Mysql 向查询添加第二个联接会使值重复

我有一个返回用户信息的sql查询，将他的权限与另一个注册了他的角色权限的表连接起来：

SELECT U.*, array_agg(RP.permission) as permission
FROM user U
JOIN role_permissions RP ON RP.role = F.role
WHERE U.email=$1
GROUP BY U.cpf -- U.cpf is the primary key

这返回了以下结果：

{
  cpf: '11111111111',
  role: 'c-level',
  area: 'Administrativo Financeiro',
  ...,
  permission: [
    'can_edit_all_metrics',
    'can_edit_all_plans',
    'can_view_all_metrics',
    'can_generate_reports',
    ...
  ]
}

现在我尝试将他有权访问的区域添加到查询中，并尝试使用以下查询：

SELECT U.*, array_agg(RP.permission) as permission, 
        array_agg(AC.has_access_to) as areas_with_access
FROM user U
JOIN role_permissions RP ON RP.role = F.role
JOIN area_access AC ON AC.area = U.area
WHERE  U.email=$1
GROUP BY U.cpf

我得到的结果是：

{
  cpf: '11111111111',
  role: 'c-level',
  area: 'Administrativo Financeiro',
  ...,
  permission: [
    'can_edit_all_metrics',
    'can_edit_all_metrics',
    'can_edit_all_plans',
    'can_edit_all_plans',
    'can_view_all_metrics',
    'can_view_all_metrics',
    'can_generate_reports',
    'can_generate_reports',
    ...
  ],
  areas_with_access: [
    'Financeiro',
    'Test',
    'Financeiro',
    'Test',
    'Financeiro',
    'Test',
    'Financeiro',
    'Test',
    ...,
  ]
}

我已经尝试过左/右加入第二个表，将AC.area和AC.has\u access\u添加到GROUPBY子句中，但什么也没有发生。 通过将RP.permission添加到组中，使其仅返回两个权限，因为他只有两个可以访问的区域

我可以只向聚合函数添加

DISTINCT

，但我想知道是否有什么地方我做错了

这是角色权限表：

create table if not exists role_permissions (
    role character varying(50),
    permission character varying(50),
    
    constraint pk_role_permissions primary key (role, permission),
    constraint fk_role_permissions_role foreign key (role) references roles(role) on update cascade on delete cascade,
    constraint fk_role_permissions_permission foreign key (permission) references permissions(permission) on update cascade on delete cascade,
);

这是区域访问表：

create table if not exists area_access (
    area character varying(50) NOT NULL,
    has_access_to character varying(50) NOT NULL,
    
    constraint pk_area_access primary key (area, has_access_to),

    constraint fk_area foreign key (area) references area(area) on update cascade on delete cascade,
    constraint fk_has_access_to foreign key (has_access_to) references area(area) on update cascade on delete cascade,

    constraint ck_area_different CHECK (area != has_access_to)
);

---

@RiggsFolly所以在这种情况下我必须使用

DISTINCT

？