跳转至

nGQL风格指南Graph

nGQL没有严格的构建格式要求,但根据恰当而统一的风格创建nGQL语句有利于提高可读性、避免歧义。在同一组织或项目中使用相同的nGQL风格有利于降低维护成本,规避因格式混乱或误解造成的问题。本文为写作nGQL语句提供了风格参考。

Compatibility

nGQL 风格与Graph不同。

换行Graph

  1. 换行写子句。

    不推荐:

    GO FROM "player100" OVER follow REVERSELY YIELD follow._dst AS id;
    

    推荐:

    GO FROM "player100" \
    OVER follow REVERSELY \
    YIELD follow._dst AS id;
    
  2. 换行写复合语句中的不同语句。

    不推荐:

    GO FROM "player100" OVER follow REVERSELY YIELD follow._dst AS id | GO FROM $-.id \
    OVER serve WHERE $^.player.age > 20 YIELD $^.player.name AS FriendOf, $$.team.name AS Team;
    

    推荐:

    GO FROM "player100" \
    OVER follow REVERSELY \
    YIELD follow._dst AS id | \
    GO FROM $-.id OVER serve \
    WHERE $^.player.age > 20 \
    YIELD $^.player.name AS FriendOf, $$.team.name AS Team;
    
  3. 子句长度超过80个字符时,在合适的位置换行。

    不推荐:

    MATCH (v:player{name:"Tim Duncan"})-[e]->(v2) \
    WHERE (v2.name STARTS WITH "Y" AND v2.age > 35 AND v2.age < v.age) OR (v2.name STARTS WITH "T" AND v2.age < 45 AND v2.age > v.age) \
    RETURN v2;
    

    推荐:

    MATCH (v:player{name:"Tim Duncan"})-[e]->(v2) \
    WHERE (v2.name STARTS WITH "Y" AND v2.age > 35 AND v2.age < v.age) \
    OR (v2.name STARTS WITH "T" AND v2.age < 45 AND v2.age > v.age) \
    RETURN v2;
    

Note

即使子句不超过80个字符,如需换行后有助于理解,也可将子句再次分行。

标识符命名Graph

在nGQL语句中,关键字、标点符号、空格以外的字符内容都是标识符。推荐的标识符命名方式如下。

  1. 使用单数名词命名Tag,用原型动词或动词短语构成Edge type。

    不推荐:

    MATCH p=(v:players)-[e:are_following]-(v2) \
    RETURN nodes(p);
    

    推荐:

    MATCH p=(v:player)-[e:follow]-(v2) \
    RETURN nodes(p);
    
  2. 标识符用蛇形命名法,以下划线(_)连接单词,且所有字母小写。

    不推荐:

    MATCH (v:basketballTeam) \
    RETURN v;
    

    推荐:

    MATCH (v:basketball_team) \
    RETURN v;
    
  3. 语法关键词大写,变量小写。

    不推荐:

    go from "player100" over Follow
    

    推荐:

    GO FROM "player100" OVER follow
    

PatternGraph

  1. 分行写Pattern时,在表示边的箭头右侧换行,而不是左侧。

    不推荐:

    MATCH (v:player{name: "Tim Duncan", age: 42}) \
    -[e:follow]->()-[e:serve]->()<--(v3) \
    RETURN v, e, v2;
    

    推荐:

    MATCH (v:player{name: "Tim Duncan", age: 42})-[e:follow]-> \
    ()-[e:serve]->()<--(v3) \
    RETURN v, e, v2;
    
  2. 将无需查询的点和边匿名化。

    不推荐:

    MATCH (v:player)[e:follow]->(v2) \
    RETURN v;
    

    推荐:

    MATCH (v:player)-[:follow]->() \
    RETURN v;
    
  3. 将非匿名点放在匿名点的前面。

    不推荐:

    MATCH ()-[:follow]->(v) \
    RETURN v;
    

    推荐:

    MATCH (v)<-[:follow]-() \
    RETURN v;
    

字符串Graph

字符串用双引号包围。

不推荐:

RETURN 'Hello Nebula!';

推荐:

RETURN "Hello Nebula!\"123\"";

Note

字符串中需要嵌套单引号或双引号时,用反斜线(\)转义。例如:

RETURN "\"NebulaGraph is amazing,\" the user says.";

结束语句Graph

  1. 用英文分号(;)结束nGQL语句。

    不推荐:

    FETCH PROP ON player "player100"
    

    推荐:

    FETCH PROP ON player "player100";
    
  2. 使用管道符(|)分隔的复合语句,仅在最后一行末用英文分号结尾。在管道符前使用英文分号会导致语句执行失败。

    不支持:

    GO FROM "player100" \
    OVER follow \
    YIELD follow._dst AS id; | \
    GO FROM $-.id \
    OVER serve \
    YIELD $$.team.name AS Team, $^.player.name AS Player;
    

    支持:

    GO FROM "player100" \
    OVER follow \
    YIELD follow._dst AS id | \
    GO FROM $-.id \
    OVER serve \
    YIELD $$.team.name AS Team, $^.player.name AS Player;
    
  3. 在包含自定义变量的复合语句中,用英文分号结束定义变量的语句。不按规则加分号或使用管道符结束该语句会导致执行失败。

    不支持:

    $var = GO FROM "player100" \
    OVER follow \
    YIELD follow._dst AS id \
    GO FROM $var.id \
    OVER serve \
    YIELD $$.team.name AS Team, $^.player.name AS Player;
    

    也不支持:

    $var = GO FROM "player100" \
    OVER follow \
    YIELD follow._dst AS id | \
    GO FROM $var.id \
    OVER serve \
    YIELD $$.team.name AS Team, $^.player.name AS Player;
    

    支持:

    $var = GO FROM "player100" \
    OVER follow \
    YIELD follow._dst AS id; \
    GO FROM $var.id \
    OVER serve \
    YIELD $$.team.name AS Team, $^.player.name AS Player;
    

最后更新: September 14, 2021
Back to top