跳转至

WHERE

WHERE子句可以根据条件过滤输出结果。

WHERE子句通常用于如下查询:

  • 原生 nGQL,例如GOLOOKUP语句。
  • openCypher 方式,例如MATCHWITH语句。

openCypher 兼容性

  • 不支持在WHERE子句中使用 Pattern(TODO: planning),例如WHERE (v)-->(v2)
  • 过滤 Rank 是原生 nGQL 功能。如需在 openCypher 兼容语句中直接获取 Rank 值,可以使用 rank() 函数,例如MATCH (:player)-[e:follow]->() RETURN rank(e);

基础用法

Note

下文示例中的$$$^等是引用符号,详情请参见引用符

用布尔运算符定义条件

WHERE子句中使用布尔运算符NOTANDORXOR定义条件。关于运算符的优先级,请参见运算符优先级

nebula> MATCH (v:player) \
        WHERE v.name == "Tim Duncan" \
        XOR (v.age < 30 AND v.name == "Yao Ming") \
        OR NOT (v.name == "Yao Ming" OR v.name == "Tim Duncan") \
        RETURN v.name, v.age;
+-------------------------+-------+
| v.name                  | v.age |
+-------------------------+-------+
| "Marco Belinelli"       | 32    |
| "Aron Baynes"           | 32    |
| "LeBron James"          | 34    |
| "James Harden"          | 29    |
| "Manu Ginobili"         | 41    |
+-------------------------+-------+
...
nebula> GO FROM "player100" \
        OVER follow \
        WHERE follow.degree > 90 \
        OR properties($$).age != 33 \
        AND properties($$).name != "Tony Parker" \
        YIELD properties($$);
+----------------------------------+
| properties($$)                   |
+----------------------------------+
| {age: 41, name: "Manu Ginobili"} |
+----------------------------------+

过滤属性

WHERE子句中使用点或边的属性定义条件。

  • 过滤点属性:
    nebula> MATCH (v:player)-[e]->(v2) \
            WHERE v2.age < 25 \
            RETURN v2.name, v2.age;
    +----------------------+--------+
    | v2.name              | v2.age |
    +----------------------+--------+
    | "Luka Doncic"        | 20     |
    | "Kristaps Porzingis" | 23     |
    | "Ben Simmons"        | 22     |
    +----------------------+--------+
    
    nebula> GO FROM "player100" \
            OVER follow \
            WHERE $^.player.age >= 42;
    +-------------+
    | follow._dst |
    +-------------+
    | "player101" |
    | "player125" |
    +-------------+
    
  • 过滤边属性:
    nebula> MATCH (v:player)-[e]->() \
            WHERE e.start_year < 2000 \
            RETURN DISTINCT v.name, v.age;
    +--------------------+-------+
    | v.name             | v.age |
    +--------------------+-------+
    | "Shaquille O'Neal" | 47    |
    | "Steve Nash"       | 45    |
    | "Ray Allen"        | 43    |
    | "Grant Hill"       | 46    |
    | "Tony Parker"      | 36    |
    +--------------------+-------+
    ...
    
    nebula> GO FROM "player100" \
            OVER follow \
            WHERE follow.degree > 90;
    +-------------+
    | follow._dst |
    +-------------+
    | "player101" |
    | "player125" |
    +-------------+
    

过滤动态计算属性

nebula> MATCH (v:player) \
        WHERE v[toLower("AGE")] < 21 \
        RETURN v.name, v.age;
+---------------+-------+
| v.name        | v.age |
+---------------+-------+
| "Luka Doncic" | 20    |
+---------------+-------+

过滤现存属性

nebula> MATCH (v:player) \
        WHERE exists(v.age) \
        RETURN v.name, v.age;
+-------------------------+-------+
| v.name                  | v.age |
+-------------------------+-------+
| "Boris Diaw"            | 36    |
| "DeAndre Jordan"        | 30    |
+-------------------------+-------+
...

过滤 rank

在 nGQL 中,如果多个边拥有相同的起始点、目的点和属性,则它们的唯一区别是 rank 值。在WHERE子句中可以使用 rank 过滤边。

# 创建测试数据。
nebula> CREATE SPACE IF NOT EXISTS test (vid_type=FIXED_STRING(30));
nebula> USE test;
nebula> CREATE EDGE IF NOT EXISTS e1(p1 int);
nebula> CREATE TAG IF NOT EXISTS person(p1 int);
nebula> INSERT VERTEX person(p1) VALUES "1":(1);
nebula> INSERT VERTEX person(p1) VALUES "2":(2);
nebula> INSERT EDGE e1(p1) VALUES "1"->"2"@0:(10);
nebula> INSERT EDGE e1(p1) VALUES "1"->"2"@1:(11);
nebula> INSERT EDGE e1(p1) VALUES "1"->"2"@2:(12);
nebula> INSERT EDGE e1(p1) VALUES "1"->"2"@3:(13);
nebula> INSERT EDGE e1(p1) VALUES "1"->"2"@4:(14);
nebula> INSERT EDGE e1(p1) VALUES "1"->"2"@5:(15);
nebula> INSERT EDGE e1(p1) VALUES "1"->"2"@6:(16);

# 通过 rank 过滤边,查找 rank 大于 2 的边。
nebula> GO FROM "1" \
        OVER e1 \
        WHERE rank(edge) > 2 \
        YIELD src(edge), dst(edge), rank(edge) AS Rank, properties(edge).p1 | \
        ORDER BY $-.Rank DESC;
+-----------+-----------+------+---------------------+
| src(EDGE) | dst(EDGE) | Rank | properties(EDGE).p1 |
+-----------+-----------+------+---------------------+
| "1"       | "2"       | 6    | 16                  |
| "1"       | "2"       | 5    | 15                  |
| "1"       | "2"       | 4    | 14                  |
| "1"       | "2"       | 3    | 13                  |
+-----------+-----------+------+---------------------+

过滤字符串

WHERE子句中使用STARTS WITHENDS WITHCONTAINS可以匹配字符串的特定部分。匹配时区分大小写。

STARTS WITH

STARTS WITH会从字符串的起始位置开始匹配。

# 查询姓名以 T 开头的 player 信息。
nebula> MATCH (v:player) \
        WHERE v.name STARTS WITH "T" \
        RETURN v.name, v.age;
+------------------+-------+
| v.name           | v.age |
+------------------+-------+
| "Tracy McGrady"  | 39    |
| "Tony Parker"    | 36    |
| "Tim Duncan"     | 42    |
| "Tiago Splitter" | 34    |
+------------------+-------+

如果使用小写tSTARTS WITH "t"),会返回空集,因为数据库中没有以小写t开头的姓名。

nebula> MATCH (v:player) \
        WHERE v.name STARTS WITH "t" \
        RETURN v.name, v.age;
Empty set (time spent 5080/6474 us)

ENDS WITH

ENDS WITH会从字符串的结束位置开始匹配。

nebula> MATCH (v:player) \
        WHERE v.name ENDS WITH "r" \
        RETURN v.name, v.age;
+------------------+-------+
| v.name           | v.age |
+------------------+-------+
| "Vince Carter"   | 42    |
| "Tony Parker"    | 36    |
| "Tiago Splitter" | 34    |
+------------------+-------+

CONTAINS

CONTAINS会检查关键字是否匹配字符串的某一部分。

nebula> MATCH (v:player) \
        WHERE v.name CONTAINS "Pa" \
        RETURN v.name, v.age;
+---------------+-------+
| v.name        | v.age |
+---------------+-------+
| "Paul George" | 28    |
| "Tony Parker" | 36    |
| "Paul Gasol"  | 38    |
| "Chris Paul"  | 33    |
+---------------+-------+

结合 NOT 使用

用户可以结合布尔运算符NOT一起使用,否定字符串匹配条件。

nebula> MATCH (v:player) \
        WHERE NOT v.name ENDS WITH "R" \
        RETURN v.name, v.age;
+-------------------------+-------+
| v.name                  | v.age |
+-------------------------+-------+
| "Rajon Rondo"           | 33    |
| "Rudy Gay"              | 32    |
| "Dejounte Murray"       | 29    |
| "Chris Paul"            | 33    |
| "Carmelo Anthony"       | 34    |
+-------------------------+-------+
...

过滤列表

匹配列表中的值

使用IN运算符检查某个值是否在指定列表中。

nebula> MATCH (v:player) \
        WHERE v.age IN range(20,25) \
        RETURN v.name, v.age;
+-------------------------+-------+
| v.name                  | v.age |
+-------------------------+-------+
| "Ben Simmons"           | 22    |
| "Kristaps Porzingis"    | 23    |
| "Luka Doncic"           | 20    |
| "Kyle Anderson"         | 25    |
| "Giannis Antetokounmpo" | 24    |
| "Joel Embiid"           | 25    |
+-------------------------+-------+

nebula> LOOKUP ON player \
        WHERE player.age IN [25,28]  \
        YIELD properties(vertex).name, properties(vertex).age;
+-------------+-------------------------+------------------------+
| VertexID    | properties(VERTEX).name | properties(VERTEX).age |
+-------------+-------------------------+------------------------+
| "player106" | "Kyle Anderson"         | 25                     |
| "player135" | "Damian Lillard"        | 28                     |
| "player130" | "Joel Embiid"           | 25                     |
| "player131" | "Paul George"           | 28                     |
| "player123" | "Ricky Rubio"           | 28                     |
+-------------+-------------------------+------------------------+

结合 NOT 使用

nebula> MATCH (v:player) \
        WHERE v.age NOT IN range(20,25) \
        RETURN v.name AS Name, v.age AS Age \
        ORDER BY Age;
+---------------------+-----+
| Name                | Age |
+---------------------+-----+
| "Kyrie Irving"      | 26  |
| "Cory Joseph"       | 27  |
| "Damian Lillard"    | 28  |
| "Paul George"       | 28  |
| "Ricky Rubio"       | 28  |
+---------------------+-----+
...

最后更新: March 7, 2023