Filtering Columnar Data

[Gregory Lypny]

Hello Josh,

Here's the final version of the function.  Thanks to everyone for  
their suggestions.  I know there are other features and better error- 
checking that can be added.  Maybe later.  It can extract roughly 75  
records from 700 in 1 tick on an Intel-based iMac.

Regards,

	Gregory

function filteredByColumn theData,theColNum,theSearchString,isWithout
   -- This function filters tab-delimited data by column
   -- By default the filter extracts lines containing the search  
string.  Make isWithout = "without" to extract lines without it.
   -- Possible mods: options to make "begins with" or "contains"  
searches
   if theColNum is an integer
   then
     set the itemDelimiter to tab
     if isWithout is empty
     then
       sort lines of theData by (item theColNum of each is not  
theSearchString)  -- Brings lines with theSearchString to the top
       repeat for each line thisLine in theData
         if item theColNum of thisLine is theSearchString
         then
           put thisLine & return after filteredData
         else
           exit repeat
         end if
       end repeat
     else
       sort lines of theData by (item theColNum of each is  
theSearchString)  -- Brings lines without theSearchString to the top
       repeat for each line thisLine in theData
         if item theColNum of thisLine is not theSearchString
         then
           put thisLine & return after filteredData
         else
           exit repeat
         end if
       end repeat
     end if
     delete the last character of filteredData
     return filteredData
   else
     return "The column number must be an integer."
   end if
end filteredByColumn






On Mon, Jul 16, 2007, at 7:49 PM, Josh Mellicker responded:

> On Jul 16, 2007, at 2:04 PM, Gregory Lypny wrote:
>
>> I tested the function below, which is much like Josh's, and it
>> filters more than 300 lines in 1 tick; the same code as a message
>> handler rather than a function takes 5 ticks; and the filter
>> command on the same data takes 39 ticks.
>
>
> Wow, I thought that handler was fast, but never tested/compared it.
>
> I wonder if sorting the lines by columnNum, then exiting on the first
> non-matching line would be faster or slower?
>