SQL查询基于时间戳计算时间跨度

Question

In my project, I have need of calculating a time span. Currently, I'm retrieving every TimeStamp that matches my query and storing them in a List<> . Then, I iterate through the list to see if any intervals are 10 seconds or less, and then I add those together. Anything greater than 10 seconds is ignored. I'm wondering if there is a SQL query that I can do that will do this for me? I've done some searching, but didn't find anything. Essentially, I'd like to not have to store so much information in memory if I don't have to. Here's the method I'm using to iterate through my List<> :

private static TimeSpan TimeCalculations(IList<DateTime> timeStamps)
{
    var interval = new TimeSpan(0, 0, 10);
    var totalTime = new TimeSpan();

    for (var j = 0; j < timeStamps.Count - 1; j++)
    {
        if (timeStamps[j + 1].Subtract(timeStamps[j]) > interval) continue;
        var timeDifference = timeStamps[j + 1].Subtract(timeStamps[j]);
        totalTime = totalTime.Add(timeDifference);
    }

    return totalTime;
}

The data that is being retrieved currently can be anywhere from 10 to 400k rows worth of data. Here is a sample:

2006-09-07 11:46:09
2006-09-07 11:46:19 - 10 seconds
2006-09-07 11:46:20 - 1 second
2006-09-07 11:46:36

2006-09-07 11:47:49
2006-09-07 11:47:53 - 4 seconds
2006-09-07 11:48:02 - 9 seconds
2006-09-07 11:48:15
2006-09-07 11:48:29
2006-09-07 11:48:34 - 5 seconds

2006-09-07 11:54:29
2006-09-07 11:54:39 - 10 seconds
2006-09-07 11:54:49 - 10 seconds
2006-09-07 11:54:59 - 10 seconds

This would result in about 59 seconds. This is the kind of result I'm looking for.

The database I'm using is SQLite.

EDIT

Looking at the answers, I can tell that my question wasn't quite thorough enough. My current query to get the TimeStamps is sufficient. What I'm looking for is a query to add the difference between them together, if the interval is 10 seconds or less.

Answer 1

Used your sample data to create a sqlfiddle and this query works against your sample data:

SELECT DISTINCT tbl.timestamp FROM main_tbl tbl
INNER JOIN
(
 SELECT temp.ID, temp.timestamp FROM main_tbl temp
)test 
ON tbl.timestamp <= datetime(test.timestamp, '+10 seconds')
AND tbl.timestamp >= datetime(test.timestamp, '-10 seconds')
AND tbl.ID <> test.ID
ORDER BY tbl.timestamp

http://sqlfiddle.com/#!7/049f5/3

EDIT 2:

SELECT
sum(
  strftime('%s',
    (
    SELECT min(temp.timestamp)
        FROM main_tbl temp
        WHERE temp.timestamp > tbl.timestamp
    )
  ) - strftime('%s',tbl.timestamp)
) as total_sum
FROM main_tbl tbl
WHERE (
  strftime('%s',
    (
      SELECT min(temp.timestamp)
      FROM main_tbl temp
      WHERE temp.timestamp > tbl.timestamp
    )
  ) - strftime('%s',tbl.timestamp)
) <= 10 
AND date = "2013-05-13" 
AND col1 = col2

http://sqlfiddle.com/#!7/049f5/55

Answer 2

I played with a simple table with int cols t1 and t2 and got the right results from this query, I think. Adapt the penultimate line as needed!

SELECT sum(diff) FROM ( 
SELECT  t_1.rowid AS this_id, other_t1 - t_1.t1 as diff
FROM temp AS t_1
JOIN
 (SELECT t_2.t1 AS other_t1, t_2.rowid AS other_id
  FROM temp t_2  )
ON this_id = other_id-1
WHERE other_t1 - t_1.t1 = 1
);

It's a triple-nested select. The outer one sums all the differences found. and is just the first and last lines. The second level - from the 2nd line - does most of the work. The inner select delivers a list of table row and timestamp values for the second level to play with.

The action is to sum the differences of all rows that have a "t1" that differs by 1 from the next higher row.

To see the differences themselves, omit the first and last lines and replace the ";"

... forgot to say. t1 was supposed to be the timestamp. t2 was representing "other data".