Performance of Extract-Transform operations

Question

I'm doing simple data analysis, where I load some "date-time" data from a file. The data is a simple list, but there are 41,000 values in total. Here's a short sample:

"26.11.2014 13:56:17" "26.11.2014 13:56:18" "26.11.2014 13:56:20"

Here is my code, which extracts a list of date-times from a file and transform the results into list of seconds (integers):

l = Import["c:\\tmp\\1.csv", "List"];   
l2 = AbsoluteTime /@ l;

The problem is that it takes about 15 seconds and 100% of a CPU core to do the step 2 (AbsoluteTime) for a mere 41k elements.

Are there ways to speed up this operation?

Maybe Import into a list of Dates, not into a generic list? Or should I change the strategy altogether and prepare data before loading into the Mathematica?

Answer 1

l = ConstantArray[{"26.11.2014 13:56:17", "26.11.2014 13:56:18" , 
                   "26.11.2014 13:56:20"}, {10000}] // Flatten;

I can get 20x speed up by converting it to DateList form:

AbsoluteTime[
  ToExpression@StringSplit[#, " " | "." | ":"][[{3, 2, 1, 4, 5, 6}]] 
  ]& /@  l; // Timing

{1.684811, Null}