集合运算符

合并多个请求时，可以使用集合运算符，包括UNION、UNION ALL、INTERSECT和MINUS。

所有集合运算符的优先级相同，如果一个nGQL语句中有多个集合运算符，Nebula Graph会从左到右进行计算，除非用括号指定顺序。

openCypher兼容性

集合运算符仅适用于原生nGQL。

UNION、UNION DISTINCT、UNION ALL

  1. <left> UNION [DISTINCT | ALL] <right> [ UNION [DISTINCT | ALL] <right> ...]

  • 运算符UNION DISTINCT（或使用缩写UNION）返回两个集合A和B的并集，不包含重复的元素。

  • 运算符UNION ALL返回两个集合A和B的并集，包含重复的元素。

  • left和right必须有相同数量的列和数据类型。如果需要转换数据类型，请参见类型转换。

示例

  1. # 返回两个查询结果的并集，不包含重复的元素。
  2. nebula> GO FROM "player102" OVER follow \
  3.         UNION \
  4.         GO FROM "player100" OVER follow;
  5. +-------------+
  6. | follow._dst |
  7. +-------------+
  8. | "player101" |
  9. +-------------+
  10. | "player102" |
  11. +-------------+
  13. # 返回两个查询结果的并集，包含重复的元素。
  14. nebula> GO FROM "player102" OVER follow \
  15.         UNION ALL \
  16.         GO FROM "player100" OVER follow;
  17. +-------------+
  18. | follow._dst |
  19. +-------------+
  20. | "player101" |
  21. +-------------+
  22. | "player101" |
  23. +-------------+
  24. | "player102" |
  25. +-------------+
  27. # UNION也可以和YIELD语句一起使用，去重时会检查每一行的所有列，每列都相同时才会去重。
  28. nebula> GO FROM "player102" OVER follow \
  29.         YIELD follow._dst AS id, follow.degree AS Degree, $$.player.age AS Age \
  30.         UNION /* DISTINCT */ \
  31.         GO FROM "player100" OVER follow \
  32.         YIELD follow._dst AS id, follow.degree AS Degree, $$.player.age AS Age;
  33. +-------------+--------+-----+
  34. | id          | Degree | Age |
  35. +-------------+--------+-----+
  36. | "player101" | 75     | 36  |     
  37. +-------------+--------+-----+
  38. | "player101" | 96     | 36  |       
  39. +-------------+--------+-----+
  40. | "player102" | 90     | 33  |     
  41. +-------------+--------+-----+

INTERSECT

  1. <left> INTERSECT <right>

  • 运算符INTERSECT返回两个集合A和B的交集。

  • left和right必须有相同数量的列和数据类型。如果需要转换数据类型，请参见类型转换。

示例

  1. nebula> GO FROM "player102" OVER follow \
  2.         YIELD follow._dst AS id, follow.degree AS Degree, $$.player.age AS Age \
  3.         INTERSECT \
  4.         GO FROM "player100" OVER follow \
  5.         YIELD follow._dst AS id, follow.degree AS Degree, $$.player.age AS Age;
  6. Empty set (time spent 2990/3511 us)

MINUS

  1. <left> MINUS <right>

运算符MINUS返回两个集合A和B的差异，即A-B。请注意left和right的顺序，A-B表示在集合A中，但是不在集合B中的元素。

示例

  1. nebula> GO FROM "player100" OVER follow \
  2.         MINUS \
  3.         GO FROM "player102" OVER follow;
  4. +-------------+
  5. | follow._dst |
  6. +-------------+
  7. | "player102" |
  8. +-------------+
  10. nebula> GO FROM "player102" OVER follow \
  11.         MINUS \
  12.         GO FROM "player100" OVER follow;
  13. Empty set (time spent 2243/3259 us)

集合运算符和管道符的优先级

当查询包含集合运算符和管道符（|）时，管道符的优先级高。例如GO FROM 1 UNION GO FROM 2 | GO FROM 3相当于GO FROM 1 UNION (GO FROM 2 | GO FROM 3)。

示例

  1. nebula> GO FROM "player102" OVER follow \
  2.         YIELD follow._dst AS play_dst  \
  3.         UNION \
  4.         GO FROM "team200" OVER serve REVERSELY \
  5.         YIELD serve._dst AS play_dst \
  6.         | GO FROM $-.play_dst OVER follow YIELD follow._dst AS play_dst;
  8. +-------------+
  9. | play_dst    |
  10. +-------------+
  11. | "player101" |
  12. +-------------+
  13. | "player102" |
  14. +-------------+

image

该查询会先执行红框内的语句，然后执行绿框的UNION操作。

圆括号可以修改执行的优先级，例如：

  1. nebula> (GO FROM "player102" OVER follow \
  2.         YIELD follow._dst AS play_dst  \
  3.         UNION \
  4.         GO FROM "team200" OVER serve REVERSELY \
  5.         YIELD serve._dst AS play_dst) \
  6.         | GO FROM $-.play_dst OVER follow YIELD follow._dst AS play_dst;

该查询中，圆括号包裹的部分先执行，即先执行UNION操作，再将结果结合管道符进行下一步操作。