From b7984c5138baabb3cc4db1752725e7f054279ae7 Mon Sep 17 00:00:00 2001
From: synhershko <synhershko@users.sourceforge.net>
Date: Wed, 3 Aug 2011 07:48:31 +0000
Subject: [PATCH] Updating CLucene search and index routines

---
 src/common/kiwix/cluceneIndexer.cpp  |  8 +++---
 src/common/kiwix/cluceneSearcher.cpp | 37 +++++++++++-----------------
 2 files changed, 19 insertions(+), 26 deletions(-)

diff --git a/src/common/kiwix/cluceneIndexer.cpp b/src/common/kiwix/cluceneIndexer.cpp
index bc3d9ab9c..e66ce596d 100644
--- a/src/common/kiwix/cluceneIndexer.cpp
+++ b/src/common/kiwix/cluceneIndexer.cpp
@@ -28,6 +28,7 @@ namespace kiwix {
 
     this->dir = FSDirectory::getDirectory(cluceneDirectoryPath.c_str(), true);
     this->writer = new IndexWriter(dir, &analyzer, true);
+    writer->setUseCompoundFile(false);
   }
   
   void CluceneIndexer::indexNextPercentPre() {
@@ -47,7 +48,7 @@ namespace kiwix {
     /* Not indexed but stored */
     //STRCPY_AtoT(buffer, title.c_str(), MAX_BUFFER_SIZE);
     ::mbstowcs(buffer,title.c_str(),MAX_BUFFER_SIZE);
-    doc.add(*_CLNEW Field(_T("title"), buffer, Field::STORE_YES | Field::INDEX_UNTOKENIZED));
+    doc.add(*_CLNEW Field(_T("title"), buffer, Field::STORE_YES | Field::INDEX_UNTOKENIZED)); // TODO: Why store, not analyzed? what is utitle?
 
     //STRCPY_AtoT(buffer, url.c_str(), MAX_BUFFER_SIZE);
     ::mbstowcs(buffer,url.c_str(),MAX_BUFFER_SIZE);
@@ -68,7 +69,7 @@ namespace kiwix {
 
     //STRCPY_AtoT(buffer, content.c_str(), MAX_BUFFER_SIZE);
     ::mbstowcs(buffer,content.c_str(),MAX_BUFFER_SIZE);
-    doc.add(*_CLNEW Field(_T("content"), buffer, Field::STORE_NO | Field::INDEX_TOKENIZED));
+    doc.add(*_CLNEW Field(_T("content"), buffer, Field::STORE_NO | Field::INDEX_TOKENIZED)); // TODO: TermVectors if you want to highlight
 
     /* Add the document to the index */
     this->writer->addDocument(&doc);
@@ -78,9 +79,10 @@ namespace kiwix {
   }
   
   void CluceneIndexer::stopIndexing() {
+    writer->setUseCompoundFile(true);
     this->writer->optimize();
     this->writer->close();
     delete this->writer;
-    delete this->dir;
+    _CLDECDELETE(this->dir);
   }
 }
diff --git a/src/common/kiwix/cluceneSearcher.cpp b/src/common/kiwix/cluceneSearcher.cpp
index 8a6ce714d..62f414677 100644
--- a/src/common/kiwix/cluceneSearcher.cpp
+++ b/src/common/kiwix/cluceneSearcher.cpp
@@ -70,50 +70,41 @@ std::string toString(const TCHAR* s){
     QueryParser* parser = new QueryParser(_T("content"), analyzer);
     STRCPY_AtoT(buffer, search.c_str(), MAX_BUFFER_SIZE);
     
-    //lucene::util::Misc::_cpycharToWide();
-    //::mbstowcs(buffer,search.c_str(),MAX_BUFFER_SIZE);
-    //search.c_str();
-    Query* query = parser->parse(_T("between"));
+    Query* query = parser->parse(buffer);
+    delete parser;
+    delete analyzer;
     
     cout << "Query: " << search << endl;
-    wcout << "Buffer: " << buffer;
-    cout << endl;
+    wcout << "Buffer: " << buffer << endl;
+    
+    if (query == NULL){
+      cout << "Hits length:0 (null query)" << endl;
+      return;
+    }
     
     const wchar_t* querystring = query->toString();
     wcout << L"Query2string: " << querystring << endl;
-    
-    //string q = toString(querystring);
-    //STRCPY_TtoA(querystring, buffer, MAX_BUFFER_SIZE);
-    //cout << "Query object: " << q << endl;
     delete[] querystring;
     
-    delete parser;
-    delete analyzer;
-
     // Search
     Hits* hits = searcher->search(query);
     cout << "Hits length:" << hits->length() << endl;
     
-    /*
-    for (size_t i=0; i < hits->length() && i<10; i++) {
+    for (int32_t i=0; i < hits->length() && i<10; i++) {
       Document* d = &hits->doc(i);
-      _tprintf(_T("#%d. %s (score: %f)\n"),
-	       i, d->get(_T("contents")),
+      _tprintf(_T("#%d. %s, url: %s (score: %f)\n"),
+	       i + 1, d->get(_T("title")), d->get(_T("url")),
 	       hits->score(i));
     }
-    */
-    /*
+/*
       Result result;
       result.url = doc.get_data();
       result.title = doc.get_value(0);
       result.score = i.get_percent();
       
       this->results.push_back(result);
-    */
+*/
   
-    searcher->close();
-    
-    delete searcher;
     delete hits;
     delete query;