时间戳

首页 > 解决方案 > T-SQL FTS 包含单个字符“n”匹配数字

tsql - T-SQL FTS 包含单个字符“n”匹配数字

问题描述

我正在使用 SQL Server 2016 并面临一个非常奇怪的行为。

我有一个为 FTS 编制索引的表。

像这样按单个字符搜索时，... CONTAINS(N'"a*"')它会返回所有包含以a- 开头的单词的记录。

但是，如果我像这样按单个字符n进行搜索，它不仅... CONTAINS(N'"n*"')会返回包含单词的记录，而且还会返回包含数字的记录。n

我不能谷歌没有提到这种行为。我相信这是一种通配符，但文档中对此没有任何内容。

为什么会这样？有转义序列还是什么？

UPD：测试环境

CREATE FULLTEXT CATALOG [ContentCatalog] WITH ACCENT_SENSITIVITY = ON
GO
CREATE TABLE [dbo].[Content]([id] [uniqueidentifier] NOT NULL, [content] [nvarchar](max) NOT NULL,
  CONSTRAINT [PK_Content] PRIMARY KEY CLUSTERED ([id] ASC))
GO
ALTER TABLE [dbo].[Content] ADD CONSTRAINT [DF_Content_id]  DEFAULT (newid()) FOR [id]
GO
INSERT [Content] ([content]) VALUES
    (N'abc def nop'),
    (N'zxc asd qwe'),
    (N'zaq nhy bgt'),
    (N'123 456 789'),
    (N'z12 x34 c56')
GO
CREATE FULLTEXT INDEX ON [Content]([content] LANGUAGE 'English')
KEY INDEX [PK_Content]ON ([ContentCatalog], FILEGROUP [PRIMARY])
WITH (CHANGE_TRACKING = AUTO, STOPLIST = SYSTEM)

SELECT * FROM [dbo].[Content]
SELECT FULLTEXTSERVICEPROPERTY('IsFullTextInstalled') -- should be 1
SELECT * FROM [dbo].[Content] WHERE CONTAINS([Content], N'"a*"') -- returns 2 records
SELECT * FROM [dbo].[Content] WHERE CONTAINS([Content], N'"n*"') -- returns extra record with digits only

标签： tsqlfull-text-searchcontainssql-server-2016

解决方案

这不是我知道的违反规则的答案，但我认为它添加了有用的附加信息，OP 可能希望将这些信息纳入他们的问题中。

有关更多信息，这似乎仅适用n于contains. 下面使用问题中的脚本来构建环境，但使用不同的测试值。然后select将包含返回的所有值整理到一行中以便于比较：

insert [content] ([content]) values
    (N'abc nop abc'),
    (N'abc nno abc'),
    (N'abc nnn abc'),
    (N'abc nnnnbc'),
    (N'abc nnnnnc'),
    (N'1 2 3'),
    (N'10 11 12'),
    (N'100 101 102'),
    (N'1000 1001 1002'),
    (N'10000 10001 10002')
go
          select 'a*' as ContainsSearch,    stuff((select ' || ' + content from [dbo].[content] where contains([content], N'"a*"') for xml path('')),1,4,'') as ReturnedValues
union all select 'n*' as ContainsSearch,    stuff((select ' || ' + content from [dbo].[content] where contains([content], N'"n*"') for xml path('')),1,4,'') as ReturnedValues
union all select 'nn*' as ContainsSearch,   stuff((select ' || ' + content from [dbo].[content] where contains([content], N'"nn*"') for xml path('')),1,4,'') as ReturnedValues
union all select 'nnn*' as ContainsSearch,  stuff((select ' || ' + content from [dbo].[content] where contains([content], N'"nnn*"') for xml path('')),1,4,'') as ReturnedValues
union all select 'nnnn*' as ContainsSearch, stuff((select ' || ' + content from [dbo].[content] where contains([content], N'"nnnn*"') for xml path('')),1,4,'') as ReturnedValues
union all select 'nnnnn*' as ContainsSearch,stuff((select ' || ' + content from [dbo].[content] where contains([content], N'"nnnnn*"') for xml path('')),1,4,'') as ReturnedValues

结果是：

+----------------+-----------------------------------------------------------------------------------------------------------------------------------------+
| ContainsSearch |                                                             ReturnedValues                                                              |
+----------------+-----------------------------------------------------------------------------------------------------------------------------------------+
| a*             | abc nop abc || abc nno abc || abc nnn abc || abc nnnnbc || abc nnnnnc                                                                   |
| n*             | abc nop abc || abc nno abc || abc nnn abc || abc nnnnbc || abc nnnnnc || 10 11 12 || 100 101 102 || 1000 1001 1002 || 10000 10001 10002 |
| nn*            | abc nno abc || abc nnn abc || abc nnnnbc || abc nnnnnc || 10 11 12 || 100 101 102 || 1000 1001 1002 || 10000 10001 10002                |
| nnn*           | abc nnn abc || abc nnnnbc || abc nnnnnc                                                                                                 |
| nnnn*          | abc nnnnbc || abc nnnnnc                                                                                                                |
| nnnnn*         | abc nnnnnc                                                                                                                              |
+----------------+-----------------------------------------------------------------------------------------------------------------------------------------+

推荐阅读