sql-server - 将 SQL Server 中的 XML 列分解为新表

Question

我有一个源表，其中每一行都有一个 XML 字段，其中包含我需要分解并放入输出表中的未知数量的调查回复。我已经尝试过节点和 OPENXML，但似乎无法正常工作。根据我的研究，我认为我需要使用 CROSS APPLY，但我似乎无法以我理解的方式将它们放在一起并让它发挥作用。如果我遗漏了一些简单的东西，我深表歉意，但我真的被困住了，如果能提供任何帮助来解决这个问题，我将不胜感激。

这是包含导入数据(我无法控制其格式)的源表，如下所示:

ImportId    SurveyId    DateImported              ResponseData
--------    --------    ------------              ------------
1           11223344    2017-05-21 18:00:00.00    <survey><data><result>...
2           55667788    2017-05-21 18:01:00.00    <survey><data><result>...
3           99009988    2017-05-21 18:02:00.00    <survey><data><result>...

这是 ResponseData 列中 XML 的简化示例:

<survey>
    <data>
        <result>
            <id>12345</id>
            <date_submitted>2017-05-10 09:30:15</date_submitted>
            <url_variables>
                <respondent_id>
                    <key>respondent_id</key>
                    <value>987654</value>
                    <type>url</type>
                </respondent_id>
                <respondent_level>
                    <key>respondent_level</key>
                    <value>5</value>
                    <type>url</type>
                </respondent_level>
            </url_variables>
            <survey_data>
                <question>
                    <id>1</id>
                    <answer>Yes</answer>
                </question>
                <question>
                    <id>2</id>
                    <answer>No</answer>
                </question>
                <question>
                    <id>3</id>
                    <subquestions>
                        <subquestion>
                            <id>4</id>                           
                            <answer>Maybe</answer>
                        </subquestion>
                        <subquestion>
                            <id>5</id>                           
                            <answer>I don't know</answer>
                        </subquestion>
                    </subquestions>
                </question>
                ... more questions ...
            </survey_data>
        </result>
        <result>
            <id>67890</id>
            <date_submitted>2017-05-11 10:00:00</date_submitted>
            <url_variables>
                <respondent_id>
                    <key>respondent_id</key>
                    <value>34567</value>
                    <type>url</type>
                </respondent_id>
                <respondent_level>
                    <key>respondent_level</key>
                    <value>10</value>
                    <type>url</type>
                </respondent_level>
            </url_variables>
            <survey_data>
                <question>
                    <id>1</id>
                    <answer>No</answer>
                </question>
                <question>
                    <id>2</id>
                    <answer>Yes</answer>
                </question>
                <question>
                    <id>3</id>
                    <subquestions>
                        <subquestion>
                            <id>4</id>                           
                            <answer>Definitely not</answer>
                        </subquestion>
                        <subquestion>
                            <id>5</id>                           
                            <answer>I object</answer>
                        </subquestion>
                    </subquestions>
                </question>
                ... more questions ...
            </survey_data>
        </result>
        ... more results ...
    <data>
</survey>

我需要获取这些数据，分解 XML 以在每个 ResponseData 字段中获得多个结果，并将其放入如下两个表中:

ResultId    SurveyId    RespondentId    RespondentLevel    DateSubmitted
--------    --------    ------------    ---------------    -------------
12345       11223344    987654          5                  2017-05-10 09:30:15
67890       11223344    34567           10                 2017-05-11 10:00:00
... 
(Data extracted from the rest of ImportId 1 followed by ImportId 2, 3, etc)

ResultId      QuestionId    SubquestionId    Answer
----------    ----------    -------------    ------
12345         1             0                Yes
12345         2             0                No
12345         3             4                Maybe
12345         3             5                I don't know
67890         1             0                No
67890         2             0                Yes
67890         3             4                Definitely not
67890         3             5                I object
 ... 
(Data extracted from the rest of ImportId 1 followed by ImportId 2, 3, etc)

Answer 1

对于两个表，您需要两个查询

示例(我只使用了 1 条记录，但将应用于多条记录)

Declare @YourTable Table ([ImportId] int,[SurveyId] int,[DateImported] datetime,[ResponseData] xml)
Insert Into @YourTable Values
(1,11223344,'2017-05-21 18:00:00.00','<survey><data><result><id>12345</id><date_submitted>2017-05-10 09:30:15</date_submitted><url_variables><respondent_id><key>respondent_id</key><value>987654</value><type>url</type></respondent_id><respondent_level><key>respondent_level</key><value>5</value><type>url</type></respondent_level></url_variables><survey_data><question><id>1</id><answer>Yes</answer></question><question><id>2</id><answer>No</answer></question><question><id>3</id><subquestions><subquestion><id>4</id><answer>Maybe</answer></subquestion><subquestion><id>5</id><answer>I don''t know</answer></subquestion></subquestions></question></survey_data></result><result><id>67890</id><date_submitted>2017-05-11 10:00:00</date_submitted><url_variables><respondent_id><key>respondent_id</key><value>34567</value><type>url</type></respondent_id><respondent_level><key>respondent_level</key><value>10</value><type>url</type></respondent_level></url_variables><survey_data><question><id>1</id><answer>No</answer></question><question><id>2</id><answer>Yes</answer></question><question><id>3</id><subquestions><subquestion><id>4</id><answer>Definitely not</answer></subquestion><subquestion><id>5</id><answer>I object</answer></subquestion></subquestions></question></survey_data></result></data></survey>')

Select B.*
 From  @YourTable A
 Cross Apply (
                Select [ResultID]        = r.n.value('(id)[1]','int')
                      ,A.[SurveyId]
                      ,[RespondentId]    = r.n.value('(url_variables/respondent_id/value)[1]','int')
                      ,[RespondentLevel] = r.n.value('(url_variables/respondent_level/value)[1]','int')
                      ,[DateSubmitted]   = r.n.value('(date_submitted)[1]','datetime')
                 From  A.[ResponseData].nodes('survey/data/result') r(n)
             ) B


Select B.*
 From  @YourTable A
 Cross Apply (
                Select [ResultID]      = r.n.value('(id)[1]','int')
                      ,[QuestionId]    = IsNull(q.n.value('(id)[1]','int'),0)
                      ,[SubquestionId] = IsNull(s.n.value('(id)[1]','int'),0)
                      ,[answer]        = concat(q.n.value('(answer)[1]','varchar(50)') 
                                               ,s.n.value('(answer)[1]','varchar(50)') 
                                               )
                 From  A.[ResponseData].nodes('survey/data/result') r(n)
                 Cross Apply r.n.nodes('survey_data/question')      q(n)
                 Outer Apply q.n.nodes('subquestions/subquestion')  s(n)
             ) B

返回