SQL Server 2017 图形表

 

SQL SERVER 2017 提供了图形数据库功能（实际是在表级别提供），图形数据库集成到关系型数据库中，便于关系型数据库使用SQL操作。图形数据库属于NOSQL类型，如Neo4j 等。

 

图形数据库是一个集合节点（或顶点） 和边缘 （或关系）。 节点表示实体 （例如，一个人或组织），边缘表示连接 （如组件或好友） 的两个节点之间的关系。 节点和边缘可能具有与之关联的属性。总的来说，图形数据库就是存储实体与关系的数据库。

 

基础概念：

节点表：存储一类节点信息（一般实体表），如“人员表”、“城市表”、“餐馆表”等

边缘表：关系表，存储节点之间的有向关系。

说到关系，我们一直都接触过。比如部门人员关系、物品层级分类等，都用到父级别，这是比较简单的关系模型。但以往我们用的这种关系，我们默认是知道父级别和子级别对应的字段，这样才能关联。因为只有实体，没有关系，当我们需要对关系标注一些额外信息时（如我们之间关系友好程度、权重等），就可能需要单独的一张关系表了。多麻烦，原本有id和父级id就体现关系了，现在又得多家一张表。查询的时候还要按顺序递归父节点。

 

节点表和边缘表都可以添加自己的属性（字段），创建节点表需指定“NODE”，创建边缘表需制定“EDGE”。节点表和边缘表都会添加隐藏的字段，如下图出属性之外的字段。

 

 

示例：

创建三个节点对象表： Person（人）、Restaurant（餐馆）、City（城市）

它们之前的边缘表（关系表）：likes（人喜欢的城市、餐馆）、friendOf（人与人之间的好友关系）、livesIn（人生活在哪个城市）、locatedIn（餐馆位于哪个城市）

脚本如下：

-- 创建一个数据库
CREATE DATABASE graphdemo;
go
USE  graphdemo;
go

-- 创建节点表（声明为 NODE）
CREATE TABLE Person (
  ID INTEGER PRIMARY KEY, 
  name VARCHAR(100)
) AS NODE;
go
CREATE TABLE Restaurant (
  ID INTEGER NOT NULL, 
  name VARCHAR(100), 
  city VARCHAR(100)
) AS NODE;
go
CREATE TABLE City (
  ID INTEGER PRIMARY KEY, 
  name VARCHAR(100), 
  stateName VARCHAR(100)
) AS NODE;
go
-- 创建边缘表（声明为 EDGE）
CREATE TABLE likes (rating INTEGER) AS EDGE;
CREATE TABLE friendOf AS EDGE;
CREATE TABLE livesIn AS EDGE;
CREATE TABLE locatedIn AS EDGE;
go
-- 插入节点数据
INSERT INTO Person VALUES (1,'John');
INSERT INTO Person VALUES (2,'Mary');
INSERT INTO Person VALUES (3,'Alice');
INSERT INTO Person VALUES (4,'Jacob');
INSERT INTO Person VALUES (5,'Julie');
go
INSERT INTO Restaurant VALUES (1,'Taco Dell','Bellevue');
INSERT INTO Restaurant VALUES (2,'Ginger and Spice','Seattle');
INSERT INTO Restaurant VALUES (3,'Noodle Land', 'Redmond');
go
INSERT INTO City VALUES (1,'Bellevue','wa');
INSERT INTO City VALUES (2,'Seattle','wa');
INSERT INTO City VALUES (3,'Redmond','wa');
go
-- 插入边缘数据，需要节点提供 $node_id 和 $from_id 两个字段（属于隐藏字段）
INSERT INTO likes VALUES ((SELECT $node_id FROM Person WHERE id = 1), (SELECT $node_id FROM Restaurant WHERE id = 1),9);
INSERT INTO likes VALUES ((SELECT $node_id FROM Person WHERE id = 2), (SELECT $node_id FROM Restaurant WHERE id = 2),9);
INSERT INTO likes VALUES ((SELECT $node_id FROM Person WHERE id = 3), (SELECT $node_id FROM Restaurant WHERE id = 3),9);
INSERT INTO likes VALUES ((SELECT $node_id FROM Person WHERE id = 4), (SELECT $node_id FROM Restaurant WHERE id = 3),9);
INSERT INTO likes VALUES ((SELECT $node_id FROM Person WHERE id = 5), (SELECT $node_id FROM Restaurant WHERE id = 3),9);
go
INSERT INTO livesIn VALUES ((SELECT $node_id FROM Person WHERE id = 1),(SELECT $node_id FROM City WHERE id = 1));
INSERT INTO livesIn VALUES ((SELECT $node_id FROM Person WHERE id = 2),(SELECT $node_id FROM City WHERE id = 2));
INSERT INTO livesIn VALUES ((SELECT $node_id FROM Person WHERE id = 3),(SELECT $node_id FROM City WHERE id = 3));
INSERT INTO livesIn VALUES ((SELECT $node_id FROM Person WHERE id = 4),(SELECT $node_id FROM City WHERE id = 3));
INSERT INTO livesIn VALUES ((SELECT $node_id FROM Person WHERE id = 5),(SELECT $node_id FROM City WHERE id = 1));
go
INSERT INTO locatedIn VALUES ((SELECT $node_id FROM Restaurant WHERE id = 1),(SELECT $node_id FROM City WHERE id =1));
INSERT INTO locatedIn VALUES ((SELECT $node_id FROM Restaurant WHERE id = 2),(SELECT $node_id FROM City WHERE id =2));
INSERT INTO locatedIn VALUES ((SELECT $node_id FROM Restaurant WHERE id = 3),(SELECT $node_id FROM City WHERE id =3));
go
INSERT INTO friendof VALUES ((SELECT $NODE_ID FROM person WHERE ID = 1), (SELECT $NODE_ID FROM person WHERE ID = 2));
INSERT INTO friendof VALUES ((SELECT $NODE_ID FROM person WHERE ID = 2), (SELECT $NODE_ID FROM person WHERE ID = 3));
INSERT INTO friendof VALUES ((SELECT $NODE_ID FROM person WHERE ID = 3), (SELECT $NODE_ID FROM person WHERE ID = 1));
INSERT INTO friendof VALUES ((SELECT $NODE_ID FROM person WHERE ID = 4), (SELECT $NODE_ID FROM person WHERE ID = 2));
INSERT INTO friendof VALUES ((SELECT $NODE_ID FROM person WHERE ID = 5), (SELECT $NODE_ID FROM person WHERE ID = 4));
go

当查询关系的时候，需要新的条件关键字  MATCH （更多参考​​MATCH (Transact SQL)​​）

-- 找到 John 喜欢的餐馆
SELECT Restaurant.name as Restaurant
FROM Person, likes, Restaurant
WHERE MATCH (Person-(likes)->Restaurant)
AND Person.name = 'John';
go
-- 找到 John 的朋友喜欢的餐馆
SELECT person1.name,person2.name as friend,Restaurant.name as Restaurant
FROM Person person1, Person person2, likes, friendOf, Restaurant
WHERE MATCH(person1-(friendOf)->person2-(likes)->Restaurant) 
AND person1.name='John';
go
--有哪些人生活与喜欢的餐馆在同一个城市
SELECT Person.name
FROM Person, likes, Restaurant, livesIn, City, locatedIn
WHERE MATCH (Person-(likes)->Restaurant-(locatedIn)->City AND Person-(livesIn)->City);
go

查询结果：

 

其他元素据查看：

--找出节点表和边缘表
select * from sys.tables where is_node =1
select * from sys.tables where is_edge =1
go
--查看节点表和边缘表的字段（有较多隐藏字段）
select [object_id],name,is_hidden,graph_type,graph_type_desc,*
from sys.columns where [object_id] = OBJECT_ID('Person')
go
select [object_id],name,is_hidden,graph_type,graph_type_desc
from sys.columns where [object_id] = OBJECT_ID('friendOf')
go

最后，删除测试数据：

USE graphdemo;
go
DROP TABLE IF EXISTS likes;
DROP TABLE IF EXISTS Person;
DROP TABLE IF EXISTS Restaurant;
DROP TABLE IF EXISTS City;
DROP TABLE IF EXISTS friendOf;
DROP TABLE IF EXISTS livesIn;
DROP TABLE IF EXISTS locatedIn;
go
USE master;
go
DROP DATABASE graphdemo;
go

注意：字段 $node_id、$edge_id、$from_id、$to_id 是不可更改的，若要更改，只能删除重新添加

 

 

参考：​​Graph processing with SQL Server and Azure SQL Database​​