Springboot管理系统数据权限过滤（三）——0业务入侵实现部门数据权限过滤

上一章，讲解了SQL拦截器的，做了一个简单的SQL改造。本章要实现：

仅对指定的service方法内执行的SQL进行处理；
完成对部门权限的过滤；

以下简单用一个图说明了整个处理过程，红色框住的部分，就是本章要实现的内容：

Spring注解拦截器，该拦截器的目标是对添加了@DataScope注解的方法，作用是解析中DataScope对象，将放到线程变量的权限过滤对象栈中。以便在SQL拦截器中获取权限对象进行解析；
DataScope栈，需要实现一个线程栈，没有业务逻辑，仅为传递作用；
SQL拦截器内，解析DataScope对象获得有权限的部门ID，然后拼接过滤条件；
另外DataScope权限注解也需要添加；

直接上干货：

定义DataSopce对象

DataScope.java
DataScope对象里面设置了用于数据权限规则数组。在SQL拦截器中将从这些数据权限规则中获取条件表达式。

package com.luo.chengrui.labs.lab02.annotation;import com.luo.chengrui.labs.lab02.datapermission.DataPermissionRule;import java.lang.annotation.*;/*** 数据权限过滤注解** @author ruoyi*/
@Target({ElementType.TYPE, ElementType.METHOD})
@Retention(RetentionPolicy.RUNTIME)
@Inherited
@Documented
public @interface DataScope
{/*** 当前类或方法是否开启数据权限* 即使不添加 @DataPermission 注解，默认是开启状态* 可通过设置 enable 为 false 禁用*/boolean enable() default true;/*** 生效的数据权限规则数组，为了以后方便扩展，所以定义为权限解析对象数组*/Class<? extends DataPermissionRule>[] includeRules() default {};
}

DataPermissionRule.java
这里定义了一个权限解析接口，如此即可以扩展很多不同类型的权限，如：对部门过滤，本章我们仅实现对部门权限过滤。（若要实现对如预算科目、项目等过滤，可实现该接口，另外还得单独实现对这些数据权限的分配功能。若依仅实现了对部门和人员控制）

package com.luo.chengrui.labs.lab02.datapermission;import com.baomidou.mybatisplus.core.metadata.TableInfoHelper;
import net.sf.jsqlparser.expression.Alias;
import net.sf.jsqlparser.expression.Expression;import java.util.Set;/*** 数据权限规则接口* 通过实现接口，自定义数据规则。例如说，** @author yudao*/
public interface DataPermissionRule {/*** 根据表名和别名，生成对应的 WHERE / OR 过滤条件** @param tableName 表名* @param tableAlias 别名，可能为空* @return 过滤条件 Expression 表达式*/Expression getExpression(String tableName, Alias tableAlias);}

DeptDataPermissionRule.java
部门权限解析实现类，只要SQL语句中有sys_dept表或者dept_id字段，则均可以通过getExpression（String tableName, Alias tableAlias）获取部门数据权限，并生成Express表达式。

ruoyi框架中数据权限分配和权限获取实现，可参考：Springboot管理系统数据权限过滤——ruoyi实现方案

package com.luo.chengrui.labs.lab02.datapermission.dept;import com.baomidou.mybatisplus.core.toolkit.StringPool;
import com.luo.chengrui.labs.lab02.datapermission.DataPermissionRule;
import lombok.AllArgsConstructor;
import lombok.extern.slf4j.Slf4j;
import net.sf.jsqlparser.expression.*;
import net.sf.jsqlparser.expression.operators.relational.ExpressionList;
import net.sf.jsqlparser.expression.operators.relational.InExpression;
import net.sf.jsqlparser.schema.Column;import java.util.*;
import java.util.stream.Collectors;/*** 基于部门的 {@link DataPermissionRule} 数据权限规则实现* <p>* 注意，使用 DeptDataPermissionRule 时，需要保证表中有 dept_id 部门编号的字段，可自定义。* <p>*/
@AllArgsConstructor
@Slf4j
public class DeptDataPermissionRule implements DataPermissionRule {static final Expression EXPRESSION_NULL = new NullValue();/*** 基于部门的表字段配置* 一般情况下，每个表的部门编号字段是 dept_id，通过该配置自定义。* <p>* key：表名* value：字段名*/private final Map<String, String> deptColumns = new HashMap<>();/*** 所有表名，需要进行权限过滤的表名*/private final Set<String> TABLE_NAMES = new HashSet<>();/*** 添加需要过滤部门权限的表名和部门ID字段名* @param tableName* @param columnName*/public void addDeptColumn(String tableName, String columnName) {deptColumns.put(tableName, columnName);TABLE_NAMES.add(tableName);}/*** 获取所有需要按部门权限过滤的表* @return*/@Overridepublic Set<String> getTableNames() {return TABLE_NAMES;}@Overridepublic Expression getExpression(String tableName, Alias tableAlias) {// 情况三，拼接 Dept 和 User 的条件，最后组合Set<Long> deptIds = new HashSet<>();//模拟数据，实现业务中需要获取当前用户部门权限，获取到部门id。deptIds.add(1L);deptIds.add(2L);// 配置中包含表时，进行过滤if (Objects.nonNull(deptColumns.get(tableName))) {return new InExpression(buildColumn(tableName, tableAlias, deptColumns.get(tableName)),new ExpressionList(deptIds.stream().map(LongValue::new).collect(Collectors.toList())));}return EXPRESSION_NULL;//}/*** 构建 Column 对象** @param tableName  表名* @param tableAlias 别名* @param column     字段名* @return Column 对象*/public static Column buildColumn(String tableName, Alias tableAlias, String column) {if (tableAlias != null) {tableName = tableAlias.getName();}return new Column(tableName + StringPool.DOT + column);}
}

数据权限线程变量

DataPermissionContextHolder.java

package com.luo.chengrui.labs.lab02.interceptor;import com.luo.chengrui.labs.lab02.annotation.DataScope;import java.util.LinkedList;
import java.util.List;/*** {@link DataScope} 注解的 Context 上下文* 将方法上的注解对象设置到 线程变量里面，在SQL执行拦截器中获取注解对象，根据内容生成相应的权限。* 告诉SQL执行{DataPermissionDatabaseInterceptor}拦截器，如此方法需要添加权限。*/
public class DataPermissionContextHolder {/*** 使用 List 的原因，可能存在方法的嵌套调用*/private static final ThreadLocal<LinkedList<DataScope>> DATA_PERMISSIONS =ThreadLocal.withInitial(LinkedList::new);/*** 获得当前的 DataPermission 注解** @return DataPermission 注解*/public static DataScope get() {return DATA_PERMISSIONS.get().peekLast();}/*** 入栈 DataPermission 注解** @param dataPermission DataPermission 注解*/public static void add(DataScope dataPermission) {DATA_PERMISSIONS.get().addLast(dataPermission);}/*** 出栈 DataPermission 注解** @return DataPermission 注解*/public static DataScope remove() {DataScope dataPermission = DATA_PERMISSIONS.get().removeLast();// 无元素时，清空 ThreadLocalif (DATA_PERMISSIONS.get().isEmpty()) {DATA_PERMISSIONS.remove();}return dataPermission;}/*** 获得所有 DataPermission** @return DataPermission 队列*/public static List<DataScope> getAll() {return DATA_PERMISSIONS.get();}/*** 清空上下文* <p>* 目前仅仅用于单测*/public static void clear() {DATA_PERMISSIONS.remove();}}

定义DataScope注解拦截器

DataScopeAnnotationInterceptor.java
方法拦截器：作用是拦截添加了@DataScope注解的方法，获取DataScope对象，放入线程变量。

package com.luo.chengrui.labs.lab02.interceptor;import com.luo.chengrui.labs.lab02.annotation.DataScope;
import lombok.Getter;
import org.aopalliance.intercept.MethodInterceptor;
import org.aopalliance.intercept.MethodInvocation;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.core.MethodClassKey;
import org.springframework.core.annotation.AnnotationUtils;import java.lang.reflect.Method;
import java.util.Map;
import java.util.concurrent.ConcurrentHashMap;/*** {@link DataScope} 注解的拦截器* 1. 在执行方法前，将 @DataPermission 注解入栈* 2. 在执行方法后，将 @DataPermission 注解出栈** @author yudao*/
public class DataScopeAnnotationInterceptor implements MethodInterceptor {/*** DataPermission 空对象，用于方法无 {@link import cn.iocoder.yudao.framework.datapermission.core.annotation.DataPermission;} 注解时，使用 DATA_PERMISSION_NULL 进行占位*/static final DataScope DATA_PERMISSION_NULL = DataScopeAnnotationInterceptor.class.getAnnotation(DataScope.class);@Getterprivate final Map<MethodClassKey, DataScope> dataPermissionCache = new ConcurrentHashMap<>();@Overridepublic Object invoke(MethodInvocation methodInvocation) throws Throwable {Logger log = LoggerFactory.getLogger(DataScopeAnnotationInterceptor.class);log.debug("DataScopeAnnotationInterceptor 拦截器：" + methodInvocation.getMethod().getName());// 入栈DataScope dataPermission = this.findAnnotation(methodInvocation);if (dataPermission != null) {DataPermissionContextHolder.add(dataPermission);}try {// 执行逻辑return methodInvocation.proceed();} finally {// 出栈if (dataPermission != null) {DataPermissionContextHolder.remove();}}}private DataScope findAnnotation(MethodInvocation methodInvocation) {// 1. 从缓存中获取Method method = methodInvocation.getMethod();Object targetObject = methodInvocation.getThis();Class<?> clazz = targetObject != null ? targetObject.getClass() : method.getDeclaringClass();MethodClassKey methodClassKey = new MethodClassKey(method, clazz);DataScope dataPermission = dataPermissionCache.get(methodClassKey);if (dataPermission != null) {return dataPermission != DATA_PERMISSION_NULL ? dataPermission : null;}// 2.1 从方法中获取dataPermission = AnnotationUtils.findAnnotation(method, DataScope.class);// 2.2 从类上获取if (dataPermission == null) {dataPermission = AnnotationUtils.findAnnotation(clazz, DataScope.class);}// 2.3 添加到缓存中dataPermissionCache.put(methodClassKey, dataPermission != null ? dataPermission : DATA_PERMISSION_NULL);return dataPermission;}}

DataPermissionAnnotationAdvisor.java
定义拦截点，对添加了DataScope注解的类和方法都添加了拦截点。

package com.luo.chengrui.labs.lab02.interceptor;import com.luo.chengrui.labs.lab02.annotation.DataScope;
import lombok.EqualsAndHashCode;
import lombok.Getter;
import org.aopalliance.aop.Advice;
import org.springframework.aop.Pointcut;
import org.springframework.aop.support.AbstractPointcutAdvisor;
import org.springframework.aop.support.ComposablePointcut;
import org.springframework.aop.support.annotation.AnnotationMatchingPointcut;
import org.springframework.context.annotation.Role;
import org.springframework.stereotype.Component;/*** {@link DataScope} 注解的 Advisor 实现类**/
@Getter
@EqualsAndHashCode(callSuper = true)
public class DataPermissionAnnotationAdvisor extends AbstractPointcutAdvisor {private final Advice advice;private final Pointcut pointcut;public DataPermissionAnnotationAdvisor() {this.advice = new DataScopeAnnotationInterceptor();this.pointcut = this.buildPointcut();}protected Pointcut buildPointcut() {Pointcut classPointcut = new AnnotationMatchingPointcut(DataScope.class, true);Pointcut methodPointcut = new AnnotationMatchingPointcut(null, DataScope.class, true);return new ComposablePointcut(classPointcut).union(methodPointcut);}}

SQL拦截器改造sql。

DataPermissionDatabaseInterceptor.java
代码比较多，主要是对sql各种情况的解析和处理，与实现业务关联性很弱。在实际业务扩展时，该类基本不需要进行修改。可能搜索对 getExpression 方法的调用，改造sql的位置。

package com.luo.chengrui.labs.lab02.datapermission;import com.baomidou.mybatisplus.core.toolkit.CollectionUtils;
import com.baomidou.mybatisplus.core.toolkit.PluginUtils;
import com.baomidou.mybatisplus.extension.parser.JsqlParserSupport;
import com.baomidou.mybatisplus.extension.plugins.inner.InnerInterceptor;
import lombok.Getter;
import lombok.RequiredArgsConstructor;
import net.sf.jsqlparser.expression.*;
import net.sf.jsqlparser.expression.operators.conditional.AndExpression;
import net.sf.jsqlparser.expression.operators.conditional.OrExpression;
import net.sf.jsqlparser.expression.operators.relational.ExistsExpression;
import net.sf.jsqlparser.expression.operators.relational.ExpressionList;
import net.sf.jsqlparser.expression.operators.relational.InExpression;
import net.sf.jsqlparser.schema.Table;
import net.sf.jsqlparser.statement.delete.Delete;
import net.sf.jsqlparser.statement.select.*;
import net.sf.jsqlparser.statement.update.Update;
import org.apache.ibatis.executor.Executor;
import org.apache.ibatis.executor.statement.StatementHandler;
import org.apache.ibatis.mapping.BoundSql;
import org.apache.ibatis.mapping.MappedStatement;
import org.apache.ibatis.mapping.SqlCommandType;
import org.apache.ibatis.session.ResultHandler;
import org.apache.ibatis.session.RowBounds;import java.sql.Connection;
import java.util.*;
import java.util.concurrent.ConcurrentHashMap;
import java.util.stream.Collectors;/*** 数据权限拦截器，通过 {@link DataPermissionRule} 数据权限规则，重写 SQL 的方式来实现* 主要的 SQL 重写方法，可见 {@link #builderExpression(Expression, List)} 方法* 主要是在执行SQL前拦截器，在执行之前可重写SQL** @author yudao*/
@RequiredArgsConstructor
public class DataPermissionDatabaseInterceptor extends JsqlParserSupport implements InnerInterceptor {private static final String MYSQL_ESCAPE_CHARACTER = "`";private final List<DataPermissionRule> dataPermissionRule;@Getterprivate final MappedStatementCache mappedStatementCache = new MappedStatementCache();@Override // SELECT 场景public void beforeQuery(Executor executor, MappedStatement ms, Object parameter, RowBounds rowBounds, ResultHandler resultHandler, BoundSql boundSql) {// 获得 Mapper 对应的数据权限的规则if (mappedStatementCache.noRewritable(ms, dataPermissionRule)) { // 如果无需重写，则跳过return;}PluginUtils.MPBoundSql mpBs = PluginUtils.mpBoundSql(boundSql);try {// 初始化上下文ContextHolder.init(dataPermissionRule);// 处理 SQLmpBs.sql(parserSingle(mpBs.sql(), null));} finally {// 添加是否需要重写的缓存addMappedStatementCache(ms);// 清空上下文ContextHolder.clear();}}@Override // 只处理 UPDATE / DELETE 场景，不处理 INSERT 场景（因为 INSERT 不需要数据权限)public void beforePrepare(StatementHandler sh, Connection connection, Integer transactionTimeout) {PluginUtils.MPStatementHandler mpSh = PluginUtils.mpStatementHandler(sh);MappedStatement ms = mpSh.mappedStatement();SqlCommandType sct = ms.getSqlCommandType();if (sct == SqlCommandType.UPDATE || sct == SqlCommandType.DELETE) {// 获得 Mapper 对应的数据权限的规则if (mappedStatementCache.noRewritable(ms, dataPermissionRule)) { // 如果无需重写，则跳过return;}PluginUtils.MPBoundSql mpBs = mpSh.mPBoundSql();try {// 初始化上下文ContextHolder.init(dataPermissionRule);// 处理 SQLmpBs.sql(parserMulti(mpBs.sql(), null));} finally {// 添加是否需要重写的缓存addMappedStatementCache(ms);// 清空上下文ContextHolder.clear();}}}@Overrideprotected void processSelect(Select select, int index, String sql, Object obj) {processSelectBody(select.getSelectBody());List<WithItem> withItemsList = select.getWithItemsList();if (!CollectionUtils.isEmpty(withItemsList)) {withItemsList.forEach(this::processSelectBody);}}/*** update 语句处理*/@Overrideprotected void processUpdate(Update update, int index, String sql, Object obj) {final Table table = update.getTable();update.setWhere(this.builderExpression(update.getWhere(), table));}/*** delete 语句处理*/@Overrideprotected void processDelete(Delete delete, int index, String sql, Object obj) {delete.setWhere(this.builderExpression(delete.getWhere(), delete.getTable()));}// ========== 和 TenantLineInnerInterceptor 一致的逻辑 ==========protected void processSelectBody(SelectBody selectBody) {if (selectBody == null) {return;}if (selectBody instanceof PlainSelect) {processPlainSelect((PlainSelect) selectBody);} else if (selectBody instanceof WithItem) {WithItem withItem = (WithItem) selectBody;processSelectBody(withItem.getSubSelect().getSelectBody());} else {SetOperationList operationList = (SetOperationList) selectBody;List<SelectBody> selectBodyList = operationList.getSelects();if (CollectionUtils.isNotEmpty(selectBodyList)) {selectBodyList.forEach(this::processSelectBody);}}}/*** 处理 PlainSelect*/protected void processPlainSelect(PlainSelect plainSelect) {//#3087 githubList<SelectItem> selectItems = plainSelect.getSelectItems();if (CollectionUtils.isNotEmpty(selectItems)) {selectItems.forEach(this::processSelectItem);}// 处理 where 中的子查询Expression where = plainSelect.getWhere();processWhereSubSelect(where);// 处理 fromItemFromItem fromItem = plainSelect.getFromItem();List<Table> list = processFromItem(fromItem);List<Table> mainTables = new ArrayList<>(list);// 处理 joinList<Join> joins = plainSelect.getJoins();if (CollectionUtils.isNotEmpty(joins)) {mainTables = processJoins(mainTables, joins);}// 当有 mainTable 时，进行 where 条件追加if (CollectionUtils.isNotEmpty(mainTables)) {plainSelect.setWhere(builderExpression(where, mainTables));}}private List<Table> processFromItem(FromItem fromItem) {// 处理括号括起来的表达式while (fromItem instanceof ParenthesisFromItem) {fromItem = ((ParenthesisFromItem) fromItem).getFromItem();}List<Table> mainTables = new ArrayList<>();// 无 join 时的处理逻辑if (fromItem instanceof Table) {Table fromTable = (Table) fromItem;mainTables.add(fromTable);} else if (fromItem instanceof SubJoin) {// SubJoin 类型则还需要添加上 where 条件List<Table> tables = processSubJoin((SubJoin) fromItem);mainTables.addAll(tables);} else {// 处理下 fromItemprocessOtherFromItem(fromItem);}return mainTables;}/*** 处理where条件内的子查询* <p>* 支持如下:* 1. in* 2. =* 3. >* 4. <* 5. >=* 6. <=* 7. <>* 8. EXISTS* 9. NOT EXISTS* <p>* 前提条件:* 1. 子查询必须放在小括号中* 2. 子查询一般放在比较操作符的右边** @param where where 条件*/protected void processWhereSubSelect(Expression where) {if (where == null) {return;}if (where instanceof FromItem) {processOtherFromItem((FromItem) where);return;}if (where.toString().indexOf("SELECT") > 0) {// 有子查询if (where instanceof BinaryExpression) {// 比较符号 , and , or , 等等BinaryExpression expression = (BinaryExpression) where;processWhereSubSelect(expression.getLeftExpression());processWhereSubSelect(expression.getRightExpression());} else if (where instanceof InExpression) {// inInExpression expression = (InExpression) where;Expression inExpression = expression.getRightExpression();if (inExpression instanceof SubSelect) {processSelectBody(((SubSelect) inExpression).getSelectBody());}} else if (where instanceof ExistsExpression) {// existsExistsExpression expression = (ExistsExpression) where;processWhereSubSelect(expression.getRightExpression());} else if (where instanceof NotExpression) {// not existsNotExpression expression = (NotExpression) where;processWhereSubSelect(expression.getExpression());} else if (where instanceof Parenthesis) {Parenthesis expression = (Parenthesis) where;processWhereSubSelect(expression.getExpression());}}}protected void processSelectItem(SelectItem selectItem) {if (selectItem instanceof SelectExpressionItem) {SelectExpressionItem selectExpressionItem = (SelectExpressionItem) selectItem;if (selectExpressionItem.getExpression() instanceof SubSelect) {processSelectBody(((SubSelect) selectExpressionItem.getExpression()).getSelectBody());} else if (selectExpressionItem.getExpression() instanceof Function) {processFunction((Function) selectExpressionItem.getExpression());}}}/*** 处理函数* <p>支持: 1. select fun(args..) 2. select fun1(fun2(args..),args..)<p>* <p> fixed gitee pulls/141</p>** @param function*/protected void processFunction(Function function) {ExpressionList parameters = function.getParameters();if (parameters != null) {parameters.getExpressions().forEach(expression -> {if (expression instanceof SubSelect) {processSelectBody(((SubSelect) expression).getSelectBody());} else if (expression instanceof Function) {processFunction((Function) expression);}});}}/*** 处理子查询等*/protected void processOtherFromItem(FromItem fromItem) {// 去除括号while (fromItem instanceof ParenthesisFromItem) {fromItem = ((ParenthesisFromItem) fromItem).getFromItem();}if (fromItem instanceof SubSelect) {SubSelect subSelect = (SubSelect) fromItem;if (subSelect.getSelectBody() != null) {processSelectBody(subSelect.getSelectBody());}} else if (fromItem instanceof ValuesList) {logger.debug("Perform a subQuery, if you do not give us feedback");} else if (fromItem instanceof LateralSubSelect) {LateralSubSelect lateralSubSelect = (LateralSubSelect) fromItem;if (lateralSubSelect.getSubSelect() != null) {SubSelect subSelect = lateralSubSelect.getSubSelect();if (subSelect.getSelectBody() != null) {processSelectBody(subSelect.getSelectBody());}}}}/*** 处理 sub join** @param subJoin subJoin* @return Table subJoin 中的主表*/private List<Table> processSubJoin(SubJoin subJoin) {List<Table> mainTables = new ArrayList<>();if (subJoin.getJoinList() != null) {List<Table> list = processFromItem(subJoin.getLeft());mainTables.addAll(list);mainTables = processJoins(mainTables, subJoin.getJoinList());}return mainTables;}/*** 处理 joins** @param mainTables 可以为 null* @param joins      join 集合* @return List<Table> 右连接查询的 Table 列表*/private List<Table> processJoins(List<Table> mainTables, List<Join> joins) {// join 表达式中最终的主表Table mainTable = null;// 当前 join 的左表Table leftTable = null;if (mainTables == null) {mainTables = new ArrayList<>();} else if (mainTables.size() == 1) {mainTable = mainTables.get(0);leftTable = mainTable;}//对于 on 表达式写在最后的 join，需要记录下前面多个 on 的表名Deque<List<Table>> onTableDeque = new LinkedList<>();for (Join join : joins) {// 处理 on 表达式FromItem joinItem = join.getRightItem();// 获取当前 join 的表，subJoint 可以看作是一张表List<Table> joinTables = null;if (joinItem instanceof Table) {joinTables = new ArrayList<>();joinTables.add((Table) joinItem);} else if (joinItem instanceof SubJoin) {joinTables = processSubJoin((SubJoin) joinItem);}if (joinTables != null) {// 如果是隐式内连接if (join.isSimple()) {mainTables.addAll(joinTables);continue;}// 当前表是否忽略Table joinTable = joinTables.get(0);List<Table> onTables = null;// 如果不要忽略，且是右连接，则记录下当前表if (join.isRight()) {mainTable = joinTable;if (leftTable != null) {onTables = Collections.singletonList(leftTable);}} else if (join.isLeft()) {onTables = Collections.singletonList(joinTable);} else if (join.isInner()) {if (mainTable == null) {onTables = Collections.singletonList(joinTable);} else {onTables = Arrays.asList(mainTable, joinTable);}mainTable = null;}mainTables = new ArrayList<>();if (mainTable != null) {mainTables.add(mainTable);}// 获取 join 尾缀的 on 表达式列表Collection<Expression> originOnExpressions = join.getOnExpressions();// 正常 join on 表达式只有一个，立刻处理if (originOnExpressions.size() == 1 && onTables != null) {List<Expression> onExpressions = new LinkedList<>();onExpressions.add(builderExpression(originOnExpressions.iterator().next(), onTables));join.setOnExpressions(onExpressions);leftTable = joinTable;continue;}// 表名压栈，忽略的表压入 null，以便后续不处理onTableDeque.push(onTables);// 尾缀多个 on 表达式的时候统一处理if (originOnExpressions.size() > 1) {Collection<Expression> onExpressions = new LinkedList<>();for (Expression originOnExpression : originOnExpressions) {List<Table> currentTableList = onTableDeque.poll();if (CollectionUtils.isEmpty(currentTableList)) {onExpressions.add(originOnExpression);} else {onExpressions.add(builderExpression(originOnExpression, currentTableList));}}join.setOnExpressions(onExpressions);}leftTable = joinTable;} else {processOtherFromItem(joinItem);leftTable = null;}}return mainTables;}// ========== 和 TenantLineInnerInterceptor 存在差异的逻辑：关键，实现权限条件的拼接 ==========/*** 处理条件** @param currentExpression 当前 where 条件* @param table             单个表*/protected Expression builderExpression(Expression currentExpression, Table table) {return this.builderExpression(currentExpression, Collections.singletonList(table));}/*** 处理条件** @param currentExpression 当前 where 条件* @param tables            多个表*/protected Expression builderExpression(Expression currentExpression, List<Table> tables) {// 没有表需要处理直接返回if (CollectionUtils.isEmpty(tables)) {return currentExpression;}// 第一步，获得 Table 对应的数据权限条件Expression dataPermissionExpression = null;for (Table table : tables) {// 构建每个表的权限 Expression 条件Expression expression = buildDataPermissionExpression(table);if (expression == null) {continue;}// 合并到 dataPermissionExpression 中dataPermissionExpression = dataPermissionExpression == null ? expression: new AndExpression(dataPermissionExpression, expression);}// 第二步，合并多个 Expression 条件if (dataPermissionExpression == null) {return currentExpression;}if (currentExpression == null) {return dataPermissionExpression;}// ① 如果表达式为 Or，则需要 (currentExpression) AND dataPermissionExpressionif (currentExpression instanceof OrExpression) {return new AndExpression(new Parenthesis(currentExpression), dataPermissionExpression);}// ② 如果表达式为 And，则直接返回 where AND dataPermissionExpressionreturn new AndExpression(currentExpression, dataPermissionExpression);}/*** 构建指定表的数据权限的 Expression 过滤条件** @param table 表* @return Expression 过滤条件*/private Expression buildDataPermissionExpression(Table table) {// 生成条件Expression allExpression = null;for (DataPermissionRule rule : ContextHolder.getRules()) {// 如果有匹配的规则，说明可重写。// 为什么不是有 allExpression 非空才重写呢？在生成 column = value 过滤条件时，会因为 value 不存在，导致未重写。// 这样导致第一次无 value，被标记成无需重写；但是第二次有 value，此时会需要重写。ContextHolder.setRewrite(true);// 单条规则的条件String tableName = getTableName(table);Expression oneExpress = rule.getExpression(tableName, table.getAlias());if (oneExpress == null) {continue;}// 拼接到 allExpression 中allExpression = allExpression == null ? oneExpress: new AndExpression(allExpression, oneExpress);}return allExpression;}/*** 判断 SQL 是否重写。如果没有重写，则添加到 {@link MappedStatementCache} 中** @param ms MappedStatement*/private void addMappedStatementCache(MappedStatement ms) {if (ContextHolder.getRewrite()) {return;}// 无重写，进行添加mappedStatementCache.addNoRewritable(ms, ContextHolder.getRules());}/*** SQL 解析上下文，方便透传 {@link DataPermissionRule} 规则** @author yudao*/static final class ContextHolder {/*** 该 {@link MappedStatement} 对应的规则*/private static final ThreadLocal<List<DataPermissionRule>> RULES = ThreadLocal.withInitial(Collections::emptyList);/*** SQL 是否进行重写*/private static final ThreadLocal<Boolean> REWRITE = ThreadLocal.withInitial(() -> Boolean.FALSE);public static void init(List<DataPermissionRule> rules) {RULES.set(rules);REWRITE.set(false);}public static void clear() {RULES.remove();REWRITE.remove();}public static boolean getRewrite() {return REWRITE.get();}public static void setRewrite(boolean rewrite) {REWRITE.set(rewrite);}public static List<DataPermissionRule> getRules() {return RULES.get();}}/*** {@link MappedStatement} 缓存* 目前主要用于，记录 {@link DataPermissionRule} 是否对指定 {@link MappedStatement} 无效* 如果无效，则可以避免 SQL 的解析，加快速度** @author yudao*/static final class MappedStatementCache {/*** 指定数据权限规则，对指定 MappedStatement 无需重写（不生效)的缓存* <p>* value：{@link MappedStatement#getId()} 编号*/@Getterprivate final Map<Class<? extends DataPermissionRule>, Set<String>> noRewritableMappedStatements = new ConcurrentHashMap<>();/*** 判断是否无需重写* ps：虽然有点中文式英语，但是容易读懂即可** @param ms    MappedStatement* @param rules 数据权限规则数组* @return 是否无需重写*/public boolean noRewritable(MappedStatement ms, List<DataPermissionRule> rules) {// 如果规则为空，说明无需重写if (org.springframework.util.CollectionUtils.isEmpty(rules)) {return true;}// 任一规则不在 noRewritableMap 中，则说明可能需要重写for (DataPermissionRule rule : rules) {Set<String> mappedStatementIds = noRewritableMappedStatements.get(rule.getClass());if (mappedStatementIds != null && !mappedStatementIds.stream().anyMatch(item -> item.equals(ms.getId()))) {return false;}return false;}return true;}/*** 添加无需重写的 MappedStatement** @param ms    MappedStatement* @param rules 数据权限规则数组*/public void addNoRewritable(MappedStatement ms, List<DataPermissionRule> rules) {for (DataPermissionRule rule : rules) {Set<String> mappedStatementIds = noRewritableMappedStatements.get(rule.getClass());if (CollectionUtils.isEmpty(mappedStatementIds)) {mappedStatementIds.add(ms.getId());} else {noRewritableMappedStatements.put(rule.getClass(), Arrays.stream(new String[]{ms.getId()}).collect(Collectors.toSet()));}}}/*** 清空缓存* 目前主要提供给单元测试*/public void clear() {noRewritableMappedStatements.clear();}}/*** 获得 Table 对应的表名* <p>* 兼容 MySQL 转义表名 `t_xxx`** @param table 表* @return 去除转移字符后的表名*/public static String getTableName(Table table) {String tableName = table.getName();if (tableName.startsWith(MYSQL_ESCAPE_CHARACTER) && tableName.endsWith(MYSQL_ESCAPE_CHARACTER)) {tableName = tableName.substring(1, tableName.length() - 1);}return tableName;}}

配置SpringBean

将三个Bean进行添加：

MybatisPlusInterceptor ，Sql拦截器；
DataPermissionAnnotationAdvisor ，DataScope拦截器；
DeptDataPermissionRule ，部门权限解析对象，同时注册需要添加部门权限的表名和部门ID字段名，有新增表时需要在这里设置。

关于springboot 拦截器之Advisor不生效问题可参考该文章。

package com.luo.chengrui.labs.lab02.config;import com.baomidou.mybatisplus.extension.plugins.MybatisPlusInterceptor;
import com.baomidou.mybatisplus.extension.plugins.inner.InnerInterceptor;
import com.baomidou.mybatisplus.extension.plugins.inner.PaginationInnerInterceptor;
import com.luo.chengrui.labs.lab02.datapermission.*;
import com.luo.chengrui.labs.lab02.datapermission.dept.DeptDataPermissionRule;
import com.luo.chengrui.labs.lab02.interceptor.DataPermissionAnnotationAdvisor;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.context.annotation.Role;import java.util.ArrayList;
import java.util.List;/*** 数据权限的自动配置类** @author yudao*/
@Configuration
public class DataPermissionConfiguration {@Beanpublic MybatisPlusInterceptor mybatisPlusInterceptor(List<DataPermissionRule> dataPermissionRule) {MybatisPlusInterceptor mybatisPlusInterceptor = new MybatisPlusInterceptor();// 分页插件
//        mybatisPlusInterceptor.addInnerInterceptor(new PaginationInnerInterceptor());//添加权限拦截器。DataPermissionDatabaseInterceptor inner = new DataPermissionDatabaseInterceptor(dataPermissionRule);List<InnerInterceptor> inners = new ArrayList<>(mybatisPlusInterceptor.getInterceptors());inners.add(0, inner);mybatisPlusInterceptor.setInterceptors(inners);//        MybatisDatabaseInterceptor mybatisDatabaseInterceptor = new MybatisDatabaseInterceptor();
//        List<InnerInterceptor> inners = new ArrayList<>(mybatisPlusInterceptor.getInterceptors());
//        inners.add(0, mybatisDatabaseInterceptor);
//        mybatisPlusInterceptor.setInterceptors(inners);return mybatisPlusInterceptor;}/*** 初始化部门权限 bean 。** @return*/@Beanpublic DeptDataPermissionRule deptDataPermissionRule() {// 创建 DeptDataPermissionRule 对象DeptDataPermissionRule rule = new DeptDataPermissionRule();// 用户表 需要作权限过滤，用户表中部门字段为dept_id。rule.addDeptColumn("users","dept_id");// 请假流程表，也需要按部门权限过滤；但wf_leave表中部门字段为：deptid，则应按如下配置rule.addDeptColumn("wf_leave","deptid");return rule;}/*** 权限注解拦截器。** @return*/@Bean@Role(2)public DataPermissionAnnotationAdvisor dataPermissionAnnotationAdvisor() {return new DataPermissionAnnotationAdvisor();}}

数据权限应用

UserService.java
定义了用户列表查询接口，添加了@DataScope注解

package com.luo.chengrui.labs.lab02.service;import com.luo.chengrui.labs.lab02.annotation.DataScope;
import com.luo.chengrui.labs.lab02.dataobject.UserDO;
import com.luo.chengrui.labs.lab02.mapper.UserMapper;
import org.springframework.aop.framework.AopContext;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.stereotype.Service;import java.util.List;/*** @author* @version 1.0.0* @description* @createTime 2023/07/21*/
@Service
public class UserService {@Autowiredprivate UserMapper userMapper;private UserService self() {return (UserService) AopContext.currentProxy();}@DataScopepublic List<UserDO> selectList() {return userMapper.selectList();}
}

UserMapper.java

package com.luo.chengrui.labs.lab02.mapper;import com.luo.chengrui.labs.lab02.dataobject.UserDO;
import org.apache.ibatis.annotations.Mapper;
import org.apache.ibatis.annotations.Param;import java.util.List;@Mapper
public interface UserMapper {UserDO selectById(@Param("id") Integer id);List<UserDO> selectList();
}

UserMapper.xml

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE mapper PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN" "http://mybatis.org/dtd/mybatis-3-mapper.dtd">
<mapper namespace="com.luo.chengrui.labs.lab02.mapper.UserMapper"><sql id="FIELDS">id, username</sql><select id="selectById" parameterType="Integer" resultType="UserDO">SELECT<include refid="FIELDS"/>FROM usersWHERE id = #{id}</select><select id="selectList" resultType="UserDo">SELECT<include refid="FIELDS"/>FROM users</select></mapper>

功能验证1：

DataPermissionConfiguration.java 类中 DeptDataPermissionRule 对象初始化设置用户表权限，配置如下：

 @Beanpublic DeptDataPermissionRule deptDataPermissionRule() {// 创建 DeptDataPermissionRule 对象DeptDataPermissionRule rule = new DeptDataPermissionRule();// 用户表 需要作权限过滤，用户表中部门字段为dept_id。rule.addDeptColumn("users","dept_id");// 请假流程表，也需要按部门权限过滤；但wf_leave表中部门字段为：deptid，则应按如下配置return rule;}

运行结果如下：
在这里插入图片描述

功能测试2：不添加表和字段

不添加表字段，或者添加的表不是users表时，再执行用户查询。

@Beanpublic DeptDataPermissionRule deptDataPermissionRule() {// 创建 DeptDataPermissionRule 对象DeptDataPermissionRule rule = new DeptDataPermissionRule();// 用户表 需要作权限过滤，用户表中部门字段为dept_id。//rule.addDeptColumn("users","dept_id");// 请假流程表，也需要按部门权限过滤；但wf_leave表中部门字段为：deptid，则应按如下配置rule.addDeptColumn("wf_leavel","deptid");return rule;}