matdiff

#!/Users/andrewyale/.virtualenvs/default/bin/python
"""compares two matrices files"""

import sys
import os
import numpy as np

if len(sys.argv) != 3:
    print("Usage: matdiff <filename> <filename2>")
    sys.exit()

match = True

if sys.argv[1].endswith(".mat"):
    import scipy.io as sio

    data1 = sio.loadmat(os.path.join(os.getcwd(), sys.argv[1]))
    data2 = sio.loadmat(os.path.join(os.getcwd(), sys.argv[2]))

    keys1 = sorted(data1.keys())
    keys2 = sorted(data2.keys())
    if len(set(keys1).difference(keys2)):
        print("Variables are different")
        print(sys.argv[1] + " keys: " + " ".join(keys1))
        print(sys.argv[2] + " keys: " + " ".join(keys2))
        sys.exit()
    for k in keys1:
        if k[:2] != "__":
            if not np.allclose(data1[k], data2[k]):
                print("Variable " + k + " does not match")
                match = False

elif sys.argv[1].endswith(".csv"):
    import pandas as pd

    data1 = pd.read_csv(os.path.join(os.getcwd(), sys.argv[1]))
    data2 = pd.read_csv(os.path.join(os.getcwd(), sys.argv[2]))
    if data1.shape != data2.shape:
        print("Shapes are different")
        sys.exit()

    if all(data1 == data2):
        print("All variables match")
        sys.exit()

    try:
        if not np.allclose(data1, data2):
            print("Doesn't match")
            match = False
    except TypeError as te:
        if not np.allclose(
                data1.select_dtypes(exclude='object'),
                data2.select_dtypes(exclude='object')):
            print("Numbers don't match")
            match = False
        if not (data1.select_dtypes(
                include='object') == data2.select_dtypes(
                    include='object')).values.all():
            print("Objects don't match")
            match = False

if match:
    print("All variables match")