use gflags in bloom_filter

Author: Leowner

src/flags.cc

@@ -121,6 +121,9 @@ DEFINE_int32(snapshot_pool_size, 1, "the size of tablet thread pool for making s
 
 DEFINE_uint32(load_index_max_wait_time, 120 * 60 * 1000, "config the max wait time of load index");
 
+DEFINE_uint32(bloom_filter_bitset_size, 10000, "config the size of bitset in bloom filter");
+DEFINE_uint32(bloom_filter_hash_seed, 7, "config the count of hash seed in bloom filter, max 7");
+
 DEFINE_string(recycle_bin_root_path, "/tmp/recycle", "specify the root path of recycle bin");
 DEFINE_string(recycle_bin_ssd_root_path, "", "specify the root path of recycle bin in ssd");
 DEFINE_string(recycle_bin_hdd_root_path, "", "specify the root path of recycle bin in hdd");

src/storage/disk_table.cc

@@ -31,6 +31,9 @@ DECLARE_uint32(write_buffer_mb);
 DECLARE_uint32(block_cache_shardbits);
 DECLARE_bool(verify_compression);
 
+DECLARE_uint32(bloom_filter_bitset_size);
+DECLARE_uint32(bloom_filter_hash_seed);
+
 namespace openmldb {
 namespace storage {
 
@@ -154,7 +157,7 @@ bool DiskTable::InitColumnFamilyDescriptor() {
     auto inner_indexs = table_index_.GetAllInnerIndex();
     for (uint32_t i = 0; i < inner_indexs->size(); i++) {
         pk_cnt_vec_.push_back(std::make_shared<std::atomic<uint64_t>>(0));
-        bloom_filter_vec_.push_back(BloomFilter(1000, 100));
+        bloom_filter_vec_.push_back(BloomFilter());
     }
     auto indexs = table_index_.GetAllIndex();
     for (uint32_t i = 0; i < indexs.size(); i++) {
@@ -1359,15 +1362,15 @@ bool BloomFilter::getBit(uint32_t bit) {
 }
 
 void BloomFilter::Set(const char* str) {
-    for (uint32_t i = 0; i < k_; ++i) {
-        uint32_t p = hash(str, base_[i]) % bitset_size_;
+    for (uint32_t i = 0; i < FLAGS_bloom_filter_hash_seed; ++i) {
+        uint32_t p = hash(str, base_[i]) % FLAGS_bloom_filter_bitset_size;
         setBit(p);
     }
 }
 
 bool BloomFilter::Valid(const char* str) {
-    for (uint32_t i = 0; i < k_; ++i) {
-        uint32_t p = hash(str, base_[i]) % bitset_size_;
+    for (uint32_t i = 0; i < FLAGS_bloom_filter_hash_seed; ++i) {
+        uint32_t p = hash(str, base_[i]) % FLAGS_bloom_filter_bitset_size;
         if (!getBit(p)) {
             return false;
         }

src/storage/disk_table.h

@@ -41,6 +41,9 @@
 #include "storage/table.h"
 #include "base/glog_wapper.h"  // NOLINT
 
+DECLARE_uint32(bloom_filter_bitset_size);
+DECLARE_uint32(bloom_filter_hash_seed);
+
 namespace openmldb {
 namespace storage {
 
@@ -129,9 +132,8 @@ class KeyTSComparator : public rocksdb::Comparator {
 
 class BloomFilter {
  public:
-    BloomFilter(uint32_t bitset_size, uint32_t string_cnt) : k_(5), bitset_size_(bitset_size), string_cnt_(string_cnt) {
-        bits_.reserve(10000);
-        for (uint32_t i = 0; i < 10000; i++) {
+    BloomFilter() {
+        for (uint32_t i = 0; i < FLAGS_bloom_filter_bitset_size; i++) {
             bits_.push_back(std::make_shared<std::atomic<uint64_t>>(0));
         }
     }
@@ -146,10 +148,8 @@ class BloomFilter {
     void setBit(uint32_t bit);
     bool getBit(uint32_t bit);
 
-    uint32_t k_, bitset_size_,
-        string_cnt_;
     std::vector<std::shared_ptr<std::atomic<uint64_t>>> bits_;
-    uint32_t base_[100] = {5, 7, 11, 13, 31, 37, 61};
+    uint32_t base_[7] = {5, 7, 11, 13, 31, 37, 61};
 };
 
 class KeyTsPrefixTransform : public rocksdb::SliceTransform {