liferay 权限策略

    先上一张大家都很熟悉的图



这张图可以说是整个liferay的核心，他和权限密不可分，在权限体系里，每一个箭头和UI设置和API一一对应，今天讲权限，不讲代码只讲理论。liferay的权限策略分为五种，默认是采用第五种，因为第五种性能最好，如果要修改则在portal-ext.properies加上permissions.user.check.algorithm=3

讲策略之前我们先讲一下，liferay权限分为几种

     A:Group Role Permission: 用户加入的Group(Community/Organization)所拥有的角色的权限
Is the user connected to one of the permissions via group or organization roles?

<sql id="com.liferay.portal.service.persistence.PermissionFinder.countByGroupsRoles">
	<![CDATA[
		SELECT
			COUNT(*) AS COUNT_VALUE
		FROM
			Groups_Roles
		INNER JOIN
			Roles_Permissions ON
				(Roles_Permissions.roleId = Groups_Roles.roleId)
		INNER JOIN
			Permission_ ON
				(Permission_.permissionId = Roles_Permissions.permissionId)
		WHERE
			([$PERMISSION_IDS$]) AND
			([$GROUP_IDS$])
	]]>
</sql>

B:Group Permission: 分配给Group(包括Community/Organization/User Group三类Group)的权限
Is the user associated with groups or organizations that are directly connected to one of the permissions?

<sql id="com.liferay.portal.service.persistence.PermissionFinder.countByGroupsPermissions">
	<![CDATA[
		SELECT
			COUNT(*) AS COUNT_VALUE
		FROM
			Permission_
		INNER JOIN
			Groups_Permissions ON
				(Groups_Permissions.permissionId = Permission_.permissionId)
		WHERE
			([$PERMISSION_IDS$]) AND
			([$GROUP_IDS$])
	]]>
</sql>

C:User Role Permission: 分配给用户Regular类型角色的权限
Is the user connected to one of the permissions via user roles?

<sql id="com.liferay.portal.service.persistence.PermissionFinder.countByUsersRoles">
	<![CDATA[
		SELECT
			COUNT(*) AS COUNT_VALUE
		FROM
			Users_Roles
		INNER JOIN
			Roles_Permissions ON
				(Roles_Permissions.roleId = Users_Roles.roleId)
		INNER JOIN
			Permission_ ON
				(Permission_.permissionId = Roles_Permissions.permissionId)
		WHERE
			([$PERMISSION_IDS$]) AND
			(Users_Roles.userId = ?)
	]]>
</sql>

D:User Group Role Permission: 用户加Group(Community/Organization)所拥有该社区角色的权限
Is the user connected to one of the permissions via user group roles?

<sql id="com.liferay.portal.service.persistence.PermissionFinder.countByUserGroupRole">
	<![CDATA[
		SELECT
			COUNT(*) AS COUNT_VALUE
		FROM
			UserGroupRole
		INNER JOIN
			Roles_Permissions ON
				(Roles_Permissions.roleId = UserGroupRole.roleId and UserGroupRole.groupId = ?)
		INNER JOIN
			Permission_ ON
				(Permission_.permissionId = Roles_Permissions.permissionId)
		WHERE
			([$PERMISSION_IDS$]) AND
			(UserGroupRole.userId = ?)
	]]>
</sql>

E:User Permission: 直接分配给user的权限
Is the user directly connected to one of the permissions?

<sql id="com.liferay.portal.service.persistence.PermissionFinder.countByUsersPermissions">
	<![CDATA[
		SELECT
			COUNT(*) AS COUNT_VALUE
		FROM
			Permission_
		INNER JOIN
			Users_Permissions ON
				(Users_Permissions.permissionId = Permission_.permissionId)
		WHERE
			([$PERMISSION_IDS$]) AND
			(Users_Permissions.userId = ?)
	]]>
</sql>

F:Role Permission: 分配给角色(包括Regular/Community/Organization三种类型角色)的权限
Is the user associated with a role that is directly connected to one of the permissions?

<sql id="com.liferay.portal.service.persistence.PermissionFinder.countByUsersPermissions">
	<![CDATA[
		SELECT
			COUNT(*) AS COUNT_VALUE
		FROM
			Permission_
		INNER JOIN
			Users_Permissions ON
				(Users_Permissions.permissionId = Permission_.permissionId)
		WHERE
			([$PERMISSION_IDS$]) AND
			(Users_Permissions.userId = ?)
	]]>
</sql>

讲完了这check权限的分类，我们现在开始讲liferay的权限策略
Algorithm 1. A or B or C or D or E
Algorithm 2. A and B and C and D and E
Algorithm 3. B or E or F
Algorithm 4. B and E and F
Algorithm 5. F

     注意：liferay推荐使用第五种是因为他效率最高，第一、二种是不可能会用到的，因为那是早期版本的设计，有些表例如：Groups_Roles都已经放弃了。如果不是必须不要去修改策略。


     现在说说他的缺点：用过权限开发的人都会发现，他的权限粒度很高，这样带来了一个问题就是效率会变低（虽然他用了大量的cache），最讨厌的是liferay权限体系总是，先把数据全搜索出来，当你要进行操作（VIEW）时，他才会跳出个红框框告诉你，你没有权限，这样很不友好。最典型的是list分页，点其中一条才会告诉你没权限，客户居然投诉：说我们有诱导之嫌。liferay现有的api是肯定做不到直接搜索出合法数据，虽然liferay在有的元素的service层，先把数据全取出来然后再一个一个的过滤来达到这样的效果，这样我觉得实在无法容忍，为了一页20条的数据，你居然要搜索出所有数据。为此我在项目中专门写了一个类，把一些共用的权限sql写了出来

public static String getPrimkeyByRolePermission(long companyId, 
			String name, int scope, String actionId, String roleIds){		
		StringBuffer sql = new StringBuffer();
		sql.append("SELECT DISTINCT RESOURCE_.PRIMKEY FROM RESOURCE_ "+
			"LEFT JOIN RESOURCECODE ON (RESOURCE_.CODEID = RESOURCECODE.CODEID) "+
			"LEFT JOIN PERMISSION_ ON (PERMISSION_.RESOURCEID = RESOURCE_.RESOURCEID) "+
			"LEFT JOIN ROLES_PERMISSIONS ON (PERMISSION_.PERMISSIONID = ROLES_PERMISSIONS.PERMISSIONID) ");
		sql.append("WHERE RESOURCECODE.COMPANYID = "+ companyId +" " +
				"AND RESOURCECODE.NAME = '"+ name +"' AND RESOURCECODE.SCOPE = "+ scope +" ");
		sql.append("AND  PERMISSION_.COMPANYID = "+ companyId +
				" AND PERMISSION_.ACTIONID = '"+ actionId +"' AND " +
				"ROLES_PERMISSIONS.ROLEID IN ("+ roleIds +")");
		return sql.toString();
	}
	
	public static String getPrimkeyByRolePermission(long companyId,
			String name, int scope, String actionId, long userId){		
		StringBuffer sql = new StringBuffer();
		sql.append("SELECT DISTINCT RESOURCE_.PRIMKEY FROM RESOURCE_ "+
			"LEFT JOIN RESOURCECODE ON (RESOURCE_.CODEID = RESOURCECODE.CODEID) "+
			"LEFT JOIN PERMISSION_ ON (PERMISSION_.RESOURCEID = RESOURCE_.RESOURCEID) "+
			"LEFT JOIN USERS_PERMISSIONS ON (PERMISSION_.PERMISSIONID = USERS_PERMISSIONS.PERMISSIONID) ");
		sql.append("WHERE RESOURCECODE.COMPANYID = "+ companyId +" " +
				"AND RESOURCECODE.NAME = '"+ name +"' AND RESOURCECODE.SCOPE = "+ scope +" ");
		sql.append("AND  PERMISSION_.COMPANYID = "+ companyId +
				" AND PERMISSION_.ACTIONID = '"+ actionId +"' AND " +
				"USERS_PERMISSIONS.USERID = " + userId) ;
		return sql.toString();
	}
	
	public static String getPrimkeyByGroupPermission(long companyId,
			String name, int scope, String actionId, long userId, long groupId){		
		StringBuffer sql = new StringBuffer();
		sql.append("SELECT DISTINCT RESOURCE_.PRIMKEY FROM RESOURCE_ "+
			"LEFT JOIN RESOURCECODE ON (RESOURCE_.CODEID = RESOURCECODE.CODEID) "+
			"LEFT JOIN PERMISSION_ ON (PERMISSION_.RESOURCEID = RESOURCE_.RESOURCEID) "+
			"LEFT JOIN GROUPS_PERMISSIONS ON (PERMISSION_.PERMISSIONID = GROUPS_PERMISSIONS.PERMISSIONID) +" +
			"LEFT JOIN USERS_GROUPS ON (GROUPS_PERMISSIONS.GROUPID = USERS_GROUPS.GROUPID) ");
		sql.append("WHERE RESOURCECODE.COMPANYID = "+ companyId +" " +
				"AND RESOURCECODE.NAME = '"+ name +"' AND RESOURCECODE.SCOPE = "+ scope +" ");
		sql.append("AND  PERMISSION_.COMPANYID = "+ companyId +
				" AND PERMISSION_.ACTIONID = '"+ actionId +"' AND " +
				"GROUPS_PERMISSIONS.GROUPID = " + groupId +" AND " +
				"USERS_GROUPS.USERID = " + userId) ;
		return sql.toString();
	}
	
	public static String getPrimkeyByPermissionAlgorithm3(long companyId,
			String name, int scope, String actionId, long userId, long groupId){
		StringBuffer sql = new StringBuffer();
		sql.append("SELECT DISTINCT RESOURCE_.PRIMKEY ID FROM RESOURCE_, RESOURCECODE, PERMISSION_, ");
		
		sql.append("(");

		sql.append("SELECT ROLES_PERMISSIONS.PERMISSIONID FROM ROLES_PERMISSIONS " +
				"JOIN USERS_ROLES ON (ROLES_PERMISSIONS.ROLEID = USERS_ROLES.USERID) " +
				"WHERE USERS_ROLES.USERID = " + userId);
		
		sql.append(") AS R, ");
		
		sql.append("(");

		sql.append("SELECT USERS_PERMISSIONS.PERMISSIONID FROM USERS_PERMISSIONS " +
				"WHERE USERS_PERMISSIONS.USERID = " + userId);
		
		sql.append(") AS U, ");
		
		sql.append("(");

		sql.append("SELECT GROUPS_PERMISSIONS.PERMISSIONID FROM GROUPS_PERMISSIONS " +
				"JOIN USERS_GROUPS ON (GROUPS_PERMISSIONS.GROUPID = USERS_GROUPS.GROUPID) " +
				"WHERE USERS_GROUPS.GROUPID = " + groupId);	
		
		sql.append(") AS G ");

		sql.append("WHERE RESOURCECODE.COMPANYID = "+ companyId +" " +
				"AND RESOURCECODE.NAME = '"+ name +"' AND RESOURCECODE.SCOPE = "+ scope +" ");
		sql.append("AND  PERMISSION_.COMPANYID = "+ companyId +
				" AND PERMISSION_.ACTIONID = '"+ actionId + "' ") ;
		
		sql.append("AND RESOURCE_.CODEID = RESOURCECODE.CODEID "+
			"AND PERMISSION_.RESOURCEID = RESOURCE_.RESOURCEID ");	
		
		sql.append("AND (PERMISSION_.PERMISSIONID = R.PERMISSIONID OR " +
				"PERMISSION_.PERMISSIONID = U.PERMISSIONID OR " +
				"PERMISSION_.PERMISSIONID = G.PERMISSIONID)" );

		return sql.toString();
	}

注意：我们到的是策略3,所以我分别写了B,E,F权限SQL,及一个总的Algorithm3SQL,这些SQL返回的都是你要搜索元素的key，这些SQL在liferay中是通用的，只要你是用的liferay的权限他都是合适的，你要把他放在你sql合适的位置，注意最好不要使用in，这样效率是比较低的。