PostgreSQL's date_trunc in mySQL

Recently, I have been getting familiar with PostgreSQL(using 8.2) and found the date_trunc function extremely useful for easily matching time stamps between certain days/months/etc. The real usefulness of the function, I believe, comes from the fact that it keeps the output in the format of a timestamp.

I have had to switch to mySQL(5.0) and find some of the date functions rather lacking in comparison. The extract function seems useful and the date function I have found solves some of my problems, but is there any way to replicate PostgreSQL's date_trunc?

Following is an example of how I used to use date_trunc to match queried timestamps to only the last 4 months including the current month, but only if a week has passed into this month already:

WHERE date_trunc('month', QUERY_DATE) BETWEEN 
    date_trunc('month', now()) - INTERVAL '4 MONTH' AND 
    date_trunc('month', now() - INTERVAL '1 WEEK')

I have no idea how to recreate such a stipulati开发者_如何学JAVAon in mySQL. So, my question at the end of the day, is whether this type of query can be accomplished in mySQL by trying replicate date_trunc(and how) or whether I need to start looking at these types of queries in a different way to make them work in mySQL(and suggestions on how to do that)?

---

The extract function seems useful and the date function I have found solves some of my problems, but is there any way to replicate PostgreSQL's date_trunc?

Indeed, EXTRACT looks like it's going to be the closest match for this specific case.

Your original code in PG:

WHERE date_trunc('month', QUERY_DATE) BETWEEN 
    date_trunc('month', now()) - INTERVAL '4 MONTH' AND 
    date_trunc('month', now() - INTERVAL '1 WEEK')

Using EXTRACT:

WHERE EXTRACT(YEAR_MONTH FROM QUERY_DATE)
      BETWEEN
          EXTRACT(YEAR_MONTH FROM NOW() - INTERVAL 4 MONTH)
      AND
          EXTRACT(YEAR_MONTH FROM NOW() - INTERVAL 1 WEEK)

While it should be functionally identical, this is actually mangling the dates into a YYYYMM string before doing the comparison.

Another option would be using DATE_FORMAT to rebuild the date string and force it to the beginning of the month:

WHERE DATE_FORMAT(QUERY_DATE, '%Y-%m-01')
      BETWEEN
          DATE_FORMAT(NOW() - INTERVAL 4 MONTH, '%Y-%m-01')
      AND
          DATE_FORMAT(NOW() - INTERVAL 1 WEEK, '%Y-%m-01')

Also, be aware that MySQL is really poor at dealing with date ranges, even when the field is indexed. You're probably going to end up with a full table scan if you aren't careful.

---

late to the party, but...

there is a way to get truncated date given you know the interval. For example, if the interval is MONTH, you could get today's date (now()) truncated to the month using the following:

select date_add('1900-01-01', interval TIMESTAMPDIFF(MONTH, '1900-01-01', now()) MONTH);

Given the above, one could create a function to take care of the other intervals as well:

DELIMITER //
create function date_trunc(vInterval varchar(7), vDate timestamp)
returns timestamp
begin
    declare toReturn timestamp;

    if vInterval = 'year' then set toReturn = date_add('1900-01-01', interval TIMESTAMPDIFF(YEAR, '1900-01-01', vDate) YEAR);
    elseif vInterval = 'quarter' then set toReturn = date_add('1900-01-01', interval TIMESTAMPDIFF(QUARTER, '1900-01-01', vDate) QUARTER);
    elseif vInterval = 'month' then set toReturn = date_add('1900-01-01', interval TIMESTAMPDIFF(MONTH, '1900-01-01', vDate) MONTH);
    elseif vInterval = 'week' then set toReturn = date_add('1900-01-01', interval TIMESTAMPDIFF(WEEK, '1900-01-01', vDate) WEEK);
    elseif vInterval = 'day' then set toReturn = date_add('1900-01-01', interval TIMESTAMPDIFF(DAY, '1900-01-01', vDate) DAY);
    elseif vInterval = 'hour' then set toReturn = date_add('1900-01-01', interval TIMESTAMPDIFF(HOUR, '1900-01-01', vDate) HOUR);
    elseif vInterval = 'minute' then set toReturn = date_add('1900-01-01', interval TIMESTAMPDIFF(MINUTE, '1900-01-01', vDate) MINUTE);
    END IF;

    return toReturn;
end//
DELIMITER ;

Use it like so:

select date_trunc('quarter', now())

---

Here is a function that mimics postgres' DATE_TRUNC contract using the DATE_FORMAT mysql function that @Charles has recommended above.

DROP FUNCTION IF EXISTS DATE_TRUNC;
CREATE FUNCTION DATE_TRUNC(
  in_granularity ENUM('hour', 'day', 'month', 'year'),
  in_datetime datetime(6)
)
RETURNS datetime(6)
DETERMINISTIC
BEGIN
  IF (in_granularity = 'hour') THEN
    RETURN DATE_FORMAT(in_datetime, '%Y-%m-%d %H:00:00.0000');
  END IF;
  IF (in_granularity = 'day') THEN
    RETURN DATE_FORMAT(in_datetime, '%Y-%m-%d 00:00:00.0000');
  END IF;
  IF (in_granularity = 'month') THEN
    RETURN DATE_FORMAT(in_datetime, '%Y-%m-01 00:00:00.0000');
  END IF;
  IF (in_granularity = 'year') THEN
    RETURN DATE_FORMAT(in_datetime, '%Y-01-01 00:00:00.0000');
  END IF;
END;

---

Here are some MySQL equivalents for the most common Redshift/Postgres date_trunc expressions, using date arithmetic:

select 
null

-- RS date_trunc('hour', now()) = '2023-01-13 17:00:00.000 +0100'
,date_add(date(now()), interval hour(now()) hour) as trunc_hour

-- RS date_trunc('week', now()) = '2023-01-09 00:00:00.000 +0100'
,date_sub(cast(date(now()) as datetime), interval weekday(now()) day) as trunc_week

-- RS date_trunc('month', now()) = '2023-01-01 00:00:00.000 +0100'
,date_add(cast(makedate(year(now()), 1) as datetime), interval month(now())-1 month) as trunc_month

-- RS date_trunc('year', now()) = '2023-01-01 00:00:00.000 +0100'
,cast(makedate(year(now()), 1) as datetime) as trunc_year