使用python中的大数据，比map（）更快的方法（通过stdin.readlines（））

Question

I've written a piece of code where i have to handle a lot of lines coming from stdin, split them, and then convert the values to integers (it's for a typical programming challenge).

My script needs ~20 sec for ~50,000 lines, but with cProfile i found out that all the map calls allready take ~8 seconds (i call split and map for every line, looks inefficient to me).

Here's what i do:

inp = [i[0:-1] for i in stdin.readlines()][1:]
inp = [map(int, i.split()) for i in inp]

What's the more pythonic way to do this?

Input looks like:

2        # number of testcases
2        # depth of following numeric triangle
1
1 2
3        # depth of next numeric triangle
1
1 2
1 2 3

thanks a lot for help!

Answer 1

Use numpy.loadtxt or numpy.genfromtxt. That will load the numbers into an array far more quickly than you can load them with loops in Python, because the implementation is in a compiled language and it's highly optimized.