Fixing bug 3465649. I still need to make CMPH_BRZ work for small sets.

configure.ac

@@ -31,7 +31,7 @@ AC_CHECK_HEADERS([getopt.h math.h])
 dnl Checks for libraries.
 LT_LIB_M  
 LDFLAGS="$LIBM $LDFLAGS"
-CFLAGS="-Wall"
+#CFLAGS="-Wall -g"
 
 AC_PROG_CXX
 CXXFLAGS="-Wall -Wno-unused-function -DNDEBUG -O3 -fomit-frame-pointer $CXXFLAGS"

examples/Makefile.am

@@ -1,4 +1,4 @@
-noinst_PROGRAMS = vector_adapter_ex1 file_adapter_ex2 struct_vector_adapter_ex3
+noinst_PROGRAMS = vector_adapter_ex1 file_adapter_ex2 struct_vector_adapter_ex3 small_set_test
 
 INCLUDES = -I../src/
 
@@ -10,3 +10,6 @@ file_adapter_ex2_SOURCES   = file_adapter_ex2.c
 
 struct_vector_adapter_ex3_LDADD     = ../src/libcmph.la
 struct_vector_adapter_ex3_SOURCES   = struct_vector_adapter_ex3.c
+
+small_set_test_LDADD     = ../src/libcmph.la
+small_set_test_SOURCES   = small_set_test.c

examples/small_set_test.c

@@ -0,0 +1,76 @@
+#include <cmph.h>
+
+int test(cmph_uint32* items_to_hash, cmph_uint32 items_len, CMPH_ALGO alg_n)
+{
+    cmph_t *hash;
+    cmph_config_t *config;
+    cmph_io_adapter_t *source;
+    cmph_uint32 i;
+
+    printf("%s (%u)\n", cmph_names[alg_n], alg_n);
+
+    source = cmph_io_struct_vector_adapter(items_to_hash,
+                                           (cmph_uint32)sizeof(cmph_uint32),
+                                           0,
+                                           (cmph_uint32)sizeof(cmph_uint32),
+                                           items_len);
+    config = cmph_config_new(source);
+    cmph_config_set_algo(config, alg_n);
+    hash = cmph_new(config);
+    cmph_config_destroy(config);
+
+    printf("packed_size %u\n",cmph_packed_size(hash));
+
+    for (i=0; i<items_len; ++i)
+        printf("%d -> %u\n",
+               items_to_hash[i],
+               cmph_search(hash,
+                           (char*)(items_to_hash+i), 
+                           (cmph_uint32)sizeof(cmph_uint32)));
+    printf("\n");
+
+    cmph_io_vector_adapter_destroy(source);   
+    cmph_destroy(hash);	
+    return 0;
+}
+
+int main (void)
+{
+    cmph_uint32 vec1[] = {1,2,3,4,5};
+    cmph_uint32 vec1_len = 5;
+
+    cmph_uint32 vec2[] = {7576423, 7554496}; //CMPH_FCH, CMPH_BDZ, CMPH_BDZ_PH (4,5,6)
+    cmph_uint32 vec2_len = 2;
+    cmph_uint32 vec3[] = {2184764, 1882984, 1170551}; // CMPH_CHD_PH, CMPH_CHD (7,8)
+    cmph_uint32 vec3_len = 3;
+    cmph_uint32 i;
+
+    // Testing with vec1
+    cmph_uint32* values = (cmph_uint32*)vec1;
+    cmph_uint32 length = vec1_len;
+    printf("TESTING VECTOR WITH %u INTEGERS\n", length);
+    for (i = 0; i < CMPH_COUNT; i++)
+    {
+        test(values, length, i);
+    }
+
+    // Testing with vec2
+    values = (cmph_uint32*)vec2;
+    length = vec2_len;
+    printf("TESTING VECTOR WITH %u INTEGERS\n", length);
+    for (i = 0; i < CMPH_COUNT; i++)
+    {
+        test(values, length, i);
+    }
+
+    // Testing with vec2
+    values = (cmph_uint32*)vec3;
+    length = vec3_len;
+    printf("TESTING VECTOR WITH %u INTEGERS\n", length);
+    for (i = 0; i < CMPH_COUNT; i++)
+    {
+        test(values, length, i);
+    }
+
+    return 0;
+}

src/bdz.c

@@ -288,6 +288,11 @@ cmph_t *bdz_new(cmph_config_t *mph, double c)
 	bdz->m = mph->key_source->nkeys;	
 	bdz->r = (cmph_uint32)ceil((c * mph->key_source->nkeys)/3);
 	if ((bdz->r % 2) == 0) bdz->r+=1;
+
+    if (bdz->r == 1) { // workaround for small key sets
+        bdz->r = 3;
+    }
+
 	bdz->n = 3*bdz->r;
 
 	bdz->k = (1U << bdz->b);

src/bdz_ph.c

@@ -254,6 +254,11 @@ cmph_t *bdz_ph_new(cmph_config_t *mph, double c)
 	bdz_ph->m = mph->key_source->nkeys;
 	bdz_ph->r = (cmph_uint32)ceil((c * mph->key_source->nkeys)/3);
 	if ((bdz_ph->r % 2) == 0) bdz_ph->r += 1;
+
+    if (bdz_ph->r == 1) { // workaround for small key sets
+        bdz_ph->r = 3;
+    }
+
 	bdz_ph->n = 3*bdz_ph->r;