patch 4.0

This commit is contained in:
wangzelin.wzl
2022-10-24 10:34:53 +08:00
parent 4ad6e00ec3
commit 93a1074b0c
10533 changed files with 2588271 additions and 2299373 deletions

View File

@ -19,32 +19,35 @@ namespace oceanbase {
namespace sql {
class ObExprEstimateNdv : public ObFuncExprOperator {
public:
explicit ObExprEstimateNdv(common::ObIAllocator& alloc);
explicit ObExprEstimateNdv(common::ObIAllocator &alloc);
virtual ~ObExprEstimateNdv();
virtual int calc_result_type1(
ObExprResType& type, ObExprResType& type1, common::ObExprTypeCtx& type_ctx) const override;
virtual int calc_result1(common::ObObj& result, const common::ObObj& obj, common::ObExprCtx& expr_ctx) const override;
static int llc_estimate_ndv(common::ObObj& result, const common::ObObj& obj, common::ObExprCtx& expr_ctx);
static void llc_estimate_ndv(int64_t& result, const common::ObString& bitmap_str);
static int llc_estimate_ndv(double& estimate_ndv, const common::ObString& bitmap_buf);
// high several bits of hash value are used to store bucket_id, the param value must
// remove these bits by left shift, the count of valid bits after removing is bit_width.
virtual int calc_result_type1(ObExprResType &type,
ObExprResType &type1,
common::ObExprTypeCtx &type_ctx) const;
static void llc_estimate_ndv(int64_t &result, const common::ObString &bitmap_str);
static int llc_estimate_ndv(double &estimate_ndv, const common::ObString &bitmap_buf);
// 计算value的leading zeros。在HyperLogLogCount中,一个hash值的前面若干位要用来做分桶,
// 这里的传入参数value是通过左移移除掉分桶部分后的部分,它的实际有效位数是高bit_width位。
static uint64_t llc_leading_zeros(uint64_t value, uint64_t bit_width);
static bool llc_is_num_buckets_valid(int64_t num_buckets);
// for engine 3.0
virtual int cg_expr(ObExprCGCtx& expr_cg_ctx, const ObRawExpr& raw_expr, ObExpr& rt_expr) const override;
static int calc_estimate_ndv_expr(const ObExpr& expr, ObEvalCtx& ctx, ObDatum& res_datum);
virtual int cg_expr(ObExprCGCtx &expr_cg_ctx, const ObRawExpr &raw_expr,
ObExpr &rt_expr) const override;
static int calc_estimate_ndv_expr(const ObExpr &expr, ObEvalCtx &ctx,
ObDatum &res_datum);
private:
// 计算HyperLogLogCount中的 alpha * m^2 的函数。计算涉及变量m(uint64_t)
// 转double的步骤,调用者需要考虑可能的精度损失(目前m通常不超过4096,无损失)。
static inline double llc_alpha_times_m_square(const uint64_t m);
// the count of buckets should be between 16 and 65536, according to Google's HLLC paper.
// 根据Google的HLLC论文桶数至少取2^4(16)个,至多取2^16(65536)个。
static const int LLC_NUM_BUCKETS_MIN = (1 << 4);
static const int LLC_NUM_BUCKETS_MAX = (1 << 16);
private:
DISALLOW_COPY_AND_ASSIGN(ObExprEstimateNdv);
};
} /* namespace sql */
} /* namespace oceanbase */
#endif /* OCEANBASE_SQL_ENGINE_EXPR_ESTIMATE_NDV_H_ */