万字长文揭秘：阿里如何实现海量数据实时分析？(3)

在结合AnalyticDB的优化器的SQL优化实践中，FastSQL不断将SQL Rewrite的优化能力前置化到SQL Parser中实现，通过与优化器的SQL优化能力协商，将尽可能多的表达式级别优化前置化到SQL Parser中，使得优化器能更加专注于基于代价和成本的优化（CBO，Cost-Based Optimization）上，让优化器能更多的集中在理解计算执行计划优化上。FastSQL在AST Tree上实现了许多SQL Rewrite的能力，例如： [好文分享：www.11jj.com]

[原文来自：www.11jj.com]

常量折叠：

SELECT * FROM t1 tWHERE comm_week   BETWEEN CAST(date_format(date_add("day",-day_of_week("20180605"),                             date("20180605")),"%Y%m%d") AS bigint)        AND CAST(date_format(date_add("day",-day_of_week("20180605")                            ,date("20180605")),"%Y%m%d") AS bigint)------>SELECT * FROM t1 tWHERE comm_week BETWEEN20180602AND20180602

函数变换：

SELECT * FROM t1 t
WHERE DATE_FORMAT(t."pay_time","%Y%m%d")>="20180529"    AND DATE_FORMAT(t."pay_time","%Y%m%d")<="20180529"------>SELECT * FROM t1 tWHERE t."pay_time">= TIMESTAMP"2018-05-29 00:00:00"AND t."pay_time"< TIMESTAMP"2018-05-30 00:00:00"