X-Git-Url: http://git.tuebingen.mpg.de/?p=adu.git;a=blobdiff_plain;f=bloom.c;h=ec12d512d942d17a7aa6c8edd55b73b2e5f5b60e;hp=2aff535e833f75ba178bfcbf0524e0bf19d0cca7;hb=3a51a868784e755c34f19a16970118936fbf299f;hpb=4074e7034641cbbbfc861961d216d58a503d0aee

diff --git a/bloom.c b/bloom.c
index 2aff535..ec12d51 100644
--- a/bloom.c
+++ b/bloom.c
@@ -1,8 +1,10 @@
-#include <stdio.h>
-#include <stdlib.h>
-#include <inttypes.h>
-#include <string.h>
-#include <assert.h>
+/*
+ * Copyright (C) 2008 Andre Noll <maan@systemlinux.org>
+ *
+ * Licensed under the GPL v2. For licencing details see COPYING.
+ */
+
+/** \file bloom.c Simple bloom filter implementation. */
 
 #include "adu.h"
 #include "string.h"
@@ -32,13 +34,14 @@ static inline uint64_t filter_bits(struct bloom *b)
  * http://www.azillionmonkeys.com/qed/hash.html
  */
 
+/** \cond */
 #if (defined(__GNUC__) && defined(__i386__))
 #define get16bits(d) (*((const uint16_t *) (d)))
 #else
 #define get16bits(d) ((((uint32_t)(((const uint8_t *)(d))[1])) << 8)\
                        +(uint32_t)(((const uint8_t *)(d))[0]) )
 #endif
-
+/** \endcond */
 
 static uint32_t super_fast_hash(const uint8_t *data, uint32_t len, uint32_t hash)
 {
@@ -96,7 +99,28 @@ static int test_and_set_bit(uint64_t bitnum, struct bloom *b)
 	return ret;
 }
 
-int bloom_test_and_insert(const uint8_t *data, size_t len, struct bloom *b)
+/**
+ * Insert data to the given bloom filter.
+ *
+ * This function computes \a k hashes from the given data where \a k is the
+ * number of hash functions of the filter \a b. Each hash value corresponds to
+ * a position in the bit array of the filter and each of these bits are being
+ * tested and set.  If not all \a k bits were already set, the given data was
+ * not yet contained in the filter and the function returns non-zero.
+ *
+ * Otherwise either (a) the same data has already been inserted previously or
+ * (b) a hash collision occurred or (c) the \a k bits are set due to previous
+ * insertion of other data (i.e. a false positive occurred). It is impossible
+ * to distinguish these cases.
+ *
+ * \param data The data to insert.
+ * \param len Number of bytes of \a data.
+ * \param b The filter to insert to.
+ *
+ * \return Zero if the entry was already contained in the filter (or in case of
+ * false positives), non-zero otherwise.
+ */
+int bloom_insert(const uint8_t *data, size_t len, struct bloom *b)
 {
 	int i, ret = 0;
 	uint32_t h = 0xb11924e1; /* some arbitrary value */
@@ -153,11 +177,11 @@ struct bloom *bloom_new(unsigned order, unsigned num_hash_functions)
 
 #ifdef TEST_BLOOM
 
-int bloom_test_and_insert_string(const char *str, struct bloom *b)
+int bloom_insert_string(const char *str, struct bloom *b)
 {
 	uint32_t len = strlen(str);
 
-	return bloom_test_and_insert((const uint8_t *)str, len, b);
+	return bloom_insert((const uint8_t *)str, len, b);
 }
 
 void add_stdin(struct bloom *b)
@@ -171,7 +195,7 @@ void add_stdin(struct bloom *b)
 			continue;
 		if (buf[len - 1] == '\n')
 			buf[len - 1] = '\0';
-		if (bloom_test_and_insert_string(buf, b))
+		if (bloom_insert_string(buf, b))
 			false_positives++;
 	}
 	INFO_LOG("filter contains %llu entries\n", b->num_entries);