write() vs. writelines()

[Gregory Petrosyan]

My question is: why write(''.join(...)) works slowly than
writelines(...)? Here's the code:

import sys
import random

#try:
#import psyco
#psyco.full()
#except ImportError:
#pass


def encrypt(key, infile, outfile, bufsize=65536):
random.seed(key)
in_buf = infile.read(bufsize)
while in_buf:
outfile.write(''.join(chr(ord(char) ^ random.randint(0, 255))
for char in in_buf))
in_buf = infile.read(bufsize)


if __name__ == '__main__':
import time
key = int(sys.argv[1])
infile = open(sys.argv[2], 'rb')
outfile = open(sys.argv[3], 'wb')
t = time.time()
encrypt(key, infile, outfile)
print time.time() - t

 

[Fredrik Lundh]

My question is: why write(''.join(...)) works slowly than
writelines(...)? Here's the code:

the first copies all the substring to a single string large enough to
hold all the data, before handing it over to the file object, while the
second just writes the substrings to the file object.

</F>

 

[Gregory Petrosyan]

Thanks for your reply. I understand this fact, but I wonder why
writelines() works slowly -- I think C code can be optimised to work
faster than Python one. Is it correct that writelines(...) is just a
shorthand for

for ch in ...:
file.write(ch)
?

 

[Marc 'BlackJack' Rintsch]

Thanks for your reply. I understand this fact, but I wonder why
writelines() works slowly -- I think C code can be optimised to work
faster than Python one. Is it correct that writelines(...) is just a
shorthand for

for ch in ...:
file.write(ch)
?

Depends on `...`. If it's a string then yes because a string is a
sequence of characters. But `writelines()` is ment for a sequence of
strings. It's the counterpart of `readlines()`. Then it's just a
shorthand for::

for line in lines:
file.write(line)

Ciao,
Marc 'BlackJack' Rintsch