Algorithm time...

[Dick Kriesel]

On Dec 9, 2012, at 12:33 AM, Glen Bojsza wrote:

> I believe that this should be doable less than 1 second

Hi, Glen.  Here's a draft you could try.  It works for the sample data you posted.

If you have questions, please ask.  If you try it, please report your timings.

-- Dick



<postScript>
function foo @pLines
   put the number of lines in pLines div 1000 into tGroupSize
   set the itemDelimiter to tab
   repeat for each line tLine in pLines
      add 1 to tLineNumber
      put item 2 of tLine into tValue
      if tIndex[ tValue ] is empty then
         put tLineNumber into tIndex[ tValue ]
      end if
      if tLineNumber is tGroupSize then
         put the keys of tIndex into tValues
         replace cr with comma in tValues
         put min( tValues ) into tMin
         put max( tValues ) into tMax
         if tIndex[ tMin ] < tIndex[ tMax ] then
            put tIndex[ tMin ] & tab & tMin & cr after tResult
            put tIndex[ tMax ] & tab & tMax & cr after tResult
         else
            put tIndex[ tMax ] & tab & tMax & cr after tResult
            put tIndex[ tMin ] & tab & tMin & cr after tResult
         end if
         put empty into tIndex
         put 0 into tLineNumber
      end if
   end repeat
   return tResult
end foo
</postScript>