Solved

# how to weekdiff, monthdiff, yeardiff

Posted on 2015-01-19
344 Views
I want to extend the answer of this question to week, month and year

http://www.experts-exchange.com/Database/MySQL/Q_28599321.html#a40557723

There used datediff function but I want  find a solution to the cases  week, month and year.
eg: recurring every other week, recurring every 3 months, recurring every 5 years, etc.
0
Question by:myyis
• 4
• 3
• 2
• +2

LVL 5

Assisted Solution

danirk2 earned 100 total points
Using the same sample data as in the referenced solution:

SELECT TIMESTAMPDIFF(WEEK,now(),d), mod(TIMESTAMPDIFF(MONTH,now(),d), 10)
SELECT TIMESTAMPDIFF(MONTH,now(),d), mod(TIMESTAMPDIFF(MONTH,now(),d), 10)
SELECT TIMESTAMPDIFF(YEAR,now(),d), mod(TIMESTAMPDIFF(MONTH,now(),d), 10)
0

Author Comment

should be like this?

SELECT TIMESTAMPDIFF(WEEK,now(),d), mod(TIMESTAMPDIFF(WEEK,now(),d), 10)
SELECT TIMESTAMPDIFF(MONTH,now(),d), mod(TIMESTAMPDIFF(MONTH,now(),d), 10)
SELECT TIMESTAMPDIFF(YEAR,now(),d), mod(TIMESTAMPDIFF(YEAR,now(),d), 10)
0

Author Comment

Hi there are some problems with the query

SELECT CASE WHEN RIGHT( RECUR, 1 ) =  'd'
THEN MOD( DATEDIFF( NOW( ) , DATETIMEDUE ) , SUBSTRING( RECUR, 1, LENGTH( RECUR ) -1 ) )
WHEN RIGHT( RECUR, 1 ) =  'w'
THEN MOD( TIMESTAMPDIFF( WEEK, NOW( ) , DATETIMEDUE ) , SUBSTRING( RECUR, 1, LENGTH( RECUR ) -1 ) )
WHEN RIGHT( RECUR, 1 ) =  'm'
THEN MOD( TIMESTAMPDIFF(MONTH , NOW( ) , DATETIMEDUE ) , SUBSTRING( RECUR, 1, LENGTH( RECUR ) -1 ) )
WHEN RIGHT( RECUR, 1 ) =  'y'
THEN MOD( TIMESTAMPDIFF( YEAR, NOW( ) , DATETIMEDUE ) , SUBSTRING( RECUR, 1, LENGTH( RECUR ) -1 ) )
END AS T, RECUR AS R, DATETIMEDUE

T      R      DATETIMEDUE
0      1d      2015-01-19 20:40:59
0      1m      2015-01-19 23:59:59
1      2d      2015-01-18 20:48:59
1      2d      2015-01-18 06:00:59
-0      2m      2014-10-19 23:59:59 not correct
-0      2w      2015-01-05 03:10:59
-1      2w      2015-01-06 23:59:59
0      2y      2014-01-19 23:59:59 not correct
-0      2y      2013-01-18 20:40:59 not correct
0      3d      2015-01-16 22:38:59
-1      3m      2014-11-19 23:59:59 conflicts with below
-0      3m      2014-10-19 14:23:59
-2      3m      2014-11-19 20:44:59 conflicts
-2      4w      2014-11-10 20:49:59
-0      5w      2014-12-15 20:42:59
-4      5w      2014-12-16 23:59:59
0      8d      2015-01-11 20:39:59
0

LVL 45

Expert Comment

First parameter of the Datediff function is the datepart that you want to be returned:
year  (yy, yyyy )
month  (mm, m)
week  (wk, ww)

So you just need to change your query to return what you need. Example for month:
``````select d
, datediff(month, now(), d)
, mod(datediff(month,now(), d), 10)
from test;
``````
0

LVL 34

Expert Comment

When you posted sample results and said that some were wrong, it would have helped if you said what you wanted to see instead.

I don't have MySQL, so I can't test this, but I suspect that the time may be a factor.  For example, if DATETIMEDUE is 2014-10-19 23:59:59 and the current time is 2015-01-19 20:00:00, and you ask for the difference in months, the result may be 2, because it wouldn't be 3 full months until 2015-01-19 23:59:59.

If you want to ignore the times and just look at the dates, I think you can use CURDATE() instead of NOW(), and use DATE on DATETIMEDUE, to extract the date without the time.

You also need some kind of check to make sure that the basic interval has elapsed.  For example, if RECUR is 2y, but it hasn't even been 1 year yet, TIMESTAMPDIFF will give you 0 (years), so the MOD function will give you 0 (0 mod 2 = 0).

But a more basic problem may be the "rounding" on the TIMESTAMPDIFF results.  For example:

-0      2y      2013-01-18 20:40:59 not correct

I assume that you're saying that 0 is wrong because you ran that query on 01/19, so it had been 2 years and 1 day, not exactly 2 years.  But TIMESTAMPDIFF presumably produces integer results, so if you ask for the difference in years, any date from 01/18/2015 to 01/17/2016 will produce the same result -- 2 years.

You could do a simple test by running a query like the following and checking the results:

SELECT TIMESTAMPDIFF(YEAR,'2013-01-18 20:40:59','2015-06-30 20:40:59');

The difference is around 2 and a half years, but I'm guessing that the result will just be 2.

James
0

LVL 34

Assisted Solution

James0628 earned 100 total points
After a little more thinking, I think the best answer might be to sort of reverse your approach.  Instead of using TIMESTAMPDIFF to calculate the difference between DATETIMEDUE and today and trying to check the interval, I think you could use DATE_ADD to add the interval to DATETIMEDUE, and then compare that to today's date.

James
0

LVL 48

Accepted Solution

PortletPaul earned 300 total points
Trying to match calculations using NOW() is a bit frustrating so below I have used @NOW which I have set to 2015-01-19 instead. This can be replaced by CURDATE(), or set @NOW equal to CURDATE().

Note I don't believe there is any point in using NOW() if the units are days, weeks, months, years, and in fact I think it;s useful to reduce the DATETIMEDUE to just a date (at 00:00:00) as well.

``````SET @NOW := '2015-01-19 00:00:00';

SELECT
*
FROM (
SELECT
RECUR AS R
, SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1) as num
, RIGHT(RECUR, 1) as unit
, DATETIMEDUE
, CASE
WHEN RIGHT(RECUR, 1) = 'd' and @now > date(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(DAY, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
WHEN RIGHT(RECUR, 1) = 'w' and @now > date(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(DAY, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1)*7)
WHEN RIGHT(RECUR, 1) = 'm' and @now > date(DATETIMEDUE) and dayofmonth(@now) = dayofmonth(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(MONTH, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
WHEN RIGHT(RECUR, 1) = 'y' and @now > date(DATETIMEDUE) and dayofmonth(@now) = dayofmonth(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(YEAR, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
ELSE
999.9
END AS T1

, TIMESTAMPDIFF(DAY, date(DATETIMEDUE), @NOW) ndays
, TIMESTAMPDIFF(week, date(DATETIMEDUE), @NOW) nweeks
, TIMESTAMPDIFF(month, date(DATETIMEDUE), @NOW) nmonths
, TIMESTAMPDIFF(year, date(DATETIMEDUE), @NOW) nyears
FROM table1
) AS derived
WHERE T1 = 0
ORDER BY T1, date(DATETIMEDUE)
;
``````

nb, there are some columns just there for reference.
For the sample data given above this query returns:
``````|  R | NUM | UNIT |                     DATETIMEDUE | T1 | NDAYS | NWEEKS | NMONTHS | NYEARS |
|----|-----|------|---------------------------------|----|-------|--------|---------|--------|
| 3m |   3 |    m |  October, 19 2014 14:23:59+0000 |  0 |    92 |     13 |       3 |      0 |
| 5w |   5 |    w | December, 15 2014 20:42:59+0000 |  0 |    35 |      5 |       1 |      0 |
| 2w |   2 |    w |  January, 05 2015 03:10:59+0000 |  0 |    14 |      2 |       0 |      0 |
| 8d |   8 |    d |  January, 11 2015 20:39:59+0000 |  0 |     8 |      1 |       0 |      0 |
| 3d |   3 |    d |  January, 16 2015 22:38:59+0000 |  0 |     3 |      0 |       0 |      0 |
``````

and correspondinly these rows are ignored (T1 <> 0)
``````|  R | NUM | UNIT |                     DATETIMEDUE |    T1 | NDAYS | NWEEKS | NMONTHS | NYEARS |
|----|-----|------|---------------------------------|-------|-------|--------|---------|--------|
| 2y |   2 |    y |  January, 19 2014 23:59:59+0000 |     1 |   365 |     52 |      12 |      1 |
| 2m |   2 |    m |  October, 19 2014 23:59:59+0000 |     1 |    92 |     13 |       3 |      0 |
| 2d |   2 |    d |  January, 18 2015 20:48:59+0000 |     1 |     1 |      0 |       0 |      0 |
| 2d |   2 |    d |  January, 18 2015 06:00:59+0000 |     1 |     1 |      0 |       0 |      0 |
| 3m |   3 |    m | November, 19 2014 20:44:05+0000 |     2 |    61 |      8 |       2 |      0 |
| 3m |   3 |    m | November, 19 2014 23:59:59+0000 |     2 |    61 |      8 |       2 |      0 |
| 2w |   2 |    w |  January, 06 2015 23:59:59+0000 |    13 |    13 |      1 |       0 |      0 |
| 4w |   4 |    w | November, 10 2014 20:49:59+0000 |    14 |    70 |     10 |       2 |      0 |
| 5w |   5 |    w | December, 16 2014 23:59:59+0000 |    34 |    34 |      4 |       1 |      0 |
| 2y |   2 |    y |  January, 18 2013 20:40:59+0000 | 999.9 |   731 |    104 |      24 |      2 |
| 1m |   1 |    m |  January, 19 2015 23:59:59+0000 | 999.9 |     0 |      0 |       0 |      0 |
| 1d |   1 |    d |  January, 19 2015 20:40:59+0000 | 999.9 |     0 |      0 |       0 |      0 |
``````

Details:
``````**MySQL 5.5.32 Schema Setup**:

CREATE TABLE Table1
(`T` int, `RECUR` varchar(2), `DATETIMEDUE` datetime, `note` varchar(100))
;

INSERT INTO Table1
(`T`, `RECUR`, `DATETIMEDUE`, `note`)
VALUES
(0, '1d', '2015-01-19 20:40:59',null),
(0, '1m', '2015-01-19 23:59:59',null),
(1, '2d', '2015-01-18 20:48:59',null),
(1, '2d', '2015-01-18 06:00:59',null),
(-0, '2m', '2014-10-19 23:59:59','not correct'),
(-0, '2w', '2015-01-05 03:10:59',null),
(-1, '2w', '2015-01-06 23:59:59',null),
(0, '2y', '2014-01-19 23:59:59','not correct'),
(-0, '2y', '2013-01-18 20:40:59','not correct'),
(0, '3d', '2015-01-16 22:38:59',null),
(-1, '3m', '2014-11-19 23:59:59','conflicts with below'),
(-0, '3m', '2014-10-19 14:23:59',null),
(-2, '3m', '2014-11-19 20:44:05','conflicts'),
(-2, '4w', '2014-11-10 20:49:59',null),
(-0, '5w', '2014-12-15 20:42:59',null),
(-4, '5w', '2014-12-16 23:59:59',null),
(0, '8d', '2015-01-11 20:39:59',null)
;

**Query 1**:

SET @NOW := '2015-01-19 00:00:00'

**[Results][2]**:

**Query 2**:

SELECT
*
FROM (
SELECT
RECUR AS R
, SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1) as num
, RIGHT(RECUR, 1) as unit
, DATETIMEDUE
, CASE
WHEN RIGHT(RECUR, 1) = 'd' and @now > date(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(DAY, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
WHEN RIGHT(RECUR, 1) = 'w' and @now > date(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(DAY, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1)*7)
WHEN RIGHT(RECUR, 1) = 'm' and @now > date(DATETIMEDUE) and dayofmonth(@now) = dayofmonth(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(MONTH, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
WHEN RIGHT(RECUR, 1) = 'y' and @now > date(DATETIMEDUE) and dayofmonth(@now) = dayofmonth(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(YEAR, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
ELSE
999.9
END AS T1

, TIMESTAMPDIFF(DAY, date(DATETIMEDUE), @NOW) ndays
, TIMESTAMPDIFF(week, date(DATETIMEDUE), @NOW) nweeks
, TIMESTAMPDIFF(month, date(DATETIMEDUE), @NOW) nmonths
, TIMESTAMPDIFF(year, date(DATETIMEDUE), @NOW) nyears
FROM table1
) AS derived
WHERE T1 = 0
ORDER BY T1, date(DATETIMEDUE)

**[Results][3]**:

|  R | NUM | UNIT |                     DATETIMEDUE | T1 | NDAYS | NWEEKS | NMONTHS | NYEARS |
|----|-----|------|---------------------------------|----|-------|--------|---------|--------|
| 3m |   3 |    m |  October, 19 2014 14:23:59+0000 |  0 |    92 |     13 |       3 |      0 |
| 5w |   5 |    w | December, 15 2014 20:42:59+0000 |  0 |    35 |      5 |       1 |      0 |
| 2w |   2 |    w |  January, 05 2015 03:10:59+0000 |  0 |    14 |      2 |       0 |      0 |
| 8d |   8 |    d |  January, 11 2015 20:39:59+0000 |  0 |     8 |      1 |       0 |      0 |
| 3d |   3 |    d |  January, 16 2015 22:38:59+0000 |  0 |     3 |      0 |       0 |      0 |

**Query 3**:

SET @NOW := '2015-01-19 00:00:00'

**[Results][4]**:

**Query 4**:

SELECT
*
FROM (
SELECT
RECUR AS R
, SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1) as num
, RIGHT(RECUR, 1) as unit
, DATETIMEDUE
, CASE
WHEN RIGHT(RECUR, 1) = 'd' and @now > date(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(DAY, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
WHEN RIGHT(RECUR, 1) = 'w' and @now > date(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(DAY, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1)*7)
WHEN RIGHT(RECUR, 1) = 'm' and @now > date(DATETIMEDUE) and dayofmonth(@now) = dayofmonth(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(MONTH, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
WHEN RIGHT(RECUR, 1) = 'y' and @now > date(DATETIMEDUE) and dayofmonth(@now) = dayofmonth(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(YEAR, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
ELSE
999.9
END AS T1

, TIMESTAMPDIFF(DAY, date(DATETIMEDUE), @NOW) ndays
, TIMESTAMPDIFF(week, date(DATETIMEDUE), @NOW) nweeks
, TIMESTAMPDIFF(month, date(DATETIMEDUE), @NOW) nmonths
, TIMESTAMPDIFF(year, date(DATETIMEDUE), @NOW) nyears
FROM table1
) AS derived
WHERE T1 <> 0
ORDER BY T1, date(DATETIMEDUE)

**[Results][5]**:

|  R | NUM | UNIT |                     DATETIMEDUE |    T1 | NDAYS | NWEEKS | NMONTHS | NYEARS |
|----|-----|------|---------------------------------|-------|-------|--------|---------|--------|
| 2y |   2 |    y |  January, 19 2014 23:59:59+0000 |     1 |   365 |     52 |      12 |      1 |
| 2m |   2 |    m |  October, 19 2014 23:59:59+0000 |     1 |    92 |     13 |       3 |      0 |
| 2d |   2 |    d |  January, 18 2015 20:48:59+0000 |     1 |     1 |      0 |       0 |      0 |
| 2d |   2 |    d |  January, 18 2015 06:00:59+0000 |     1 |     1 |      0 |       0 |      0 |
| 3m |   3 |    m | November, 19 2014 20:44:05+0000 |     2 |    61 |      8 |       2 |      0 |
| 3m |   3 |    m | November, 19 2014 23:59:59+0000 |     2 |    61 |      8 |       2 |      0 |
| 2w |   2 |    w |  January, 06 2015 23:59:59+0000 |    13 |    13 |      1 |       0 |      0 |
| 4w |   4 |    w | November, 10 2014 20:49:59+0000 |    14 |    70 |     10 |       2 |      0 |
| 5w |   5 |    w | December, 16 2014 23:59:59+0000 |    34 |    34 |      4 |       1 |      0 |
| 2y |   2 |    y |  January, 18 2013 20:40:59+0000 | 999.9 |   731 |    104 |      24 |      2 |
| 1m |   1 |    m |  January, 19 2015 23:59:59+0000 | 999.9 |     0 |      0 |       0 |      0 |
| 1d |   1 |    d |  January, 19 2015 20:40:59+0000 | 999.9 |     0 |      0 |       0 |      0 |

[1]: http://sqlfiddle.com/#!2/f0128/26

[2]: http://sqlfiddle.com/#!2/f0128/26/0
``````
0

LVL 34

Expert Comment

Re: My last post, where I suggested using DATE_ADD

FWIW, I realized later that I wasn't thinking of multiples of the interval.  For example, if the interval was 2y, I was just thinking of 2 years, not 4, 6, etc.  So that idea wasn't as straightforward as I thought it was.

Hopefully Paul's code will give you what you need.

Paul,

When it's a year interval, shouldn't you also check the month, along with the day of the month?

myyis,

I kind of hate to even bring it up, because it could really complicate things, but there are potential problems because different months have a different numbers of days.  For a simple example, let's say that DATETIMEDUE was on 03/31 and the interval was 1 month.  There is no 04/31.  How would you want to handle that?

Similarly, you have February and leap years.  What if DATETIMEDUE is on 02/29 and the interval is years?

James
0

Author Comment

Hi James,
I can manipulate that kind of cases you've written, I see that Paul's code is ok. Thank you
0

LVL 48

Expert Comment

>>"When it's a year interval, shouldn't you also check the month..."
good point, thanks

WHEN RIGHT(RECUR, 1) = 'y' and @now > date(DATETIMEDUE) and month(@now) = month(DATETIMEDUE) and dayofmonth(@now) = dayofmonth(DATETIMEDUE)
THEN MOD(TIMESTAMPDIFF(YEAR, date(DATETIMEDUE), @NOW), SUBSTRING(RECUR, 1, LENGTH(RECUR) - 1))
0

LVL 34

Expert Comment

I can manipulate that kind of cases you've written, I see that Paul's code is ok.
Glad to hear it.  Handling the different number of days per month seemed like something that could get really messy.

James
0

## Featured Post

Read about achieving the basic levels of HRIS security in the workplace.
Using SQL Scripts we can save all the SQL queries as files that we use very frequently on our database later point of time. This is one of the feature present under SQL Workshop in Oracle Application Express.
Using examples as well as descriptions, step through each of the common simple join types, explaining differences in syntax, differences in expected outputs and showing how the queries run along with the actual outputs based upon a simple set of dem…
Polish reports in Access so they look terrific. Take yourself to another level. Equations, Back Color, Alternate Back Color. Write easy VBA Code. Tighten space to use less pages. Launch report from a menu, considering criteria only when it is filled…