MySQL Count data for last 7 days

Question

I have the following schema.

Table votes

+------------------+--------------+------+-----+---------------------+----------------+
| Field            | Type         | Null | Key | Default             | Extra          |
+------------------+--------------+------+-----+---------------------+----------------+
| id               | int(10)      | NO   | PRI | NULL                | auto_increment |
| aid              | varchar(10)  | NO   |     |                     |                |
| ip               | varchar(100) | NO   |     |                     |                |
| host             | varchar(200) | NO   |     |                     |                |
| timestamp        | varchar(20)  | NO   |     | 0000-00-00 00:00:00 |                |
| user             | tinytext     | NO   |     | NULL                |                |
| userid           | int(10)      | NO   |     | 0                   |                |
+------------------+--------------+------+-----+---------------------+----------------+

Here I want to get the count of each aid on a day for the last 7 days with "0"s for the dates where there a no votes for aid. timestamp is unix timestamp here.

Any help is highly appreciated.

Answer 1

MySQL doesn't have recursive functionality, so you're left with using the NUMBERS table trick -

1. Create a table that only holds incrementing numbers - easy to do using an auto_increment:

   DROP TABLE IF EXISTS `example`.`numbers`;
   CREATE TABLE  `example`.`numbers` (
     `id` int(10) unsigned NOT NULL auto_increment,
      PRIMARY KEY  (`id`)
   ) ENGINE=InnoDB DEFAULT CHARSET=latin1;
   

2. Populate the table using:

   INSERT INTO NUMBERS
     (id)
   VALUES
     (NULL)
   

   ...for as many values as you need.

3. Use DATE_ADD to construct a list of dates, increasing the days based on the NUMBERS.id value. Replace "2010-01-01" and "2010-01-02" with your respective start and end dates (but use the same format, YYYY-MM-DD HH:MM:SS). In this example, I subtracted the NUMBERS.id value from the CURRENT_DATE to get a list of sequential date values for the last week -

   SELECT x.dt
     FROM (SELECT DATE_SUB(CURRENT_DATE, INTERVAL (n.id - 1) DAY) AS dt
             FROM numbers n
            WHERE n.id <= 7 ) x
   

4. LEFT JOIN onto your table of data based on the datetime portion.

      SELECT x.dt,
              COUNT(v.aid) AS num
        FROM (SELECT DATE_SUB(CURRENT_DATE, INTERVAL (n.id - 1) DAY) AS dt
                FROM numbers n
               WHERE n.id <= 7 ) x
   LEFT JOIN VOTES v ON DATE(FROM_UNIXTIME(v.timestamp)) = DATE(x.dt)
    GROUP BY x.dt
    ORDER BY x.dt
   

Why Numbers, not Dates?

Simple - dates can be generated based on the number, like in the example I provided. It also means using a single table, vs say one per data type.

Previously:

  SELECT DATE(FROM_UNIXTIME(v.timestamp)) AS dt,
         COUNT(v.aid)
    FROM VOTES v
   WHERE DATE(FROM_UNIXTIME(v.timestamp)) BETWEEN DATE_SUB(CURRENT_DATE, INTERVAL 7 DAY)
                                              AND CURRENT_DATE
GROUP BY DATE(FROM_UNIXTIME(v.timestamp))

Answer 2

might want to try converting your varchar(20) to a DATETIME, since that is what you are using anyways. But for the last 7 days just do WHERE timestamp > @7DaysAgo. In your explanation you said your using a unix timestamp, so u will need to send the timestamp from now - 7 days. You will also want to change your default to 0, since your sending timestamps you said, not DATETIME values.

SELECT COUNT(*) as votes, aid FROM table WHERE timestamp>@7DaysAgo GROUP BY aid