add a new test-add method that doesn't write unless it has to

axiak · Sep 23, 2015 · 39fac58 · 39fac58
1 parent 458e554
commit 39fac58
Show file tree

Hide file tree

Showing 5 changed files with 3,819 additions and 2,125 deletions.
diff --git a/src/bloomfilter.h b/src/bloomfilter.h
@@ -93,6 +93,23 @@ static inline int bloomfilter_Test(BloomFilter * bf, Key * key)
 __attribute__((always_inline))
 
 
+static inline int bloomfilter_Test_Add(BloomFilter * bf, Key * key)
+{
+    BTYPE (*hashfunc) (uint32_t, Key *) = key->shash == NULL ? _hash_long : _hash_char;
+    BTYPE mod = bf->array->bits;
+    int i;
+    int result = 1;
+    BTYPE hash_res;
 
+    for (i = bf->num_hashes - 1; i >= 0; --i) {
+        hash_res = (*hashfunc)(bf->hash_seeds[i], key) % mod;
+        result &= mbarray_Test_Set(bf->array, hash_res);
+    }
+    if (!result && bf->count_correct) {
+        bf->elem_count ++;
+    }
+    return result;
+}
+__attribute__((always_inline))
 
 #endif
diff --git a/src/cbloomfilter.pxd b/src/cbloomfilter.pxd
@@ -41,13 +41,14 @@ cdef extern from "bloomfilter.h":
                                       double error_rate,
                                       char * fname, long num_bits,
                                       int oflags, int perms,
-                                      int * hash_seeds, int num_hashes)
+                                      int * hash_seeds, int num_hashes) nogil
      BloomFilter * bloomfilter_Create_Malloc(long max_num_elem,
                                       double error_rate,
                                       long num_bits,
-                                      int * hash_seeds, int num_hashes)
-     void bloomfilter_Destroy(BloomFilter * bf)
-     int bloomfilter_Add(BloomFilter * bf, Key * key)
-     int bloomfilter_Test(BloomFilter * bf, Key * key)
-     int bloomfilter_Update(BloomFilter * bf, char * data, int size)
-     BloomFilter * bloomfilter_Copy_Template(BloomFilter * src, char * filename, int perms)
+                                      int * hash_seeds, int num_hashes) nogil
+     void bloomfilter_Destroy(BloomFilter * bf) nogil
+     int bloomfilter_Add(BloomFilter * bf, Key * key) nogil
+     int bloomfilter_Test(BloomFilter * bf, Key * key) nogil
+     int bloomfilter_Test_Add(BloomFilter * bf, Key * key) nogil
+     int bloomfilter_Update(BloomFilter * bf, char * data, int size) nogil
+     BloomFilter * bloomfilter_Copy_Template(BloomFilter * src, char * filename, int perms) nogil
diff --git a/src/mmapbitarray.h b/src/mmapbitarray.h
@@ -115,4 +115,21 @@ static inline int mbarray_Test(MBArray * array, BTYPE bit)
 __attribute__((always_inline))
 
 
+static inline int mbarray_Test_Set(MBArray * array, BTYPE bit)
+{
+    if (bit > array->bits) {
+        errno = EINVAL;
+        return -1;
+    }
+    DTYPE *chunk = array->vector + (size_t)(array->preamblesize + bit / (sizeof(DTYPE) << 3));
+    int byte = 1 << (bit % (sizeof(DTYPE) << 3));
+    int result = (*chunk & byte) != 0;
+    if (!result) {
+        *chunk |= byte;
+    }
+    return result;
+}
+__attribute__((always_inline))
+
+
 #endif