检索所有PK和FK

我有一个很大的数据库，需要从每个表中提取所有主键和外键。

我有pgAdmin III。

有没有一种方法可以自动执行此操作，而不必手动检查每个表？

您可以pg_get_constraintdef(constraint_oid)在查询中使用该函数，如下所示：

SELECT conrelid::regclass AS table_from
     , conname
     , pg_get_constraintdef(oid)
FROM   pg_constraint
WHERE  contype IN ('f', 'p ')
AND    connamespace = 'public'::regnamespace  -- your schema here
ORDER  BY conrelid::regclass::text, contype DESC;

结果：

 table_from | conname    | pg_get_constraintdef
------------+------------+----------------------
 tbl        | tbl_pkey   | PRIMARY KEY (tbl_id)
 tbl        | tbl_col_fk | FOREIGN KEY (col) REFERENCES tbl2(col) ON UPDATE CASCADE
...

返回给定模式中所有表的所有主键和外键，按表名顺序排列，首先是PK。

有关的手册pg_constraint。

该手册关于对象标识符类型（regclass，regnamespace，...）。

基于Erwin解决方案：

SELECT conrelid::regclass AS "FK_Table"
      ,CASE WHEN pg_get_constraintdef(c.oid) LIKE 'FOREIGN KEY %' THEN substring(pg_get_constraintdef(c.oid), 14, position(')' in pg_get_constraintdef(c.oid))-14) END AS "FK_Column"
      ,CASE WHEN pg_get_constraintdef(c.oid) LIKE 'FOREIGN KEY %' THEN substring(pg_get_constraintdef(c.oid), position(' REFERENCES ' in pg_get_constraintdef(c.oid))+12, position('(' in substring(pg_get_constraintdef(c.oid), 14))-position(' REFERENCES ' in pg_get_constraintdef(c.oid))+1) END AS "PK_Table"
      ,CASE WHEN pg_get_constraintdef(c.oid) LIKE 'FOREIGN KEY %' THEN substring(pg_get_constraintdef(c.oid), position('(' in substring(pg_get_constraintdef(c.oid), 14))+14, position(')' in substring(pg_get_constraintdef(c.oid), position('(' in substring(pg_get_constraintdef(c.oid), 14))+14))-1) END AS "PK_Column"
FROM   pg_constraint c
JOIN   pg_namespace n ON n.oid = c.connamespace
WHERE  contype IN ('f', 'p ')
AND pg_get_constraintdef(c.oid) LIKE 'FOREIGN KEY %'
ORDER  BY pg_get_constraintdef(c.oid), conrelid::regclass::text, contype DESC;

将返回一个表格形式：

| FK_Table | FK_Column | PK_Table | PK_Column |

无需解析pg_get_constraintdef()，只需使用表的pg_constraint列即可获取其他详细信息（docs）。

这里constraint_type可以是：

• p 主键，
• f- 外键，
• ü - 独特的，
• c- check约束，
• X - 排除，
• ...

根据Erwin的答案：

SELECT c.conname                                     AS constraint_name,
       c.contype                                     AS constraint_type,
       sch.nspname                                   AS "schema",
       tbl.relname                                   AS "table",
       ARRAY_AGG(col.attname ORDER BY u.attposition) AS columns,
       pg_get_constraintdef(c.oid)                   AS definition
FROM pg_constraint c
       JOIN LATERAL UNNEST(c.conkey) WITH ORDINALITY AS u(attnum, attposition) ON TRUE
       JOIN pg_class tbl ON tbl.oid = c.conrelid
       JOIN pg_namespace sch ON sch.oid = tbl.relnamespace
       JOIN pg_attribute col ON (col.attrelid = tbl.oid AND col.attnum = u.attnum)
GROUP BY constraint_name, constraint_type, "schema", "table", definition
ORDER BY "schema", "table";

结果由schema和排序table。

技术说明：请参阅有关的问题with ordinality。

最近不得不为基于信息模式构建CRUD实用程序的数据访问层实现此功能，最终解决了这个问题。

SELECT

    current_schema() AS "schema",
    current_catalog AS "database",
    "pg_constraint".conrelid::regclass::text AS "primary_table_name",
    "pg_constraint".confrelid::regclass::text AS "foreign_table_name",

    (
        string_to_array(
            (
                string_to_array(
                    pg_get_constraintdef("pg_constraint".oid),
                    '('
                )
            )[2],
            ')'
        )
    )[1] AS "foreign_column_name",

    "pg_constraint".conindid::regclass::text AS "constraint_name",

    TRIM((
        string_to_array(
            pg_get_constraintdef("pg_constraint".oid),
            '('
        )
    )[1]) AS "constraint_type",

    pg_get_constraintdef("pg_constraint".oid) AS "constraint_definition"

FROM pg_constraint AS "pg_constraint"

JOIN pg_namespace AS "pg_namespace" ON "pg_namespace".oid = "pg_constraint".connamespace

WHERE

    "pg_constraint".contype IN ( 'f', 'p' )
    AND
    "pg_namespace".nspname = current_schema()
    AND
    "pg_constraint".conrelid::regclass::text IN ('whatever_table_name')