在PostgreSQL提示符上发出\d+ tablename，除了显示表列的数据类型外，它还会显示索引和外键。

PSQL就是这样做的，如果你用:

psql -E

它将准确地显示执行了哪些查询。在查找外键的情况下，它是:

SELECT conname,
  pg_catalog.pg_get_constraintdef(r.oid, true) as condef
FROM pg_catalog.pg_constraint r
WHERE r.conrelid = '16485' AND r.contype = 'f' ORDER BY 1

在这种情况下，16485是我正在寻找的表的oid -你可以通过将你的表名转换为regclass来获得它:

WHERE r.conrelid = 'mytable'::regclass

如果表名不是唯一的(或者是搜索路径中的第一个)，则对表名进行模式限定:

WHERE r.conrelid = 'myschema.mytable'::regclass

注意:在读取约束列时，不要忘记列的顺序!

SELECT conname, attname
  FROM pg_catalog.pg_constraint c 
  JOIN pg_catalog.pg_attribute a ON a.attrelid = c.conrelid AND a.attnum = ANY (c.conkey)
 WHERE attrelid = 'schema.table_name'::regclass
 ORDER BY conname, array_position(c.conkey, a.attnum)

Ollyc的答案很好，因为它不是特定于postgres的，但是，当外键引用多个列时，它就会崩溃。以下查询适用于任意数量的列，但它严重依赖于Postgres扩展:

select 
    att2.attname as "child_column", 
    cl.relname as "parent_table", 
    att.attname as "parent_column",
    conname
from
   (select 
        unnest(con1.conkey) as "parent", 
        unnest(con1.confkey) as "child", 
        con1.confrelid, 
        con1.conrelid,
        con1.conname
    from 
        pg_class cl
        join pg_namespace ns on cl.relnamespace = ns.oid
        join pg_constraint con1 on con1.conrelid = cl.oid
    where
        cl.relname = 'child_table'
        and ns.nspname = 'child_schema'
        and con1.contype = 'f'
   ) con
   join pg_attribute att on
       att.attrelid = con.confrelid and att.attnum = con.child
   join pg_class cl on
       cl.oid = con.confrelid
   join pg_attribute att2 on
       att2.attrelid = con.conrelid and att2.attnum = con.parent

我自己的贡献。目标是备份所有外键的定义:

SELECT
    'ALTER TABLE ' || tc.table_schema || '.' || tc.table_name || E'\n
    ADD FOREIGN KEY (' || kcu.column_name || ')' || E'\n
    REFERENCES ' || ccu.table_schema || '.' || ccu.table_name ||
    ' (' || ccu.column_name || ') ' || E'\n    ' ||
    CASE WHEN rc.match_option <> 'NONE' THEN E'\n
    MATCH ' || rc.match_option ELSE '' END ||
    CASE WHEN rc.update_rule <> 'NO ACTION' THEN E'\n
    ON UPDATE ' || rc.update_rule || ' ' ELSE '' END ||
    CASE WHEN rc.delete_rule <> 'NO ACTION'
    THEN 'ON DELETE ' || rc.delete_rule ELSE '' END || ';'
AS add_constraint
FROM
    information_schema.table_constraints AS tc
    JOIN information_schema.key_column_usage AS kcu
        ON tc.constraint_name = kcu.constraint_name
        AND tc.table_schema = kcu.table_schema
    JOIN information_schema.constraint_column_usage AS ccu
        ON ccu.constraint_name = tc.constraint_name
        AND ccu.table_schema = tc.table_schema
    JOIN information_schema.referential_constraints AS rc
        ON tc.constraint_name=rc.constraint_name
WHERE tc.constraint_type = 'FOREIGN KEY'
\t\a\g\a\ta

我升级了@ollyc的答案，目前在顶部。 我同意@fionbio，因为key_column_usage和constraint_column_usage在列级上没有相关信息。

如果constraint_column_usage具有像key_column_usage一样的ordinal_position列，则可以将其与该列连接。所以我做了一个ordinal_position到constraint_column_usage如下所示。

我无法确认手动创建的ordinal_position与key_column_usage的顺序完全相同。但我检查了一下，至少在我的箱子里是完全一样的顺序。

SELECT
    tc.table_schema, 
    tc.constraint_name, 
    tc.table_name, 
    kcu.column_name, 
    ccu.table_schema AS foreign_table_schema,
    ccu.table_name AS foreign_table_name,
    ccu.column_name AS foreign_column_name
FROM 
    information_schema.table_constraints AS tc 
    JOIN information_schema.key_column_usage AS kcu
      ON tc.constraint_name = kcu.constraint_name
      AND tc.table_schema = kcu.table_schema
    JOIN (select row_number() over (partition by table_schema, table_name, constraint_name order by row_num) ordinal_position,
                 table_schema, table_name, column_name, constraint_name
          from   (select row_number() over (order by 1) row_num, table_schema, table_name, column_name, constraint_name
                  from   information_schema.constraint_column_usage
                 ) t
         ) AS ccu
      ON ccu.constraint_name = tc.constraint_name
      AND ccu.table_schema = tc.table_schema
      AND ccu.ordinal_position = kcu.ordinal_position
WHERE tc.constraint_type = 'FOREIGN KEY' AND tc.table_name = 'mytable'