00001 #include <boost/test/unit_test.hpp>
00002 #include <boost/filesystem/fstream.hpp>
00003 #include "MPHashFunction.hpp"
00004 #include "Vocabulary.hpp"
00005 #include "Converter.hpp"
00006 #include <vector>
00007
00008 using namespace aitools::invertedindex;
00009
00010 BOOST_AUTO_TEST_SUITE(mp_hash_function);
00011
00012 BOOST_AUTO_TEST_CASE(default_ctor)
00013 {
00014 MPHashFunction mphf;
00015 BOOST_CHECK_EQUAL(mphf.size(), 0);
00016 BOOST_CHECK_EQUAL(mphf.hash("key"), 0);
00017 }
00018
00019 BOOST_AUTO_TEST_CASE(create_a)
00020 {
00021 bfs::path vocabulary("aitools.invertedindex.vocabulary.txt");
00022 bfs::ofstream ofs(vocabulary);
00023 unsigned key_count(100);
00024 for (unsigned i(0); i != key_count; ++i)
00025 {
00026 ofs << "key" << i << '\n';
00027 }
00028 ofs.close();
00029
00030 uint32_t hash;
00031 MPHashFunction mphf;
00032 mphf.create(vocabulary, MPHashFunction::BDZ);
00033 BOOST_CHECK_EQUAL(mphf.size(), key_count);
00034 std::vector<bool> slots(key_count, true);
00035 for (unsigned i(0); i != key_count; ++i)
00036 {
00037 hash = mphf.hash("key" + Converter::ui32_to_str(i));
00038 BOOST_CHECK(slots.at(hash));
00039 slots.at(hash) = false;
00040 }
00041 bfs::remove(vocabulary);
00042 }
00043
00044 BOOST_AUTO_TEST_CASE(create_b)
00045 {
00046 Vocabulary vocabulary;
00047 unsigned key_count(100);
00048 for (unsigned i(0); i != key_count; ++i)
00049 {
00050 vocabulary.insert("key" + Converter::ui32_to_str(i));
00051 }
00052
00053 uint32_t hash;
00054 MPHashFunction mphf;
00055 mphf.create(vocabulary, MPHashFunction::BDZ);
00056 BOOST_CHECK_EQUAL(mphf.size(), key_count);
00057 std::vector<bool> slots(key_count, true);
00058 for (unsigned i(0); i != key_count; ++i)
00059 {
00060 hash = mphf.hash("key" + Converter::ui32_to_str(i));
00061 BOOST_CHECK(slots.at(hash));
00062 slots.at(hash) = false;
00063 }
00064 }
00065
00066 BOOST_AUTO_TEST_CASE(save_load)
00067 {
00068 Vocabulary vocabulary;
00069 unsigned key_count(100);
00070 for (unsigned i(0); i != key_count; ++i)
00071 {
00072 vocabulary.insert("key" + Converter::ui32_to_str(i));
00073 }
00074
00075 uint32_t hash;
00076 MPHashFunction mphf;
00077 mphf.create(vocabulary, MPHashFunction::BDZ);
00078
00079 bfs::path invalid_path("/directory/does/not/exist");
00080 BOOST_CHECK_THROW(mphf.save(invalid_path), std::invalid_argument);
00081 bfs::path path("aitools.invertedindex.mphf");
00082 mphf.save(path);
00083
00084 mphf = MPHashFunction();
00085 BOOST_CHECK_THROW(mphf.load(invalid_path), std::invalid_argument);
00086 mphf.load(path);
00087 BOOST_CHECK_EQUAL(mphf.size(), key_count);
00088 std::vector<bool> slots(key_count, true);
00089 for (unsigned i(0); i != key_count; ++i)
00090 {
00091 hash = mphf.hash("key" + Converter::ui32_to_str(i));
00092 BOOST_CHECK(slots.at(hash));
00093 slots.at(hash) = false;
00094 }
00095 bfs::remove(path);
00096 }
00097
00098 BOOST_AUTO_TEST_SUITE_END();
