Question

简短而简单，我希望获得一个独特小时列表及其计数代表。

SELECT DISTINCT(date_trunc('hour', occurred)) as time, COUNT(*)
  FROM log
  WHERE occurred BETWEEN '2014-01-01 00:00' AND '2014-01-01 23:59'
GROUP BY occurred;

这不起作用，因为我获得了比24更多的结果所以我继续尝试：

SELECT DISTINCT(occurred), COUNT(*)
    FROM log
    WHERE occurred BETWEEN .. AND ..
GROUP BY date_trunc('hour', occurred);

这是一种无效的语法，显然我只是在扯动它，因此我需要帮助。如何解决这个问题？

Answer 1

您根本不需要DISTINCT 。

如果您只对发生任何事情的“小时”列表感兴趣（适用于任何时间范围）：

SELECT date_trunc('hour', occurred) AS hour, COUNT(*) AS ct
FROM   log
WHERE  occurred >= '2014-01-01 0:0'
AND    occurred <  '2014-01-02 0:0'  -- do not use between for this
GROUP  BY 1
ORDER  BY 1;

BETWEEN包括上限和下限，通常不适用于时间戳范围：

Optimize BETWEEN date statement

如果一个多小时没有任何反应怎么办？如果您仍然希望结果中的行没有活动几小时（忽略此变体中的日期！）：

SELECT hour, COUNT(l.occurred) AS ct
FROM   generate_series (0,23) h(hour)
LEFT   JOIN log l ON occurred >= '2014-01-01 0:0'
                 AND occurred <  '2014-01-02 0:0'  -- do not use between for this
                 AND EXTRACT(hour FROM occurred)::int = h.hour
GROUP  BY hour
ORDER  BY hour;

generate_series()每天提供完整的小时数 LEFT JOIN保留了所有这些内容仅计算实际日志条目。

对列值的更改date_trunc进行不同的搜索

1 个答案: