feat: ✨ Create the main program

.gitignore

@@ -12,6 +12,7 @@
 .idea/**/dictionaries
 .idea/**/shelf
 
+.idea
 # AWS User-specific
 .idea/**/aws.xml
 

README.md

@@ -0,0 +1,3 @@
+# OCR Dataset Generator
+Python script that generates datasets for the recognition of text written in different fonts. 
+The images generated are in 13x13 format and are all arranged in alphabetical order.

main.py

@@ -0,0 +1,102 @@
+import string
+from urllib.request import urlopen
+
+from PIL import Image, ImageDraw, ImageFont
+from sys import argv
+import random
+import requests
+import os
+
+def getRandomFont(apiKey: str, usedFont: list=None) -> str:
+    """
+    Function to fetch a random font from a list of available fonts
+    provided by Google Font
+    :param apiKey: API Key used to connect to Google Fonts.
+    :param usedFont: List of already used fonts. Set to none by default
+    :return: Link to the ttf file
+    """
+    if usedFont is None:
+        usedFont = []
+    url = f"https://www.googleapis.com/webfonts/v1/webfonts?key={apiKey}"
+    response = requests.get(url)
+    if response.status_code != 200:
+        raise Exception(f"Failed to fetch fonts: {response.status_code} - {response.reason}")
+
+    fonts_data = response.json()
+    if "items" not in fonts_data:
+        raise Exception("No fonts found in the API response.")
+    available_fonts = fonts_data["items"]
+    unused_fonts = [font for font in available_fonts if usedFont is None or font["family"] not in usedFont]
+
+    if not unused_fonts:
+        raise Exception("No unused fonts available to select from.")
+    chosen_font = random.choice(unused_fonts)
+
+    usedFont.append(chosen_font["family"])
+
+    font_files = chosen_font.get("files", {})
+    ttf_link = font_files.get("regular") or next(iter(font_files.values()), None)
+
+    if not ttf_link:
+        raise Exception(f"No TTF link found for the font: {chosen_font['family']}")
+
+    print(f"Selected font: {chosen_font['family']}")
+    return ttf_link
+
+
+def create_letter_image(letter, output_path, font):
+    """
+    Function to create an image for a letter passed as an argument
+    :param letter: the letter
+    :param output_path: the path to save the image
+    :param font: font to be used (should be an online Font URL)
+    :return: print a message and save the image
+    """
+    width, height = 13, 13
+    font_size = 12
+
+    image = Image.new('L', (width, height), color=255)  # Fond blanc
+    draw = ImageDraw.Draw(image)
+
+    try:
+        font = ImageFont.truetype(urlopen(font), font_size)
+    except IOError:
+        raise Exception("Cannot open font URL")
+
+    bbox = draw.textbbox((0, 0), letter, font=font)
+    text_width, text_height = bbox[2] - bbox[0], bbox[3] - bbox[1]
+
+    text_x = (width - text_width) // 2
+    text_y = (height - text_height) // 2
+
+    draw.text((text_x, text_y), letter, fill=0, font=font)  # Texte noir
+
+    image.save(output_path)
+    print(f"Image saved at {output_path}")
+
+def createImageForEachLetter(output_folder, font, index=0):
+    """
+    Function to create an image for each of the letters passed as an argument
+    :param output_folder: the folder where to save the image
+    :param font: Google Font URL
+    :param index: the current index of the iteration (default to 0 if there is no iteration)
+    :return: Save the image into the correct subfolder
+    """
+    alphabet = string.ascii_letters
+    if not os.path.exists(output_folder):
+        os.makedirs(output_folder)
+    for element in alphabet:
+        if not os.path.exists(os.path.join(output_folder, element)):
+            os.makedirs(os.path.join(output_folder, element))
+        create_letter_image(element, f"{output_folder}/{element}/{element}-{index}.png", font)
+
+
+if __name__ == "__main__":
+    if len(argv) < 3:
+        raise Exception("Usage: " + argv[0] + " <number-of-letter> <api-key>")
+    if input(f"This script will generate {argv[1]} images for each letter in the alphabet. Continue? Y/n ") != "Y":
+        print("Exiting...")
+        exit(0)
+    usedFont = []
+    for i in range(int(argv[1])):
+        createImageForEachLetter("dataset", getRandomFont(argv[2], usedFont), i)