speed bottleneck within decode_mapi

[TZanke]

Hello,

today i had a problem with a customer tnef mail. It took around 150ms to call decode_mapi. Because it is called several times within tnefparse and within my code, it took around 2500ms to handle a request to this file.

I searched for the bottleneck with cProfile and found the problem:

In decode_mapi the var num_vals is a int bigger then several millions. Python will generate a list this big... Solution: do not generate this big list, just use a iterator: xrange()

Speed change: 2500ms > 60ms

[TZanke]

I do not know how you handle python2/3 compatibility, so i do not implement a patch atm.

http://python-future.org/compatible_idioms.html#xrange

[jrideout]

This was fixed in 0898eb8

[petri]

thx @jrideout