// validate block size control for 512 and max count of 1
void test_block_size_512_and_count_1() {
// clean up from any previous run
rm_hashdb_dir(temp_dir1);
rm_hashdb_dir(temp_dir2);
remove(temp_dfxml_file);
// create new hashdb
hashdb_settings_t settings;
settings.hash_block_size = 512;
settings.maximum_hash_duplicates = 1;
commands_t::create(settings, temp_dir1);
commands_t::create(settings, temp_dir2);
// import
commands_t::import(temp_dir1, sample_dfxml512, "test_repository_name");
commands_t::import(temp_dir1, sample_dfxml512, "test_repository_name2");
BOOST_TEST_EQ(map_size(temp_dir1), 24);
// export
commands_t::do_export(temp_dir1, "temp_dfxml_out.xml");
// import
commands_t::import(temp_dir2, "temp_dfxml_out.xml", "test_repository_name");
BOOST_TEST_EQ(map_size(temp_dir1), 24);
}
// import/export: import1, export, import2 should retain size of 74
void test_import_export() {
hashdb_settings_t settings;
rm_hashdb_dir(temp_dir1);
rm_hashdb_dir(temp_dir2);
remove(temp_dfxml_file);
// import
commands_t::create(settings, temp_dir1);
commands_t::import(temp_dir1, sample_dfxml4096, "repository1");
commands_t::do_export(temp_dir1, "temp_dfxml_out.xml");
commands_t::create(settings, temp_dir2);
commands_t::import(temp_dir2, temp_dfxml_file, "repository2");
BOOST_TEST_EQ(map_size(temp_dir2), 74);
}
void run_setup() {
rm_hashdb_dir(temp_dir);
make_dir_if_not_there(temp_dir);
source_lookup_store_t store(temp_store, RW_NEW);
store.insert_value(std::pair<uint64_t,uint64_t>(2,3));
}
// validate use of multiple repository names
void test_multiple_repository_names() {
// clean up from any previous run
rm_hashdb_dir(temp_dir);
// create new hashdb
hashdb_settings_t settings;
commands_t::create(settings, temp_dir);
// test ability to manage multiple repository names
for (int i=0; i<12; i++) {
// generate unique repository name
std::ostringstream ss;
ss << "test_repository_name_" << i;
// import
commands_t::import(temp_dir, sample_dfxml4096, ss.str());
}
// duplicate import should not add elements
commands_t::import(temp_dir, sample_dfxml4096, "test_repository_name_0");
// validate correct size
BOOST_TEST_EQ(map_size(temp_dir), 74*12);
}
void rw_new() {
// clean up from any previous run
rm_hashdb_dir(temp_dir);
make_dir_if_not_there(temp_dir);
// create each db
lmdb_hash_store_t hash_store(temp_dir, RW_NEW, 512, 0);
lmdb_name_store_t name_store(temp_dir, RW_NEW);
lmdb_source_store_t source_store(temp_dir, RW_NEW);
}
int cpp_main(int argc, char* argv[]) {
rm_hashdb_dir(temp_dir);
make_dir_if_not_there(temp_dir);
run_setup();
run_test();
// done
int status = boost::report_errors();
return status;
}
// tuning
void test_tuning() {
// setup
hashdb_settings_t settings;
rm_hashdb_dir(temp_dir1);
commands_t::create(settings, temp_dir1);
commands_t::import(temp_dir1, sample_dfxml4096, "repository1");
// rebuild_bloom
settings.bloom1_M_hash_size = 20;
commands_t::rebuild_bloom(settings, temp_dir1);
}
void write_settings() {
// clean up from any previous run
rm_hashdb_dir(temp_dir);
// create the hashdb directory
hashdb_directory_manager_t::create_new_hashdb_dir(temp_dir);
// create working settings
hashdb_settings_t settings;
settings.maximum_hash_duplicates = 4;
hashdb_settings_store_t::write_settings(temp_dir, settings);
}
// validate block size control for 4096
void test_block_size_4096() {
// clean up from any previous run
rm_hashdb_dir(temp_dir1);
rm_hashdb_dir(temp_dir2);
remove(temp_dfxml_file);
// create new hashdb
hashdb_settings_t settings;
commands_t::create(settings, temp_dir1);
commands_t::create(settings, temp_dir2);
// import
commands_t::import(temp_dir1, sample_dfxml4096, "test_repository_name");
BOOST_TEST_EQ(map_size(temp_dir1), 74);
// export
commands_t::do_export(temp_dir1, "temp_dfxml_out.xml");
// import
commands_t::import(temp_dir2, "temp_dfxml_out.xml", "test_repository_name");
BOOST_TEST_EQ(map_size(temp_dir2), 74);
}
// scan services
void test_scan_services() {
// setup
hashdb_settings_t settings;
rm_hashdb_dir(temp_dir1);
commands_t::create(settings, temp_dir1);
commands_t::import(temp_dir1, sample_dfxml4096, "repository1");
// scan
commands_t::scan(temp_dir1, sample_dfxml4096);
hash_t k1;
to_key(1, k1);
commands_t::scan_hash(temp_dir1, k1.hexdigest());
}
// test check for valid hash block size
void test_block_size_0() {
// clean up from any previous run
rm_hashdb_dir(temp_dir1);
// create new hashdb
hashdb_settings_t settings;
settings.hash_block_size = 0;
commands_t::create(settings, temp_dir1);
// import
commands_t::import(temp_dir1, sample_dfxml4096, "test_repository_name");
// with hash_block_size=0, total should be every hash, including remainder
BOOST_TEST_EQ(map_size(temp_dir1), 75);
}
// performance analysis
void test_performance_analysis() {
// setup
hashdb_settings_t settings;
rm_hashdb_dir(temp_dir1);
commands_t::create(settings, temp_dir1);
// add_random
// not tested because it requires user input "q" to complete.
// commands_t::add_random("repo_random", temp_dir1, "100000");
// scan_random
// not tested because it takes time to run
// commands_t::scan_random(temp_dir1);
}
// statistics
void test_statistics() {
// setup
hashdb_settings_t settings;
rm_hashdb_dir(temp_dir1);
commands_t::create(settings, temp_dir1);
commands_t::import(temp_dir1, sample_dfxml4096, "repository1");
// size
commands_t::size(temp_dir1);
// sources
commands_t::sources(temp_dir1);
// histogram
commands_t::histogram(temp_dir1);
// duplicates
commands_t::duplicates(temp_dir1, "2");
// hash_table
commands_t::hash_table(temp_dir1, "repository1", sample_dfxml4096);
commands_t::hash_table(temp_dir1, "invalid_repository", sample_dfxml4096);
}
// database manipulation commands
void test_database_manipulation() {
hashdb_settings_t settings;
rm_hashdb_dir(temp_dir1);
rm_hashdb_dir(temp_dir2);
rm_hashdb_dir(temp_dir3);
rm_hashdb_dir(temp_dir4);
rm_hashdb_dir(temp_dir5);
rm_hashdb_dir(temp_dir6);
rm_hashdb_dir(temp_dir7);
rm_hashdb_dir(temp_dir8);
commands_t::create(settings, temp_dir1);
commands_t::import(temp_dir1, sample_dfxml4096, "repository1");
commands_t::create(settings, temp_dir2);
commands_t::import(temp_dir2, sample_dfxml4096, "repository2");
commands_t::add_multiple(temp_dir1, temp_dir2, temp_dir3);
// add
commands_t::add(temp_dir1, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 74);
commands_t::add(temp_dir1, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 74);
commands_t::add(temp_dir2, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 2*74);
rm_hashdb_dir(temp_dir4);
// add_multiple
std::cout << "add_multiple\n";
commands_t::add_multiple(temp_dir1, temp_dir2, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir3), 2*74);
commands_t::add_multiple(temp_dir1, temp_dir2, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir3), 2*74);
rm_hashdb_dir(temp_dir4);
// add_repository
std::cout << "add_repository\n";
commands_t::add_repository(temp_dir1, temp_dir4, "invalid_repository");
BOOST_TEST_EQ(map_size(temp_dir4), 0);
commands_t::add_repository(temp_dir1, temp_dir4, "repository1");
BOOST_TEST_EQ(map_size(temp_dir4), 74);
commands_t::add_repository(temp_dir1, temp_dir4, "repository1");
BOOST_TEST_EQ(map_size(temp_dir4), 74);
commands_t::add_repository(temp_dir2, temp_dir4, "repository1");
BOOST_TEST_EQ(map_size(temp_dir4), 74);
commands_t::add_repository(temp_dir2, temp_dir4, "repository2");
BOOST_TEST_EQ(map_size(temp_dir4), 2*74);
rm_hashdb_dir(temp_dir4);
// intersect
std::cout << "intersect\n";
commands_t::intersect(temp_dir1, temp_dir2, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 0);
commands_t::intersect(temp_dir1, temp_dir3, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 74);
rm_hashdb_dir(temp_dir4);
// intersect hash
std::cout << "intersect_hash\n";
commands_t::intersect_hash(temp_dir1, temp_dir2, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 2*74);
commands_t::intersect_hash(temp_dir1, temp_dir2, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 2*74);
rm_hashdb_dir(temp_dir4);
// subtract
std::cout << "subtract\n";
commands_t::subtract(temp_dir3, temp_dir1, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 74);
commands_t::subtract(temp_dir3, temp_dir1, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 74);
commands_t::add(temp_dir2, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 74);
commands_t::add(temp_dir1, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 2*74);
commands_t::subtract(temp_dir3, temp_dir2, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 2*74);
commands_t::subtract(temp_dir3, temp_dir4, temp_dir5);
BOOST_TEST_EQ(map_size(temp_dir5), 0);
rm_hashdb_dir(temp_dir4);
rm_hashdb_dir(temp_dir5);
// subtract_hash
std::cout << "subtract_hash\n";
commands_t::subtract_hash(temp_dir3, temp_dir1, temp_dir4);
BOOST_TEST_EQ(map_size(temp_dir4), 0);
rm_hashdb_dir(temp_dir4);
// deduplicate
std::cout << "deduplicate1\n";
commands_t::create(settings, temp_dir6);
commands_t::import(temp_dir6, sample_dfxml4096, "repository1");
commands_t::deduplicate(temp_dir6, temp_dir7);
BOOST_TEST_EQ(map_size(temp_dir7), 74);
std::cout << "deduplicate2\n";
commands_t::import(temp_dir6, sample_dfxml4096, "repository2");
commands_t::deduplicate(temp_dir6, temp_dir8);
// check ability to open existing target
commands_t::deduplicate(temp_dir6, temp_dir8);
BOOST_TEST_EQ(map_size(temp_dir8), 0);
}