Question

我正在尝试执行以下查询：

INSERT 
INTO rooms(
    id,
    name,
    body,
    parents,
    tags,
    createtime,
    creator,
    deletetime,
    meta,
    params,
    terms,
    updater,
    updatetime,
    counts,
    identities) 
SELECT *
FROM dblink ('dbname=oldsb',
    'SELECT '
        '(SELECT newid FROM id_map WHERE oldid = entities.id) AS id, '
        'id AS name, '
        'description AS body, '
        'NULL AS parents, '
        'NULL AS tags, '
        'ROUND(EXTRACT(EPOCH FROM current_timestamp)*1000) AS createtime, '
        'NULL AS creator, '
        'ROUND(EXTRACT(EPOCH FROM deletetime)*1000) AS deletetime, '
        'json_build_object(''picture'', picture) AS meta, '
        'jsonb_object_agg(
            (SELECT * '
            'FROM jsonb_each(params) '
            'AS fields (name, value) '
            'WHERE name <> ''places'')) AS params, '
        'terms AS terms, '
        'NULL AS updater, '
        'EXTRACT(EPOCH FROM lAStseentime)*1000 AS updatetime, '
        'NULL AS counts, '
        'NULL AS identities '
    'FROM entities WHERE type=''room''') 
AS t(
    id uuid,
    name text,
    body text,
    parents uuid[],
    tags smallint[],
    createtime bigint,
    creator text,
    deletetime bigint,
    meta jsonb,
    params jsonb,
    terms tsvector,
    updater text,
    updatetime bigint,
    counts jsonb,
    identities text[]);

我收到以下错误：

执行客房迁移查询错误：子查询必须只返回一列上下文：名为“unnamed”的dblink连接发生错误：无法执行查询。用于更新房间身份的表格。

我无法理解查询出错的地方。

Answer 1

您的一个子查询返回多个结果。正是这个。

SELECT newid FROM id_map WHERE oldid = entities.id

将LIMIT 1添加到结尾并再试一次。实际上，您可以将LIMIT 1添加到每个子查询中，看看它是否有帮助。

Answer 2

jsonb_each（params）有可能有多个列吗？

jsonb_object_agg(
        (SELECT * '
        'FROM jsonb_each(params) '
        'AS fields (name, value) '
        'WHERE name <> ''places'')) AS params,

我不知道jsonb_object_agg（）的作用，但我不认为它是为了处理作为表的输入而设计的。通常，聚合函数将一个或多个行结果合并为一个值。

Answer 3

是的。查看this文档以获取json_each函数。 json_each和jsonb_each的工作方式相同。

错误：子查询必须只返回一列

3 个答案: