--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/MLaaS/mnist_img.py	Tue Mar 28 15:07:30 2023 +0000
@@ -0,0 +1,83 @@
+#!/usr/bin/env python
+#-*- coding: utf-8 -*-
+#pylint: disable=
+"""
+File       : mnist_img.py
+Author     : Valentin Kuznetsov <vkuznet AT gmail dot com>
+Description: 
+"""
+
+import json
+import gzip
+import argparse
+# from itertools import chain
+
+import numpy as np
+import matplotlib.pyplot as plt
+
+
+def readImage(fname, fout, num_images=5, imgId=2):
+    """
+    Helper function to read MNIST image
+    """
+    image_size = 28
+    with gzip.open(fname, 'r') as fstream:
+        fstream.read(16)
+        buf = fstream.read(image_size * image_size * num_images)
+        data = np.frombuffer(buf, dtype=np.uint8).astype(np.float32)
+        data = data.reshape(num_images, image_size, image_size, 1)
+        image = np.asarray(data[imgId]).squeeze()
+        plt.imsave(fout, image)
+        print("read:", fname, "wrote:", fout, "image:", type(image), "shape:", image.shape)
+
+def img2json(image):
+    """
+    Convert given image to JSON data format used by TFaaS
+    """
+    # values = [int(i) for i in list(chain.from_iterable(image))]
+    # values = image.tolist()
+    values = []
+    for row in image.tolist():
+        row = [int(i) for i in row]
+        vals = [[i] for i in row]
+        values.append(vals)
+    # final values should be an array of elements, e.g. single image representation
+    values = [values]
+    keys = [str(i) for i in range(0, 10)]
+    meta = {
+        'keys': keys,
+        'values': values,
+        'model': 'mnist'
+    }
+    with open('img.json', 'w') as ostream:
+        ostream.write(json.dumps(meta))
+
+
+class OptionParser():
+    def __init__(self):
+        "User based option parser"
+        fname = "train-images-idx3-ubyte.gz"
+        self.parser = argparse.ArgumentParser(prog='PROG')
+        self.parser.add_argument("--fin", action="store",
+            dest="fin", default=fname, help=f"Input MNIST file, default {fname}")
+        self.parser.add_argument("--fout", action="store",
+            dest="fout", default="img.png", help="Output image fila name, default img.png")
+        self.parser.add_argument("--nimages", action="store",
+            dest="nimages", default=5, help="number of images to read, default 5")
+        self.parser.add_argument("--imgid", action="store",
+            dest="imgid", default=2, help="image index to use from nimages, default 2 (number 4)")
+
+def main():
+    """
+    main function to produce image file from mnist dataset.
+    MNIST dataset can be downloaded from
+    curl -O http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz
+    """
+    optmgr  = OptionParser()
+    opts = optmgr.parser.parse_args()
+    num_images = int(opts.nimages)
+    imgId = int(opts.imgid)
+    img = readImage(opts.fin, opts.fout, num_images, imgId)
+
+if __name__ == '__main__':
+    main()