由于开发的系统为多租户系统并且，技术采用了jpa+hibernate。查阅了hibernate的官方文档，并不支持sql方式对tenant_id 的数据隔离。所以无奈只能自行实现：

项目源码：https://gitee.com/97wx/nm-datasource

一、修改druid 项目与 druid-spring-boot-starter项目源码

项目源码地址：

https://github.com/alibaba/druid

项目源码版本：

<dependency>   <groupId>com.alibaba</groupId><artifactId>druid</artifactId><version>1.1.13-preview_3</version>
</dependency>

1、druid源码变更部分：

1）新增扩展接口

/*** 租户sql 服务提供都*/
public interface TenantSqlProvider {/*** sql 替换操作* @param sql* @return*/String prepareStatement(String sql);
}

2）DriudDataSource类与DruidConnectionHolder同时添加成员变量

    /*** 租户sql*/private TenantSqlProvider tenantSqlProvider;public TenantSqlProvider getTenantSqlProvider() {return tenantSqlProvider;}public void setTenantSqlProvider(TenantSqlProvider tenantSqlProvider) {this.tenantSqlProvider = tenantSqlProvider;}

3）修改DriudDataSource类在 getConnectionInternal方法最后将tenantSqlProvider 设置到DruidConnectionHolder中

holder.setTenantSqlProvider(tenantSqlProvider);

4）在DruidPooledConnection类下所有prepareStatement方法最开始的位置添加如下代码

if(holder.getTenantSqlProvider()!=null){sql = holder.getTenantSqlProvider().prepareStatement(sql);
}

2、druid-spring-boot-starter源码变更部分：

1）DruidDataSourceAutoConfigure类注入TenantSqlProvider并设置到DruidDataSourceWrapper中

public class DruidDataSourceAutoConfigure {private static final Logger LOGGER = LoggerFactory.getLogger(DruidDataSourceAutoConfigure.class);@Autowired(required = false)private TenantSqlProvider tenantSqlProvider;@Bean(initMethod = "init")@ConditionalOnMissingBeanpublic DataSource dataSource() {LOGGER.info("Init DruidDataSource");DruidDataSourceWrapper wrapper = new DruidDataSourceWrapper();wrapper.setTenantSqlProvider(tenantSqlProvider);return wrapper;}
}

二、整体思路如下：

1、通过栈找到匹配的(),根据栈的先进后出特性总能找到匹配成一对的()并记录“(”与“(”的位置

正则表达式：

 "(\\(|\\))"

数据结构：

@Data
public class OffsetItem {public OffsetItem(){}public OffsetItem(Integer start) {this.start = start;}private String tableName;private String values;private String fields;private Integer start;private Integer end;private String subSql;
}

2、当找到")"时，出栈，并使用截取子串方式获得配对的内容，如果字串内容包含select 则可以认为此子串为一个sql 语句,并且将子串保到父节点中

最后提取出来的sql 应该是这样的:

select u.* from user left join member m on m.userId=u.id where u.name=123 and u.p=ii and u.id in(select j.id from job limit 1) and m.id in (select d.id from deparment d where d.userId=u.id and d.jobId in (select j.id from job j)) order by id descselect j.id from job limit 1select j.id from job j

3、递归分解子句

若存在子句则进行递归调用，继续分解子句。直到最后一个子句不存在子句时，则可以进行替换

4、提取别名

将子句替换后返回到上一层语句，先将当前层的语句，将子句剔除出去，用来提取别名，应该是这样的如果不存在别名则用null代替

将子句剔除后应该是这样的：

select d.id from deparment d where d.userId=u.id and d.jobId in ()

正则表达式：

1）提取表名、左右连接、全连接部分

"((from|update)\\s+\\w+(((\\s+as|\\s+AS)*\\s+(?<as>\\w+))*?)\\s*(where|order|group|set|limit|$))"

2）提取别名通过find 可以获得一个别名数组

((?<=from|join|,|update)\s*\w+(((\s+as|\s+AS)*\s+(?<as>\w+))*?)\s*(?=where|left|right|inner|on|order|group|limit|$|,|set))

5、语句与子句拼接

最后通过子句的位置可以将当前层的语句完整的拼接出来，然后返回

三、具体实现

1、用到的java Bean

import lombok.Data;/***** @Auther: guzhangwen* @BusinessCard https://blog.csdn.net/gu_zhang_w* @Date: 2019/1/24* @Time: 14:01* @version : V1.0*/
@Data
public class OffsetItem {public OffsetItem() {}public OffsetItem(Integer start) {this.start = start;}private String tableName;private String values;private String fields;private Integer start;private Integer end;private String subSql;
}/*** sql 关键字汇总*/
public interface SqlConstant {public static final String SELECT = "select";public static final String UPDATE = "update";public static final String DELETE = "delete";public static final String INSERT = "insert";public static final String WHERE = "where";public static final String ORDER = "order";public static final String GROUP = "group";public static final String LIMIT = "limit";public static final String AS = "AS";public static final String as = "as";public static final String JOIN = "join";public static final String ON = "on";public static final String LEFT = "left";public static final String RIGHT = "right";public static final String INNER = "inner";public static final String FROM = "from";public static final String AND = "and";}

2、sql 置换默认实现抽象类TenantSqlSupport

import com.lvbang.erp.common.support.bean.ConditionItem;
import com.lvbang.erp.common.support.bean.OffsetItem;
import com.lvbang.erp.common.support.bean.SqlConstant;
import lombok.extern.slf4j.Slf4j;
import org.apache.commons.collections4.CollectionUtils;
import org.apache.commons.lang3.StringUtils;import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;
import java.util.Stack;
import java.util.regex.Matcher;
import java.util.regex.Pattern;/**** 租户sql 帮助类* @Auther: guzhangwen* @BusinessCard https://blog.csdn.net/gu_zhang_w* @Date: 2019/1/24* @Time: 13:59* @version : V1.0*/
@Slf4j
public abstract class TenantSqlSupport {/*** 提取配对的"("")"*/private static final Pattern PATTERN = Pattern.compile("(\\(|\\))");/*** 提取from 到where 之间的别名或者表名*/private static final Pattern ALALIS_PATTERN = Pattern.compile("((?<=from|join|,|update)\\s*\\w+(((\\s+as|\\s+AS)*\\s+(?<as>\\w+))*?)\\s*(?=where|left|right|inner|on|order|group|limit|$|,|set))", Pattern.CASE_INSENSITIVE);/*** 保存sql 匹配，当前仅仅匹配带字段的sql即：* insert into tmp(name,age)values (0,'gone'),(1,'123') insert into tmp(name,age)values (0,'gone'),(1,'123')*/private static final Pattern INSERT_PATTERN = Pattern.compile("(\\s*insert into\\s+(?<tableName>\\w+)\\s*\\((?<fields>.*?)\\)\\s*value(s)?\\s*(?<values>(\\(.*?\\),?)+?)(?=insert|$|;))", Pattern.CASE_INSENSITIVE);/*** 保存的值正则*/private static final Pattern INSERT_VALUES_PATTERN = Pattern.compile("((?<=\\().*?(?=\\)))", Pattern.CASE_INSENSITIVE);/*** 表的正则*/private static final Pattern TABLE_PATTERN = Pattern.compile("((from|update)\\s+\\w+(((\\s+as|\\s+AS)*\\s+(?<as>\\w+))*?)\\s*(where|order|group|set|limit|$))");/*** 构建查询语句** @param sql* @return*/public String rebuilderSelectSql(String sql) {/*** 不加自定义处理*/if (CollectionUtils.isEmpty(getCondtion())) {return sql;}Matcher matcher = PATTERN.matcher(sql);Stack<OffsetItem> offsetItemStack = new Stack<>();List<OffsetItem> offsetItemStackReverse = new ArrayList<>();boolean flag = false;while (matcher.find()) {String operator = matcher.group(1);if ("(".equals(operator)) {OffsetItem offsetItem = new OffsetItem(matcher.end());offsetItemStack.push(offsetItem);} else if (")".equals(operator)) {OffsetItem offsetItem = offsetItemStack.pop();offsetItem.setEnd(matcher.start());String subSql = sql.substring(offsetItem.getStart(), offsetItem.getEnd());if (subSql.trim().startsWith(SqlConstant.SELECT)) {flag = true;if (offsetItemStack.empty()) {checkedSubSql(offsetItem, subSql);offsetItemStackReverse.add(offsetItem);} else {OffsetItem prevOffsetItem = offsetItemStack.peek();checkedSubSql(prevOffsetItem, subSql);}}}}String newSql = null;if (flag) {newSql = mergeSql(sql, offsetItemStackReverse);} else {List<String> alaisList = getSqlAlais(sql);newSql = appendToCondition(sql, alaisList);}log.debug("==>>构建查询语句：置换后的sql={}", newSql);return newSql;}/*** 重构更新语句** @param sql* @return*/public String rebuilderUpdateSql(String sql) {/*** 不加自定义处理*/if (CollectionUtils.isEmpty(getCondtion())) {return sql;}List<String> alaisList = getSqlAlais(sql);if (CollectionUtils.isEmpty(alaisList)) {log.error("无法获取别名、请检查sql 匹配规则是否正确,语句={}", sql);throw new RuntimeException("无法获取别名、请检查sql 匹配规则是否正确");}String newSql = appendToCondition(sql, Arrays.asList(alaisList.get(0)));log.debug("==>>重构更新语句：置换后的sql={}", newSql);return newSql;}/*** 重构删除语句** @param sql* @return*/public String rebuilderDeletedSql(String sql) {/*** 不加自定义处理*/if (CollectionUtils.isEmpty(getCondtion())) {return sql;}List<String> alaisList = getSqlAlais(sql);if (CollectionUtils.isEmpty(alaisList)) {log.error("无法获取别名、请检查sql 匹配规则是否正确，语句={}", sql);throw new RuntimeException("无法获取别名、请检查sql 匹配规则是否正确");}String newSql = appendToCondition(sql, Arrays.asList(alaisList.get(0)));log.debug("==>>重构删除语句：置换后的sql={}", newSql);return newSql;}/*** 构建插入语句** @param sql* @return*/public String rebuilderInsertSql(String sql) {/*** 不加自定义处理*/List<ConditionItem> conditionItemList = getCondtion();if (CollectionUtils.isEmpty(conditionItemList)) {return sql;}try {Matcher matcher = INSERT_PATTERN.matcher(sql);List<OffsetItem> offsetItemList = new ArrayList<>();while (matcher.find()) {OffsetItem offsetItem = new OffsetItem();offsetItem.setTableName(matcher.group("tableName"));offsetItem.setFields(matcher.group("fields").trim());offsetItem.setValues(matcher.group("values").trim());offsetItemList.add(offsetItem);}if (CollectionUtils.isNotEmpty(offsetItemList)) {StringBuffer sb = new StringBuffer();for (OffsetItem item : offsetItemList) {if (sb.length() > 0) {sb.append(";");}sb.append("insert into ").append(item.getTableName()).append(" (").append(item.getFields());String fields = item.getFields();for (ConditionItem conditionItem : conditionItemList) {if (fields.contains(conditionItem.getName()) || (StringUtils.isBlank(conditionItem.getName()) || StringUtils.isBlank(conditionItem.getValue()))) {continue;}sb.append(",").append(conditionItem.getName());}sb.append(")");List<String> valueList = getInsertSqlValues(item.getValues());if (CollectionUtils.isEmpty(valueList)) {log.error("插入语句重构失败，请检查匹配规则与语法是否正确，sql={},正则表达式={}", sql, INSERT_VALUES_PATTERN.pattern());throw new RuntimeException("插入语句重构失败，请检查匹配规则与语法是否正确");}sb.append("value");if (valueList.size() > 1) {sb.append("s");}for (int i = 0; i < valueList.size(); i++) {if (i > 0) {sb.append(",");}sb.append("(").append(valueList.get(i));for (ConditionItem conditionItem : conditionItemList) {if (fields.contains(conditionItem.getName()) || (StringUtils.isBlank(conditionItem.getName()) || StringUtils.isBlank(conditionItem.getValue()))) {continue;}sb.append(",").append(conditionItem.getValue());}sb.append(")");}}String newSql = sb.toString();log.debug("==>>构建插入语句：置换后的sql={}", newSql);return newSql;}} catch (Exception e) {}return sql;}/*** 获取插入语句的sql** @param values* @return*/private List<String> getInsertSqlValues(String values) {List<String> valueList = new ArrayList<>();Matcher matcher = INSERT_VALUES_PATTERN.matcher(values);while (matcher.find()) {valueList.add(matcher.group(1));}return valueList;}/*** 合并sql** @param originalSql            原sql* @param offsetItemStackReverse sql子句对象{@link OffsetItem}* @return*/private String mergeSql(String originalSql, List<OffsetItem> offsetItemStackReverse) {StringBuffer sb = new StringBuffer();List<String> subSqlList = new ArrayList<>();if (CollectionUtils.isNotEmpty(offsetItemStackReverse)) {OffsetItem prevItem = null;for (int i = 0; i < offsetItemStackReverse.size(); i++) {OffsetItem item = offsetItemStackReverse.get(i);if (sb.length() <= 0) {sb.append(originalSql.substring(0, item.getStart()));}if (prevItem != null && prevItem.getEnd() != item.getStart()) {sb.append(originalSql.substring(prevItem.getEnd(), item.getStart()));}sb.append("%s");if ((i + 1) == offsetItemStackReverse.size()) {sb.append(originalSql.substring(item.getEnd(), originalSql.length()));}subSqlList.add(item.getSubSql());prevItem = item;}}//例如 select * from a where uid in(%s) and  did in(%s)String newSql = sb.toString();List<String> alaisList = getSqlAlais(newSql);newSql = appendToCondition(newSql, alaisList);if (CollectionUtils.isNotEmpty(subSqlList)) {for (String s : subSqlList) {newSql = newSql.replaceFirst("%s", s);}}return newSql;}/*** 检查subSql 是否还存在子查询** @param offsetItem sql子句对象{@link OffsetItem}* @param subSql     子查询sql*/private void checkedSubSql(OffsetItem offsetItem, String subSql) {if (appearNumber(subSql, SqlConstant.SELECT) > 1) {offsetItem.setSubSql(rebuilderSelectSql(subSql));} else {List<String> alaisList = getSqlAlais(subSql);String newSql = appendToCondition(subSql, alaisList);offsetItem.setSubSql(newSql);}}/*** 获取指定字符串出现的次数** @param srcText  源字符串* @param findText 要查找的字符串* @return*/public int appearNumber(String srcText, String findText) {int count = 0;Pattern p = Pattern.compile(findText);Matcher m = p.matcher(srcText);while (m.find()) {count++;}return count;}/*** 获取sql 语句的别名** @param sql* @return*/public List<String> getSqlAlais(String sql) {Matcher tableMatcher = TABLE_PATTERN.matcher(sql);StringBuffer stringBuffer = new StringBuffer();while (tableMatcher.find()) {stringBuffer.append(tableMatcher.group(1));}Matcher matcher = ALALIS_PATTERN.matcher(stringBuffer.toString());List<String> alaisList = new ArrayList<>();while (matcher.find()) {alaisList.add(matcher.group(SqlConstant.as));}return alaisList;}/*** 添加查询过滤条件** @param sql* @param alaisList* @return*/private String appendToCondition(String sql, List<String> alaisList) {StringBuffer sb = new StringBuffer();int offset = 0;offset = sql.lastIndexOf(SqlConstant.WHERE);if (offset > 0) {offset += 5;sb.append(sql.substring(0, offset)).append(" ");boolean flag = appendTenant(sb, alaisList);if (flag) {sb.append(" and ");}sb.append(sql.substring(offset, sql.length()));} else {try {offset = sql.lastIndexOf(SqlConstant.ORDER);if (offset > 0) {throw new Exception();}offset = sql.lastIndexOf(SqlConstant.GROUP);if (offset > 0) {throw new Exception();}offset = sql.lastIndexOf(SqlConstant.LIMIT);if (offset > 0) {throw new Exception();}} catch (Exception e) {}if (offset > 0) {sb.append(sql.substring(0, offset)).append(SqlConstant.WHERE).append(" ");appendTenant(sb, alaisList);sb.append(" ").append(sql.substring(offset, sql.length()));} else {sb.append(sql).append(" ").append(SqlConstant.WHERE + " ");appendTenant(sb, alaisList);}}return sb.toString();}/*** 添加租户id** @param sb        拼接的字符串* @param alaisList sql 中包含的别名*/private boolean appendTenant(StringBuffer sb, List<String> alaisList) {boolean flag = false;for (int i = 0; i < alaisList.size(); i++) {String alais = alaisList.get(i);if (i >= 1) {sb.append(" " + SqlConstant.AND + " ");}List<ConditionItem> conditionItemList = getCondtion();if (CollectionUtils.isNotEmpty(conditionItemList)) {for (ConditionItem item : conditionItemList) {//已经包含指定字段名称if (StringUtils.isEmpty(item.getName()) || StringUtils.isEmpty(item.getValue())) {continue;}flag = true;if (StringUtils.isBlank(item.getName()) || StringUtils.isBlank(item.getValue())) {log.error("condition name 与 value 都不能为null");throw new RuntimeException("condition name 与 value 都不能为null");}if (alais == null) {sb.append(String.format("%s=%s", item.getName(), item.getValue()));} else {sb.append(String.format("%s.%s=%s", alais, item.getName(), item.getValue()));}}}}return flag;}/*** 获取需要附加的查询条件** @return*/public abstract List<ConditionItem> getCondtion();
}

3、TenantId 扩展接口的实现者与默认sql 置换默认实现抽象类的实现者

import com.alibaba.druid.tenant.TenantSqlProvider;
import com.lvbang.erp.common.support.bean.ConditionItem;
import com.lvbang.erp.common.support.bean.SqlConstant;import java.util.Arrays;
import java.util.List;
import java.util.Optional;
import java.util.regex.Matcher;
import java.util.regex.Pattern;/***** @Auther: guzhangwen* @BusinessCard https://blog.csdn.net/gu_zhang_w* @Date: 2019/1/22* @Time: 18:01* @version : V1.0*/
public class DefaultTenantSqlProvider extends TenantSqlSupport implements TenantSqlProvider {@Overridepublic String prepareStatement(String sql) {sql = sql.replaceAll("/\\*.*?\\*/","").trim();Optional<String> optional = Optional.of(sql);if (sql.startsWith(SqlConstant.SELECT)) {optional = Optional.ofNullable(getSelectSql(sql));} else if (sql.startsWith(SqlConstant.UPDATE)) {optional = Optional.ofNullable(getUpdateSql(sql));} else if (sql.startsWith(SqlConstant.DELETE)) {optional = Optional.ofNullable(getDeletedSql(sql));} else if (sql.startsWith(SqlConstant.INSERT)) {optional = Optional.ofNullable(getInsertSql(sql));}return optional.orElse(sql);}private String getUpdateSql(String sql) {return rebuilderUpdateSql(sql);}private String getSelectSql(String sql) {return rebuilderSelectSql(sql);}private String getDeletedSql(String sql) {return rebuilderDeletedSql(sql);}private String getInsertSql(String sql) {return rebuilderInsertSql(sql);}private static final Pattern alalisPattern = Pattern.compile("((?<=from|join|update|,)\\s*\\w+(((\\s+as|\\s+AS)*\\s+(?<as>\\w+))*?)\\s*(?=where|left|right|inner|on|order|group|limit|$|set|,))",Pattern.CASE_INSENSITIVE);private static final Pattern insertPattern = Pattern.compile("(\\s*insert into\\s+\\w+\\s*(?<fields>\\(.*?\\))\\s*value(s)?\\s*(?<values>(\\(.*?\\),?)+?)(?=insert|$))",Pattern.CASE_INSENSITIVE);private static final Pattern insertValuesPattern = Pattern.compile("(\\(.*?\\))",Pattern.CASE_INSENSITIVE);//private static final Pattern pattern = Pattern.compile("(from\\s*\\w+(((\\s+as|\\s+AS)*\\s+(?<as>\\w+))*?)\\s*(where|order|group|limit|$))");private static final Pattern TABLE_PATTERN = Pattern.compile("((from|update)\\s+\\w+(((\\s+as|\\s+AS)*\\s+(?<as>\\w+))*?)\\s*(where|order|group|set|limit|$))");public static void main(String[] args) {//String sql = "select u.* from user,car c,left join member m on m.userId=u.id join tree t on t.userId=u.id where u.name=123 and u.p=ii and u.id in(select d.id from deparment d where d.userId=u.id and d.jobId in (select j.id from job j)) and m.id in (select p.id from position p where p.userId=u.id) order by id desc";//String sql = "update customer_customer_type set status=? where id=? and tenant_id=?";//String sql = "delete from financial_bank where id=4";String sql = "select customerty0_.id as id1_7_, customerty0_.create_host as create_h2_7_, customerty0_.create_time as create_t3_7_, customerty0_.create_user_id as create_u4_7_, customerty0_.create_user_name as create_u5_7_, customerty0_.tenant_id as tenant_i6_7_, customerty0_.update_host as update_h7_7_, customerty0_.update_time as update_t8_7_, customerty0_.update_user_id as update_u9_7_, customerty0_.update_user_name as update_10_7_, customerty0_.deleted as deleted11_7_, customerty0_.remark as remark12_7_, customerty0_.status as status13_7_, customerty0_.type_name as type_na14_7_ from customer_customer_type customerty0_ where 1=1 and customerty0_.deleted=? and customerty0_.tenant_id=10 order by customerty0_.update_time desc limit ?";//String sql = "insert into tmp(name,age)values (0,'gone'),(1,'123')insert into tmp(name,age)values (0,'gone'),(1,'123')";Matcher matcher = TABLE_PATTERN.matcher(sql);StringBuffer stringBuffer = new StringBuffer();while (matcher.find()) {//stringBuffer.append();System.out.println(matcher.group(1));}/*DefaultTenantSqlProvider provider = new DefaultTenantSqlProvider();System.out.println(provider.getSqlAlais(sql));*/}@Overridepublic List<ConditionItem> getCondtion() {//StringHelper.getObjectValue(BaseContextHandler.getTenantId())List<ConditionItem> list = Arrays.asList(new ConditionItem("tenant_id", "10"));return list;}
}

Druid数据源MySql语句，添加租户(tenant_id)id相关推荐

mysql语句添加、删除索引（转）
转自:mysql语句添加索引参考: mysql索引学习----2----创建索引.修改索引.删除索引的命令语句 mysql语句添加索引创建或添加索引可以使用如下语句. 一.使用ALTER TABL ...
mysql语句添加索引
1.PRIMARY KEY(主键索引) mysql>ALTER TABLE `table_name` ADD PRIMARY KEY ( `column` ) ...
mysql语句添加字段和注释
简单点的:工作中需要对某个表进行添加字段,输出脚本添加字段关键字 alter 注释关键字 comment ALTER TABLE 表名 ADD 字段名 varchar(20) COMMENT ' ...
使用MySQL语句给数据库的表和字段添加注释
给MySQL语句添加注释 MySQL中使用 "comment" 来给数据库表和字段添加注释,以下是添加注释的几种情况: 一.给表添加注释 1)创建表时添加注释 CREATE ...
php用到的mysql语句_PHP中常用到的一些MySQL语句_php
在php开发中,经常会使用到mysql语句,下面就为您列举了一些经常使用的MySQL语句,希望对您平时的学习和开发工作能起到些许的作用. MySQL语句显示数据库或表: show databases; ...
Druid拦截sql语句，实现在添加一个查询条件
Druid拦截sql语句,实现在添加一个查询条件这里就不详细描述原理了. 首先需要重写一下FilterEventAdapter里的connection_prepareStatement方法,然后对s ...
mybatis mysql merge into_整合DRUID数据源+MyBatis
整合DRUID数据源实际的开发中,我们很少用上面那种方式的数据源. 引入DruidDataSource: com.alibaba druid 1.1.8 配置: type: com.alibaba. ...
Springboot + mybatis + druid 整合 (Mysql单数据源)
Druid多数据源整合前言为什么要使用Druid连接池? 通常我们是直接通过mybatis与数据库建立连接,而创建连接的过程是在发起请求和接受请求之间进行的,这样请求就会消耗更多的时间.并且在大型 ...
mybatis配置mysql数据源_springboot+mybatis+Druid配置多数据源(mysql+postgre)
springboot+mybatis+Druid配置多数据源(mysql+postgre) 引入pom依赖 org.mybatis.spring.boot mybatis-spring-boot-st ...

Druid数据源MySql语句，添加租户(tenant_id)id