void testIndexData()
{
/// Create Schema
Schema *schema = Schema::create(srch2::instantsearch::DefaultIndex);
schema->setPrimaryKey("article_id"); // integer, not searchable
schema->setSearchableAttribute("article_id"); // convert id to searchable text
schema->setSearchableAttribute("article_authors", 2); // searchable text
schema->setSearchableAttribute("article_title", 7); // searchable text
/// Create Analyzer
SynonymContainer *syn = SynonymContainer::getInstance("", SYNONYM_DONOT_KEEP_ORIGIN);
syn->init();
Analyzer *analyzer = new Analyzer(NULL, NULL, NULL, syn, "");
/// Create IndexData
string INDEX_DIR = ".";
IndexData *indexData = IndexData::create(INDEX_DIR,
analyzer,
schema,
srch2::instantsearch::DISABLE_STEMMER_NORMALIZER);
Record *record = new Record(schema);
record->setPrimaryKey(1001);
record->setSearchableAttributeValue("article_authors", "Tom Smith and Jack Lennon");
record->setSearchableAttributeValue("article_title", "come Yesterday Once More");
record->setRecordBoost(10);
indexData->_addRecord(record, analyzer);
record->clear();
record->setPrimaryKey(1008);
record->setSearchableAttributeValue(0, "Jimi Hendrix");
record->setSearchableAttributeValue(1, "Little wing");
record->setRecordBoost(90);
indexData->_addRecord(record, analyzer);
indexData->finishBulkLoad();
//index->print_Index();
record->clear();
record->setPrimaryKey(1007);
record->setSearchableAttributeValue(0, "Jimaai Hendaarix");
record->setSearchableAttributeValue(1, "Littaale waaing");
record->setRecordBoost(90);
indexData->_addRecord(record, analyzer);
//index->print_Index();
/// test Trie
Trie_Internal *trie = indexData->trie;
typedef boost::shared_ptr<TrieRootNodeAndFreeList > TrieRootNodeSharedPtr;
TrieRootNodeSharedPtr rootSharedPtr;
trie->getTrieRootNode_ReadView(rootSharedPtr);
TrieNode *root = rootSharedPtr->root;
(void)(root);
ASSERT( trie->getTrieNodeFromUtf8String( root, "and")->getId() < trie->getTrieNodeFromUtf8String( root, "come")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "come")->getId() < trie->getTrieNodeFromUtf8String( root, "hendrix")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "hendrix")->getId() < trie->getTrieNodeFromUtf8String( root, "jack")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "jack")->getId() < trie->getTrieNodeFromUtf8String( root, "jimi")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "jimi")->getId() < trie->getTrieNodeFromUtf8String( root, "lennon")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "lennon")->getId() < trie->getTrieNodeFromUtf8String( root, "little")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "little")->getId() < trie->getTrieNodeFromUtf8String( root, "more")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "more")->getId() < trie->getTrieNodeFromUtf8String( root, "once")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "once")->getId() < trie->getTrieNodeFromUtf8String( root, "smith")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "smith")->getId() < trie->getTrieNodeFromUtf8String( root, "tom")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "tom")->getId() < trie->getTrieNodeFromUtf8String( root, "wing")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "wing")->getId() < trie->getTrieNodeFromUtf8String( root, "yesterday")->getId() );
// we assume that there is no background thread does merge,
// or even if there is such a background thread, it didn't have a chance to do the merge
ASSERT( trie->getTrieNodeFromUtf8String( root, "jimaai") == NULL );
ASSERT( trie->getTrieNodeFromUtf8String( root, "Hendaarix") == NULL );
ASSERT( trie->getTrieNodeFromUtf8String( root, "Littaale") == NULL );
ASSERT( trie->getTrieNodeFromUtf8String( root, "waaing") == NULL );
ASSERT( trie->getTrieNodeFromUtf8String( root, "j")->getMinId() == trie->getTrieNodeFromUtf8String( root, "jack")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "j")->getMaxId() == trie->getTrieNodeFromUtf8String( root, "jimi")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "ja")->getMinId() == trie->getTrieNodeFromUtf8String( root, "jack")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "ja")->getMaxId() == trie->getTrieNodeFromUtf8String( root, "jack")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "win")->getMinId() == trie->getTrieNodeFromUtf8String( root, "wing")->getId() );
ASSERT( trie->getTrieNodeFromUtf8String( root, "win")->getMaxId() == trie->getTrieNodeFromUtf8String( root, "wing")->getId() );
/// test ForwardIndex
ForwardIndex *forwardIndex = indexData->forwardIndex;
shared_ptr<vectorview<ForwardListPtr> > forwardListDirectoryReadView;
forwardIndex->getForwardListDirectory_ReadView(forwardListDirectoryReadView);
float score = 0;
unsigned keywordId = 1;
// define the attributeBitmap only in debug mode
#if ASSERT_LEVEL > 0
vector<unsigned> attributeBitmap;
#endif
ASSERT( forwardIndex->haveWordInRange(forwardListDirectoryReadView, 0,
trie->getTrieNodeFromUtf8String( root, "jack")->getId(),
trie->getTrieNodeFromUtf8String( root, "lennon")->getId(),
vector<unsigned>(), ATTRIBUTES_OP_AND,
keywordId, attributeBitmap, score) == true );
ASSERT( forwardIndex->haveWordInRange(forwardListDirectoryReadView, 0,
trie->getTrieNodeFromUtf8String( root, "smith")->getId() + 1,
trie->getTrieNodeFromUtf8String( root, "tom")->getId() - 1,
vector<unsigned>(), ATTRIBUTES_OP_AND,
keywordId, attributeBitmap, score) == false );
ASSERT( forwardIndex->haveWordInRange(forwardListDirectoryReadView, 1,
trie->getTrieNodeFromUtf8String( root, "hendrix")->getId(),
trie->getTrieNodeFromUtf8String( root, "jimi")->getId(),
vector<unsigned>(), ATTRIBUTES_OP_AND,
keywordId, attributeBitmap, score) == true );
ASSERT( forwardIndex->haveWordInRange(forwardListDirectoryReadView, 1,
trie->getTrieNodeFromUtf8String( root, "wing")->getId() + 1,
trie->getTrieNodeFromUtf8String( root, "wing")->getId() + 2,
vector<unsigned>(), ATTRIBUTES_OP_AND,
keywordId, attributeBitmap, score) == false );
/// test InvertedIndex
InvertedIndex *invertedIndex = indexData->invertedIndex;
(void)(forwardIndex);
(void)(invertedIndex);
(void)score;
(void)keywordId;
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "and")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "come")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "hendrix")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "jack")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "jimi")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "lennon")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "little")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "more")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "once")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "smith")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "tom")->getInvertedListOffset() ) == 1);
ASSERT(invertedIndex->getInvertedListSize_ReadView( trie->getTrieNodeFromUtf8String( root, "wing")->getInvertedListOffset() ) == 1);
delete schema;
delete record;
delete analyzer;
delete indexData;
syn->free();
}
