I want to convert a = [1,2,3,4,5] into a_string = "1 2 3 4 5". The real numpy array is quite big (50000x200) so I assume using for loops is too slow.
7 Answers
You can use the join method from string:
>>> a = [1,2,3,4,5]
>>> ' '.join(map(str, a))
"1 2 3 4 5"
5 Comments
Framester
thanks, I tried
join unsuccessfully before. Map is the missing piece for me.
alexis
a simpler variant (IMHO) would be with an iterator:
' '.join(str(n) for n in a)
ben26941
Something like this looks good: def float2str(f): return '{:.5f}'.format(f) ' '.join(map(float2str, embeddings_fb.embed_token('.')))
shaifali Gupta
@tito Is there any fast alternative to do the same thing. It is very slow for a big array.
Mateen Ulhaq
@shaifaliGupta Printing is more bounded by I/O than CPU so I doubt alternative methods are going to be much faster.
np.savetxt
Python 3 (see also):
import numpy as np
import sys
a = np.array([0.0, 1.0, 2.0, 3.0])
np.savetxt(sys.stdout.buffer, a)
Python 2:
import numpy as np
import sys
a = np.array([0.0, 1.0, 2.0, 3.0])
np.savetxt(sys.stdout, a)
Output:
0.000000000000000000e+00
1.000000000000000000e+00
2.000000000000000000e+00
3.000000000000000000e+00
Control the precision
Use fmt:
np.savetxt(sys.stdout, a, fmt="%.3f")
output:
0.000
1.000
2.000
3.000
or:
np.savetxt(sys.stdout, a, fmt="%i")
output:
0
1
2
3
Get a string instead of printing
Python 3:
import io
bio = io.BytesIO()
np.savetxt(bio, a)
mystr = bio.getvalue().decode('latin1')
print(mystr, end='')
We use latin1 because the docs tell us that it is the default encoding used.
Python 2:
import StringIO
sio = StringIO.StringIO()
np.savetxt(sio, a)
mystr = sio.getvalue()
print mystr
All in one line
Or if you really want all in one line:
a = np.array([0.0, 1.0, 2.0, 3.0])
np.savetxt(sys.stdout, a, newline=' ')
print()
Output:
0.000000000000000000e+00 1.000000000000000000e+00 2.000000000000000000e+00 3.000000000000000000e+00
TODO: there is a trailing space. The only solution I see is to save to a string and strip.
Tested on Python 2.7.15rc1 and Python 3.6.6, numpy 1.13.3
answered Feb 4, 2017 at 23:17
Ciro Santilli OurBigBook.com
392k120 gold badges1.3k silver badges1.1k bronze badges
Comments
Maybe a bit hacky, but I just slice them off after using np.array2string so:
import numpy as np
a = np.arange(0,10)
a_str = np.array2string(a, precision=2, separator=', ')
print(a_str[1:-1])
Result:
0, 1, 2, 3, 4, 5, 6, 7, 8, 9
np.array2string has lots of options also, so you can set your column width which can be very helpful with lots of data:
a = np.arange(0,15)
a_str = np.array2string(a, precision=2, separator=', ', max_line_width=15)
print(' ' + a_str[1:-1])
Gives:
0, 1, 2,
3, 4, 5,
6, 7, 8,
9, 10, 11,
12, 13, 14
And it will smartly split at the array elements. Note the space appended to the beginning of the string to account for aligning the first row after removing the initial bracket.
Comments
If you have a numpy array to begin with rather than a list (since you mention a "real numpy array" in your post) you could use re.sub on the string representation of the array:
print(re.sub('[\[\]]', '', np.array_str(a)))
Again, this is assuming your array a was a numpy array at some point. This has the advantage of working on matrices as well.
Comments
Numpy provides two functions for this array_str and array_repr -- either of which should fit your needs. Since you could use either, here's an example of each:
>>> from numpy import arange, reshape, array_str
>>> M = arange(10).reshape(2,5)
>>> M
array([[0, 1, 2, 3, 4],
[5, 6, 7, 8, 9]])
>>> array_str(M)
'[[0 1 2 3 4]\n [5 6 7 8 9]]'
>>> array_repr(M)
'array([[0, 1, 2, 3, 4],\n [5, 6, 7, 8, 9]])'
These two functions are both highly optimized and, as such, should be preferred over a function you might write yourself. When dealing with arrays this size, I'd imagine you'd want all the speed you can get.
2 Comments
Framester
Using build-in functions is definitely of advantage, but it still leaves the brackets in. So this method superior for other, but similar, cases. +1
mau5padd
You said you need save the string to a file. If you have plans on later retrieving the string from that file it could be useful to have the brackets. Also, a combination of sub-strings and splits() would remove the brackets.
>>> a=np.array([1,2,3,4,5])
>>> print(*a)
1 2 3 4 5
>>> print(str(a)[1:-1])
1 2 3 4 5
same with lists
1 Comment
ZK xxxxx
The str() will not work when the length of the array is long. np.array2string is an alternative.
If you are dealing with floating-point numbers and two dimensional arrays you can also do the following:
import numpy as np
np.random.seed(77)
def my_print(x):
for row in x:
print(' '.join(map(lambda x: "{:.3f}\t".format(x), row)))
if __name__ == '__main__':
x = np.random.random(size=(3, 9))
my_print(x)
which prints:
0.919 0.642 0.754 0.139 0.087 0.788 0.326 0.541 0.240
0.545 0.401 0.715 0.837 0.588 0.296 0.281 0.706 0.423
0.057 0.747 0.452 0.176 0.049 0.292 0.067 0.751 0.064
writeline.