自动生成测试数据加强版

---

加强版存储过程

转载请标明出处，引用请保留头注视。第一版链接 ：

http://topic.csdn.net/u/20080516/15/3fcf4880-67e9-4a28-844d-05985db51215.html

---

-------------------------------------------------------------------------------
-- 自动生成测试数据共享加强版
--
-- 基本特性
-- * 表前缀：@表变量、#临时表、$实体表，支持混合多表。
-- * 无表名：@1...@n　#1...#n　T1...Tn
-- * 无列名：C1...Cn
-- * 少列名：Cm...Cn
-- * 列分隔：空格和TAB键。
-- * 列填充：数据左边对齐原则，数据不足右向填充NULL。
-- * 占位符：分号(;)，控制数据位置，左向填充NULL。
-- * 字符集：不考虑UNICODE/NCHAR，网页能够正常显示的应该没问题，网页不能正常显示的，UNICODE也没用。
--
-- 加强特性
-- * 优化数据类型解析，基本上涵盖了所有的原始测试数据形式。
-- * 逗号(,)不再作为列分隔符，添加指定列分隔符控制，具体方法见示例。
-- * 连续(;;)或离散(;)的分号才会被解析为占位符，维持数据中的分号。
--
-- 解析类型
-- * int
-- * bigint
-- * money (New)
-- * numeric
-- * float
-- * datetime
-- * varchar
-- * varbinary (New)
-- * uniqueidentifier (New)
-- * sql_variant
--
-- Limpire(昨夜小楼)
-------------------------------------------------------------------------------
CREATE PROCEDURE CS#
(
  @Input varchar(8000)
)
AS

SET NOCOUNT ON

declare @tid int,@cid int,@pos int,@num int,@p int,@s int,@tb sysname,@spt varchar(20),@inf varchar(2000),@col varchar(4000),@def sysname,@ent varchar(2),@sql varchar(8000)
declare @tables table(id int identity primary key,name sysname)
declare @data table(id int identity primary key,data varchar(8000),fc as nullif(left(nullif(data,''),charindex(' ',nullif(data,''))-1),'null'))
declare @temp table(id int primary key,temp varchar(4000))
declare @code table(id int primary key,code varchar(8000))

--> 格式整理
set @Input=replace(replace(replace(@Input collate Chinese_PRC_CS_AS_KS_WS,'　',' '),char(9),' '),char(13),char(10))
set @Input=char(10)+ltrim(rtrim(@Input))+char(10)
while charindex(char(10)+char(10),@Input)>0 set @Input=replace(@Input,char(10)+char(10),char(10))
set @Input=stuff(@Input,1,1,'')
if @Input not like '[@#$]%' and @Input not like '0[@#$]%' return
set @Input=replace(replace(@Input,'。','.'),'·','.')
select top 94 n=identity(tinyint,33,1) into # from syscolumns
select @Input=replace(@Input collate Chinese_PRC_CS_AS_KS_WS,nchar(n+65248),char(n)) from #
while charindex('  ',@Input)>0 set @Input=replace(@Input,'  ',' ')
set @Input=replace(@Input,char(10)+' ',char(10))
set @Input=replace(@Input,' '+char(10),char(10))
select @ent=char(13)+char(10),@pos=charindex(char(10),@Input)

--> 解析数据
while @Input<>''
  begin
    select @tb=left(@Input,@pos-1),@tid=isnull(@tid,0)+1,@Input=stuff(@Input,1,@pos,''),@pos=charindex(char(10),@Input)
    if @tb like '0%' set @tb=stuff(@tb,1,1,'')
    else select @inf=left(@Input,@pos-1)+' ',@Input=stuff(@Input,1,@pos,''),@pos=charindex(char(10),@Input)
    if @tb like '%/%' select @spt=stuff(@tb,1,charindex('/',@tb),''),@tb=left(@tb,charindex('/',@tb)-1)
    if @tb like '$%' set @tb=stuff(@tb,1,1,'')
    if @tb='' set @tb='T'
    if len(@tb)=1 set @tb=@tb+ltrim(@tid)
    if @tb like '[^@#]%' set @tb='['+@tb+']'
    insert @tables select @tb
    while @pos>0
      begin
        insert @data select left(@Input,@pos-1)+' '
        select @Input=stuff(@Input,1,@pos,''),@pos=charindex(char(10),@Input)
        if @Input like '[@#$]%' or @Input like '0[@#$]%' break
      end
    if @spt is not null
      begin
        set @inf=replace(@inf,@spt,' ')
        while charindex('  ',@inf)>0 set @inf=replace(@inf,'  ',' ')
        update @data set data=replace(data,@spt,' ')
        while exists (select 1 from @data where charindex('  ',data)>0) update @data set data=replace(data,'  ',' ') where charindex('  ',data)>0
      end
    delete @data where data not like '%[^ -]%'
    insert @code select id,null from @data
    while exists (select 1 from @data where data<>'')
      begin
        set @cid=isnull(@cid,0)+1
        insert @temp select id,case when fc not like '%[^;]%' then null else fc end from @data
        update @data set data=case when fc not like '%[^;]%' and len(fc)>1 then stuff(fc,1,1,'')+' ' else '' end+stuff(data,1,charindex(' ',data),'')
        if exists (select 1 from @temp a join @data b on a.id=b.id and temp is not null and fc is not null and isdate(temp)=1 and isdate(fc)=1) and not exists (select 1 from @temp a join @data b on a.id=b.id and temp is not null and fc is not null and isdate(temp+case when fc not like '%[^;]%' then '' else ' '+fc end)=0)
        begin
          update a set temp=temp+case when fc not like '%[^;]%' then '' else ' '+fc end from @temp a join @data b on a.id=b.id where temp is not null and fc is not null
          update @data set data=case when fc not like '%[^;]%' and len(fc)>1 then stuff(fc,1,1,'')+' ' else '' end+stuff(data,1,charindex(' ',data),'')
        end
        select @num=max(datalength(temp)) from @temp where temp is not null
        if @num is null set @def='sql_variant'
        else if not exists (select 1 from @temp where temp is not null and isnumeric(temp)=0)
          begin
            if exists (select 1 from @temp where (len(temp)>1 and temp like '0%' and temp not like '%[^0-9]%') or temp like '%,%') set @def=@num
            else if exists (select 1 from @temp where patindex('%[Ee]%',temp)>0) set @def='float'
            else if exists (select 1 from @temp where len(replace(replace(replace(temp,'+',''),'-',''),'.',''))>@@max_precision) set @def=@num
            else if exists (select 1 from @temp where temp like '%[^0-9.+-]%') set @def='money'
            else if exists (select 1 from @temp where temp like '%.%')
              begin
                select @p=max(len(parsename(n,2))),@s=max(len(parsename(n,1))) from (select n+case when n like '%.' then ' ' when n like '%.%' then '' else '. ' end n from (select replace(replace(temp,'+',''),'-','')n from @temp where temp is not null) a) b
                if @p+@s>@@max_precision set @def='float'
                else set @def='numeric('+ltrim(@p+@s)+','+ltrim(@s)+')'
              end
            else if exists (select 1 from @temp where temp is not null and isdate(temp)=0 or len(temp)<>8)
              begin
                if exists (select 1 from @temp where cast(temp as numeric(38,0)) not between -9223372036854775808 and 9223372036854775807) select @def='numeric('+ltrim(max(len(replace(replace(temp,'+',''),'-',''))))+',0)' from @temp where temp is not null
                else if exists (select 1 from @temp where cast(temp as numeric(38,0)) not between -2147483648 and 2147483647) set @def='bigint'
                else set @def='int'
              end
            else set @def='datetime'
          end
        else if not exists (select 1 from @temp where temp is not null and isdate(temp)=0) set @def='datetime'
        else if not exists (select 1 from @temp where temp not like '0x%' or temp like '0x%[^0-9a-f]%') select @def='varbinary('+ltrim((@num-3)/2+1)+')'
        else if not exists (select 1 from @temp where temp not like replicate('[0-9a-z]',8)+replicate('[-]'+replicate('[0-9a-z]',4),3)+'[-]'+replicate('[0-9a-z]',12)) set @def='uniqueidentifier'
        else set @def=@num
        if isnumeric(@def)=1 set @def='varchar('+@def+')'
        set @col=isnull(@col+',','')+'['+isnull(left(@inf,charindex(' ',@inf)-1),'C'+ltrim(@cid))+'] '+@def
        select @inf=nullif(stuff(@inf,1,charindex(' ',@inf),''),''),@def=left(@def+'(',charindex('(',@def+'(')-1)
        update a set code=isnull(code+',','select ')+case when @def in ('datetime','varchar','uniqueidentifier') then isnull(''''+temp+'''','null') else isnull(temp,'null') end from @code a join @temp b on a.id=b.id
        delete @temp
      end
    if @inf is not null
      begin
        update @code set code=code+replicate(',null',len(@inf)-len(replace(@inf,' ',''))+1)
        set @col=@col+',['+replace(rtrim(@inf),' ','] sql_variant,[')+'] sql_variant'
      end
    select @sql=isnull(@sql+' union all'+@ent,'')+code from @code
    print '--> 测试数据：'+@tb
    print case when @tb like '@%' then 'declare '+@tb+' table' else 'if object_id('''+case when @tb like '#%' then 'tempdb.dbo.' else '' end+@tb+''') is not null drop table '+@tb+@ent+'create table '+@tb end+'('+@col+')'+@ent+'insert '+@tb
    print @sql
    delete @data
    delete @code
    select @cid=null,@spt=null,@inf=null,@col=null,@sql=null
  end

--> 智能代码(略)
select @sql=isnull(@sql+@ent,@ent)+'select * from '+name from @tables
print @sql

SET NOCOUNT OFF

GO

---

加强版使用示例

--> 多表
exec cs# '@var/,
id data
1, 表变量@
2, 指定逗号为列分隔符
3  空格和TAB是默认列分隔符
#tmp
id data
-- -------
1  临时表#
2  忽略分隔横线
3  默认逗号不是列分隔符：a,b,c,d
$tab
id   data
---- -------
1    实体表$
2    保留null值
null null'
go

--> 无表名/无列名(表前缀@#$前用0修饰)
exec cs# '@
id data
1 无表名
0#tmp
2 无列名
0$
3 双无'
go

--> 动态列名：列名不足动态添加Cm...Cn，数据左边对齐原则。
exec cs# '@dynamic_column_name
a
11 12 13
   21 22
      31'
go

--> 动态填充：数据列不足，右向填充NULL值。
exec cs# '@dynamic_fill_null
a  b  c  d  e
11 12 13
   21 22
      31'
go

--> 占位符：控制数据的位置，左向填充NULL值。
exec cs# '@semicolon
a  b  c  d  e
11 12 13
;  21 22
;  ;  31'
go

--> 占位符增强特性：数据中的分号不会解析为占位符。
exec cs# '@semicolon
a b c d
; ;; 连续;;或离散;才会被解析为占位符'
go

--> 时间解析1
exec cs# '@time
id date boy
1 2001-1-1 Mark
2 20020101 John
3 23:15:39 Paul'
go

--> 时间解析2
exec cs# '@time
id date boy
1 2001-1-1 12:28:47 Mark
2 20020101 17:30:00 John
3 2003/1/1 23:15:39 Paul'
go

--> 基本上可以覆盖所有的原始测试数据形式
exec cs# '0@data_type
001 1,2,3 1.0E10 123456789012345678901234567890123456789 $9.99  9.99 9223372036854775808                    -9223372036854775808 -2147483648 20080101 2008-07-05 23:59:59.000 0x1234 681000D3-0E3A-49ED-8F59-80973A021B8E A
002 4,5,6 2.0E11 111111111111111111111111111111111111111 99.99 99.99 99999999999999999999999999999999999999  9223372036854775807  2147483647 20081231 2008-07-06 00:00:00.000 0xabcd 09E80A5C-7699-4BC9-AEF4-B0BCCF2E03E9 B'
go

/*
PS
表前缀$和美元符号$冲突，如美元货币类型出现在第一列会报错。
$99.9这种原始测试数据形式实属罕见，有需要者自行修改表前缀定义。
*/

Limpire

2008-07-07