Padding MYSQL data with missing dates when comparing year over year stats? - mysql

Question

I have a table that tracks emails sent. It is pretty simple.
ID | DATETIME | E-MAIL | SUBJECT | MESSAGE
I have been collecting data for several years. Some days I don't have any entries in the table.
query1:
SELECT COUNT(ID) FROM emails
WHERE DATE(datetime) >= 'XXXX-XX-XX'
AND DATE(datetime) is <= 'ZZZZ-ZZ-ZZ'
GROUP BY DATE(datetime)
I then use a some php to get one year prior for both XXXX and YYYY and run the second query which is the same as the first...
query2:
SELECT COUNT(ID) from emails
WHERE DATE(datetime) >= 'XXXX-XX-XX'
AND DATE(datetime) is <= 'ZZZZ-ZZ-ZZ'
GROUP BY DATE(datetime)
I am using a charting package to compare how many emails I got for a date range and then I overlay how many emails I got for the same range only one year prior. This is two queries right now and I chart the results.
The issue is where mysql does not have any emails for 2011 for a day in question, but has a few in 2012 for the same day.
Combining the results and graphing them skews the results since I am missing a date and a 0 value for last year for that day, effectively making all my values no longer match up.
2011-03-01 10 2012-03-01 4
2011-03-02 4 2012-03-02 2
2011-03-03 6 2012-03-04 1 <---- see where the two queries
end up diverging? (I had nothing
logged for 2012-03-03 so naturally
it was not in the results.
Is there a way I can get mysql to output the data I need including dates where value appear in one year but not another OR if no values appear in either year (still need date and 0) so my chart works?
I cannot seem to figure out how to do this...
Thanks!

Answer 1

There are a few different ways to get the results for a contiguous set of dates. My favourite one is to create the full set that is required using a dummy table or an existing contiguous set of ids from an AI PK. Something like this -
SELECT '2011-01-01' + INTERVAL (id -1) DAY
FROM dummy
WHERE id BETWEEN 1 AND 365
This will return a full set of days for 2011 which can then be LEFT JOINed to your emails table to get the counts -
SELECT `dates`.`date`, COUNT(emails.id)
FROM (
SELECT '2011-01-01' + INTERVAL (id - 1) DAY AS `date`, '2011-01-01 23:59:59' + INTERVAL (id - 1) DAY AS `end_of_day`
FROM dummy
WHERE id BETWEEN 1 AND 365
) `dates`
LEFT JOIN emails
ON `emails`.`datetime` BETWEEN `dates`.`date` AND `dates`.`end_of_day`
GROUP BY `dates`.`date`
To populate your dummy / seq table you can insert the first ten values manually and then use INSERT ... SELECT to add the rest -
CREATE TABLE dummy (id INTEGER NOT NULL PRIMARY KEY);
INSERT INTO dummy VALUES (0),(1),(2),(3),(4),(5),(6),(7),(8),(9),(10);
SET #tmp := (SELECT MAX(id) FROM dummy) + 1;
INSERT INTO dummy
SELECT #tmp + id
FROM dummy;
You need to execute the SET query before each run of the INSERT ... SELECT query.