Question

我有一个包含设置列的表，它们的格式如下：

SETTING_NAME = setting_value | SETTING_NAME = setting_value | SETTING_NAME = setting_value

问题在于，填充了哪些设置会有很大差异。我想拆分所有值并以更好的方式存储它们。

目前它看起来像这样：

我希望如此：

为了达到目的，我使用了一个函数来分割值。然后我将它们组合在一起并使用子字符串来获取属于setting_name的setting_value。这是我到目前为止所得到的：

/*
create function [dbo].[split_to_columns](@text varchar(8000)
                                            , @column tinyint
                                            , @separator char(1))
returns varchar(8000)
as
  begin
       declare @pos_start int = 1
       declare @pos_end int = charindex(@separator, @text, @pos_start)
       while (@column > 1 and @pos_end > 0)
         begin
             set @pos_start = @pos_end + 1
             set @pos_end = charindex(@separator, @text, @pos_start)
             set @column = @column - 1
         end 
       if @column > 1  set @pos_start = len(@text) + 1
       if @pos_end = 0 set @pos_end = len(@text) + 1 
       return substring(@text, @pos_start, @pos_end - @pos_start)
  end
*/

create table #settings(id int, setting varchar(255))
 insert into #settings(id, setting) values(1,'setting1=a|setting2=b|setting3=c')
 insert into #settings(id, setting) values(2,'setting1=d|setting2=e')
 insert into #settings(id, setting) values(3,'setting1=f|setting3=g')
 insert into #settings(id, setting) values(4,'setting2=h')

;
with cte as ( 
      select id, dbo.split_to_columns(setting, 1, '|') as setting from #settings
union select id, dbo.split_to_columns(setting, 2, '|') from #settings
union select id, dbo.split_to_columns(setting, 3, '|') from #settings
)
select distinct 
       x.id
     , (select substring(setting, charindex('=', setting) + 1, 255) from cte where setting like 'setting1=%' and id = x.id) as setting1 
     , (select substring(setting, charindex('=', setting) + 1, 255) from cte where setting like 'setting2=%' and id = x.id) as setting2  
     , (select substring(setting, charindex('=', setting) + 1, 255) from cte where setting like 'setting3=%' and id = x.id) as setting3 
  from cte x

drop table #settings

我是以正确的方式做到这一点的吗？我不禁想到自己太复杂了。虽然我不是现在设置格式的忠实粉丝，但我确实经常看到它。这意味着更多的人必须做这个伎俩...

修改

我将图片属性导入数据库。上面提到的设置是图片属性，id是设置所属图片的名称。

一栏中的设置示例：

全名= d：\ 8.JPG |文件夹名称= d：\ |文件名= 8.JPG |大小= 7284351 |扩展= .JPG | datePictureTaken = 2017年10月3日 11：53：38 | ApertureValue = 2 |日期时间= 2017年10月3日 11：53：38 |数字化日期时间= 2017年10月3日 11：53：38 |原始日期时间= 2017年10月3日 11：53：38 | ExposureTime = 0,0025706940874036 |长焦点= 3,65 | GPSAltitude = 43 | GPSDateStamp = 2017年10月3日 0：00：00 |型号= qcam的-AA | ShutterSpeedValue = 8604

这就是我希望以上述方式进行格式化的原因。

Answer 1

我会将文本转换为XML的基本块，以便我们可以采用基于集合的方法将数据转换为您想要的结果：

declare @settings table(id int, setting varchar(255))
insert into @settings (id,setting) values
(1,'setting1=a|setting2=b|setting3=c'),
(2,'setting1=d|setting2=e'),
(3,'setting1=f|setting3=g'),
(4,'setting2=h')

;with Xmlised (id,detail) as (
    select id,CONVERT(xml,'<prob><setting name="' +
        REPLACE(
            REPLACE(setting,'=','">'),
        '|','</setting><setting name="') + '</setting></prob>')
    from @settings
), shredded as (
    select
        x.id,
        S.value('./@name','varchar(50)') as name,
        S.value('./text()[1]','varchar(100)') as value
    from
        Xmlised x
            cross apply
        detail.nodes('prob/setting') as T(S)
)
select
    id,setting1,setting2,setting3
from
    shredded
        pivot (MAX(value) for name in (setting1,setting2,setting3)) u

希望我已经把它分解成足够的步骤，你可以看到它正在做什么以及如何做。

结果：

id          setting1  setting2  setting3
----------- --------- --------- ---------
1           a         b         c
2           d         e         NULL
3           f         NULL      g
4           NULL      h         NULL

正如Sean在评论中所建议的那样，我通常不会考虑存储枢轴结果，并且通常会跳过该步骤

Answer 2

WITH很慢。我建议存储设置名称，值和某种组ID的表。例如：

CREATE TABLE [dbo].[settings_table](
    [id] [int] NULL,
    [group] [int] NULL,
    [name] [nchar](10) NULL,
    [value] [nchar](10) NOT NULL
) ON [PRIMARY]

我不确切知道你的程序在这些设置中做了什么，但从长远来看这种结构会更有效。

Answer 3

我会做以下3个步骤：

1）创建一个通用的Split函数。这是我使用的那个：

CREATE FUNCTION Split(
     @StringToSplit VARCHAR(MAX)
    ,@Delimiter VARCHAR(10)
) 

RETURNS @SplitResult TABLE (id int, item VARCHAR(MAX))
BEGIN

    DECLARE @item VARCHAR(8000)
    DECLARE @counter int = 1

    WHILE CHARINDEX(@Delimiter, @StringToSplit,0) <> 0
    BEGIN

        SELECT
            @item = RTRIM(LTRIM(SUBSTRING(@StringToSplit,1, CHARINDEX(@Delimiter,@StringToSplit,0)-1))),
            @StringToSplit = RTRIM(LTRIM(SUBSTRING(@StringToSplit, CHARINDEX(@Delimiter,@StringToSplit,0) + LEN(@Delimiter), LEN(@StringToSplit))))

        IF LEN(@item) > 0
            INSERT INTO @SplitResult SELECT @counter, @item

        SET @counter = @counter + 1
    END

    IF LEN(@StringToSplit) > 0

        INSERT INTO @SplitResult SELECT @counter,@StringToSplit

    SET @counter = @counter + 1

    RETURN
END
GO

-- You use it like this
SELECT S.id, T.item FROM #settings AS S CROSS APPLY Split(S.setting, '|') AS T

2）拆分设置并将设置名称与其值分开。

SELECT 
    S.id, 
    T.item,
    SettingName = SUBSTRING(T.item, 1, CHARINDEX('=', T.item, 1) - 1), -- -1 to not include the "="
    SettingValue = SUBSTRING(T.item, CHARINDEX('=', T.item, 1) + 1, 100) -- +1 to not include the "="
FROM 
    #settings AS S 
    CROSS APPLY Split(S.setting, '|') AS T

3）按名称旋转已知设置：

;WITH SplitValues AS
(
    SELECT 
        S.id, 
        SettingName = SUBSTRING(T.item, 1, CHARINDEX('=', T.item, 1) - 1), -- -1 to not include the "="
        SettingValue = SUBSTRING(T.item, CHARINDEX('=', T.item, 1) + 1, 100) -- +1 to not include the "="
    FROM 
        #settings AS S 
        CROSS APPLY Split(S.setting, '|') AS T
)
SELECT
    P.id,
    P.setting1,
    P.setting2,
    P.setting3
FROM
    SplitValues AS S
    PIVOT (
        MAX(S.SettingValue) FOR SettingName IN ([setting1], [setting2], [setting3])
    ) AS P

Answer 4

对于设置列（照片属性），我同意连续列中的列使用正确的类型，例如DateTime，Int，Numeric，因为您可以搜索范围，排序，它更有效。

我知道你要求SQL，但我会在.NET中这样做，因为你需要做一些清理，比如从整数中删除逗号。在现实生活中从文件中读取行，这样就可以打开命令（插入）。

public static void ParsePhoto(string photo)
{
    if(string.IsNullOrEmpty(photo))
    {
        photo = @"FullName = D:\8.jpg | FolderName = D:\| FileName = 8.jpg | Size = 7284351 | Extension =.jpg | datePictureTaken = 10 - 3 - 2017 11:53:38 | ApertureValue = 2 | DateTime = 10 - 3 - 2017 11:53:38 | DateTimeDigitized = 10 - 3 - 2017 11:53:38 | DateTimeOriginal = 10 - 3 - 2017 11:53:38 | ExposureTime = 0,0025706940874036 | FocalLength = 3,65 | GPSAltitude = 43 | GPSDateStamp = 10 - 3 - 2017 0:00:00 | Model = QCAM - AA | ShutterSpeedValue = 8,604";
    }
    List<KeyValuePair<string, string>> kvp = new List<KeyValuePair<string, string>>();
    foreach(string s in photo.Trim().Split(new char[] {'|'}, StringSplitOptions.RemoveEmptyEntries))
    {
        string[] sp = s.Split(new char[] { '=' }, StringSplitOptions.RemoveEmptyEntries);
        if (sp.Count() == 2)
        {
            kvp.Add(new KeyValuePair<string, string>(sp[0].Trim(), sp[1].Trim()));
        }
        else
        {
            throw new IndexOutOfRangeException("bad photo");
        }
    }
    foreach(KeyValuePair<string, string> pair in kvp)
    {
        Debug.WriteLine($"{pair.Key} = {pair.Value}");
        //build up and execute insert statement here
    }
    Debug.WriteLine("Done");
}

FullName = D:\8.jpg
FolderName = D:\
FileName = 8.jpg
Size = 7284351
Extension = .jpg
datePictureTaken = 10 - 3 - 2017 11:53:38
ApertureValue = 2
DateTime = 10 - 3 - 2017 11:53:38
DateTimeDigitized = 10 - 3 - 2017 11:53:38
DateTimeOriginal = 10 - 3 - 2017 11:53:38
ExposureTime = 0,0025706940874036
FocalLength = 3,65
GPSAltitude = 43
GPSDateStamp = 10 - 3 - 2017 0:00:00
Model = QCAM - AA
ShutterSpeedValue = 8,604

Answer 5

如果性能很重要，您可以轻松完成此操作，无需拆分器功能，将数据转换为XML或进行任何转动。这种技术通常被称为Cascading CROSS APPLY。代码有点冗长，但性能回报令人惊讶。首先是解决方案：

SELECT 
  id, 
  setting1 = substring(setting, s1.p+1, x1.x),
  setting2 = substring(setting, s2.p+1, x2.x),
  setting3 = substring(setting, s3.p+1, x3.x)
FROM #settings t
CROSS APPLY (VALUES (nullif(charindex('setting1=', t.setting),0)+8)) s1(p)
CROSS APPLY (VALUES (nullif(charindex('setting2=', t.setting),0)+8)) s2(p)
CROSS APPLY (VALUES (nullif(charindex('setting3=', t.setting),0)+8)) s3(p)
CROSS APPLY (VALUES (isnull(nullif(charindex('|',t.setting,s1.p),0)-s1.p-1, 1))) x1(x)
CROSS APPLY (VALUES (isnull(nullif(charindex('|',t.setting,s2.p),0)-s2.p-1, 1))) x2(x)
CROSS APPLY (VALUES (isnull(nullif(charindex('|',t.setting,s3.p),0)-s3.p-1, 1))) x3(x);

请注意执行计划：

我没有时间进行性能测试，但根据执行计划，级联交叉应用技术的速度大约快44,000倍。

Answer 6

试试这个：

declare @table table (id int, setting varchar(100))
insert into @table values
(1,'setting1=a|setting2=b|setting3=c'),
(2,'setting1=d|setting2=e'),
(3,'setting1=f|setting3=g'),
(4,'setting2=h')

select id, 
       case when charindex('setting1=',setting) = 0 then null else SUBSTRING(setting, charindex('setting1=',setting) + 9, 1) end [setting1],
       case when charindex('setting2=',setting) = 0 then null else SUBSTRING(setting, charindex('setting2=',setting) + 9, 1) end [setting2],
       case when charindex('setting3=',setting) = 0 then null else SUBSTRING(setting, charindex('setting3=',setting) + 9, 1) end [setting3]
from @table

多个值存储在一列中

6 个答案: