他又又来了,c#开源sql解析引擎类库【SqlParser.Net 1.0】正式发布,它可以帮助你简单快速高效的解析和处理sql
背景hi 大家好,我是三合,在过往的岁月中,我曾经想过要写以下这些工具
[*]写一个通过拦截业务系统所有sql,然后根据这些sql自动分析表与表,字段与字段之间是如何关联的工具,即sql血缘分析工具
[*]想着动态改写sql,比如给where动态添加一个条件。
[*]写一个sql格式化工具
[*]写一个像mycat那样的分库分表中间件
[*]写一个sql防火墙,防止出现where 1=1后没有其他条件导致查询全表
[*]写一个数据库之间的sql翻译工具,比如把sqlserver的sql自动翻译为oracle的sql
但是无一例外,都失败了,因为要实现以上这些需求,都需要一个核心的类库,即sql解析引擎,遗憾的是,我没有找到合适的,这是我当初寻找的轨迹
[*]我发现了tsql-parser,但他只支持sql server,所以只能pass。
[*]然后我又发现了SqlParser-cs,
他的语法树解析出来像这样,
JsonConvert.SerializeObject(statements.First(), Formatting.Indented)// Elided for readability{ "Query": { "Body": { "Select": { "Projection": [ { "Expression": { "Ident": { "Value": "a", "QuoteStyle": null } } } ...额,怎么说呢,这语法树也太丑了点吧,同时非常难以理解,跟我想象中的完全不一样啊,于是也只能pass。
[*]接下来我又发现了另外一些基于antlr来解析sql的类库,比如SQLParser,因为代码是antlr自动生成的,比较难以进行手动优化,所以还是pass。
[*]最后我还发现了另外一个gsp的sqlparser,但它是收费的,而且巨贵无比,也pass。
找了一圈下来,我发现符合我要求的类库并不存在,所以我上面的那些想法,也一度搁浅了,但每一次的搁浅,都会使我内心的不甘加重一分,终于有一天,我下定决心,自己动手,丰衣足食,所以最近花了大概3个月时间,从头开始写了一个sql解析引擎,包括词法解析器到语法分析器,不依赖任何第三方组件,纯c#代码,在通过了156个各种各样场景的单元测试以及各种真实的业务环境验证后,今天它SqlParser.Net1.0.0正式发布了,本项目基于MIT协议开源,有以下优点,
[*]支持5大数据库,oracle,sqlserver,mysql,pgsql以及sqlite。
[*]极致的速度,解析普通sql,时间基本在0.3毫秒以下,当然了,sql越长,解析需要的时间就越长。
[*]文档完善,众所周知,我三合的开源项目,一向是文档齐全且简单易懂,做到看完就能上手,同时,我也会根据用户的反馈不断的补充以及完善文档。
[*]代码简洁易懂
SqlParser.Net存在的意义
SqlParser.Net是一个免费,功能全面且高性能的sql解析引擎类库,它可以帮助你简单快速高效的解析和处理sql。
Getting Started
接下来,我将介绍SqlParser.Net的用法
通过Nuget安装
你可以运行以下命令在你的项目中安装 SqlParser.Net 。
PM> Install-Package SqlParser.Net
支持框架
netstandard2.0
从最简单的demo开始
让我们一起看一个最简单的select语句是如何解析的
var sql = "select * from test";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() } }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test" } } }};以上面为例子,抽象语法树的所有叶子节点均为sqlExpression的子类,且各种sqlExpression节点可以互相嵌套,组成一颗复杂无比的树,其他sql解析引擎还分为statement和expression,我认为过于复杂,所以全部统一为sqlExpression,顶级的sqlExpression总共分为4种,
[*]查询语句(SqlSelectExpression)
[*]插入语句(SqlInsertExpression)
[*]删除语句(SqlDeleteExpression)
[*]更新语句(SqlUpdateExpression)
这4种顶级语句中,我认为最复杂的是查询语句,因为查询组合非常多,要兼容各种各样的情况,其他3种反而非常简单。现阶段,sqlExpression的子类一共有38种,我将在下面的演示中,结合例子给大家讲解一下。
1. Select查询语句
如上例子,SqlSelectExpression代表一个查询语句,SqlSelectQueryExpression则是真正具体的查询语句,他包括了
[*]要查询的所有列(Columns字段)
[*]数据源(From字段)
[*]条件过滤语句(Where字段)
[*]分组语句(GroupBy字段)
[*]排序语句(OrderBy字段)
[*]分页语句(Limit字段)
[*]Into语句(sql server专用,如SELECT id,name into test14 from TEST t)
[*]ConnectBy语句(oracle专用,如SELECT LEVEL l FROM DUALCONNECT BY NOCYCLE LEVEL<=100)
[*]WithSubQuerys语句,公用表表达式,即CTE
其中Columns是一个列表,他的每一个子项都是一个SqlSelectItemExpression,他的body代表一个逻辑子句,逻辑子句的值,可以为以下这些
[*]字段,如name,
[*]二元表达式,如t.age+3
[*]函数调用,如LOWER(t.NAME)
[*]一个完整的查询语句,如SELECT name FROM TEST2 t2
包括order by,partition by,group by,between,in,case when后面跟着的都是逻辑子句,这个稍后会演示,在这个例子中,因为是要查询所有列,所以仅有一个SqlSelectItemExpression,他的body是SqlAllColumnExpression(代表所有列),From代表要查询的数据源,在这里仅单表查询,所以From的值为SqlTableExpression(代表单表),表名是一个SqlIdentifierExpression,即标识符表达式,表示这是一个标识符,在SQL中,标识符(Identifier)是用于命名数据库对象的名称。这些对象可以包括表、列、索引、视图、模式、数据库等。标识符使得我们能够引用和操作这些对象,在这里,标识符的值为test,表示表名为test。
1.1 查询返回列的各种情形
1.1.1 查询指定字段
var sql = "select id AS bid,t.NAME testNamefrom test t";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "id", }, Alias = new SqlIdentifierExpression() { Value = "bid", }, }, new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "NAME", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, Alias = new SqlIdentifierExpression() { Value = "testName", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, },};在上面这个例子中,我们指定了要查询2个字段,分别是id和t.NAME,此时Columns列表里有2个值,
第一个SqlSelectItemExpression包含了
[*]主体,即body字段,在本例子中他的值是一个SqlIdentifierExpression表达式,值为id,表示列名为id,
[*]别名,即Alias字段,在本例子中他也是一个SqlIdentifierExpression,值为bid,代表列别名为bid,
第二个SqlSelectItemExpression的body里是一个SqlPropertyExpression,代表这是一个属性表达式,SqlPropertyExpression它包含了
[*]表名,即Table字段,值为t,即表名为t
[*]属性名,即Name字段,值为Name,即属性名为name
合起来则代表t表的name字段,而第二个SqlSelectItemExpression也有列别名,即testName,这个查询也是单表查询,但SqlTableExpression他多了一个Alias别名字段,即表示,表别名为t。
1.1.2 查询列为二元表达式的情况
var sql = "select 1+2 from test";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlBinaryExpression() { Left = new SqlNumberExpression() { Value = 1M, }, Operator = SqlBinaryOperator.Add, Right = new SqlNumberExpression() { Value = 2M, }, }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, }, },};在这个例子中,要查询的字段的值为一个二元表达式SqlBinaryExpression,他包含了
[*]左边部分,即Left字段,值为一个SqlNumberExpression,即数字表达式,它的值为1
[*]右边部分,即Right字段,值为一个SqlNumberExpression,即数字表达式,它的值为2
[*]中间符号,即Operator字段,值为add,即加法
这个例子证明了,SqlSelectItemExpression代表一个逻辑子句,而不仅仅是某个字段。
1.1.3 查询列为字符串/数字/布尔值的情况
var sql = "select ''' ''',3,true FROM test";var sqlAst = DbUtils.Parse(sql, DbType.MySql);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlStringExpression() { Value = "' '" }, }, new SqlSelectItemExpression() { Body = new SqlNumberExpression() { Value = 3M, }, }, new SqlSelectItemExpression() { Body = new SqlBoolExpression() { Value = true }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, }, },};在这个例子中,要查询的3个字段为字符串,数字和布尔值,字符串表达式即SqlStringExpression,body里即字符串的值' ',数字表达式即SqlNumberExpression,值为3,布尔表达式即SqlBoolExpression,值为true;
1.1.4 查询列为函数调用的情况
1.1.4.1 简单的函数调用
var sql = "select LOWER(name)FROM test";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlFunctionCallExpression() { Name = new SqlIdentifierExpression() { Value = "LOWER", }, Arguments = new List<SqlExpression>() { new SqlIdentifierExpression() { Value = "name", }, }, }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, }, },};在这个例子中,要查询的表达式是一个函数调用,函数调用表达式即SqlFunctionCallExpression,它包含了
[*]函数名,即Name字段,值为LOWER,
[*]函数参数列表,即Arguments字段,列表里只有一个值,即函数只有一个参数,且参数的值为name
1.1.4.2 带有over子句的函数调用
var sql = "SELECT t.*, ROW_NUMBER() OVER ( PARTITION BY t.IDORDER BY t.NAME,t.ID) as rnum FROM TEST t";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "*", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, }, new SqlSelectItemExpression() { Body = new SqlFunctionCallExpression() { Name = new SqlIdentifierExpression() { Value = "ROW_NUMBER", }, Over = new SqlOverExpression() { PartitionBy = new SqlPartitionByExpression() { Items = new List<SqlExpression>() { new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "ID", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, }, }, OrderBy = new SqlOrderByExpression() { Items = new List<SqlOrderByItemExpression>() { new SqlOrderByItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "NAME", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, }, new SqlOrderByItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "ID", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, }, }, }, }, }, Alias = new SqlIdentifierExpression() { Value = "rnum", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, },};在这个例子中,SqlFunctionCallExpression它除了常规字段外,还包含了Over子句,具体有以下这些
[*]函数名,即Name字段,值为ROW_NUMBER,
[*]函数参数列表,即Arguments字段,值为null,即无参数
[*]Over子句,即Over字段,他的值为一个SqlOverExpression表达式,SqlOverExpression本身又包含了以下内容
[*]PartitionBy分区子句,值为一个SqlPartitionByExpression表达式,表达式的内容也非常简单,只有一个Items,即一个分区表达式的列表,在这个例子中,列表里只有一个值SqlPropertyExpression,即根据t.id分区
[*]OrderBy排序子句,值为SqlOrderByExpression表达式,表达式的内容也非常简单,只有一个Items,即一个排序表达式的列表,列表里的值为SqlOrderByItemExpression,即排序子项表达式,排序子项表达式里又包含了以下内容
[*]排序依据,即Body字段,在这个例子中,排序依据是2个SqlPropertyExpression表达式,即根据t.NAME,t.ID排序
[*]排序类型,即OrderByType字段,值为Asc或者Desc,默认为asc,在这2个例子中,默认排序类型都是asc
1.1.4.3 带有within group子句的函数调用
var sql = "select name,PERCENTILE_CONT(0.5) within group(order by \"number\") from TEST5 group by name";var sqlAst = DbUtils.Parse(sql, DbType.Pgsql);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "name", }, }, new SqlSelectItemExpression() { Body = new SqlFunctionCallExpression() { Name = new SqlIdentifierExpression() { Value = "PERCENTILE_CONT", }, WithinGroup = new SqlWithinGroupExpression() { OrderBy = new SqlOrderByExpression() { Items = new List<SqlOrderByItemExpression>() { new SqlOrderByItemExpression() { Body = new SqlIdentifierExpression() { Value = "number", LeftQualifiers = "\"", RightQualifiers = "\"", }, }, }, }, }, Arguments = new List<SqlExpression>() { new SqlNumberExpression() { Value = 0.5M, }, }, }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST5", }, }, GroupBy = new SqlGroupByExpression() { Items = new List<SqlExpression>() { new SqlIdentifierExpression() { Value = "name", }, }, }, },};在这个例子中,SqlFunctionCallExpression它除了常规字段外,还包含了within group子句,具体有以下这些
[*]函数名,即Name字段,值为PERCENTILE_CONT,
[*]函数参数列表,即Arguments字段,列表里只有一项,表示只有1个参数,参数是SqlNumberExpression表达式,值为0.5
[*]within group子句,即WithinGroup字段,他的值为一个SqlWithinGroupExpression表达式,SqlWithinGroupExpression又包含了OrderBy排序子句,这里根据number字段排序
1.1.5 查询列为子查询的情况
var sql = "select c.*, (select a.name as province_name from portal_area a where a.id = c.province_id) as province_name, (select a.name as city_name from portal_area a where a.id = c.city_id) as city_name, (CASE WHEN c.area_id IS NULL THEN NULL ELSE (select a.name as area_name from portal_area a where a.id = c.area_id)END )as area_namefrom portal.portal_company c";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "*", }, Table = new SqlIdentifierExpression() { Value = "c", }, }, }, new SqlSelectItemExpression() { Body = new SqlSelectExpression() { Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "a", }, }, Alias = new SqlIdentifierExpression() { Value = "province_name", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "portal_area", }, Alias = new SqlIdentifierExpression() { Value = "a", }, }, Where = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "id", }, Table = new SqlIdentifierExpression() { Value = "a", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "province_id", }, Table = new SqlIdentifierExpression() { Value = "c", }, }, }, }, }, Alias = new SqlIdentifierExpression() { Value = "province_name", }, }, new SqlSelectItemExpression() { Body = new SqlSelectExpression() { Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "a", }, }, Alias = new SqlIdentifierExpression() { Value = "city_name", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "portal_area", }, Alias = new SqlIdentifierExpression() { Value = "a", }, }, Where = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "id", }, Table = new SqlIdentifierExpression() { Value = "a", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "city_id", }, Table = new SqlIdentifierExpression() { Value = "c", }, }, }, }, }, Alias = new SqlIdentifierExpression() { Value = "city_name", }, }, new SqlSelectItemExpression() { Body = new SqlCaseExpression() { Items = new List<SqlCaseItemExpression>() { new SqlCaseItemExpression() { Condition = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "area_id", }, Table = new SqlIdentifierExpression() { Value = "c", }, }, Operator = SqlBinaryOperator.Is, Right = new SqlNullExpression() }, Value = new SqlNullExpression() }, }, Else = new SqlSelectExpression() { Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "a", }, }, Alias = new SqlIdentifierExpression() { Value = "area_name", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "portal_area", }, Alias = new SqlIdentifierExpression() { Value = "a", }, }, Where = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "id", }, Table = new SqlIdentifierExpression() { Value = "a", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "area_id", }, Table = new SqlIdentifierExpression() { Value = "c", }, }, }, }, }, }, Alias = new SqlIdentifierExpression() { Value = "area_name", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "portal_company", }, Schema = new SqlIdentifierExpression() { Value = "portal", }, Alias = new SqlIdentifierExpression() { Value = "c", }, }, },};在这个例子中,要查询的列值为一个SqlSelectExpression表达式,即要查询的列是一个子查询
1.2 Where条件过滤语句
1.2.1 二元表达式
var sql = "SELECT * FROM test WHERE ID =1";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, }, Where = new SqlBinaryExpression() { Left = new SqlIdentifierExpression() { Value = "ID", }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlNumberExpression() { Value = 1M, }, }, },};在这个例子中,where字段的值是一个二元表达式SqlBinaryExpression,他包含了
[*]左边部分,即Left字段,值为一个SqlIdentifierExpression,即标识符表达式,它的值为ID
[*]右边部分,即Right字段,值为一个SqlNumberExpression,即数字表达式,它的值为1
[*]中间符号,即Operator字段,值为EqualTo,即等号,当然了,还可以是大于号,小于号,大于等于号,小于等于号,不等号等
二元表达式的两边可以非常灵活,可以是各种其他表达式,同时也可以自我嵌套另一个二元表达式,组成一个非常复杂的二元表达式
1.2.2 between/not between子句
var sql = "SELECT * FROM test WHERE ID BETWEEN 1 AND 2";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, }, Where = new SqlBetweenAndExpression() { Body = new SqlIdentifierExpression() { Value = "ID", }, Begin = new SqlNumberExpression() { Value = 1M, }, End = new SqlNumberExpression() { Value = 2M, }, }, },};between子句包含了
[*]Begin部分,即Begin字段,在这个例子中,值为一个SqlNumberExpression,,它的值为1
[*]End部分,即End字段,在这个例子中,值为一个SqlNumberExpression,它的值为2
[*]Body主体部分,即Body字段,值为SqlIdentifierExpression,即标识符表达式,值为id
[*]取反部分,即IsNot字段,如果是not between,则IsNot=true
1.2.3 is null/is not null子句
var sql = "select * from test rd where rd.name is null";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, Alias = new SqlIdentifierExpression() { Value = "rd", }, }, Where = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "rd", }, }, Operator = SqlBinaryOperator.Is, Right = new SqlNullExpression() }, },};is null/is not null子句主要体现在二元表达式里,Operator字段为Is/IsNot,right字段为SqlNullExpression,即null表达式,代表值为null
1.2.4 exists/not exists子句
var sql = "select * from TEST t where EXISTS(select * from TEST2 t2)";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, Where = new SqlExistsExpression() { Body = new SqlSelectExpression() { Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST2", }, Alias = new SqlIdentifierExpression() { Value = "t2", }, }, }, }, }, },};exists/not exists子句,主要体现为SqlExistsExpression表达式,
[*]主体,即body字段,本例子中值为一个SqlSelectExpression表达式
[*]取反部分,即IsNot字段,如果是not exists,则IsNot=true
1.2.5 like/not like子句
var sql = "SELECT * from TEST t WHERE name LIKE '%a%'";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, Where = new SqlBinaryExpression() { Left = new SqlIdentifierExpression() { Value = "name", }, Operator = SqlBinaryOperator.Like, Right = new SqlStringExpression() { Value = "%a%" }, }, },};like子句,主要体现在二元表达式里,Operator字段为Like/NotLike,本例子中right字段为字符串表达式,即SqlStringExpression表达式,值为%a%。
1.2.6 all/any子句
var sql = "select * from customer c where c.Age >all(select o.Quantityfrom orderdetail o)";var sqlAst = DbUtils.Parse(sql, DbType.MySql);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "customer", }, Alias = new SqlIdentifierExpression() { Value = "c", }, }, Where = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "Age", }, Table = new SqlIdentifierExpression() { Value = "c", }, }, Operator = SqlBinaryOperator.GreaterThen, Right = new SqlAllExpression() { Body = new SqlSelectExpression() { Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "Quantity", }, Table = new SqlIdentifierExpression() { Value = "o", }, }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "orderdetail", }, Alias = new SqlIdentifierExpression() { Value = "o", }, }, }, }, }, }, },};all/any子句,主要体现在SqlAllExpression/SqlAnyExpression表达式,它的body里是另一个SqlSelectExpression表达式
1.2.7 in/ not in子句
var sql = "SELECT* from TEST t WHERE t.NAME IN ('a','b','c')";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, Where = new SqlInExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "NAME", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, TargetList = new List<SqlExpression>() { new SqlStringExpression() { Value = "a" }, new SqlStringExpression() { Value = "b" }, new SqlStringExpression() { Value = "c" }, }, }, },};in/not in子句,主要体现在SqlInExpression表达式,它包含了
[*]body字段,即in的主体,在这里是SqlPropertyExpression,值为t.NAME
[*]TargetList字段,即in的目标列表,在这里是一个SqlExpression的列表,里面包括3个SqlStringExpression,即字符串表达式,a,b,c
[*]取反部分,即IsNot字段,如果是not in,则IsNot=true
当然了,in也有另一种子查询的类型,即
var sql = "select * from TEST5WHERE NAME IN (SELECT NAMEFROM TEST3)";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST5", }, }, Where = new SqlInExpression() { Body = new SqlIdentifierExpression() { Value = "NAME", }, SubQuery = new SqlSelectExpression() { Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "NAME", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST3", }, }, }, }, }, },};在这里的SqlInExpression表达式中,它包含了
[*]body字段,即in的主体,在这里是SqlIdentifierExpression,值为NAME
[*]SubQuery字段,即子查询,值为一个SqlSelectExpression
[*]IsNot字段,如果是not in,则IsNot=true
1.2.8 case when子句
var sql = "SELECT CASE WHEN t.name='1' THEN 'a' WHEN t.name='2' THEN 'b' ELSE 'c' END AS v from TEST t";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlCaseExpression() { Items = new List<SqlCaseItemExpression>() { new SqlCaseItemExpression() { Condition = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlStringExpression() { Value = "1" }, }, Value = new SqlStringExpression() { Value = "a" }, }, new SqlCaseItemExpression() { Condition = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlStringExpression() { Value = "2" }, }, Value = new SqlStringExpression() { Value = "b" }, }, }, Else = new SqlStringExpression() { Value = "c" }, }, Alias = new SqlIdentifierExpression() { Value = "v", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, },};case when子句,主要体现在SqlCaseExpression表达式里,他包含了
[*]各种case when键值对的列表,即Items字段,列表里的每一个元素都是SqlCaseItemExpression表达式,SqlCaseItemExpression表达式,又包含了
[*]条件,即Condition字段,在本例子中是二元表达式,即SqlBinaryExpression表达式,值为t.name ='1'
[*]值,即value字段,在本例子中值为字符串a
[*]Else字段,即默认值,本例子中为字符串c
case when还有另外一种句式,如下:
var sql = "select case t.name when 'a' then 1 else 2 endfrom test t ";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlCaseExpression() { Items = new List<SqlCaseItemExpression>() { new SqlCaseItemExpression() { Condition = new SqlStringExpression() { Value = "a" }, Value = new SqlNumberExpression() { Value = 1M, }, }, }, Else = new SqlNumberExpression() { Value = 2M, }, Value = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, },};在这种SqlCaseExpression表达式里,他包含了
[*]case条件的主体变量,即Value字段,本例子中值为SqlPropertyExpression,它的值为t.name
[*]各种when then键值对的列表,即Items字段,列表里的每一个元素都是SqlCaseItemExpression表达式,SqlCaseItemExpression表达式,又包含了
[*]条件,即Condition字段,在本例子中是字符串表达式SqlStringExpression,它的值为a
[*]值,即Value字段,在本例子中值为SqlNumberExpression,它的值为1
[*]Else字段,即默认值,本例子中为数字2
1.2.9 not子句
var sql = "select * from TEST t WHERE not t.NAME ='abc'";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, Where = new SqlNotExpression() { Body = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "NAME", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlStringExpression() { Value = "abc" }, }, }, },};not子句,主要体现在SqlNotExpression表达式里,它只有一个body字段,即代表否定的部分
1.2.10 变量子句
var sql = "select * from TEST t WHERE not t.NAME =:name";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, Where = new SqlNotExpression() { Body = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "NAME", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlVariableExpression() { Name = "name", Prefix = ":", }, }, }, },};变量子句,主要体现在SqlVariableExpression表达式里,它包括以下部分:
[*]变量名,即字段Name,这里值为name
[*]变量前缀,这里值为:
1.3 From数据源
在sql中,From关键字后面有多种形式来指定数据源。主要有以下几种
1.3.1 表名或者视图
select * from test这个解析结果上面已经演示了。
1.3.2 子查询(子表)
var sql = "select * from (select * from test) t";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlSelectExpression() { Alias = new SqlIdentifierExpression() { Value = "t", }, Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, }, }, }, },};在这个例子中,数据源From的值为一个SqlSelectExpression,即SqlSelectExpression中可以嵌套SqlSelectExpression,同时我们注意到内部的SqlSelectExpression有一个表别名的字段Alias,标识符的值为t,表示表别名为t;
1.3.3 连表查询(JOIN)
var sql = "select t1.id from test t1 left join test2 t2 on t1.id=t2.id right join test3 t3 on t2.id=t3.id";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "id", }, Table = new SqlIdentifierExpression() { Value = "t1", }, }, }, }, From = new SqlJoinTableExpression() { Left = new SqlJoinTableExpression() { Left = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, Alias = new SqlIdentifierExpression() { Value = "t1", }, }, JoinType = SqlJoinType.LeftJoin, Right = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test2", }, Alias = new SqlIdentifierExpression() { Value = "t2", }, }, Conditions = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "id", }, Table = new SqlIdentifierExpression() { Value = "t1", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "id", }, Table = new SqlIdentifierExpression() { Value = "t2", }, }, }, }, JoinType = SqlJoinType.RightJoin, Right = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test3", }, Alias = new SqlIdentifierExpression() { Value = "t3", }, }, Conditions = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "id", }, Table = new SqlIdentifierExpression() { Value = "t2", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "id", }, Table = new SqlIdentifierExpression() { Value = "t3", }, }, }, }, },};在上面这个例子中,我们演示了连表查询是如何解析的,From字段的值为一个SqlJoinTableExpression,即连表查询表达式,他包含了
[*]左边部分,即Left字段
[*]右边部分,即Right字段
[*]连接方式,即JoinType字段,值包括InnerJoin,LeftJoin,RightJoin,FullJoin,CrossJoin,CommaJoin
[*]表关联条件,即Conditions字段。在这里,Conditions字段的值为一个二元表达式SqlBinaryExpression
在这个例子中,总共3张表联查,SqlJoinTableExpression中得left字段又是一个SqlJoinTableExpression,即SqlJoinTableExpression中可以嵌套SqlJoinTableExpression,无限套娃。
1.3.4 公用表表达式(CTE)
var sql = "with c1 as (select name from test t) , c2(name) AS (SELECT name FROM TEST2 t3 ) select *from c1 JOIN c2 ON c1.name=c2.name";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { WithSubQuerys = new List<SqlWithSubQueryExpression>() { new SqlWithSubQueryExpression() { Alias = new SqlIdentifierExpression() { Value = "c1", }, FromSelect = new SqlSelectExpression() { Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "name", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, }, }, }, new SqlWithSubQueryExpression() { Alias = new SqlIdentifierExpression() { Value = "c2", }, FromSelect = new SqlSelectExpression() { Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "name", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST2", }, Alias = new SqlIdentifierExpression() { Value = "t3", }, }, }, }, Columns = new List<SqlIdentifierExpression>() { new SqlIdentifierExpression() { Value = "name", }, }, }, }, Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlJoinTableExpression() { Left = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "c1", }, }, JoinType = SqlJoinType.InnerJoin, Right = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "c2", }, }, Conditions = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "c1", }, }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "c2", }, }, }, }, },};公用表表达式(CTE),主要体现在SqlSelectQueryExpression的WithSubQuerys字段,他是一个SqlWithSubQueryExpression表达式列表,即公用表列表,它里面的每一个元素都是SqlWithSubQueryExpression表达式,此表达式,包含了
[*]公共表的来源部分,即FromSelect字段,在本例子中,他的值是一个SqlSelectExpression表达式,即一个查询
[*]公共表的表别名,即Alias字段,在本例子中,他的值是c1
[*]公共表的列部分,即Columns字段,在本例子中只有一个列名,即name
1.3.5 函数返回的结果集
特定数据库支持从返回结果集的函数中查询,比如oracle中添加一个自定义函数splitstr,他的作用是将一个字符串根据;号进行分割,返回多行数据
var sql = "SELECT * FROM TABLE(splitstr('a;b',';'))";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlReferenceTableExpression() { FunctionCall = new SqlFunctionCallExpression() { Name = new SqlIdentifierExpression() { Value = "TABLE", }, Arguments = new List<SqlExpression>() { new SqlFunctionCallExpression() { Name = new SqlIdentifierExpression() { Value = "splitstr", }, Arguments = new List<SqlExpression>() { new SqlStringExpression() { Value = "a;b" }, new SqlStringExpression() { Value = ";" }, }, }, }, }, } },};函数返回的结果集主要体现在SqlReferenceTableExpression表达式,他的内部包含了一个FunctionCall字段,值为SqlFunctionCallExpression表达式,代表从函数调用的结果集中进行查询。
1.4 OrderBy排序语句
var sql = "select fa.FlowIdfrom FlowActivity fa order by fa.FlowId desc,fa.Id asc";var sqlAst = DbUtils.Parse(sql, DbType.SqlServer);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "FlowId" }, Table = new SqlIdentifierExpression() { Value = "fa" }, }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "FlowActivity" }, Alias = new SqlIdentifierExpression() { Value = "fa" }, }, OrderBy = new SqlOrderByExpression() { Items = new List<SqlOrderByItemExpression>() { new SqlOrderByItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "FlowId" }, Table = new SqlIdentifierExpression() { Value = "fa" }, }, OrderByType = SqlOrderByType.Desc }, new SqlOrderByItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "Id" }, Table = new SqlIdentifierExpression() { Value = "fa" }, }, OrderByType = SqlOrderByType.Asc }, }, }, },};OrderBy排序子句,值为SqlOrderByExpression表达式,表达式的内容也非常简单,只有一个Items,即一个排序子项表达式的列表,列表里的值为SqlOrderByItemExpression,即排序子项表达式,排序子项表达式里又包含了以下内容
[*]排序依据,即Body字段,在这个例子中,排序依据是2个SqlPropertyExpression表达式,即根据fa.FlowId,fa.Id排序
[*]排序类型,即OrderByType字段,值为Asc或者Desc,默认为asc,在这2个例子中,有asc和Desc
[*]决定null排在前面或后面的NullsType字段,在oracle,pgsql,sqlite中我们可以指定null在排序中的位置,如以下sql
select * from TEST5 t order by t.NAMEdesc nulls FIRST,t.AGE ASC NULLSlast 那么我们的NullsType字段,他的值有SqlOrderByNullsType.First和SqlOrderByNullsType.Last,与之对应。
1.5 GroupBy分组语句
var sql = "select fa.FlowIdfrom FlowActivity fa group by fa.FlowId,fa.Id HAVING count(fa.Id)>1";var sqlAst = DbUtils.Parse(sql, DbType.SqlServer);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "FlowId" }, Table = new SqlIdentifierExpression() { Value = "fa" }, }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "FlowActivity" }, Alias = new SqlIdentifierExpression() { Value = "fa" }, }, GroupBy = new SqlGroupByExpression() { Items = new List<SqlExpression>() { new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "FlowId" }, Table = new SqlIdentifierExpression() { Value = "fa" }, }, new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "Id" }, Table = new SqlIdentifierExpression() { Value = "fa" }, }, }, Having = new SqlBinaryExpression() { Left = new SqlFunctionCallExpression() { Name = new SqlIdentifierExpression() { Value = "count" }, Arguments = new List<SqlExpression>() { new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "Id" }, Table = new SqlIdentifierExpression() { Value = "fa" }, }, }, }, Operator = SqlBinaryOperator.GreaterThen, Right = new SqlNumberExpression() { Value = 1M }, }, }, },};GroupBy分组语句,值为SqlGroupByExpression表达式,他的内容如下
[*]分组子项表达式的列表,即Items字段,列表里的值为SqlExpression,他的值是一个逻辑子句
[*]分组过滤子句,即Having字段,他的值是一个逻辑子句,在本例子中,逻辑子句的值为一个SqlBinaryExpression
1.5 Limit分页子句
1.5.1 mysql,sqlite
var sql = "select * from test t limit 1,5";var sqlAst = DbUtils.Parse(sql, DbType.MySql);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, Limit = new SqlLimitExpression() { Offset = new SqlNumberExpression() { Value = 1M, }, RowCount = new SqlNumberExpression() { Value = 5M, }, }, },};Limit分页子句,值为SqlLimitExpression表达式,他的内容如下
[*]每页数量,即RowCount字段,这本例子中,值为5
[*]跳过数量,即Offset字段,本例子中,值为1
1.5.2 oracle
var sql = "SELECT * FROM TEST3 tORDER BY t.NAMEDESC FETCH FIRST 2 rows ONLY";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() } }, From = new SqlTableExpression() { Alias = new SqlIdentifierExpression() { Value = "t" }, Name = new SqlIdentifierExpression() { Value = "TEST3" } }, OrderBy = new SqlOrderByExpression() { Items = new List<SqlOrderByItemExpression>() { new SqlOrderByItemExpression() { OrderByType = SqlOrderByType.Desc, Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "NAME" }, Table = new SqlIdentifierExpression() { Value = "t" } } } } }, Limit = new SqlLimitExpression() { RowCount = new SqlNumberExpression() { Value = 2 } } }};1.5.3 pgsql
var sql = "select * from test5 t order by t.name limit 1 offset 10;";var sqlAst = DbUtils.Parse(sql, DbType.Pgsql);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test5", }, Alias = new SqlIdentifierExpression() { Value = "t", }, }, OrderBy = new SqlOrderByExpression() { Items = new List<SqlOrderByItemExpression>() { new SqlOrderByItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name", }, Table = new SqlIdentifierExpression() { Value = "t", }, }, }, }, }, Limit = new SqlLimitExpression() { Offset = new SqlNumberExpression() { Value = 10M, }, RowCount = new SqlNumberExpression() { Value = 1M, }, }, },};1.5.4 sqlServer
var sql = "select * from test t order by t.name OFFSET 5 ROWS FETCH NEXT 10 ROWS ONLY";var sqlAst = DbUtils.Parse(sql, DbType.SqlServer);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() } }, From = new SqlTableExpression() { Alias = new SqlIdentifierExpression() { Value = "t" }, Name = new SqlIdentifierExpression() { Value = "test" } }, OrderBy = new SqlOrderByExpression() { Items = new List<SqlOrderByItemExpression>() { new SqlOrderByItemExpression() { Body = new SqlPropertyExpression() { Name = new SqlIdentifierExpression() { Value = "name" }, Table = new SqlIdentifierExpression() { Value = "t" } } } } }, Limit = new SqlLimitExpression() { Offset = new SqlNumberExpression() { Value = 5 }, RowCount = new SqlNumberExpression() { Value = 10 } } }};1.6 ConnectBy层次查询语句(oracle专用)
var sql = "SELECT EMPLOYEEID , MANAGERID , LEVEL FROM EMPLOYEE e START WITH MANAGERID IS NULL CONNECT BY NOCYCLE PRIOR EMPLOYEEID = MANAGERID ORDER SIBLINGS BY EMPLOYEEID ";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "EMPLOYEEID", }, }, new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "MANAGERID", }, }, new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "LEVEL", }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "EMPLOYEE", }, Alias = new SqlIdentifierExpression() { Value = "e", }, }, ConnectBy = new SqlConnectByExpression() { StartWith = new SqlBinaryExpression() { Left = new SqlIdentifierExpression() { Value = "MANAGERID", }, Operator = SqlBinaryOperator.Is, Right = new SqlNullExpression() }, Body = new SqlBinaryExpression() { Left = new SqlIdentifierExpression() { Value = "EMPLOYEEID", }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlIdentifierExpression() { Value = "MANAGERID", }, }, IsNocycle = true, IsPrior = true, OrderBy = new SqlOrderByExpression() { Items = new List<SqlOrderByItemExpression>() { new SqlOrderByItemExpression() { Body = new SqlIdentifierExpression() { Value = "EMPLOYEEID", }, }, }, IsSiblings = true, }, }, },};ConnectBy层次查询子句,值为SqlConnectByExpression表达式,他的内容如下
[*]指定层次查询的根节点条件,即StartWith字段,本例子中他的值为SqlBinaryExpression二元表达式
[*]主体关联条件子句,即Body字段,本例子中他的值是一个SqlBinaryExpression二元表达式
[*]IsPrior字段用来指示层次结构中哪个列是父节点,如果sql中存在Prior关键字,则值为true
[*]IsNocycle字段,用来防止循环引用导致无限递归,如果sql中存在Nocycle则为true
[*]order by子句,用于排序
1.7 Into子句(sql server专用)
var sql = "SELECT name into test14 from TEST as t ";var sqlAst = DbUtils.Parse(sql, DbType.SqlServer);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "name" }, }, }, Into = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test14" }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST" }, Alias = new SqlIdentifierExpression() { Value = "t" }, }, },};into子句,在本例子中值为SqlTableExpression,即into到某张表里。
2. Insert插入语句
2.1 插入单个值
var sql = "insert into test11(name,id) values('a1','a2')";var sqlAst = DbUtils.Parse(sql, DbType.SqlServer);解析结果如下:
var expect = new SqlInsertExpression(){ Columns = new List<SqlExpression>() { new SqlIdentifierExpression() { Value = "name" }, new SqlIdentifierExpression() { Value = "id" }, }, ValuesList = new List<List<SqlExpression>>() { new List<SqlExpression>() { new SqlStringExpression() { Value = "a1" }, new SqlStringExpression() { Value = "a2" }, }, }, Table = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test11" }, },};如上例子,插入语句表现为一个SqlInsertExpression,他包含了
[*]要插入的字段列表,即Columns字段,值为一个SqlExpression的列表,本例子中值为2个SqlIdentifierExpression,它们的值为name和id,即插入name和id字段
[*]值列表,即ValuesList字段,值为一个List<SqlExpression>的列表,即列表里的元素是一个列表,每个元素代表一组待插入的数据,本例子中列表里只有一个List,并且子列表里的值为a1和a2,即待插入的值为a1和a2。
[*]待插入数据的表,即Table字段,本例子中值为test11
为什么ValuesList字段是列表里嵌套列表呢?主要是因为可以插入多个值列表,让我们继续往下看
2.2 插入多个值
var sql = "insert into test11(name,id) values('a1','a2'),('a3','a4')";var sqlAst = DbUtils.Parse(sql, DbType.MySql);解析结果如下:
var expect = new SqlInsertExpression(){ Columns = new List<SqlExpression>() { new SqlIdentifierExpression() { Value = "name" }, new SqlIdentifierExpression() { Value = "id" }, }, ValuesList = new List<List<SqlExpression>>() { new List<SqlExpression>() { new SqlStringExpression() { Value = "a1" }, new SqlStringExpression() { Value = "a2" }, }, new List<SqlExpression>() { new SqlStringExpression() { Value = "a3" }, new SqlStringExpression() { Value = "a4" }, }, }, Table = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test11" }, },};在本例子中,ValuesList字段中有2个子元素,即2个List列表,代表插入2组数据,值分别为('a1','a2')和('a3','a4')
2.3 待插入的值为一个子查询
var sql = "INSERT INTO TEST2(name) SELECT name AS name2 FROM TEST t";var sqlAst = DbUtils.Parse(sql, DbType.MySql);解析结果如下:
var expect = new SqlInsertExpression(){ Columns = new List<SqlExpression>() { new SqlIdentifierExpression() { Value = "name" }, }, Table = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST2" }, }, FromSelect = new SqlSelectExpression() { Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlIdentifierExpression() { Value = "name" }, Alias = new SqlIdentifierExpression() { Value = "name2" }, }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "TEST" }, Alias = new SqlIdentifierExpression() { Value = "t" }, }, }, },};如上例子,插入语句表现为一个SqlInsertExpression,他包含了
[*]要插入的字段列表,即Columns字段,值为一个SqlExpression的列表,本例子中值为name
[*]子查询来源,即FromSelect字段,值为一个SqlSelectExpression,即一个子查询
[*]待插入数据的表,即Table字段,本例子中值为TEST2
3. Update更新语句
var sql = "update test set name ='4',d='2024-11-22 08:19:47.243' where name ='1'";var sqlAst = DbUtils.Parse(sql, DbType.MySql);解析结果如下:
var expect = new SqlUpdateExpression(){ Table = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test" }, }, Where = new SqlBinaryExpression() { Left = new SqlIdentifierExpression() { Value = "name" }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlStringExpression() { Value = "1" }, }, Items = new List<SqlExpression>() { new SqlBinaryExpression() { Left = new SqlIdentifierExpression() { Value = "name" }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlStringExpression() { Value = "4" }, }, new SqlBinaryExpression() { Left = new SqlIdentifierExpression() { Value = "d" }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlStringExpression() { Value = "2024-11-22 08:19:47.243" }, }, },};如上例子,更新语句表现为一个SqlUpdateExpression,他包含了
[*]要更新的(字段-值)的列表,即Items字段,值为一个SqlExpression的列表,本例子中值为2个SqlBinaryExpression,即name='4'和d='2024-11-22 08:19:47.243'
[*]条件过滤子句,即Where字段,代表过滤条件,本例子中值为一个SqlBinaryExpression,即name ='1'
[*]待更新数据的表,即Table字段,本例子中值为test
4. Delete删除语句
var sql = "delete from test where name=4";var sqlAst = DbUtils.Parse(sql, DbType.MySql);解析结果如下:
var expect = new SqlDeleteExpression(){ Table = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test" }, }, Where = new SqlBinaryExpression() { Left = new SqlIdentifierExpression() { Value = "name" }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlNumberExpression() { Value = 4M }, },};如上例子,删除语句表现为一个SqlDeleteExpression,他包含了
[*]条件过滤子句,即Where字段,代表过滤条件,本例子中值为一个SqlBinaryExpression,即name=4
[*]待删除数据的表,即Table字段,本例子中值为test
5. 注释处理
5.1 单行注释
var sql = @"select *--abc from test lbu WHERE a ='1'--aaaaaaFROM test";var sqlAst = DbUtils.Parse(sql, DbType.SqlServer);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, }, },};如上例子,单行注释被正确忽视,解析正确。
5.2 多行注释
var sql = @"/*这 是 顶部*/ select *--abc FROM test/*这 是 底部*/";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析结果如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, }, },};如上例子,多行注释被正确忽视,解析正确。
6. 如何解析ast抽象语法树
当我们通过
var sql = @"select * from test";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析sql获取到抽象语法树以后,我们就要对这颗抽象语法树进行解析,获取我们想要的数据,此时就要用上访问者模式(visitor) .
6.1 访问者模式
访问者模式最大的特点就是结构与算法分离,结合本项目理解,就是ast抽象语法树这个结构已经解析出来了,你可以根据自己的需要写算法去任意解析这颗语法树。这是一个1-N的操作,即一个抽象语法树,可以对应N个解析算法,当我们要自定义算法去解析抽象语法树时,我们需要自定义一个Visitor类,并且实现IAstVisitor接口
public class CustomVisitor : IAcceptVisitor{ }但是实现这个接口要实现接口里的很多个方法,并且有些数据并不是我们关心的,所以我提供了一个实现了IAcceptVisitor接口的抽象类BaseAstVisitor用来简化操作,我们只需要继承这个抽象类,然后重写我们感兴趣的方法即可
public class CustomVisitor : BaseAstVisitor{ }在本项目中,我提供了2个基本的vistor供大家使用,UnitTestAstVisitor和SqlGenerationAstVisitor,大家可以参考这2个visitor去写自己的算法来解析抽象语法树。接下来,我将介绍这2个visitor的用法。
6.2 UnitTestAstVisitor
当我们通过
var sql = @"select * from test";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);解析sql获取到抽象语法树之后,sqlAst其实还是一个数据结构,我们可以通过vs监视这个变量来查看内部的结构,但是如果是非常复杂的sql,那这颗树会巨大无比,要靠我们手动去慢慢点开查看得累死,没错!写单元测试的时候,我刚开始都是用手写的结果去对比引擎解析出来的结果,后来我就被累死了,该说不说,这活真不是人干的,所以痛定思痛之后我就写了这个UnitTestAstVisitor来替我生成ast的结构字符串,接下来让我们看看用法
var sql = @"select * from test";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);var unitTestAstVisitor = new UnitTestAstVisitor();sqlAst.Accept(unitTestAstVisitor);var result = unitTestAstVisitor.GetResult();其中的result就是解析抽象语法树生成的字符串,如下:
var expect = new SqlSelectExpression(){ Query = new SqlSelectQueryExpression() { Columns = new List<SqlSelectItemExpression>() { new SqlSelectItemExpression() { Body = new SqlAllColumnExpression() }, }, From = new SqlTableExpression() { Name = new SqlIdentifierExpression() { Value = "test", }, }, },};然后把这个生成的字符串黏贴到vs里去和引擎生成的结果进行对比,
Assert.True(sqlAst.Equals(expect));至此,我写单元测试的工作量大大减轻,同时对于生成的sqlAst语法树的结构也更加一目了然了。
6.2 SqlGenerationAstVisitor
我们通过解析sql生成了抽象语法树之后,如果我们想要给这颗抽象语法树添加一个where条件,比如添加test.name ='a'
var sql = @"select * from test";var sqlAst = DbUtils.Parse(sql, DbType.Oracle);if (sqlAst is SqlSelectExpression sqlSelectExpression && sqlSelectExpression.Query is SqlSelectQueryExpression sqlSelectQueryExpression){ sqlSelectQueryExpression.Where = new SqlBinaryExpression() { Left = new SqlPropertyExpression() { Table = new SqlIdentifierExpression() { Value = "test" }, Name = new SqlIdentifierExpression() { Value = "name" } }, Operator = SqlBinaryOperator.EqualTo, Right = new SqlStringExpression() { Value = "a" } };}好了,现在我们添加完了,接下来我们肯定是想着把抽象语法树转化为sql语句,此时,就需要用上SqlGenerationAstVisitor了,他就是负责把抽象语法树转化为sql的
var sqlGenerationAstVisitor = new SqlGenerationAstVisitor(DbType.Oracle);sqlAst.Accept(sqlGenerationAstVisitor);var newSql = sqlGenerationAstVisitor.GetResult();我们获取到的newSql就是新的sql了,他的值为
select * from test where(test.name = 'a')至此,我们的目的就达到了。
7. sql解析的理论基础
sql之所以能被我们解析出来,主要是因为sql是一种形式语言,自然语言和形式语言的一个重要区别是,自然语言的一个语句,可能有多重含义,而形式语言的一个语句,只能有一个语义;形式语言的语法是人为规定的,有了一定的语法规则,语法解析器就能根据语法规则,解析出一个语句的唯一含义。
项目感悟
[*]解决嵌套问题的唯一方案,就是用递归
[*]对于基础项目,单元测试非常非常重要,因为开发的过程中可能会不断地重构,那以前跑过的测试案例就有可能失败,如果此时需要靠人手工去回归测试验证的话,那工作量是天量的,做不完,根本做不完,所以正确的解决方案是写单元测试,新添加一个功能后,为这个功能写1-N个单元测试,确保新功能对各种情况都有覆盖到,然后再跑一遍所有单元测试,确保没有影响到旧的功能。当然了,跑单元测试最让我崩溃的是,跑一遍所有单元测试,红了(即失败)几十个,天都塌了。
开源地址,欢迎star
本项目基于MIT协议开源,地址为
https://github.com/TripleView/SqlParser.Net
同时感谢以下项目
[*]阿里巴巴开源的druid
写在最后
如果各位靓仔觉得这个项目不错,欢迎一键三连(推荐,star,关注),同时欢迎加入三合的开源交流群,QQ群号:799648362
页:
[1]