python : mysql : Return 0 when no rows found

Answer 1

I see no easy way to create non-existing records out of thin air, but you could create yourself a point_dates table containing all the timestamps you're interested in, and left join it on your data:

select pd.slot, IFNULL(data_point, 0) from point_dates pd left join some_table st on st.point_date=pd.slot where point_date >= start_date AND point_date < end_date order by point_date

Answer 2

You cannot query data you do not have.

You (as a thinking person) can claim that the 00:20 data is missing; but there's no easy way to define "missing" in some more formal SQL sense.

The best you can do is create a table with all of the expected times.

Then you can do an outer join between expected times (including a 0 for 00:20) and actual times (missing the 00:20 sample) and you'll get kind of result you're expecting.

Answer 3

Yes, you can do that using SQL only. A solution would be to use a Stored Routine. The bellow Stored Procedure produces following output:

start   cnt
00:05:00   1
00:10:00   0
00:15:00   1
00:20:00   0
00:25:00   1
00:30:00   0
00:35:00   1
00:40:00   0
00:45:00   0
00:50:00   0
00:55:00   2

The table I used:

CREATE TABLE `timedata` (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `c1` datetime DEFAULT NULL,
  `c2` varchar(20) DEFAULT NULL,
  PRIMARY KEY (`id`)
)

Here the Stored Procedure (adjust for your environment):

DROP PROCEDURE IF EXISTS per5min;
DELIMITER //
CREATE PROCEDURE per5min ()
BEGIN
  DECLARE dtMin DATETIME;
  DECLARE dtMax DATETIME;
  DECLARE dtStart DATETIME;
  DECLARE dtStop DATETIME;
  DECLARE tmDiff TIME;
  DECLARE result INT UNSIGNED;
  SET @offset = 5 * 60;
  SELECT MIN(c1) into dtMin FROM timedata;
  SELECT MAX(c1) into dtMax FROM timedata;

  CREATE TEMPORARY TABLE tmp_per5min (
      start TIME,
      cnt INT UNSIGNED
  );

  SET dtStart = dtMin;
  REPEAT
    SELECT dtStart + INTERVAL @offset SECOND into dtStop;
    SELECT count(c2) into result FROM timedata WHERE c1 BETWEEN dtStart and dtStop;
    SELECT TIME(SUBTIME(dtStop,TIME(dtMin))) into tmDiff;
    INSERT INTO tmp_per5min (start,cnt) VALUES (tmDiff,result);
    SET dtStart = dtStop;
  UNTIL dtStop >= dtMax END REPEAT;

  SELECT * FROM tmp_per5min;
  DROP TABLE tmp_per5min;
END;
//
DELIMITER ;

CALL per5min();

If you save the above into a file called 'per5minproc.sql', you can load it like this:

shell> mysql -uroot test < per5minproc.sql

In Python using MySQLdb (I didn't get this working in MySQL Connector/Python, me ashamed!):

import MySQLdb as m

if __name__ == '__main__':
    db = m.connect(user='root',db='test')
    c = db.cursor()
    c.callproc("per5min")
    print(c.fetchall())
    c.close()
    db.close()

The solution above works, but probably will need some tweaking, e.g. dtStart can be an argument to the SP. And, it's indeed all SQL!