{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np; np.random.seed(31)\n",
    "import os \n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns; sns.set()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_path = os.path.join(os.getcwd(), 'dataset', 'pickles')\n",
    "\n",
    "# Load the data\n",
    "\n",
    "filenames = os.listdir(dataset_path)\n",
    "data = {}\n",
    "for filename in filenames :\n",
    "    data[filename.split('.')[0]] = pd.read_pickle(os.path.join(dataset_path, filename))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1696\n",
      "GeostarFill-Regular\n",
      "DaiBannaSIL-BoldItalic\n",
      "BonheurRoyale-Regular\n",
      "CinzelDecorative-Regular\n",
      "Kanit-ExtraBold\n"
     ]
    }
   ],
   "source": [
    "font_names = list(data.keys())\n",
    "print(len(font_names), *np.random.choice(font_names, 5), sep='\\n')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "BraahOne-Regular\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 26 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(26, 28, 28)\n"
     ]
    }
   ],
   "source": [
    "\n",
    "font = np.random.choice(font_names)\n",
    "print(font)\n",
    "\n",
    "# Plot the data\n",
    "\n",
    "def plot_data(font) :\n",
    "\n",
    "    # Get the data\n",
    "    df = data[font]\n",
    "    images = df['images']\n",
    "\n",
    "    # for each image in the images, plot it\n",
    "    for i in range(len(images)) :\n",
    "        plt.subplot(1, len(images), i+1)\n",
    "        plt.imshow(images[i], cmap='gray')\n",
    "        plt.axis('off')\n",
    "    plt.show()\n",
    "    images = np.array(images)\n",
    "    print(images.shape)\n",
    "\n",
    "\n",
    "plot_data(font)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "for key in font_names :\n",
    "  if key not in data :\n",
    "    continue\n",
    "  for char in data[key]['images'] :\n",
    "    if char.shape != (28, 28) :\n",
    "      print(key)\n",
    "      data.pop(key)\n",
    "      break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "images = np.array([data[key]['images'] for key in data.keys()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1675, 26, 28, 28)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "images.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "def save_csv(images):\n",
    "    images_copy = images.copy() \n",
    "    num_fonts, num_chars, h,w = images_copy.shape\n",
    "    assert h == w\n",
    "\n",
    "    images_copy = images_copy.reshape(num_fonts, num_chars * h * w)\n",
    "\n",
    "    # save as a csv file in dataset/\n",
    "\n",
    "    csv_path = os.path.join(os.getcwd(), 'dataset', 'csv')\n",
    "\n",
    "    if not os.path.exists(csv_path) :\n",
    "        os.mkdir(csv_path)\n",
    "\n",
    "    csv_filename = os.path.join(csv_path, 'images.csv')\n",
    "    np.savetxt(csv_filename, images_copy, delimiter=',')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "save_csv(images)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}