leveldb源码分析13

// 开始构建新的filter block，TableBuilder在构造函数和Flush中调用  

void StartBlock(uint64_tblock_offset);  

// 添加key，TableBuilder每次向data block中加入key时调用  

void AddKey(const Slice&key);  

// 结束构建，TableBuilder在结束对table的构建时调用  

Slice Finish();    

const FilterPolicy* policy_; // filter类型，构造函数参数指定  

std::string keys_;          //Flattened key contents  

std::vector<size_t> start_; // 各key在keys_中的位置  

std::string result_;        // 当前计算出的filter data  

std::vector<uint32_t>filter_offsets_; // 各个filter在result_中的位置  

std::vector<Slice> tmp_keys_;// policy_->CreateFilter()参数   

uint64_t filter_index =(block_offset / kFilterBase);  

assert(filter_index >=filter_offsets_.size());  

while (filter_index >filter_offsets_.size()) GenerateFilter();  

//S1 如果filter中key个数为0，则直接压入result_.size()并返回  

  const size_t num_keys =start_.size();  

  if (num_keys == 0) { // there are no keys for this filter  

    filter_offsets_.push_back(result_.size()); //result_.size()应该是0  

    return;  

  }  

//S2 从key创建临时key list，根据key的序列字符串kyes_和各key在keys_
//中的开始位置start_依次提取出key。  

  start_.push_back(keys_.size());  // Simplify lengthcomputation  

  tmp_keys_.resize(num_keys);  

  for (size_t i = 0; i <num_keys; i++) {  

    const char* base =keys_.data() + start_[i]; // 开始指针  

    size_t length = start_[i+1] -start_[i]; // 长度  

    tmp_keys_[i] = Slice(base,length);  

  }  

//S3 为当前的key集合生产filter，并append到result_  

filter_offsets_.push_back(result_.size());  

policy_->CreateFilter(&tmp_keys_[0], num_keys, &result_);  

//S4 清空，重置状态  

tmp_keys_.clear();  

keys_.clear();  

start_.clear();  

Slice k = key;  

start_.push_back(keys_.size());  

keys_.append(k.data(),k.size());  

//S1 如果start_数字不空，把为的key列表生产filter  

  if (!start_.empty()) GenerateFilter();  

//S2 从0开始顺序存储各filter的偏移值，见filter block data的数据格式。  

  const uint32_t array_offset =result_.size();  

  for (size_t i = 0; i < filter_offsets_.size();i++) {  

    PutFixed32(&result_,filter_offsets_[i]);  

  }  

//S3 最后是filter个数，和shift常量（11），并返回结果  

  PutFixed32(&result_,array_offset);  

  result_.push_back(kFilterBaseLg);  // Save encoding parameter in result  

  return Slice(result_);  

const FilterPolicy* policy_; // filter策略  

const char* data_;           // filter data指针 (at block-start)  

const char* offset_;        // offset array的开始地址 (at block-end)  

size_t num_;                // offsetarray元素个数  

size_t base_lg_;            // 还记得kFilterBaseLg吗  

bool KeyMayMatch(uint64_t block_offset, const Slice& key);

FilterBlockReader::FilterBlockReader(const FilterPolicy* policy, 
                                     constSlice& contents)  

    : policy_(policy),data_(NULL), offset_(NULL), num_(0), base_lg_(0) {  

  size_t n = contents.size();  

  if (n < 5) return;  // 1 byte forbase_lg_ and 4 for start of offset array  

  base_lg_ = contents[n-1]; // 最后1byte存的是base  

  uint32_t last_word =DecodeFixed32(contents.data() + n - 5); //偏移数组的位置  

  if (last_word > n - 5)return;  

  data_ = contents.data();  

  offset_ = data_ + last_word; // 偏移数组开始指针  

  num_ = (n - 5 - last_word) / 4; // 计算出filter个数  

uint64_t index = block_offset>> base_lg_; // 计算出filter index  

if (index < num_) {  
  // 解析出filter的range  
  uint32_t start =DecodeFixed32(offset_ + index*4);

  uint32_t limit =DecodeFixed32(offset_ + index*4 + 4);  

  if (start <= limit&& limit <= (offset_ - data_)) {  

    Slice filter = Slice(data_ +start, limit - start); // 根据range得到filter  

    returnpolicy_->KeyMayMatch(key, filter);  

  } else if (start == limit) {  

    return false;  // 空filter不匹配任何key  

  }  

}  

return true; // 当匹配处理