[feature](graph)Support querying data from the Nebula graph database

[chenlinzhong]

Proposed changes

Problem summary

Support querying data from the Nebula graph database
This feature comes from the needs of commercial customers who have used Doris and Nebula, hoping to connect these two databases

changes mainly include:

• add New Graph Database JDBC Type
• Adapt the type and map the graph to the Doris type

How to use ?

Nebula-graph type map

nebula	Doris
tinyint/samllint/int/int64	bigint
double/float	double
date	date
timestamp	bigint
datetime	datetime
bool	boolean
vertex/edge/path/list/set/time等	varchar

NebulaGraphTest （仅支持查询）

nebula版本	JDBC驱动版本
3.0.0	nebula-jdbc-3.0.0-jar-with-dependencies.jar

#step1.在nebula创建测试数据
#1.1 创建结点
(root@nebula) [basketballplayer]> CREATE TAG test(t_str string, 
    t_int int, 
    t_date date,
    t_datetime datetime,
    t_bool bool,
    t_timestamp timestamp,
    t_float float,
    t_double double
);

#1.2 插入数据
(root@nebula) [basketballplayer]> INSERT VERTEX test_type(t_str,t_int,t_date,t_datetime,t_bool,t_timestamp,t_float,t_double) values "zhangshan":("zhangshan",1000,date("2023-01-01"),datetime("2023-01-23 15:23:32"),true,1234242423,1.2,1.35);

#1.3 查询数据
(root@nebula) [basketballplayer]> match (v:test_type) where id(v)=="zhangshan" return v.test_type.t_str,v.test_type.t_int,v.test_type.t_date,v.test_type.t_datetime,v.test_type.t_bool,v.test_type.t_timestamp,v.test_type.t_float,v.test_type.t_double,v limit 30;
+-------------------+-------------------+--------------------+----------------------------+--------------------+-------------------------+---------------------+----------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| v.test_type.t_str | v.test_type.t_int | v.test_type.t_date | v.test_type.t_datetime     | v.test_type.t_bool | v.test_type.t_timestamp | v.test_type.t_float | v.test_type.t_double | v                                                                                                                                                                                                         |
+-------------------+-------------------+--------------------+----------------------------+--------------------+-------------------------+---------------------+----------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| "zhangshan"       | 1000              | 2023-01-01         | 2023-01-23T15:23:32.000000 | true               | 1234242423              | 1.2000000476837158  | 1.35                 | ("zhangshan" :test_type{t_bool: true, t_date: 2023-01-01, t_datetime: 2023-01-23T15:23:32.000000, t_double: 1.35, t_float: 1.2000000476837158, t_int: 1000, t_str: "zhangshan", t_timestamp: 1234242423}) |
+-------------------+-------------------+--------------------+----------------------------+--------------------+-------------------------+---------------------+----------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
Got 1 rows (time spent 1616/2048 us)

Mon, 17 Apr 2023 17:23:14 CST


#step2.在doris中创建外表
#2.1 创建一个resource
MySQL [test_db]> CREATE EXTERNAL RESOURCE gg_jdbc_resource 
properties (
   "type"="jdbc",
   "user"="root",
   "password"="123",
   "jdbc_url"="jdbc:nebula://127.0.0.1:9669/basketballplayer",
   "driver_url"="file:///home/clz/baidu/bdg/doris/be/lib/nebula-jdbc-3.0.0-jar-with-dependencies.jar",  --仅支持本地路径，需放到be/lib目录下--
   "driver_class"="com.vesoft.nebula.jdbc.NebulaDriver"
);

#2.2 创建一个外表，这个主要是告诉doris如何解析nebulagraph返回的数据
MySQL [test_db]> CREATE TABLE `test_type` ( 
 `t_str` varchar(64),
 `t_int` bigint,
 `t_date` date,
 `t_datetime` datetime,
 `t_bool` boolean,
 `t_timestamp` bigint,
 `t_float` double,
 `t_double` double,
 `t_vertx`  varchar(128) --vertex对应doris类型是varchar---
) ENGINE=JDBC
PROPERTIES (
"resource" = "gg_jdbc_resource",
"table" = "xx",  --因为graph没有表的概念，这里随便填一个值--
"table_type"="nebula"
);

#2.3 查询graph外表,用g()函数把图的nGQL透传给nebula
MySQL [test_db]> select * from test_type where g('match (v:test_type) where id(v)=="zhangshan" return v.test_type.t_str,v.test_type.t_int,v.test_type.t_date,v.test_type.t_datetime,v.test_type.t_bool,v.test_type.t_timestamp,v.test_type.t_float,v.test_type.t_double,v')\G;
*************************** 1. row ***************************
      t_str: zhangshan
      t_int: 1000
     t_date: 2023-01-01
 t_datetime: 2023-01-23 15:23:32
     t_bool: 1
t_timestamp: 1234242423
    t_float: 1.2000000476837158
   t_double: 1.35
    t_vertx: ("zhangshan" :test_type {t_datetime: utc datetime: 2023-01-23T15:23:32.000000, timezoneOffset: 0, t_timestamp: 1234242423, t_date: 2023-01-01, t_double: 1.35, t_str: "zhangshan", t_int: 1000, t_bool: true, t_float: 1.2000000476837158})
1 row in set (0.024 sec)

#2.3 与doris的其他表进行关联查询
#假设有张用户表
MySQL [test_db]> select * from t_user;
+-----------+------+---------------------------------+
| username  | age  | addr                            |
+-----------+------+---------------------------------+
| zhangshan |   26 | 北京市西二旗街道1008号          |
+-----------+------+---------------------------------+
| lisi |   29 | 北京市西二旗街道1007号          |
+-----------+------+---------------------------------+
1 row in set (0.013 sec)

#与这张用表关联查询用户相关的信息
MySQL [test_db]> select u.* from (select t_str username  from test_type where g('match (v:test_type) where id(v)=="zhangshan" return v.test_type.t_str limit 1')) g left join t_user u on g.username=u.username;
+-----------+------+---------------------------------+
| username  | age  | addr                            |
+-----------+------+---------------------------------+
| zhangshan |   26 | 北京市西二旗街道1008号          |
+-----------+------+---------------------------------+
1 row in set (0.029 sec)

Checklist(Required)

• Does it affect the original behavior
• Has unit tests been added
• Has document been added or modified
• Does it need to update dependencies
• Is this PR support rollback (If NO, please explain WHY)

Further comments

If this is a relatively large or complex change, kick off the discussion at dev@doris.apache.org by explaining why you chose the solution you did and what alternatives you considered, etc...

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[BiteTheDDDDt]

Is it too simple to use g as a function name? The function that is also located in the ES is called es_query

[zhangstar333]

maybe can't at here to convert, it's will be exectue low efficiency

[chenlinzhong]

Is it too simple to use g as a function name? The function that is also located in the ES is called es_query

In the field of graphs, the variable g is usually used to execute GQL, so it is best to keep their usage consistent and simple here