Question

我正在尝试从SQL Server 2014数据库的文本字段中提取所有“＃”注释。我正在使用我在MSDN thread找到的代码在一条记录中找到多个主题标签并稍微修改它以满足我的需要，但我在结果集中看到了一些意想不到的结果。

到目前为止我的T-SQL：

IF OBJECT_ID('tempdb..#hashtag') IS NOT NULL DROP TABLE #hashtag;

IF OBJECT_ID('tempdb..#numbers') IS NOT NULL DROP TABLE #numbers;

IF OBJECT_ID('tempdb..#hashtagcounts') IS NOT NULL DROP TABLE #hashtagcounts;

CREATE TABLE #numbers ( N INT );
DECLARE @i INT; 
SET @i = 1;
WHILE @i < 500
BEGIN
    INSERT  #numbers
    VALUES  ( @i );
    SET @i = @i + 1;
END;

CREATE TABLE #hashtag ( tkt_desc VARCHAR(MAX) );

INSERT INTO #hashtag ( tkt_desc )
SELECT  cst.ticketDescription
FROM    dbo.Trending AS [cst]
    WHERE   cst.ticketDescription LIKE '%#%'
            AND LTRIM(RTRIM(cst.ticketDescription)) NOT LIKE '%# %'
            AND cst.ticketDescription NOT LIKE '%BATCH #%'
            AND cst.ticketDescription NOT LIKE '%#[.:''1-9]%';

SELECT  LOWER ('#' + SUBSTRING(tkt_desc, N,
              CASE WHEN CHARINDEX(' ', tkt_desc, N) > 0
                   THEN CHARINDEX(' ', tkt_desc, N) - N
                   ELSE LEN(tkt_desc)
              END)) AS tkt_desc
INTO #hashtagcounts
FROM    #hashtag
    CROSS JOIN #numbers
WHERE   N <= LEN(tkt_desc) AND SUBSTRING(tkt_desc, N - 1, 1) = '#';

SELECT tkt_desc, COUNT(*) AS [Count]
FROM #hashtagcounts
GROUP BY tkt_desc
ORDER BY Count DESC, tkt_desc;

我的数据集如下：

tkt_desc              Count
#updateinfo           6
#update               4
#update update        3
#update updated       3
#reprint              2
#callback             1
#nochargereprint      1
#nocostreprint        1
#notes update         1
#paperlicense please  1

我遇到的问题是#update标签以三种不同的方式表示。理想情况下，我不希望在主题标签中包含空格后的任何内容 - 并且#update标记应该看到10的计数。

我最初的想法是，由于这是一个文本字段，因此字段中可能会有换行符或回车符，因此我尝试将SELECT cst.ticketDescription替换为REPLACE(REPLACE(cst.ticketDescription,CHAR(13),''),CHAR(10),'')来处理此问题，但这只是将单独的单词分成一个标签。见下面的例子：

#updateinfo    6
#update        4
#updateupdate  3
#updateupdated 3

有关如何达到理想结果的任何建议？我会在下面提供一些示例数据，以防有人想要进行实验。

查询我决定使用两个函数/交叉应用并清理LIKE和NOT LIKE语句：

SELECT  '#' + LOWER(B.RetVal) AS [HashTag] ,
    COUNT(*) AS [Cnt]
FROM    dbo.Common_SupportTickets AS [cst]
    CROSS APPLY [dbo].[udf-Str-Extract]([dbo].[udf-Str-Strip-Control](cst.ticketDescription) + ' ', '#', ' ') AS [B]
WHERE   cst.ticketDescription LIKE '%#%'
    AND LTRIM(RTRIM(REPLACE(cst.ticketDescription,CHAR(13)+CHAR(10),''))) NOT LIKE '%#'
    AND cst.ticketDescription NOT LIKE '%BATCH #%'
    AND cst.ticketDescription NOT LIKE '%#[.:'')1-9]%'
    AND CAST(cst.createDate AS DATE) >= CAST( @paramStartDate AS DATE )
    AND CAST(cst.createDate AS DATE) <= CAST( @paramEndDate AS DATE )
GROUP BY B.RetVal

示例数据和记录：

USE [Sandbox]
GO

SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
CREATE TABLE [dbo].[Trending](
[TicketDescription] [varchar](max) NULL
) ON [PRIMARY] TEXTIMAGE_ON [PRIMARY]

GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#registrationnotreceived     customer has not received registration for boat...')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#registrationnoreceived      Customer called and still has not received duplicate registration...')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#callback (111) 111-1111

Agent''s POS is briefly turning on before "going to sleep" and entering sleep mode. Agent claims POS will not stay active for any length of time.')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#Update
Corrected last name and driver''s license number.')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#Update
Update customer''s last name.')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#Update
Update last name, address')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#update - Profile updated. Corrected last name.')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#updateinfo')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#updateinfo')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#updateinfo')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#updateinfo')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#question')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#Update
Update residency status')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#update #SSNdiscrepancy 
John Doe called in claiming this was their SSN, please advise. Please contact John Doe at this number (111-111-2222) when the issue is resolved. He wishes to create an account once the issue is resolved.')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#Notes
Update Customer''s Hunter certificate number')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#update
Updated residency status')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#Update
Changed residency from in-state to out-of-state, likely didn''t update.')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#Update
Updated Customer''s last name')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#updateinfo')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#updateinfo')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#void - Agent called in asking to void a duplicate license sale.')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#Update - updated customer''s last name')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#permissions
Changed agent role from AGENT CLERK to AGENT MANAGER in order to order supplies.')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#registrationreprint customer didn''t receive registration I sent to Twra It for reprint. Told to call if he has not received in 10days ')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#printerissue')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#update - Profile updated. Religious Exempt.')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#reprint   this is 2nd call from customer that they have not received there boat registration...')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#registratedincorrectly  He send in check and info from Clerks office beginning of Dec, ')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#reprint #paperlicense Please reprint this license for the customer, he claims he has not yet received it. ')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#nocostreprint customer did not receive boat regst')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#Update
Updated customer''s address over the phone')
GO
INSERT [dbo].[Trending] ([TicketDescription]) VALUES (N'#nochargereprint  ')
GO

当我为不同的日期范围运行相同的脚本时，我注意到了一个奇怪的行为 - 它正在创建不存在的#hashtags。例如，它返回Order：10次，但#order：

字段中没有文本

示例数据：

CREATE TABLE #temptable ( [ticketDescription] varchar(max), [RetVal] varchar(max) )
INSERT INTO #temptable
VALUES
( 'DURABLE HARD CARD RETURN-WAS GOING TO CALL TO PICK UP  NO# NUMBER ', 'DURABLE' ), 
( 'Order: 30341143OrderItemId: 30517890License for: NATHAN TIMOTHY SUMNER  Printed on: 10 Apr 2017 06:43:57:857 Shipped to: 7650 KIOWA ST, APT #01 MILLINGTON TN 38053 - 3219', 'Order:' ), 
( 'Order: 30341143OrderItemId: 30517890License for: NATHAN TIMOTHY SUMNER  Printed on: 10 Apr 2017 06:43:57:857 Shipped to: 7650 KIOWA ST, APT #01 MILLINGTON TN 38053 - 3219', '01' ), 
( 'Order: 30346281OrderItemId: 30526511License for: STANLEY R ROWLAND  Printed on: 07 Apr 2017 06:22:23:417 Shipped to: 25 COUNTRY WOOD LN # 601 WALNUT MS 38683 - 5367', 'Order:' ), 
( 'Order: 30347906OrderItemId: 30529325License for: DOUGLAS R EASTRIDGE  Printed on: 07 Apr 2017 09:29:51:643 Shipped to: 7980 HUFFS FERRY RD N # R LOUDON TN 37774 - 5910', 'Order:' ), 
( 'Order: 30361947OrderItemId: 30554547License for: BLAKE R HADDON  Printed on: 07 Apr 2017 14:01:42:637 Shipped to: 146 14TH AVE NW # 146 WINCHESTER TN 37398 - 1079', 'Order:' ), 
( 'Order: 30362075OrderItemId: 30554740License for: KYLE  JACKSON  Printed on: 07 Apr 2017 14:04:23:473 Shipped to: 3765 E ANDREW JOHNSON HWY, APT # A4 MORRISTOWN TN 37814 - 6200', 'Order:' ), 
( 'Order: 30369152OrderItemId: 30565137License for: DANIEL JAMES SOLA  Printed on: 09 Apr 2017 07:08:28:683 Shipped to: 3212 MAYES LOOP RD # 1 PIGEON FORGE TN 37863 - 7721', 'Order:' ), 
( 'Order: 30370097OrderItemId: 30566543License for: JAMES D JOHNSON  Printed on: 09 Apr 2017 11:51:37:170 Shipped to: 230 FRANKLIN RD # 907 FRANKLIN TN 37064 - 2256', 'Order:' ), 
( 'Order: 30372876OrderItemId: 30571083License for: MARCOS  CLAUDIO P POLONIATO  Printed on: 06 Apr 2017 08:15:01:093 Shipped to: 295 WINDING RIVER DR, # J SANDY SPRINGS GA 30350 - 1926', 'Order:' ), 
( 'Order: 30396415OrderItemId: 30604206License for: GARY T GOODMAN  Printed on: 07 Apr 2017 15:11:20:317 Shipped to: 1046 GREENBRIAR RD # RD.423 TALBOTT TN 37877 - 9055', 'Order:' ), 
( 'Order: 30405689OrderItemId: 30617970License for: VANCE K JOHNSON  Printed on: 09 Apr 2017 09:25:48:670 Shipped to: 614 GARRISON HOLLOW RD, LOT # 11 ELIZABETHTON TN 37643 - 4897', 'Order:' )

SELECT * FROM #temptable;

DROP TABLE #temptable

编辑 - 修复

Select HashTag = '#'+B.RetVal
      ,Cnt      = Count(*)
 From  #temptable A
 Cross Apply [dbo].[udf-Str-Extract]([dbo].[udf-Str-Strip-Control]('#>>> '+A.TicketDescription)+' ','#',' ') B
 Where B.RetVal <> '>>>'
   and B.RetVal Not Like '[0-9][0-9]'
 Group By B.RetVal
 Order By 1

Answer 1

在两个函数和一个CROSS APPLY的帮助下，以下内容可能会有所帮助。

第一个函数将删除所有控制字符并替换为空格，以免结合John{13}{10}Smith之类的单词将返回为John Smith

第二个函数是一个修改后的解析函数，用于接受两个非类似分隔符（开始/结束）。在你的情况下＃和空格。 如果有多个主题标签，则会返回多条记录。

示例

Select HashTag = '#'+B.RetVal ,Cnt = Count(*) From trending A Cross Apply [dbo].[udf-Str-Extract]([dbo].[udf-Str-Strip-Control](A.TicketDescription)+' ','#',' ') B Group By B.RetVal Order By 1

<强>返回

HashTag Cnt #callback 1 #nochargereprint 1 #nocostreprint 1 #Notes 1 #paperlicense 1 ---<< 2nd hashtag in text #permissions 1 #printerissue 1 #question 1 #registratedincorrectly 1 #registrationnoreceived 1 #registrationnotreceived1 #registrationreprint 1 #reprint 2 #SSNdiscrepancy 1 ---<< 2nd hashtag in text #Update 12 #updateinfo 6 #void 1

如果感兴趣的是UDF

CREATE FUNCTION [dbo].[udf-Str-Strip-Control](@S varchar(max)) Returns varchar(max) Begin ;with cte1(N) As (Select 1 From (Values(1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N(N)), cte2(C) As (Select Top (32) Char(Row_Number() over (Order By (Select NULL))-1) From cte1 a,cte1 b) Select @S = Replace(@S,C,' ') From cte2 Return LTrim(RTrim(Replace(Replace(Replace(@S,' ','><'),'<>',''),'><',' '))) End --Select [dbo].[udf-Str-Strip-Control]('Michael '+char(13)+char(10)+'LastName') --Returns: Michael LastName

CREATE FUNCTION [dbo].[udf-Str-Extract] (@String varchar(max),@Delimiter1 varchar(100),@Delimiter2 varchar(100)) Returns Table As Return ( with cte1(N) As (Select 1 From (Values(1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) N(N)), cte2(N) As (Select Top (IsNull(DataLength(@String),0)) Row_Number() over (Order By (Select NULL)) From (Select N=1 From cte1 N1,cte1 N2,cte1 N3,cte1 N4,cte1 N5,cte1 N6) A ), cte3(N) As (Select 1 Union All Select t.N+DataLength(@Delimiter1) From cte2 t Where Substring(@String,t.N,DataLength(@Delimiter1)) = @Delimiter1), cte4(N,L) As (Select S.N,IsNull(NullIf(CharIndex(@Delimiter1,@String,s.N),0)-S.N,8000) From cte3 S) Select RetSeq = Row_Number() over (Order By N) ,RetPos = N ,RetVal = left(RetVal,charindex(@Delimiter2,RetVal)-1) From (Select *,RetVal = Substring(@String, N, L) From cte4) A Where charindex(@Delimiter2,RetVal)>1 ) /* Max Length of String 1MM characters Declare @String varchar(max) = 'Dear [[FirstName]] [[LastName]], ...' Select * From [dbo].[udf-Str-Extract] (@String,'[[',']]') */

编辑 - 可以帮助实现可视化

如果您运行查询而没有任何聚合

Select A.*,B.* From trending A Cross Apply [dbo].[udf-Str-Extract]([dbo].[udf-Str-Strip-Control](' '+replace(A.TicketDescription,'#','|||#'))+' .','|||',' ') B Order By 1

你会得到

从SQL Server记录

1 个答案: