public LuceneSearch() {
String dvnIndexLocation = System.getProperty("dvn.index.location");
// System.out.println("INDEX LOCATION " + dvnIndexLocation);
File locationDirectory = null;
if (dvnIndexLocation != null) {
locationDirectory = new File(dvnIndexLocation);
if (locationDirectory.exists() && locationDirectory.isDirectory()) {
indexDir = dvnIndexLocation + "/index-dir";
// System.out.println("INDEX " + indexDir);
}
}
// System.out.println("INDEX DEFAULT " + indexDir);
String dvnMaxClauseCountStr = System.getProperty("dvn.search.maxclausecount");
if (dvnMaxClauseCountStr != null) {
try {
dvnMaxClauseCount = Integer.parseInt(dvnMaxClauseCountStr);
} catch (Exception e) {
e.printStackTrace();
dvnMaxClauseCount = 1024;
}
}
try {
dir = FSDirectory.getDirectory(indexDir, false);
r = IndexReader.open(dir);
searcher = new IndexSearcher(r);
} catch (IOException ex) {
ex.printStackTrace();
}
}
/**
* check the existence of the given word in the index
*
* @param indexPath index path's
* @param word String to check
* @return
* @throws IOException
*/
private boolean checkExistingWord(String indexPath, String word) throws IOException {
File file = new File(indexPath);
FSDirectory directory = FSDirectory.getDirectory(file);
SpellChecker spellChecker = new SpellChecker(directory);
return spellChecker.exist(word);
}
/*
* index all child directories(only first level directories) in parent directory
* and indexed data is stored in the same name source directory
*/
private long indexDirectories(String parent, String[] dirs, String index, SetupParameters Pa)
throws FileHandlerException, IOException {
long sumDocs = 0;
// index each directory in parent directory
for (int i = 0; i < dirs.length; i++) {
System.out.println("\t-----FOLDER----- :" + dirs[i].toUpperCase());
String dir_index = index + "/" + dirs[i];
if ((index.endsWith("\\")) || (index.endsWith("/"))) {
dir_index = index + dirs[i];
}
Directory di = FSDirectory.getDirectory(new File(dir_index), true);
Pa.setDir(di);
Pa.setWriter(new IndexWriter(Pa.getDir(), Pa.getAnalyzer(), true));
// //get name of directory contains website to index
// int begin=dirs[i].lastIndexOf("\\");
// if(begin==-1) begin=dirs[i].lastIndexOf("/");
// int end=dirs[i].length()-1;
// String dir_site=dirs[i].substring(begin, end);
this.index(dirs[i].toLowerCase(), Pa.getWriter(), new File(parent + "\\" + dirs[i]));
Pa.getWriter().optimize();
Pa.getWriter().close();
IndexReader reader = Pa.getReader().open(Pa.getDir());
sumDocs += reader.numDocs();
reader.close();
}
return sumDocs;
}
public static void IndexInputTest() throws Exception {
String path = "D:\\Lucene Document";
directory = FSDirectory.getDirectory(path);
IndexInput indexInput = directory.openInput("segments.gen");
int version = indexInput.readInt();
System.out.println(version);
System.out.println(indexInput.readString());
}
/** 具体执行删除索引的方法，将被工作流中DeleteTask类的workrun()方法调用。 */
public void deleteIndex() throws IOException {
Directory fsDir = FSDirectory.getDirectory(indexFile);
Term t = new Term("ID", String.valueOf(this.id));
IndexReader reader = IndexReader.open(fsDir);
reader.deleteDocuments(t);
reader.close();
fsDir.close();
}
protected void openWriterForDatabase(String db) throws IOException, CorruptIndexException {
File indexDir = indexPath(db);
if (!indexDir.exists()) {
onDatabaseCreated(db, -1);
} else {
Directory directory = FSDirectory.getDirectory(indexDir.getPath());
writers.put(db, new IndexWriter(directory, true, new StandardAnalyzer(), false));
}
}
private Directory getDirectory(Path file) throws IOException {
if ("file".equals(this.fs.getUri().getScheme())) {
Path qualified = file.makeQualified(FileSystem.getLocal(conf));
File fsLocal = new File(qualified.toUri());
return FSDirectory.getDirectory(fsLocal.getAbsolutePath());
} else {
return new FsDirectory(this.fs, file, false, this.conf);
}
}
protected Hits query(String db, String defaultField, String queryString)
throws IOException, CorruptIndexException, ParseException {
Directory directory = FSDirectory.getDirectory(indexPath(db));
IndexReader reader = IndexReader.open(directory);
Searcher searcher = new IndexSearcher(reader);
Analyzer analyzer = new StandardAnalyzer();
QueryParser qp = new QueryParser(defaultField, analyzer);
Query query = qp.parse(queryString);
return searcher.search(query);
}
private RAMDirectory(FSDirectory dir, boolean closeDir, IOContext context) throws IOException {
this();
for (String file : dir.listAll()) {
if (!Files.isDirectory(dir.getDirectory().resolve(file))) {
copyFrom(dir, file, file, context);
}
}
if (closeDir) {
dir.close();
}
}
public void startSearch() {
try {
IndexReader.unlock(FSDirectory.getDirectory(ConfigHandler.indexPath));
reader = IndexReader.open(ConfigHandler.indexPath);
searcher = new IndexSearcher(reader);
analyzer = new StandardAnalyzer();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
private void unlockIndex(String indexLocation) {
if (IndexReader.indexExists(indexLocation)) {
try {
if (IndexReader.isLocked(indexLocation)) {
Directory directory = FSDirectory.getDirectory(indexLocation);
IndexReader.unlock(directory);
}
} catch (IOException e) {
log.fatal("Error trying to unlock " + indexLocation + " index.", e);
}
}
}
public void stop() {
@SuppressWarnings("unused")
int readCurrentState =
current; // Another unneeded value, to ensure visibility of state protected by memory
// barrier
timer.cancel();
task.stop();
try {
directory.close();
} catch (Exception e) {
log.unableToCloseLuceneDirectory(directory.getDirectory(), e);
}
}
public static void IndexCreate() throws Exception {
String path = "D:\\Lucene Document";
directory = FSDirectory.getDirectory(path);
Analyzer analyzer = new StandardAnalyzer();
IndexWriter iwriter = new IndexWriter(directory, analyzer, true);
iwriter.setMaxFieldLength(25000);
// make a new, empty document
Document doc = new Document();
String text = "This is the text to be indexed.";
doc.add(new Field("fieldname", text, Field.Store.YES, Field.Index.TOKENIZED));
iwriter.addDocument(doc);
iwriter.optimize();
iwriter.close();
}
public void onDatabaseCreated(String db, long seq) {
this.abort(db);
File path = indexPath(db);
onDatabaseDeleted(db, -1);
if (!path.exists()) {
log.info("Creating lucene directory {} for db {}", path.getPath(), db);
path.mkdirs();
}
try {
Directory directory = FSDirectory.getDirectory(path.getPath());
writers.put(db, new IndexWriter(directory, true, new StandardAnalyzer(), true));
} catch (CorruptIndexException e) {
} catch (LockObtainFailedException e) {
} catch (IOException e) {
}
}
private String getId(IndexCommit commit) {
StringBuilder sb = new StringBuilder();
Directory dir = commit.getDirectory();
// For anything persistent, make something that will
// be the same, regardless of the Directory instance.
if (dir instanceof FSDirectory) {
FSDirectory fsd = (FSDirectory) dir;
File fdir = fsd.getDirectory();
sb.append(fdir.getPath());
} else {
sb.append(dir);
}
sb.append('/');
sb.append(commit.getGeneration());
sb.append('_');
sb.append(commit.getVersion());
return sb.toString();
}
/**
* constructs a new IndexReader instance
*
* @param indexPath Where the index is.
* @return Constructed IndexReader instance.
* @throws IOException
*/
private static InternalIndexReader newReader(File luceneDir, IndexSignature signature)
throws IOException {
if (!luceneDir.exists() || !IndexReader.indexExists(luceneDir)) return null;
Directory dir = FSDirectory.getDirectory(luceneDir);
int numTries = INDEX_OPEN_NUM_RETRIES;
InternalIndexReader reader = null;
// try max of 5 times, there might be a case where the segment file is being updated
while (reader == null) {
if (numTries == 0) {
log.error("Problem refreshing disk index, all attempts failed.");
throw new IOException("problem opening new index");
}
numTries--;
try {
log.debug("opening index reader at: " + luceneDir.getAbsolutePath());
IndexReader srcReader = IndexReader.open(dir, true);
try {
reader = new InternalIndexReader(srcReader, signature);
} catch (IOException ioe) {
// close the source reader if InternalIndexReader construction fails
if (srcReader != null) {
srcReader.close();
}
throw ioe;
}
} catch (IOException ioe) {
try {
Thread.sleep(100);
} catch (InterruptedException e) {
log.warn("thread interrupted.");
continue;
}
}
}
return reader;
}
/*
* index specific directory-all file in one directory
*/
private long indexDirectory(String directory, String index, SetupParameters Pa)
throws FileHandlerException, IOException {
long sumDocs = 0;
Directory di = FSDirectory.getDirectory(new File(index), true);
// RAMDirectory di = new RAMDirectory(new Directory());
Pa.setDir(di);
Pa.setWriter(new IndexWriter(Pa.getDir(), Pa.getAnalyzer(), true));
// get name of directory contains website to index
int begin = directory.lastIndexOf("\\");
if (begin == -1) begin = directory.lastIndexOf("/");
int end = directory.length();
String dir_site = directory.substring(begin + 1, end).toLowerCase();
index(dir_site, Pa.getWriter(), new File(directory));
Pa.getWriter().optimize();
Pa.getWriter().close();
IndexReader reader = Pa.getReader().open(Pa.getDir());
sumDocs += reader.numDocs();
reader.close();
return sumDocs;
}
/**
* Will reindex, shift if needed and publish indexes for a "remote" repository (published over
* jetty component).
*
* @param repositoryRoot
* @param repositoryId
* @param deleteIndexFiles
* @param shiftDays
* @throws IOException
*/
protected void reindexRemoteRepositoryAndPublish(
File repositoryRoot, String repositoryId, boolean deleteIndexFiles, int shiftDays)
throws IOException, ComponentLookupException {
File indexDirectory = getIndexFamilyDirectory(repositoryId);
Directory directory = FSDirectory.getDirectory(indexDirectory);
IndexingContext ctx =
nexusIndexer.addIndexingContextForced(
repositoryId + "-temp",
repositoryId,
repositoryRoot,
directory,
null,
null,
new IndexCreatorHelper(getContainer()).getFullCreators());
// shifting if needed (very crude way to do it, but heh)
shiftContextInTime(ctx, shiftDays);
// and scan "today"
nexusIndexer.scan(ctx);
ctx.updateTimestamp(true);
// pack it up
File targetDir = new File(repositoryRoot, ".index");
targetDir.mkdirs();
IndexPackingRequest ipr = new IndexPackingRequest(ctx, targetDir);
ipr.setCreateIncrementalChunks(true);
indexPacker.packIndex(ipr);
nexusIndexer.removeIndexingContext(ctx, deleteIndexFiles);
}
static String str(IndexCommit commit) {
StringBuilder sb = new StringBuilder();
try {
sb.append("commit{");
Directory dir = commit.getDirectory();
if (dir instanceof FSDirectory) {
FSDirectory fsd = (FSDirectory) dir;
sb.append("dir=").append(fsd.getDirectory());
} else {
sb.append("dir=").append(dir);
}
sb.append(",segFN=").append(commit.getSegmentsFileName());
sb.append(",version=").append(commit.getVersion());
sb.append(",generation=").append(commit.getGeneration());
sb.append(",filenames=").append(commit.getFileNames());
} catch (Exception e) {
sb.append(e);
}
return sb.toString();
}
// CorruptIndexException, LockObtainFailedException, IOException,
// ParseException
public IndexesConfigLowVersionLucene(String indexDirectoryPath) {
// create some index
// we could also create an index in our ram ...
// Directory index = new RAMDirectory();
indexName = indexDirectoryPath;
try {
indexDirectory = FSDirectory.getDirectory(indexDirectoryPath); // "index/pages"
indexDirectory.setLockFactory(new NativeFSLockFactory(indexDirectoryPath));
w = new IndexWriter(indexDirectory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED);
} catch (CorruptIndexException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (LockObtainFailedException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
// searchIndex("vitol");
urlsFileNames = new HashMap<String, String>();
emptyFiles = new HashMap<String, String>();
}
protected void setup() throws IOException {
File indexDirectory = new File(indexDir);
dir = FSDirectory.getDirectory(indexDir, !indexDirectory.exists());
}
public void convert(Application app, File dbhome) throws Exception {
FSDirectory indexDir = FSDirectory.getDirectory(dbhome, false);
if (indexDir instanceof TransFSDirectory) {
FSDirectory.setDisableLocks(true);
TransFSDirectory d = (TransFSDirectory) indexDir;
TransSource source = app.getTransSource();
d.setDriverClass(source.getDriverClass());
d.setUrl(source.getUrl());
d.setUser(source.getUser());
d.setPassword(source.getPassword());
}
File ndbhome = new File(dbhome.getParentFile(), dbhome.getName() + "_tmp");
File olddbhome = new File(dbhome.getParentFile(), dbhome.getName() + "_old");
FSDirectory nindexDir = FSDirectory.getDirectory(ndbhome, true);
if (nindexDir instanceof TransFSDirectory) {
FSDirectory.setDisableLocks(true);
TransFSDirectory d = (TransFSDirectory) nindexDir;
TransSource source = app.getTransSource();
d.setDriverClass(source.getDriverClass());
d.setUrl(source.getUrl());
d.setUser(source.getUser());
d.setPassword(source.getPassword());
}
IndexSearcher searcher = null;
IndexWriter writer = null;
LuceneManager lmgr = null;
try {
searcher = new IndexSearcher(indexDir);
PerFieldAnalyzerWrapper a = LuceneManager.buildAnalyzer();
writer = IndexWriterManager.getWriter(nindexDir, a, true);
final int numDocs = searcher.getIndexReader().numDocs();
HashSet deldocs = new HashSet();
HashMap infos = new HashMap();
for (int i = 0; i < numDocs; i++) {
Document doc = searcher.doc(i);
String delprop = doc.get(DeletedInfos.DELETED);
String layerStr = doc.get(LuceneManager.LAYER_OF_SAVE);
int layer = -1;
try {
layer = Integer.parseInt(layerStr);
} catch (Exception ex) {
layer = -1;
}
final String id =
doc.get(LuceneManager.ID)
+ DeletedInfos.KEY_SEPERATOR
+ doc.get(LuceneManager.LAYER_OF_SAVE);
if (delprop != null && "true".equals(delprop) /* && layer == DbKey.LIVE_LAYER*/) {
deldocs.add(id);
} else {
Object v;
if ((v = infos.get(id)) == null) {
infos.put(id, new Integer(i));
} else {
final String lmod = doc.get(LuceneManager.LASTMODIFIED);
final String lmod_prev = searcher.doc(((Integer) v).intValue()).get("_lastmodified");
if (lmod_prev == null || (lmod != null && lmod.compareTo(lmod_prev) > 0)) {
infos.put(id, new Integer(i));
}
}
}
}
ArrayList listOfMaps = new ArrayList();
for (int i = 0; i < numDocs; i++) {
Document doc = searcher.doc(i);
String delprop = doc.get(DeletedInfos.DELETED);
String layerStr = doc.get(LuceneManager.LAYER_OF_SAVE);
int layer = -1;
try {
layer = Integer.parseInt(layerStr);
} catch (Exception ex) {
layer = -1;
}
final String id =
doc.get(LuceneManager.ID)
+ DeletedInfos.KEY_SEPERATOR
+ doc.get(LuceneManager.LAYER_OF_SAVE);
if (delprop != null && "true".equals(delprop)) {
continue;
} else if (id != null && deldocs.contains(id) /* && layer == DbKey.LIVE_LAYER*/) {
continue;
}
Integer idx = (Integer) infos.get(id);
if (idx != null && i != idx.intValue()) {
continue;
}
Document ndoc = convertDocument(doc);
if (this.recordNodes) {
listOfMaps.add(LuceneManager.luceneDocumentToMap(doc));
}
if (ndoc != null) {
writer.addDocument(ndoc);
}
}
if (this.recordNodes) {
lmgr = new LuceneManager(this.app, false, true);
this.allNodes = new HashMap();
final int size = listOfMaps.size();
for (int i = 0; i < size; i++) {
HashMap m = (HashMap) listOfMaps.get(i);
INode n = lmgr.mapToNode(m);
this.allNodes.put(n.getID(), getPath(n));
n = null;
}
}
} catch (Exception ex) {
ex.printStackTrace();
throw new RuntimeException(ex);
} finally {
if (searcher != null) {
try {
searcher.close();
} catch (Exception ex) {
app.logError(ErrorReporter.errorMsg(this.getClass(), "convert"), ex);
}
}
if (lmgr != null) {
lmgr.shutdown();
lmgr = null;
}
indexDir.close();
SegmentInfos sinfos = IndexObjectsFactory.getFSSegmentInfos(indexDir);
sinfos.clear();
IndexObjectsFactory.removeDeletedInfos(indexDir);
}
Connection conn = null;
boolean exceptionOccured = false;
try {
if (writer != null) {
TransSource ts = app.getTransSource();
conn = ts.getConnection();
DatabaseMetaData dmd = conn.getMetaData();
ResultSet rs = dmd.getColumns(null, null, "Lucene", "version");
if (!rs.next()) {
final String alterTbl = "ALTER TABLE Lucene ADD version INT NOT NULL DEFAULT 1";
PreparedStatement pstmt = null;
try {
pstmt = conn.prepareStatement(alterTbl);
pstmt.execute();
} catch (SQLException sqle) {
app.logError(ErrorReporter.errorMsg(this.getClass(), "convert"), sqle);
} finally {
if (pstmt != null) {
pstmt.close();
pstmt = null;
}
}
}
rs.close();
rs = null;
writer.close();
writer.flushCache(); // TODO:writer.writeSegmentsFile();
LuceneManager.commitSegments(conn, app, writer.getDirectory());
writer.finalizeTrans();
this.updateSQL(conn);
}
} catch (Exception ex) {
ex.printStackTrace();
exceptionOccured = true;
throw new RuntimeException(ex);
} finally {
if (conn != null) {
try {
if (!conn.getAutoCommit()) {
if (!exceptionOccured) {
conn.commit();
} else {
conn.rollback();
}
}
conn.close();
} catch (Exception ex) {
app.logError(ErrorReporter.errorMsg(this.getClass(), "convert"), ex);
}
conn = null;
}
nindexDir.close();
SegmentInfos sinfos = IndexObjectsFactory.getFSSegmentInfos(nindexDir);
sinfos.clear();
IndexObjectsFactory.removeDeletedInfos(nindexDir);
}
if (!dbhome.renameTo(olddbhome)) {
throw new Exception("Could not move the old version of the db into " + olddbhome);
}
if (!ndbhome.renameTo(dbhome)) {
throw new Exception("Could not move the newer version of the db into " + dbhome);
}
File oldBlobDir = new File(olddbhome, "blob");
File newBlobDir = new File(ndbhome, "blob");
oldBlobDir.renameTo(newBlobDir);
if (!FileUtils.deleteDir(olddbhome)) {
throw new Exception("Could not delete the old version of the db at " + olddbhome);
}
}
/**
* Canonicalize the first maxLineNumber lines in input, i.e., sort the tokens by document
* frequency in ascending order
*
* @param input
* @param maxLineNumber
* @param indexFolder
* @param output
* @throws Exception
*/
public static void canonicalize(
String input, int maxLineNumber, String indexFolder, String output) throws Exception {
Directory dir = FSDirectory.getDirectory(indexFolder);
IndexWriter iwriter = new IndexWriter(dir, null, true, IndexWriter.MaxFieldLength.UNLIMITED);
BufferedReader br = IOFactory.getBufferedReader(input);
int lineCount = 0;
for (String line = br.readLine(); line != null; line = br.readLine()) {
String[] tokens = Common.sortUnique(line, 1);
for (int i = 1; i < tokens.length; i++) {
String t = tokens[i];
Document doc = new Document();
doc.add(new Field("term", t, Field.Store.NO, Field.Index.NOT_ANALYZED));
iwriter.addDocument(doc);
}
lineCount++;
if (lineCount % 100 == 0)
System.out.println(new Date().toString() + " : " + lineCount + " lines indexed");
if (lineCount == maxLineNumber) break;
}
System.out.println(new Date().toString() + " : " + lineCount + " lines indexed");
br.close();
iwriter.optimize();
iwriter.close();
System.out.println(new Date().toString() + " : indexing finished");
final IndexReader ireader = IndexReader.open(dir);
br = IOFactory.getBufferedReader(input);
PrintWriter pw = IOFactory.getPrintWriter(output);
lineCount = 0;
for (String line = br.readLine(); line != null; line = br.readLine()) {
String[] tokens = Common.sortUnique(line, 1);
Arrays.sort(
tokens,
1,
tokens.length,
new Comparator<String>() {
public int compare(String a, String b) {
try {
int fa = ireader.docFreq(new Term("term", a));
int fb = ireader.docFreq(new Term("term", b));
if (fa > fb) return 1;
else if (fa < fb) return -1;
return 0;
} catch (Exception e) {
e.printStackTrace();
return 0;
}
}
});
pw.print(tokens[0]);
for (int i = 1; i < tokens.length; i++) pw.print(" " + tokens[i]);
pw.println();
lineCount++;
if (lineCount % 100000 == 0)
System.out.println(new Date().toString() + " : " + lineCount + " lines output");
}
System.out.println(new Date().toString() + " : " + lineCount + " lines output");
pw.close();
br.close();
ireader.close();
dir.close();
Common.deleteFolder(new File(indexFolder));
System.out.println(new Date().toString() + " : canonicalization finished");
}
