filling and printing numpy arrays of str

[pjfarley3]

I am new to python but experienced in several other languages. I do not understand yet why numpy.full does not fill an array of strings with the string value I supply in the code.

Example code and output follows. OS is Win10-64 Pro, Python version is 3.8.3.

1 2 3 4 5 6 7 8 9 10 11 12 13

#!/usr/bin/env python import numpy as np   class arstr:     def __init__(self, asize, aval):         self.strs = np.full([asize, asize, asize], aval, dtype=str)   Mystrs = arstr(3, " ? ")   for s in Mystrs.strs:     print (s)   exit()

Output in a cmd.exe window follows.

Output:
C:\Testdir>python atest.py
[[' ' ' ' ' ']
 [' ' ' ' ' ']
 [' ' ' ' ' ']]
[[' ' ' ' ' ']
 [' ' ' ' ' ']
 [' ' ' ' ' ']]
[[' ' ' ' ' ']
 [' ' ' ' ' ']
 [' ' ' ' ' ']]

Why are each of the array elements a single blank character and not the three-character string " ? " that I specified in the invocation of the class?

Thank you in advance for any help you can offer.

Peter

[bowlofred]

dtype=strmeans a fixed-size string. And without other information, it defaults to a one-character string. So your info is being truncated to the first character.

You can instead pass a tuple, with the padded size given as the second field.

1 2 3 4

>>> np.full([3], "ABC", dtype=(str)) array(['A', 'A', 'A'], dtype='<U1') >>> np.full([3], "ABC", dtype=(str, 3)) array(['ABC', 'ABC', 'ABC'], dtype='<U3')

[pjfarley3]

Thanks for that clarification. That cures at least a little my ignorance.

Is there a way to populate a numpy.ndarray with strings of arbitrary size?

Peter

[bowlofred]

It is possible to use the object type instead. Now instead of storing the data within the array, the array only stores references to external objects.

1 2 3 4 5 6 7 8 9

>>> x=np.full([3], "ABC") >>> x[1] = "Longerstring" >>> x array(['ABC', 'Lon', 'ABC'], dtype='<U3') #note truncation   >>> x=np.full([3], "ABC", dtype=object) >>> x[1] = "Longerstring" >>> x array(['ABC', 'Longerstring', 'ABC'], dtype=object)

[pjfarley3]

Thank you, that helps me a lot.

I will mark this thread as solved now.

Peter