file are subject to the terms of the
883N/ACommon Development and Distribution License (the "License").
883N/AYou may not use this file except in compliance with the License.
883N/A
883N/ASee LICENSE.txt included in this distribution for the specific
883N/Alanguage governing permissions and limitations under the License.
883N/A
883N/AWhen distributing Covered Code, include this CDDL HEADER in each
883N/Afile and include the License file at LICENSE.txt.
883N/AIf applicable, add the following below this CDDL HEADER, with the
883N/Afields enclosed by brackets "[]" replaced with your own identifying
883N/Ainformation: Portions Copyright [yyyy] [name of copyright owner]
883N/A
883N/ACDDL HEADER END
883N/A
883N/ACopyright (c) 2005, 2011, Oracle and/or its affiliates. All rights reserved.
883N/A
883N/A--%><%@ page import = "javax.servlet.*,
883N/Ajava.lang.Integer,
883N/Ajavax.servlet.http.*,
883N/Ajava.util.Hashtable,
883N/Ajava.util.Vector,
883N/Ajava.util.Date,
883N/Ajava.util.ArrayList,
883N/Ajava.util.List,
883N/Ajava.lang.*,
883N/Ajava.io.*,
883N/Ajava.io.StringReader,
883N/Aorg.opensolaris.opengrok.analysis.*,
883N/Aorg.opensolaris.opengrok.index.IndexDatabase,
883N/Aorg.opensolaris.opengrok.search.*,
883N/Aorg.opensolaris.opengrok.web.*,
883N/Aorg.opensolaris.opengrok.search.context.*,
883N/Aorg.opensolaris.opengrok.configuration.*,
883N/Aorg.apache.lucene.search.spell.LuceneDictionary,
883N/Aorg.apache.lucene.search.spell.SpellChecker,
883N/Aorg.apache.lucene.search.SortField,
883N/Aorg.apache.lucene.store.FSDirectory,
883N/Aorg.apache.lucene.analysis.*,
883N/Aorg.apache.lucene.document.*,
883N/Aorg.apache.lucene.search.*,
883N/Aorg.apache.lucene.queryParser.*"
883N/A%><%@ page session="false" %><%@ page errorPage="error.jsp" %><%--
883N/A--%><%@ include file="projects.jspf" %><%
883N/ADate starttime = new Date();
883N/AString q = request.getParameter("q");
883N/AString defs = request.getParameter("defs");
883N/AString refs = request.getParameter("refs");
883N/AString hist = request.getParameter("hist");
883N/AString path = request.getParameter("path");
883N/A
883N/AString sort = null;
883N/A
883N/Afinal String LASTMODTIME = "lastmodtime";
883N/Afinal String RELEVANCY = "relevancy";
883N/Afinal String BY_PATH = "fullpath";
883N/Afinal SortField S_BY_PATH = new SortField(BY_PATH,SortField.STRING);
883N/A
883N/ACookie[] cookies = request.getCookies();
883N/Aif (cookies != null) {
883N/A for (Cookie cookie : cookies) {
883N/A if (cookie.getName().equals("OpenGrokSorting")) {
883N/A sort = cookie.getValue();
883N/A if (!LASTMODTIME.equals(sort) && !RELEVANCY.equals(sort) && !BY_PATH.equals(sort)) {
883N/A sort = RELEVANCY;
883N/A }
break;
}
}
}
String sortParam = request.getParameter("sort");
if (sortParam != null) {
if (LASTMODTIME.equals(sortParam)) {
sort = LASTMODTIME;
} else if (RELEVANCY.equals(sortParam)) {
sort = RELEVANCY;
} else if (BY_PATH.equals(sortParam)) {
sort = BY_PATH;
}
if (sort != null) {
Cookie cookie = new Cookie("OpenGrokSorting", sort);
response.addCookie(cookie);
}
} else { sort = RELEVANCY; }
//List<org.apache.lucene.document.Document> docs=new ArrayList<org.apache.lucene.document.Document>();
String errorMsg = null;
if( q!= null && q.equals("")) q = null;
if( defs != null && defs.equals("")) defs = null;
if( refs != null && refs.equals("")) refs = null;
if( hist != null && hist.equals("")) hist = null;
if( path != null && path.equals("")) path = null;
if (project != null && project.size()<1) project = null;
if (q != null || defs != null || refs != null || hist != null || path != null) {
Searcher searcher = null; //the searcher used to open/search the index
ScoreDoc[] hits = null; // list of documents which result from the query
Query query = null; //the Query created by the QueryBuilder
int totalHits=0;
int start = 0; //the first index displayed on this page
//TODO deprecate max this and merge with paging and param n - TEST needed
//int max = 25; //the maximum items displayed on this page
int max=RuntimeEnvironment.getInstance().getHitsPerPage();
int thispage = 0; //used for the for/next either max or
QueryBuilder queryBuilder =
new QueryBuilder()
.setFreetext(q).setDefs(defs).setRefs(refs)
.setPath(path).setHist(hist);
try {
String DATA_ROOT = env.getDataRootPath();
if(DATA_ROOT.equals("")) {
throw new Exception("DATA_ROOT parameter is not configured in web.xml!");
}
File data_root = new File(DATA_ROOT);
if(!data_root.isDirectory()) {
throw new Exception("DATA_ROOT parameter in web.xml does not exist or is not a directory!");
}
//String date = request.getParameter("date");
try {
start = Integer.parseInt(request.getParameter("start")); //parse the max results first
max = Integer.parseInt(request.getParameter("n")); //then the start index
if(max < 0 || (max % 10 != 0) || max > 50) max = 25;
if(start < 0 ) start = 0;
} catch (Exception e) { }
query = queryBuilder.build();
File root = new File(RuntimeEnvironment.getInstance().getDataRootFile(),
"index");
if (RuntimeEnvironment.getInstance().hasProjects()) {
if (project == null) {
errorMsg = "<b>Error:</b> You must select a project!";
} else {
if (project.size() > 1) { //more projects
IndexSearcher[] searchables = new IndexSearcher[project.size()];
File droot = new File(RuntimeEnvironment.getInstance().getDataRootFile(), "index");
int ii = 0;
//TODO might need to rewrite to Project instead of String , need changes in projects.jspf too
for (String proj : project) {
FSDirectory dir = FSDirectory.open(new File(droot, proj));
searchables[ii++] = new IndexSearcher(dir);
}
if (Runtime.getRuntime().availableProcessors() > 1) {
searcher = new ParallelMultiSearcher(searchables);
} else {
searcher = new MultiSearcher(searchables);
}
} else { // just 1 project selected
root = new File(root, project.get(0));
FSDirectory dir = FSDirectory.open(root);
searcher = new IndexSearcher(dir);
}
}
} else { //no project setup
FSDirectory dir = FSDirectory.open(root);
searcher = new IndexSearcher(dir);
}
//TODO check if below is somehow reusing sessions so we don't requery again and again, I guess 2min timeout sessions could be usefull, since you click on the next page within 2mins, if not, then wait ;)
if (errorMsg == null) {
Sort sortf;
if (LASTMODTIME.equals(sort)) {
sortf = new Sort(new SortField("date", SortField.STRING, true));
} else if (BY_PATH.equals(sort)) {
sortf = new Sort(S_BY_PATH);
} else {
sortf = Sort.RELEVANCE;
}
TopFieldDocs fdocs = searcher.search(query, null, start + max, sortf);
totalHits = fdocs.totalHits;
hits = fdocs.scoreDocs;
}
thispage = max;
} catch (BooleanQuery.TooManyClauses e) {
errorMsg = "<b>Error:</b> Too many results for wildcard!";
} catch (ParseException e) {
errorMsg = "<b>Error parsing your query</b>" +
"<p/>You might try to enclose your search term in quotes, " +
"<a href=\"help.jsp#escaping\">escape special characters</a> " +
"with <b>\\</b>, or read the <a href=\"help.jsp\">Help</a> " +
"on the query language.<p/>" +
"Error message from parser:<br/>" + Util.htmlize(e.getMessage());
} catch (FileNotFoundException e) {
errorMsg = "<b>Error:</b> Index database not found";
} catch (Exception e) {
errorMsg = "<b>Error:</b> " + Util.htmlize(e.getMessage());
}
// Bug #3900: Check if this is a search for a single term, and that term
// is a definition. If that's the case, and we only have one match, we'll
// generate a direct link instead of a listing.
boolean isSingleDefinitionSearch =
(query instanceof TermQuery) && (defs != null);
// Attempt to create a direct link to the definition if we search for one
// single definition term AND we have exactly one match AND there is only
// one definition of that symbol in the document that matches.
boolean uniqueDefinition = false;
if (isSingleDefinitionSearch && hits != null && hits.length == 1) {
Document doc = searcher.doc(hits[0].doc);
if (doc.getFieldable("tags")!=null) {
byte[] rawTags = doc.getFieldable("tags").getBinaryValue();
Definitions tags = Definitions.deserialize(rawTags);
String symbol = ((TermQuery) query).getTerm().text();
if (tags.occurrences(symbol) == 1) {
uniqueDefinition = true;
}
}
}
// @TODO fix me. I should try to figure out where the exact hit is instead
// of returning a page with just _one_ entry in....
if (uniqueDefinition && request.getServletPath().equals(Constants.searchR)) {
String preFragmentPath = Util.URIEncodePath(context + Constants.xrefP + searcher.doc(hits[0].doc).get("path"));
String fragment = Util.URIEncode(((TermQuery)query).getTerm().text());
StringBuilder url = new StringBuilder(preFragmentPath);
url.append("#");
url.append(fragment);
response.sendRedirect(url.toString());
} else {
String pageTitle = "Search";
RuntimeEnvironment environment = RuntimeEnvironment.getInstance();
environment.register();
%><%@ include file="httpheader.jspf" %>
<body>
<div id="page">
<div id="header"><%@ include file="pageheader.jspf" %></div>
<div id="Masthead"></div>
<div id="bar">
<table border="0" width="100%"><tr><td><a href="<%=context%>/" id="home">Home</a></td><td align="right"><%
{
String url = "search?";
url = url + (q == null ? "" : "&amp;q=" + Util.URIEncode(q)) +
(defs == null ? "" : "&amp;defs=" + Util.URIEncode(defs)) +
(refs == null ? "" : "&amp;refs=" + Util.URIEncode(refs)) +
(path == null ? "" : "&amp;path=" + Util.URIEncode(path)) +
(hist == null ? "" : "&amp;hist=" + Util.URIEncode(hist));
if (hasProjects) {
if (project!=null) {
url = url + "&amp;project=";
for (Iterator it = project.iterator(); it.hasNext();) {
url = url + (project == null ? "" : Util.URIEncode((String) it.next()) + ",");
}
}
}
%>Sort by: <%
url=url+("&amp;sort=");
if (sort == null || RELEVANCY.equals(sort)) {
%><b>relevance</b> | <a href="<%=url+LASTMODTIME%>">last modified time</a> | <a href="<%=url+BY_PATH%>">path</a><%
} else if (LASTMODTIME.equals(sort)) {
%><a href="<%=url+RELEVANCY%>">relevance</a> | <b>last modified time</b> | <a href="<%=url+BY_PATH%>">path</a><%
} else if (BY_PATH.equals(sort)) {
%><a href="<%=url+RELEVANCY%>">relevance</a> | <a href="<%=url+LASTMODTIME%>">last modified time</a> | <b>path</b><%
} else {
%><a href="<%=url+RELEVANCY%>">relevance</a> | <a href="<%=url+LASTMODTIME%>">last modified time</a> | <a href="<%=url+BY_PATH%>">path</a><%
}
} %></td></tr></table>
</div>
<div id="menu">
<%@ include file="menu.jspf"%>
</div>
<div id="results">
<%
//TODO spellchecking cycle below is not that great and we only create suggest links for every token in query, not for a query as whole
if( hits == null || errorMsg != null) {
%><%=errorMsg%><%
} else if (hits.length == 0) {
File spellIndex = new File(env.getDataRootPath(), "spellIndex");
File[] spellIndexes=null;
if (RuntimeEnvironment.getInstance().hasProjects()) {
if (project.size() > 1) { //more projects
spellIndexes = new File[project.size()];
int ii = 0;
//TODO might need to rewrite to Project instead of String , need changes in projects.jspf too
for (String proj : project) {
spellIndexes[ii++] = new File(spellIndex,proj);
}
} else { // just 1 project selected
spellIndex = new File(spellIndex, project.get(0));
}
}
int count=1;
if (spellIndexes!=null) {count=spellIndexes.length;}
for (int idx = 0; idx < count; idx++) {
if (spellIndexes!=null) spellIndex = spellIndexes[idx];
if (spellIndex.exists()) {
FSDirectory spellDirectory = FSDirectory.open(spellIndex);
SpellChecker checker = new SpellChecker(spellDirectory);
Date sstart = new Date();
boolean printHeader = true;
String[] toks;
if(q != null) {
toks = q.split("[\t ]+");
if(toks != null){
for(int j=0; j<toks.length; j++) {
if(toks[j].length() > 3) {
String[] ret = checker.suggestSimilar(toks[j].toLowerCase(), 5);
for(int i = 0;i < ret.length; i++) {
if (printHeader) {
%><p><font color="#cc0000">Did you mean(for <%=spellIndex.getName()%>)</font>:<%
printHeader = false;
}
%> <a href=search?q=<%=ret[i]%>><%=ret[i]%></a> &nbsp; <%
}
}
}
}
}
if(refs != null) {
toks = refs.split("[\t ]+");
if(toks != null){
for(int j=0; j<toks.length; j++) {
if(toks[j].length() > 3) {
String[] ret = checker.suggestSimilar(toks[j].toLowerCase(), 5);
for(int i = 0;i < ret.length; i++) {
if (printHeader) {
%><p><font color="#cc0000">Did you mean(for <%=spellIndex.getName()%>)</font>:<%
printHeader = false;
}
%> <a href=search?refs=<%=ret[i]%>><%=ret[i]%></a> &nbsp; <%
}
}
}
}
}
//TODO it seems the only true spellchecker is for below field, see IndexDatabase createspellingsuggestions ...
if(defs != null) {
toks = defs.split("[\t ]+");
if(toks != null){
for(int j=0; j<toks.length; j++) {
if(toks[j].length() > 3) {
String[] ret = checker.suggestSimilar(toks[j].toLowerCase(), 5);
for(int i = 0;i < ret.length; i++) {
if (printHeader) {
%><p><font color="#cc0000">Did you mean(for <%=spellIndex.getName()%>)</font>:<%
printHeader = false;
}
%> <a href=search?defs=<%=ret[i]%>><%=ret[i]%></a> &nbsp; <%
}
}
}
}
}
if (printHeader) {
%></p><%
}
spellDirectory.close();
checker.close();
}
}
%><p> Your search <b><%=query.toString()%></b> did not match any files.
<br />
Suggestions:<br/><blockquote>- Make sure all terms are spelled correctly.<br/>
- Try different keywords.<br/>
- Try more general keywords.<br/>
- Use 'wil*' cards if you are looking for partial match.
</blockquote>
</p><%
} else { // We have a lots of results to show
StringBuilder slider = null;
if ( max < totalHits) {
if((start + max) < totalHits) {
thispage = max;
} else {
thispage = totalHits - start;
}
String urlp = (q == null ? "" : "&amp;q=" + Util.URIEncode(q)) +
(defs == null ? "" : "&amp;defs=" + Util.URIEncode(defs)) +
(refs == null ? "" : "&amp;refs=" + Util.URIEncode(refs)) +
(path == null ? "" : "&amp;path=" + Util.URIEncode(path)) +
(hist == null ? "" : "&amp;hist=" + Util.URIEncode(hist)) +
(sort == null ? "" : "&amp;sort=" + Util.URIEncode(sort));
if (hasProjects) {
urlp = urlp + "&amp;project=";
for (Iterator it = project.iterator(); it.hasNext();) {
urlp = urlp + (project == null ? "" : Util.URIEncode((String) it.next()) + ",");
}
}
slider = new StringBuilder();
int labelStart =1;
int sstart = start - max* (start / max % 10 + 1) ;
if(sstart < 0) {
sstart = 0;
labelStart = 1;
} else {
labelStart = sstart/max + 1;
}
int label = labelStart;
int labelEnd = label + 11;
String arr;
for(int i=sstart; i<totalHits && label <= labelEnd; i+= max) {
if (i <= start && start < i+ max) {
slider.append("<span class=\"sel\">" + label + "</span>");
} else {
if(label == labelStart && label != 1) {
arr = "&lt;&lt";
} else if(label == labelEnd && i < totalHits) {
arr = "&gt;&gt;";
} else {
arr = label < 10 ? " " + label : String.valueOf(label);
}
slider.append("<a class=\"more\" href=\"s?n=" + max + "&amp;start=" + i + urlp + "\">"+
arr + "</a>");
}
label++;
}
} else {
thispage = totalHits - start; // set the max index to max or last
}
%>&nbsp; &nbsp; Searched <b><%=query.toString()%></b> (Results <b><%=start+1%> -
<%=thispage+start%></b> of <b><%=totalHits%></b>) sorted by <%=sort%> <p><%=slider != null ?
slider.toString(): ""%></p>
<table width="100%" cellpadding="3" cellspacing="0" border="0"><%
Context sourceContext = null;
Summarizer summer = null;
if (query != null) {
try{
sourceContext =
new Context(query, queryBuilder.getQueries());
if(sourceContext != null)
summer = new Summarizer(query,
new CompatibleAnalyser());
} catch (Exception e) {
}
}
HistoryContext historyContext = null;
try {
historyContext = new HistoryContext(query);
} catch (Exception e) {
}
EftarFileReader ef = null;
try{
ef = new EftarFileReader(env.getDataRootPath() + "/index/dtags.eftar");
} catch (Exception e) {
}
//TODO also fix the way what and how it is passed to prettyprint, can improve performance! SearchEngine integration is really needed here.
Results.prettyPrintHTML(searcher,hits, start, start+thispage,
out,
sourceContext, historyContext, summer,
context,
env.getSourceRootPath(),
env.getDataRootPath(),
ef);
if(ef != null) {
try{
ef.close();
} catch (IOException e) {
}
}
%></table><br/>
<b> Completed in <%=(new Date()).getTime() - starttime.getTime()%> milliseconds </b> <br/>
<%=slider != null ? "<p>" + slider + "</p>" : ""%>
<%
}
%><br/></div><%@include file="foot.jspf"%><%
}
if (searcher != null) {
searcher.close();
}
} else { // Entry page show the map
response.sendRedirect(context + "/index.jsp");
}
%>