Question

我有一张文件表和一张标签表。文档标有各种值。

我正在尝试创建这些标记的搜索，并且大多数情况下它都在工作。但是，当我匹配任何标签时，我会收到额外的结果。我只想要匹配所有标签的结果。

我创建了这个来说明问题http://sqlfiddle.com/#!3/8b98e/11

表格和数据：

CREATE TABLE Documents
(
 DocId INT,
 DocText VARCHAR(500)
);

CREATE TABLE Tags
(
  TagId INT,
  TagName VARCHAR(50)
);

CREATE TABLE DocumentTags
(
  DocTagId INT,
  DocId INT,
  TagId INT,
  Value VARCHAR(50)
);

INSERT INTO Documents VALUES (1, 'Document 1 Text');
INSERT INTO Documents VALUES (2, 'Document 2 Text');

INSERT INTO Tags VALUES (1, 'Tag Name 1');
INSERT INTO Tags VALUES (2, 'Tag Name 2');

INSERT INTO DocumentTags VALUES (1, 1, 1, 'Value 1');
INSERT INTO DocumentTags VALUES (1, 1, 2, 'Value 2');
INSERT INTO DocumentTags VALUES (1, 2, 1, 'Value 1');

代码：

-- Set up the parameters
DECLARE @TagXml VARCHAR(max)
SET @TagXml = '<tags>
                  <tag>
                    <description>Tag Name 1</description>
                    <value>Value 1</value>
                  </tag>
                  <tag>
                    <description>Tag Name 2</description>
                    <value>Value 2</value>
                  </tag>
                </tags>'

-- Create a table to store the parsed xml in
DECLARE @XmlTagData TABLE 
(
    id varchar(20)
    ,[description] varchar(100)
    ,value varchar(250)
)

-- Populate our XML table
DECLARE @iTag int
EXEC sp_xml_preparedocument @iTag OUTPUT, @TagXml
-- Execute a SELECT statement that uses the OPENXML rowset provider
-- to produce a table from our xml structure and insert it into our temp table
INSERT INTO @XmlTagData (id, [description], value)
SELECT  id, [description], value
FROM    OPENXML (@iTag, '/tags/tag',1)
        WITH (id varchar(20),
                [description] varchar(100) 'description',
                value varchar(250) 'value')

EXECUTE sp_xml_removedocument @iTag

-- Update the XML table Id's to match existsing Tag Id's
UPDATE      @XmlTagData
SET         X.Id = T.TagId
FROM        @XmlTagData X
INNER JOIN  Tags T ON X.[description] = T.TagName

-- Check it looks right
--SELECT * 
--FROM @XmlTagData

-- This is where things do not quite work. I get both doc 1 & 2 back, 
-- but what I want is just document 1.
-- i.e. documents that have both tags with matching values
SELECT DISTINCT D.*
FROM Documents D
INNER JOIN DocumentTags T ON T.DocId = D.DocId
INNER JOIN @XmlTagData X ON X.id = T.TagId AND X.value = T.Value

（注意我不是DBA，所以可能有更好的做事方式。希望我走在正确的轨道上，但如果我的实施可以改进，我愿意接受其他建议。）

任何人都可以提供有关如何仅获取包含所有标记的结果的任何建议吗？

非常感谢。

Answer 1

在上一次查询中使用[NOT] EXISTS和EXCEPT运算符选项

SELECT *
FROM Documents D
WHERE NOT EXISTS (
                  SELECT X.ID , X.Value
                  FROM @XmlTagData X 
                  EXCEPT
                  SELECT T.TagId, T.VALUE
                  FROM DocumentTags T
                  WHERE T.DocId = D.DocId
                  )

SQLFiddle上的演示

OR

SELECT *
FROM Documents D
WHERE EXISTS (
              SELECT X.ID , X.Value
              FROM @XmlTagData X 
              EXCEPT
              SELECT T.TagId, T.VALUE
              FROM DocumentTags T
              WHERE T.DocId != D.DocId
              )

SQLFiddle上的演示

OR

此外，您可以使用XQuery方法的简单解决方案：nodes()，value()）和CTE /子查询。

-- Set up the parameters
DECLARE @TagXml XML
SET @TagXml = '<tags>
                  <tag>
                    <description>Tag Name 1</description>
                    <value>Value 1</value>
                  </tag>
                  <tag>
                    <description>Tag Name 2</description>
                    <value>Value 2</value>
                  </tag>              
                </tags>'               


;WITH cte AS
 (
  SELECT TagValue.value('(./value)[1]', 'nvarchar(100)') AS value,
         TagValue.value('(./description)[1]', 'nvarchar(100)') AS [description]       
  FROM @TagXml.nodes('/tags/tag') AS T(TagValue)
  )
  SELECT *
  FROM Documents D
  WHERE NOT EXISTS (
                    SELECT T.TagId, c.value
                    FROM cte c JOIN Tags T WITH(FORCESEEK) 
                      ON c.[description] = T.TagName
                    EXCEPT
                    SELECT T.TagId, T.VALUE
                    FROM DocumentTags T WITH(FORCESEEK)
                    WHERE T.DocId = D.DocId                          
                    )

SQLFiddle上的演示

OR

-- Set up the parameters
DECLARE @TagXml XML
SET @TagXml = '<tags>
                  <tag>
                    <description>Tag Name 1</description>
                    <value>Value 1</value>
                  </tag>
                  <tag>
                    <description>Tag Name 2</description>
                    <value>Value 2</value>
                  </tag>              
                </tags>'      

  SELECT *
  FROM Documents D
  WHERE NOT EXISTS (
                    SELECT T2.TagId,
                           TagValue.value('(./value)[1]', 'nvarchar(100)') AS value                           
                    FROM @TagXml.nodes('/tags/tag') AS T(TagValue)
                      JOIN Tags T2 WITH(FORCESEEK)
                        ON TagValue.value('(./description)[1]', 'nvarchar(100)') = T2.TagName                                        
                    EXCEPT
                    SELECT T.TagId, T.VALUE
                    FROM DocumentTags T WITH(FORCESEEK)
                    WHERE T.DocId = D.DocId                       
                    )

SQLFiddle上的演示

为了提高性能（在Tags和DocumentTags表上强制操作索引查找），使用索引和表提示（FORCESEEK提示添加到上面的查询中）：

CREATE INDEX x ON Documents(DocId) INCLUDE(DocText)
CREATE INDEX x ON Tags(TagName) INCLUDE(TagId)
CREATE INDEX x ON DocumentTags(DocId) INCLUDE(TagID, VALUE)

Answer 2

我不确定SQL Server的语法，但我觉得这样的事情应该有效

SELECT d.docId
FROM Documents D
INNER JOIN DocumentTags T ON T.DocId = D.DocId
INNER JOIN @XmlTagData X ON X.id = T.TagId AND X.value = T.Value
group by 
documents.docid
having count(*) = 2 --[total of tags to be searched]

Answer 3

添加where子句以检查不存在的条件：

SELECT DISTINCT D.*
FROM Documents D
INNER JOIN DocumentTags T ON T.DocId = D.DocId
INNER JOIN @XmlTagData X ON X.id = T.TagId AND X.value = T.Value
WHERE NOT EXISTS (SELECT 1 FROM Documents dt2
                  CROSS JOIN Tags t2 
                  LEFT JOIN DocumentTags dt3 
                  ON t2.TagId = dt3.TagId
                  AND dt2.DocId = dt3.DocId
                  WHERE dt3.DocTagId IS NULL
                  AND dt2.DocId = D.DocId)

SQL Fiddle.

从表A中选择它加入表B中的所有数据

3 个答案: