sql语句在SQLserver可以运行，跑的时候报错，说by附近有问题，求大神指点。

Question

sql语句在SQLserver可以运行，跑的时候报错，说by附近有问题，求大神指点。
下面是报错信息：```
[ERROR] 2019-05-09 16:07:06.262 [http-bio-8080-exec-2] com.alibaba.druid.filter.stat.StatFilter - merge sql error, dbType sqlserver, druid-1.1.5, sql : SELECT count(DISTINCT jhta.EXECUTION_)AS num
         
         
         
        FROM BUSINESS_WF_FILE bwf
        LEFT JOIN BUSINESS_WHGL bw ON bwf.fwzh=CAST(bw.whid AS VARCHAR(8))
        LEFT JOIN BUSINESS_GWFL bg ON bwf.gwfl=CAST(bg.gwflId AS VARCHAR(8))
        LEFT JOIN JBPM4_HIST_TASK jhta ON substring(jhta.EXECUTION_,1,patindex('%[[0-9][.]]%',jhta.EXECUTION_+'.1.'))=(bwf.lcslKey+'.'+cast(bwf.ID AS VARCHAR(8)))
        LEFT JOIN SYS_USER su ON jhta.ASSIGNEE_=(su.LOGIN_NAME+'/'+su.NAME)
        WHERE jhta.EXECUTION_ IS NOT NULL AND bw.zhName IS NOT NULL
        AND bg.mc IS NOT NULL         
         
            AND jhta.DBID_ IN (SELECT max(DBID_) FROM JBPM4_HIST_TASK WHERE SUPERTASK_ IS NULL GROUP BY EXECUTION_)
            AND jhta.EXECUTION_ IN (SELECT EXECUTION_ID_ FROM JBPM4_TASK GROUP BY EXECUTION_ID_)
            AND substring(jhta.EXECUTION_,1,5) = 'fawen'
            AND jhta.STATE_ = 'reading'
            AND datediff(DAY,jhta.CREATE_,convert(VARCHAR(10),GETDATE(),120))  <=  90
         
         
         
        GROUP BY
         
         
         
/*        GROUP BY so.ORG_NAME,so.ORG_CODE
        ORDER BY so.ORG_CODE*/
com.alibaba.druid.sql.parser.EOFParserException: EOF
at com.alibaba.druid.sql.parser.SQLExprParser.primary(SQLExprParser.java:758)
at com.alibaba.druid.sql.dialect.sqlserver.parser.SQLServerExprParser.primary(SQLServerExprParser.java:90)
at com.alibaba.druid.sql.parser.SQLExprParser.expr(SQLExprParser.java:85)
at com.alibaba.druid.sql.parser.SQLSelectParser.parseGroupByItem(SQLSelectParser.java:491)
at com.alibaba.druid.sql.parser.SQLSelectParser.parseGroupBy(SQLSelectParser.java:397)
at com.alibaba.druid.sql.dialect.sqlserver.parser.SQLServerSelectParser.query(SQLServerSelectParser.java:164)
at com.alibaba.druid.sql.dialect.sqlserver.parser.SQLServerSelectParser.select(SQLServerSelectParser.java:47)
at com.alibaba.druid.sql.parser.SQLStatementParser.parseSelect(SQLStatementParser.java:2371)
at com.alibaba.druid.sql.parser.SQLStatementParser.parseStatementList(SQLStatementParser.java:147)
at com.alibaba.druid.sql.parser.SQLStatementParser.parseStatementList(SQLStatementParser.java:70)
at com.alibaba.druid.sql.visitor.ParameterizedOutputVisitorUtils.parameterize(ParameterizedOutputVisitorUtils.java:61)
at com.alibaba.druid.sql.visitor.ParameterizedOutputVisitorUtils.parameterize(ParameterizedOutputVisitorUtils.java:37)
at com.alibaba.druid.filter.stat.StatFilter.mergeSql(StatFilter.java:147)
at com.alibaba.druid.filter.stat.StatFilter.createSqlStat(StatFilter.java:648)
at com.alibaba.druid.filter.stat.StatFilter.statementPrepareAfter(StatFilter.java:311)
```下面是第二段
```
org.springframework.jdbc.UncategorizedSQLException: 
### Error querying database.  Cause: com.microsoft.sqlserver.jdbc.SQLServerException: “BY”附近有语法错误。
### The error may exist in com/harme/jbpm4workflow/entity/mapper/BusinessTongjiMapper.xml
### The error may involve com.harme.jbpm4workflow.entity.mapper.BusinessTongjiMapper.findFaWenTongji-Inline
### The error occurred while setting parameters
### SQL: SELECT count(DISTINCT jhta.EXECUTION_)AS num                                       FROM BUSINESS_WF_FILE bwf         LEFT JOIN BUSINESS_WHGL bw ON bwf.fwzh=CAST(bw.whid AS VARCHAR(8))         LEFT JOIN BUSINESS_GWFL bg ON bwf.gwfl=CAST(bg.gwflId AS VARCHAR(8))         LEFT JOIN JBPM4_HIST_TASK jhta ON substring(jhta.EXECUTION_,1,patindex('%[[0-9][.]]%',jhta.EXECUTION_+'.1.'))=(bwf.lcslKey+'.'+cast(bwf.ID AS VARCHAR(8)))         LEFT JOIN SYS_USER su ON jhta.ASSIGNEE_=(su.LOGIN_NAME+'/'+su.NAME)         WHERE jhta.EXECUTION_ IS NOT NULL AND bw.zhName IS NOT NULL         AND bg.mc IS NOT NULL                                  AND jhta.DBID_ IN (SELECT max(DBID_) FROM JBPM4_HIST_TASK WHERE SUPERTASK_ IS NULL GROUP BY EXECUTION_)             AND jhta.EXECUTION_ IN (SELECT EXECUTION_ID_ FROM JBPM4_TASK GROUP BY EXECUTION_ID_)             AND substring(jhta.EXECUTION_,1,5) = 'fawen'             AND jhta.STATE_ = 'reading'             AND datediff(DAY,jhta.CREATE_,convert(VARCHAR(10),GETDATE(),120))  <=  90                                       GROUP BY                               /*        GROUP BY so.ORG_NAME,so.ORG_CODE         ORDER BY so.ORG_CODE*/
### Cause: com.microsoft.sqlserver.jdbc.SQLServerException: “BY”附近有语法错误。
; uncategorized SQLException for SQL []; SQL state [S0001]; error code [102]; “BY”附近有语法错误。; nested exception is com.microsoft.sqlserver.jdbc.SQLServerException: “BY”附近有语法错误。```

这是sql语句
```
   <select id="findFaWenTongji" parameterType="map" resultType="map">
     
        SELECT count(DISTINCT jhta.EXECUTION_)AS num
        <if test="flag == 'gwfl' ">
            ,bg.mc as name
        </if>
        <if test="flag == 'zh' ">
            ,bw.zhName as name
        </if>
        <if test="flag == 'isgk' ">
            ,bwf.isgk as name
        </if>
        FROM BUSINESS_WF_FILE bwf
        LEFT JOIN BUSINESS_WHGL bw ON bwf.fwzh=CAST(bw.whid AS VARCHAR(8))
        LEFT JOIN BUSINESS_GWFL bg ON bwf.gwfl=CAST(bg.gwflId AS VARCHAR(8))
        LEFT JOIN JBPM4_HIST_TASK jhta ON substring(jhta.EXECUTION_,1,patindex('%[[0-9][.]]%',jhta.EXECUTION_+'.1.'))=(bwf.lcslKey+'.'+cast(bwf.ID AS VARCHAR(8)))
        LEFT JOIN SYS_USER su ON jhta.ASSIGNEE_=(su.LOGIN_NAME+'/'+su.NAME)
        WHERE jhta.EXECUTION_ IS NOT NULL AND bw.zhName IS NOT NULL
        AND bg.mc IS NOT NULL
<!--        <if test="year != null ">
            and YEAR(bwf.fwdate) = #{year}
        </if>
        <if test="month != null ">
            and MONTH(bwf.fwdate) = #{month}
        </if>-->
        <if test="state=='ended'">
            AND jhta.execution_ NOT IN (SELECT EXECUTION_ID_ FROM JBPM4_TASK GROUP BY EXECUTION_ID_)
        </if>
        <if test="state==zaiban">
            AND jhta.DBID_ IN (SELECT max(DBID_) FROM JBPM4_HIST_TASK WHERE SUPERTASK_ IS NULL GROUP BY EXECUTION_)
            AND jhta.EXECUTION_ IN (SELECT EXECUTION_ID_ FROM JBPM4_TASK GROUP BY EXECUTION_ID_)
            AND substring(jhta.EXECUTION_,1,5) = 'fawen'
            AND jhta.STATE_ = 'reading'
            AND datediff(DAY,jhta.CREATE_,convert(VARCHAR(10),GETDATE(),120)) <![CDATA[<=]]> 90
        </if>
        <if test="state=='daiban'">
            AND jhta.DBID_ IN (SELECT max(DBID_) FROM JBPM4_HIST_TASK WHERE SUPERTASK_ IS NULL GROUP BY EXECUTION_)
            AND jhta.EXECUTION_ IN (SELECT EXECUTION_ID_ FROM JBPM4_TASK GROUP BY EXECUTION_ID_)
            AND jhta.STATE_ IS NULL
            AND datediff(DAY,jhta.CREATE_,convert(VARCHAR(10),getdate(),120)) <![CDATA[<=]]> 90
        </if>
        <if test="state=='yichang'">
            AND jhta.DBID_ IN (SELECT max(DBID_) FROM JBPM4_HIST_TASK WHERE SUPERTASK_ IS NULL GROUP BY EXECUTION_)
            AND jhta.EXECUTION_ IN (SELECT EXECUTION_ID_ FROM JBPM4_TASK GROUP BY EXECUTION_ID_)
            AND jhta.DBID_ IS NOT NULL
            AND datediff(DAY,jhta.CREATE_,convert(VARCHAR(10),getdate(),120)) <![CDATA[>]]> 90
        </if>
        GROUP BY
        <if test="flag == 'gwfl' ">
            bg.mc
        </if>
        <if test="flag == 'zh' ">
            ,bw.zhName
        </if>
        <if test="flag == 'isgk' ">
            ,bwf.isgk
        </if>
/*        GROUP BY so.ORG_NAME,so.ORG_CODE
        ORDER BY so.ORG_CODE*/
    </select>
    
```

Answer 1

  GROUP BY
        <if test="flag == 'gwfl' ">
            bg.mc
        </if>
        <if test="flag == 'zh' ">
            ,bw.zhName
        </if>
        <if test="flag == 'isgk' ">
            ,bwf.isgk
        </if>group by 后面的3个if分支 按照你的逻辑应该至多走进一个吧。
           1> 如果一个都不符合，表达式即 group by  后面没有分组字段，会报错的。
           2> 还有就是后面两个if中 也不应该有 逗号【,】吧