How do you turn a string back into an array?

Question

Firstly is there a way to turn a string into an array?
I'm looking for a way to turn this:

['[[ 18.41978673]\n', ' [ 0.34748864]\n', ' [ -9.55729142]]'] 

back into

[[ 18.41978673]
 [ 0.34748864]
 [ -9.55729142]]

Secondly, is there a was to store the array as an array in a .txt file without having to turn it into a string in the first place?
This is the current method I have been using to put them into storage:

def add_to_file(self, weights):
 path = 'C:/Users\PycharmProjects\project Current'
 file = open(path, 'w')
 file.write(weights)
 file.close()
add_to_file(str(synaptic_weights))

When trying to do it the following way:

add_to_file(synaptic_weights)

I have this return in the console:

TypeError: write() argument must be str, not numpy.ndarray

The file type doesn't have to be .txt. If there is a better way to store arrays I am all ears.

• moritzg: Have a look at serializing docs.python.org/2/library/pickle.html
• juanpa.arrivillaga: Use numpy.save

Answer 1

Since you're using a Numpy array, an easy solution to save the array is:

np.savetxt(file_name, my_array)

then:

my_array = np.loadtxt(file_name)

• obachtos: There's a typo there, the function is called np.savetxt, not np.savetext.

Answer 2

You'll need to jump through a few hoops, but you can do it with json.loads and re.sub. This is the first step to salvaging your data and storing it in a usable form.

import json
import re
data = json.loads(re.sub(r"'|\\n", '', str(l)))[0]
print(data)
# [[18.41978673], [0.34748864], [-9.55729142]]

---

Now that you've salvaged your data, you can look at some options for saving to a file.

One option (probably the best) is using np.save:

np.save('data.npy', data) 

Alternatives include either json.dump or pickle.dump. This would work:

import json
json.dump(data, open('data.json', 'w'))

Or, with pickle:

import pickle
pickle.dump(data, open('data.pkl', 'wb'))

• moritzg: Of course this works, but if he'd serialize it in the first place it would be much better.
• coldspeed95: @moritzg The list has not been stringified properly to begin with. If OP is to salvage anything, sooner or later, they'd need to do this.

Answer 3

I usually use

a = np.ones((3,3))
np.save('arr', a)
a2 = np.load('arr.npy')
print a2

See the documentation here: np.save, np.load