sql Runtime reports

Question

I have a query which displays information from a table like this.

SELECT tag.label, inputs.input_raw, inputs.unix_timestamp, tag.inverted FROM inputs 
JOIN tag 
ON tag.tag_id = inputs.tag_id
WHERE (inputs.tag_id = 92084)

AND (inputs.date_time > dateadd(day,-1,getdate()))  
ORDER BY date_time DESC

I would like to write a query which would do two things.

I need a count of every time input_raw switches from '0' to '1'.

I also need a total time of the pump running, using the unix_timestamp ie. when the input_raw = 1.

Does anyone have any ideas.

I would settle for an algorithm to use php to get the results I need but I've hit a brick wall and haven't been able to figure it out.

Thanks

EDIT: The table also contains a date_time field which matches the value of the unix_timestamp if there is a date_time method that can be used>

Answer 1

You want to use the Lead or Lag function to compare the current result with either the previous or next. These functions, however, are introduced in SQL Server 2012.

With the help of Mr. pinaldave I managed to produce the following SQL Fiddle that counts every change from 0 to 1.

;WITH x AS 
(
  SELECT
    1 AS ldOffset, -- equiv to 2nd param of LEAD
    1 AS lgOffset, -- equiv to 2nd param of LAG
    NULL AS ldDefVal, -- equiv to 3rd param of LEAD
    NULL AS lgDefVal, -- equiv to 3rd param of LAG
    ROW_NUMBER() OVER (ORDER BY unix_timestamp) AS row,
    label,
    input_raw,
    unix_timestamp,
    inverted
  FROM inputs
 )
SELECT 
    COUNT(1)
FROM x
LEFT OUTER JOIN x AS xLg 
  ON x.row = xLg.row + x.lgOffset
WHERE xLg.input_raw = 0 AND x.input_raw = 1;

You can use the same Lag function to calculate the difference between the current timestamp and the previous timestamp.

EDIT:

This SQL Fiddle should show how to get the total time the pump is running (you'll have to define running yourself. This query now assumes that going from 0 to 1 or staying 1 is running. You should also double check my timestamp calculations cause I've never used unix timestamps before.

;WITH x AS 
(
  SELECT
    1 AS ldOffset, -- equiv to 2nd param of LEAD
    1 AS lgOffset, -- equiv to 2nd param of LAG
    NULL AS ldDefVal, -- equiv to 3rd param of LEAD
    NULL AS lgDefVal, -- equiv to 3rd param of LAG
    ROW_NUMBER() OVER (ORDER BY unix_timestamp) AS row,
    label,
    input_raw,
    unix_timestamp,
    inverted
  FROM inputs
 )
SELECT 
    SUM(DATEDIFF(mi, 
             DATEADD(ss, xLg.unix_timestamp,'01/01/1970'), 
             DATEADD(ss, x.unix_timestamp,'01/01/1970')))  
FROM x
LEFT OUTER JOIN x AS xLg 
  ON x.row = xLg.row + x.lgOffset
WHERE 
  (xLg.input_raw = 0 AND x.input_raw = 1)
OR
  (xLg.input_raw = 1 AND x.input_raw = 1);

EDIT 2:

I guess the easiest way to check for inverted is the change the WHERE clause to something like:

WHERE 
  (x.inverted = 1 AND xLg.input_raw = 0 AND x.input_raw = 1)
OR
  (x.inverted = 0 AND xLg.input_raw = 1 AND x.input_raw = 0)