psycopg 有数据时返回空查询结果

Question

我正在尝试从下面的 postgressql 查询中获取记录。它确实有数据，但当我尝试用 psycopg 检索它时，奇怪的是它返回空。它可以与WITH查询一起使用还是我们只能使用SELECT来代替？任何意见将不胜感激，谢谢。

query = \
        f""" WITH a AS (
            SELECT "public"."table_a"."code", 
            "public"."table_a"."address", sum("public"."table_a"."amount") AS 
            "sum" FROM "public"."table_a" WHERE "public"."table_a"."address" <> 'KL' GROUP BY 
            "public"."table_a"."code", "public"."table_a"."address"), 
            c as (
                WITH b AS ( 
                SELECT address, code, count(*) as "total" FROM table_b where created_date between 
                ((current_date + TIME '14:00:00.000+08:00') - interval '7 days') and ((
                current_date + TIME '23:59:00.000+08:00') - interval '7 days') group by 
                address, code order by address, code) 
                select b.address, table_c.unit_code, table_c.name, sum(b.total * 
                table_c.amount) as "total" from table_c join b on table_c.code = b.code 
                group by table_c.unit_code, table_c.name, b.address 
                order by table_c.unit_code, b.address
                )
            SELECT a.address, a.code, (a.sum - c.total)::int as output
            FROM a join c on a.code = c.unit_code
            AND a.address = c.address
            ORDER BY a.address, a.code 
"""

conn = psycopg2.connect(**params)
        cur = conn.cursor()
        cur.execute(query)
rows = cur.fetchall()
for row in rows:
            print(row)

Answer 1

我花了一整天才意识到日志中的时间戳与预期不同。下面的查询有效（使用WITH AS和所有），只需要为我删除时区，因为我不需要它。学习到教训了。请务必检查服务器时区以及是否需要转换。希望这对将来的人有帮助。

WITH a AS (
            SELECT "public"."table_a"."code", 
            "public"."table_a"."address", sum("public"."table_a"."amount") AS 
            "sum" FROM "public"."table_a" WHERE "public"."table_a"."address" <> 'KL' GROUP BY 
            "public"."table_a"."code", "public"."table_a"."address"), 
            c as (
                WITH b AS ( 
                SELECT address, code, count(*) as "total" FROM table_b where created_date between 
                ((current_date + TIME '14:00:00.000') - interval '7 days') and ((
                current_date + TIME '23:59:00.000') - interval '7 days') group by 
                address, code order by address, code) 
                select b.address, table_c.unit_code, table_c.name, sum(b.total * 
                table_c.amount) as "total" from table_c join b on table_c.code = b.code 
                group by table_c.unit_code, table_c.name, b.address 
                order by table_c.unit_code, b.address
                )
            SELECT a.address, a.code, (a.sum - c.total)::int as output
            FROM a join c on a.code = c.unit_code
            AND a.address = c.address
            ORDER BY a.address, a.code 

Answer 2

我最近遇到了一个令人困惑的问题，想分享我的经验以造福社区。我在本地计算机上制作了一个查询，其中涉及与 PostgreSQL 表中的created_at 列进行时间戳比较。该查询在我的 DBMS 客户端中执行时完美运行，但在我的 Python 代码中使用 psycopg2 执行时抛出错误。

令我惊讶的是，解决方案非常简单。最初看起来不错的时间戳比较导致了差异。当用 psycopg2 执行时，它不知何故失败了。然而，我能够通过使用 PostgreSQL 的clock_timestamp() 函数和时间间隔来转换比较来解决这个问题。

这是最初导致问题的代码片段：

WHERE created_at > '2023-08-17 12:00:00' -- Example timestamp

这是与 DBMS 客户端和 psycopg2 一致工作的改进版本：

WHERE created_at > clock_timestamp() - interval '10 minutes'

这里的关键教训是，即使看似正确的查询有时在不同的上下文中（例如在服务器中）也会有不同的行为。通过使用 PostgreSQL 的内置函数和表达式，我们可以确保不同工具和接口之间的行为一致。

有关 PostgreSQL 日期时间函数的更多信息，可以参考官方文档：PostgreSQL Datetime Functions