/**
* Fetches the attachment content from the repository. Content is flat text that can be used for
* indexing/searching or display
*
* @param attachmentName Name of the attachment.
* @param version The version of the attachment.
* @return the content of the Attachment as a String.
*/
protected String getAttachmentContent(String attachmentName, int version) {
AttachmentManager mgr = m_engine.getAttachmentManager();
try {
Attachment att = mgr.getAttachmentInfo(attachmentName, version);
// FIXME: Find out why sometimes att is null
if (att != null) {
return getAttachmentContent(att);
}
} catch (ProviderException e) {
log.error("Attachment cannot be loaded", e);
}
// Something was wrong, no result is returned.
return null;
}
/** {@inheritDoc} */
public void initialize(WikiEngine engine, Properties props)
throws NoRequiredPropertyException, IOException {
m_engine = engine;
m_luceneDirectory = engine.getWorkDir() + File.separator + LUCENE_DIR;
int initialDelay =
TextUtil.getIntegerProperty(props, PROP_LUCENE_INITIALDELAY, LuceneUpdater.INITIAL_DELAY);
int indexDelay =
TextUtil.getIntegerProperty(props, PROP_LUCENE_INDEXDELAY, LuceneUpdater.INDEX_DELAY);
m_analyzerClass = TextUtil.getStringProperty(props, PROP_LUCENE_ANALYZER, m_analyzerClass);
// FIXME: Just to be simple for now, we will do full reindex
// only if no files are in lucene directory.
File dir = new File(m_luceneDirectory);
log.info("Lucene enabled, cache will be in: " + dir.getAbsolutePath());
try {
if (!dir.exists()) {
dir.mkdirs();
}
if (!dir.exists() || !dir.canWrite() || !dir.canRead()) {
log.error("Cannot write to Lucene directory, disabling Lucene: " + dir.getAbsolutePath());
throw new IOException("Invalid Lucene directory.");
}
String[] filelist = dir.list();
if (filelist == null) {
throw new IOException(
"Invalid Lucene directory: cannot produce listing: " + dir.getAbsolutePath());
}
} catch (IOException e) {
log.error("Problem while creating Lucene index - not using Lucene.", e);
}
// Start the Lucene update thread, which waits first
// for a little while before starting to go through
// the Lucene "pages that need updating".
LuceneUpdater updater = new LuceneUpdater(m_engine, this, initialDelay, indexDelay);
updater.start();
}
/**
* Adds a page-text pair to the lucene update queue. Safe to call always
*
* @param page WikiPage to add to the update queue.
*/
public void reindexPage(WikiPage page) {
if (page != null) {
String text;
// TODO: Think if this was better done in the thread itself?
if (page instanceof Attachment) {
text = getAttachmentContent((Attachment) page);
} else {
text = m_engine.getPureText(page);
}
if (text != null) {
// Add work item to m_updates queue.
Object[] pair = new Object[2];
pair[0] = page;
pair[1] = text;
m_updates.add(pair);
log.debug("Scheduling page " + page.getName() + " for index update");
}
}
}
/**
* @param att Attachment to get content for. Filename extension is used to determine the type of
* the attachment.
* @return String representing the content of the file. FIXME This is a very simple implementation
* of some text-based attachment, mainly used for testing. This should be replaced /moved to
* Attachment search providers or some other 'pluggable' wat to search attachments
*/
protected String getAttachmentContent(Attachment att) {
AttachmentManager mgr = m_engine.getAttachmentManager();
// FIXME: Add attachment plugin structure
String filename = att.getFileName();
boolean searchSuffix = false;
for (String suffix : SEARCHABLE_FILE_SUFFIXES) {
if (filename.endsWith(suffix)) {
searchSuffix = true;
}
}
if (searchSuffix) {
InputStream attStream;
try {
attStream = mgr.getAttachmentStream(att);
StringWriter sout = new StringWriter();
FileUtil.copyContents(new InputStreamReader(attStream), sout);
attStream.close();
sout.close();
return sout.toString();
} catch (ProviderException e) {
log.error("Attachment cannot be loaded", e);
return null;
} catch (IOException e) {
log.error("Attachment cannot be loaded", e);
return null;
}
}
return null;
}
/**
* Searches pages using a particular combination of flags.
*
* @param query The query to perform in Lucene query language
* @param flags A set of flags
* @return A Collection of SearchResult instances
* @throws ProviderException if there is a problem with the backend
*/
public Collection findPages(String query, int flags) throws ProviderException {
IndexSearcher searcher = null;
ArrayList<SearchResult> list = null;
Highlighter highlighter = null;
try {
String[] queryfields = {
LUCENE_PAGE_CONTENTS, LUCENE_PAGE_NAME, LUCENE_AUTHOR, LUCENE_ATTACHMENTS
};
QueryParser qp =
new MultiFieldQueryParser(Version.LUCENE_36, queryfields, getLuceneAnalyzer());
// QueryParser qp = new QueryParser( LUCENE_PAGE_CONTENTS, getLuceneAnalyzer() );
Query luceneQuery = qp.parse(query);
if ((flags & FLAG_CONTEXTS) != 0) {
highlighter =
new Highlighter(
new SimpleHTMLFormatter("<span class=\"searchmatch\">", "</span>"),
new SimpleHTMLEncoder(),
new QueryScorer(luceneQuery));
}
try {
File dir = new File(m_luceneDirectory);
Directory luceneDir = new SimpleFSDirectory(dir, null);
IndexReader reader = IndexReader.open(luceneDir);
searcher = new IndexSearcher(reader);
} catch (Exception ex) {
log.info("Lucene not yet ready; indexing not started", ex);
return null;
}
ScoreDoc[] hits = searcher.search(luceneQuery, MAX_SEARCH_HITS).scoreDocs;
list = new ArrayList<SearchResult>(hits.length);
for (int curr = 0; curr < hits.length; curr++) {
int docID = hits[curr].doc;
Document doc = searcher.doc(docID);
String pageName = doc.get(LUCENE_ID);
WikiPage page = m_engine.getPage(pageName, WikiPageProvider.LATEST_VERSION);
if (page != null) {
if (page instanceof Attachment) {
// Currently attachments don't look nice on the search-results page
// When the search-results are cleaned up this can be enabled again.
}
int score = (int) (hits[curr].score * 100);
// Get highlighted search contexts
String text = doc.get(LUCENE_PAGE_CONTENTS);
String[] fragments = new String[0];
if (text != null && highlighter != null) {
TokenStream tokenStream =
getLuceneAnalyzer().tokenStream(LUCENE_PAGE_CONTENTS, new StringReader(text));
fragments = highlighter.getBestFragments(tokenStream, text, MAX_FRAGMENTS);
}
SearchResult result = new SearchResultImpl(page, score, fragments);
list.add(result);
} else {
log.error(
"Lucene found a result page '"
+ pageName
+ "' that could not be loaded, removing from Lucene cache");
pageRemoved(new WikiPage(m_engine, pageName));
}
}
} catch (IOException e) {
log.error("Failed during lucene search", e);
} catch (ParseException e) {
log.info("Broken query; cannot parse query ", e);
throw new ProviderException(
"You have entered a query Lucene cannot process: " + e.getMessage());
} catch (InvalidTokenOffsetsException e) {
log.error("Tokens are incompatible with provided text ", e);
} finally {
if (searcher != null) {
try {
searcher.close();
} catch (IOException e) {
log.error(e);
}
}
}
return list;
}
/**
* Indexes page using the given IndexWriter.
*
* @param page WikiPage
* @param text Page text to index
* @param writer The Lucene IndexWriter to use for indexing
* @return the created index Document
* @throws IOException If there's an indexing problem
*/
protected Document luceneIndexPage(WikiPage page, String text, IndexWriter writer)
throws IOException {
if (log.isDebugEnabled()) log.debug("Indexing " + page.getName() + "...");
// make a new, empty document
Document doc = new Document();
if (text == null) return doc;
// Raw name is the keyword we'll use to refer to this document for updates.
Field field = new Field(LUCENE_ID, page.getName(), Field.Store.YES, Field.Index.NOT_ANALYZED);
doc.add(field);
// Body text. It is stored in the doc for search contexts.
field =
new Field(
LUCENE_PAGE_CONTENTS, text, Field.Store.YES, Field.Index.ANALYZED, Field.TermVector.NO);
doc.add(field);
// Allow searching by page name. Both beautified and raw
String unTokenizedTitle =
StringUtils.replaceChars(
page.getName(), MarkupParser.PUNCTUATION_CHARS_ALLOWED, c_punctuationSpaces);
field =
new Field(
LUCENE_PAGE_NAME,
TextUtil.beautifyString(page.getName()) + " " + unTokenizedTitle,
Field.Store.YES,
Field.Index.ANALYZED,
Field.TermVector.NO);
doc.add(field);
// Allow searching by authorname
if (page.getAuthor() != null) {
field =
new Field(
LUCENE_AUTHOR,
page.getAuthor(),
Field.Store.YES,
Field.Index.ANALYZED,
Field.TermVector.NO);
doc.add(field);
}
// Now add the names of the attachments of this page
try {
Collection attachments = m_engine.getAttachmentManager().listAttachments(page);
String attachmentNames = "";
for (Iterator it = attachments.iterator(); it.hasNext(); ) {
Attachment att = (Attachment) it.next();
attachmentNames += att.getName() + ";";
}
field =
new Field(
LUCENE_ATTACHMENTS,
attachmentNames,
Field.Store.YES,
Field.Index.ANALYZED,
Field.TermVector.NO);
doc.add(field);
} catch (ProviderException e) {
// Unable to read attachments
log.error("Failed to get attachments for page", e);
}
writer.addDocument(doc);
return doc;
}
/**
* Performs a full Lucene reindex, if necessary.
*
* @throws IOException If there's a problem during indexing
*/
protected void doFullLuceneReindex() throws IOException {
File dir = new File(m_luceneDirectory);
String[] filelist = dir.list();
if (filelist == null) {
throw new IOException(
"Invalid Lucene directory: cannot produce listing: " + dir.getAbsolutePath());
}
try {
if (filelist.length == 0) {
//
// No files? Reindex!
//
Date start = new Date();
IndexWriter writer = null;
log.info("Starting Lucene reindexing, this can take a couple minutes...");
Directory luceneDir = new SimpleFSDirectory(dir, null);
try {
writer = getIndexWriter(luceneDir);
Collection allPages = m_engine.getPageManager().getAllPages();
for (Iterator iterator = allPages.iterator(); iterator.hasNext(); ) {
WikiPage page = (WikiPage) iterator.next();
try {
String text =
m_engine
.getPageManager()
.getPageText(page.getName(), WikiProvider.LATEST_VERSION);
luceneIndexPage(page, text, writer);
} catch (IOException e) {
log.warn("Unable to index page " + page.getName() + ", continuing to next ", e);
}
}
Collection allAttachments = m_engine.getAttachmentManager().getAllAttachments();
for (Iterator iterator = allAttachments.iterator(); iterator.hasNext(); ) {
Attachment att = (Attachment) iterator.next();
try {
String text = getAttachmentContent(att.getName(), WikiProvider.LATEST_VERSION);
luceneIndexPage(att, text, writer);
} catch (IOException e) {
log.warn("Unable to index attachment " + att.getName() + ", continuing to next", e);
}
}
} finally {
close(writer);
}
Date end = new Date();
log.info(
"Full Lucene index finished in "
+ (end.getTime() - start.getTime())
+ " milliseconds.");
} else {
log.info("Files found in Lucene directory, not reindexing.");
}
} catch (NoClassDefFoundError e) {
log.info("Lucene libraries do not exist - not using Lucene.");
} catch (IOException e) {
log.error("Problem while creating Lucene index - not using Lucene.", e);
} catch (ProviderException e) {
log.error("Problem reading pages while creating Lucene index (JSPWiki won't start.)", e);
throw new IllegalArgumentException("unable to create Lucene index");
} catch (Exception e) {
log.error("Unable to start lucene", e);
}
}
