Merge branch 'master' of github.com:bonitao/cmph

configure.ac

@@ -1,7 +1,7 @@
 dnl Process this file with autoconf to produce a configure script.
 AC_INIT
 AC_CONFIG_SRCDIR([Makefile.am])
-AM_INIT_AUTOMAKE(cmph, 1.0)
+AM_INIT_AUTOMAKE(cmph, 2.0)
 AC_CONFIG_HEADERS([config.h])
 AC_CONFIG_MACRO_DIR([m4])
 
@@ -50,10 +50,12 @@ if test x$cxxmph = xtrue; then
   AC_SUBST([CXXMPH], "cxxmph")
 fi
 
-PKG_CHECK_MODULES([CHECK], [check >= 0.9.8]) 
+PKG_CHECK_MODULES([CHECK], [check >= 0.9.8], [HAVE_LIBCHECK=1], [HAVE_LIBCHECK=]) 
 if test "x$CHECK_LIBS" = "x" ; then
+  if test "x$cxxmph" = "xtrue"; then
     AC_MSG_WARN([Testing library not found, make check will fail.])
   fi
+fi
 AC_SUBST(CHECK_LIBS)
 AC_SUBST(CHECK_CFLAGS)
 

cxxmph/Makefile.am

@@ -4,14 +4,17 @@ noinst_PROGRAMS = bm_index bm_map
 bin_PROGRAMS = cxxmph
 
 cxxmph_includedir = $(includedir)/cxxmph/
-cxxmph_include_HEADERS = mph_map.h mph_index.h MurmurHash3.h trigraph.h seeded_hash.h stringpiece.h hollow_iterator.h
+cxxmph_include_HEADERS = mph_bits.h mph_map.h mph_index.h MurmurHash3.h trigraph.h seeded_hash.h stringpiece.h hollow_iterator.h string_util.h
 
-noinst_LTLIBRARIES = libcxxmph_test.la
+check_LTLIBRARIES = libcxxmph_test.la
+noinst_LTLIBRARIES = libcxxmph_bm.la
 lib_LTLIBRARIES = libcxxmph.la
-libcxxmph_la_SOURCES = MurmurHash3.h MurmurHash3.cpp trigragh.h trigraph.cc mph_bits.h mph_bits.cc mph_index.h mph_index.cc seeded_hash.h stringpiece.h benchmark.h benchmark.cc string_util.cc
+libcxxmph_la_SOURCES = MurmurHash3.cpp trigraph.cc mph_bits.cc mph_index.cc benchmark.h benchmark.cc string_util.cc
 libcxxmph_la_LDFLAGS = -version-info 0:0:0
 libcxxmph_test_la_SOURCES = test.h test.cc
 libcxxmph_test_la_LIBADD = libcxxmph.la
+libcxxmph_bm_la_SOURCES = benchmark.h benchmark.cc bm_common.h bm_common.cc
+libcxxmph_bm_la_LIBADD = libcxxmph.la
 
 test_test_SOURCES = test_test.cc
 test_test_LDADD = libcxxmph_test.la $(CHECK_LIBS)
@@ -22,14 +25,14 @@ mph_map_test_SOURCES = mph_map_test.cc
 mph_index_test_LDADD   = libcxxmph.la
 mph_index_test_SOURCES = mph_index_test.cc
 
-bm_index_LDADD = libcxxmph.la -lcmph
-bm_index_SOURCES = bm_common.cc bm_index.cc 
-
 trigraph_test_LDADD   = libcxxmph.la
 trigraph_test_SOURCES = trigraph_test.cc
 
-bm_map_LDADD = libcxxmph.la
+bm_index_LDADD = libcxxmph_bm.la -lcmph
-bm_map_SOURCES = bm_common.cc bm_map.cc
+bm_index_SOURCES = bm_index.cc 
+
+bm_map_LDADD = libcxxmph_bm.la
+bm_map_SOURCES = bm_map.cc
 
 cxxmph_LDADD   = libcxxmph.la
 cxxmph_SOURCES = cxxmph.cc

cxxmph/hollow_iterator.h

@@ -26,7 +26,7 @@ struct hollow_iterator_base
     : public std::iterator<std::forward_iterator_tag,
                            typename iterator::value_type> {
  public:
-  typedef hollow_iterator_base<iterator, is_empty>& self_type;
+  typedef hollow_iterator_base<iterator, is_empty> self_type;
   typedef self_type& self_reference;
   typedef typename iterator::reference reference;
   typedef typename iterator::pointer pointer;

cxxmph/map_tester.h

@@ -11,6 +11,7 @@
 
 namespace cxxmph {
 
+using namespace cxxmph;
 using namespace std;
 
 template <template<typename...> class map_type>
@@ -24,8 +25,9 @@ struct MapTester {
   static bool large_insert() {
     map_type<int64_t, int64_t> m;
     // Start counting from 1 to not touch default constructed value bugs
-    for (int i = 1; i < 12 * 256 * 256; ++i) m.insert(make_pair(i, i));
+    int nkeys = 12 * 256 * 256;
-    return m.size() == 12 * 256 * 256 - 1;
+    for (int i = 1; i < nkeys; ++i) m.insert(make_pair(i, i));
+    return static_cast<int>(m.size()) == nkeys - 0;
   }
   static bool small_search() {
     map_type<int64_t, int64_t> m;
@@ -54,7 +56,7 @@ struct MapTester {
     int nkeys = 10 * 1000;
     vector<string> keys;
     for (int i = 0; i < nkeys; ++i) {
-      keys.push_back(cxxmph::format("%v", i));
+      keys.push_back(format("%v", i));
     }
     map_type<string, int64_t> m;
     for (int i = 0; i < nkeys; ++i) m.insert(make_pair(keys[i], i));
@@ -65,6 +67,42 @@ struct MapTester {
     }
     return true;
   }
+  static bool rehash_zero() {
+    map_type<int64_t, int64_t> m;
+    m.rehash(0);
+    return m.size() == 0;
+  }
+  static bool rehash_size() {
+    map_type<int64_t, int64_t> m;
+    int nkeys = 10 * 1000;
+    for (int i = 0; i < nkeys; ++i) { m.insert(make_pair(i, i)); }
+    m.rehash(nkeys);
+    for (int i = 0; i < nkeys; ++i) { if (m.find(i) == m.end()) return false; }
+    for (int i = nkeys; i < nkeys * 2; ++i) {
+      if (m.find(i) != m.end()) return false;
+    }
+    return true;
+  }
+  static bool erase_iterator() {
+    map_type<int64_t, int64_t> m;
+    int nkeys = 10 * 1000;
+    for (int i = 0; i < nkeys; ++i) { m.insert(make_pair(i, i)); }
+    for (int i = nkeys; i >= 0; --i) { 
+      m.erase(m.find(i));
+      if (static_cast<int>(m.size()) != i) return false;
+    }
+    return true;
+  }
+ static bool erase_value() {
+    map_type<int64_t, int64_t> m;
+    int nkeys = 10 * 1000;
+    for (int i = 0; i < nkeys; ++i) { m.insert(make_pair(i, i)); }
+    for (int i = nkeys; i >= 0; --i) { 
+      m.erase(i);
+      if (static_cast<int>(m.size()) != i) return false;
+    }
+    return true;
+  }
 };
 
 }  // namespace cxxxmph

cxxmph/map_tester_test.cc

@@ -11,3 +11,7 @@ CXXMPH_CXX_TEST_CASE(small_search, Tester::small_search);
 CXXMPH_CXX_TEST_CASE(default_search, Tester::default_search);
 CXXMPH_CXX_TEST_CASE(large_search, Tester::large_search);
 CXXMPH_CXX_TEST_CASE(string_search, Tester::string_search);
+CXXMPH_CXX_TEST_CASE(rehash_zero, Tester::rehash_zero);
+CXXMPH_CXX_TEST_CASE(rehash_size, Tester::rehash_size);
+CXXMPH_CXX_TEST_CASE(erase_value, Tester::erase_value);
+CXXMPH_CXX_TEST_CASE(erase_iterator, Tester::erase_iterator);

cxxmph/mph_map_test.cc

@@ -17,58 +17,7 @@ CXXMPH_CXX_TEST_CASE(small_search, Tester::small_search);
 CXXMPH_CXX_TEST_CASE(default_search, Tester::default_search);
 CXXMPH_CXX_TEST_CASE(large_search, Tester::large_search);
 CXXMPH_CXX_TEST_CASE(string_search, Tester::string_search);
-
+CXXMPH_CXX_TEST_CASE(rehash_zero, Tester::rehash_zero);
-/*
+CXXMPH_CXX_TEST_CASE(rehash_size, Tester::rehash_size);
-using std::make_pair;
+CXXMPH_CXX_TEST_CASE(erase_value, Tester::erase_value);
-using std::string;
+CXXMPH_CXX_TEST_CASE(erase_iterator, Tester::erase_iterator);
-using cxxmph::mph_map;
-
-int main(int argc, char** argv) {
-  mph_map<int64_t, int64_t> b;
-  int32_t num_keys = 1000*10;
-  for (int i = 0; i < num_keys; ++i) {
-    b.insert(make_pair(i, i));
-  }
-  b.rehash(b.size());
-  for (int i = 0; i < 1000000; ++i) {
-    auto it = b.find(i % num_keys);
-    if (it == b.end()) {
-      std::cerr << "Failed to find " << i << std::endl;
-      exit(-1);
-    }
-    if (it->first != it->second || it->first != i % num_keys) {
-      std::cerr << "Found " << it->first << " looking for " << i << std::endl;
-      exit(-1);
-    }
-  }
-  mph_map<string, int> h;
-  h.insert(std::make_pair("-1",-1));
-  mph_map<string, int>::const_iterator it;
-  for (it = h.begin(); it != h.end(); ++it) {
-    if (it->second != -1) exit(-1);
-  }
-  int32_t num_valid = 100;
-  for (int i = 0; i < num_valid; ++i) {
-     char buf[10];    
-     snprintf(buf, 10, "%d", i);
-     h.insert(std::make_pair(buf, i));
-  }
-  for (int j = 0; j < 100; ++j) {
-    for (int i = 1000; i > 0; --i) {
-       char buf[10];    
-       snprintf(buf, 10, "%d", i - 1);
-       auto it = h.find(buf);
-       if (i < num_valid && it->second != i - 1) exit(-1);
-    }
-  }
-  for (int j = 0; j < 100; ++j) {
-    for (int i = 1000; i > 0; --i) {
-       char buf[10];    
-       int key = i*100 - 1;
-       snprintf(buf, 10, "%d", key);
-       auto it = h.find(buf);
-       if (key < num_valid && it->second != key) exit(-1);
-    }
-  }
-}
-*/

src/Makefile.am

@@ -3,7 +3,7 @@ noinst_PROGRAMS = bm_numbers
 lib_LTLIBRARIES = libcmph.la
 include_HEADERS = cmph.h cmph_types.h cmph_time.h chd_ph.h
 libcmph_la_SOURCES =  hash.h hash.c \
-		      jenkins_hash.h jenkins_hash.c MurmurHash2.h\
+		      jenkins_hash.h jenkins_hash.c \
 		      hash_state.h debug.h \
 		      vstack.h vstack.c vqueue.h vqueue.c\
 		      graph.h graph.c bitbool.h \