Pinned python package versions

.gitignore

@@ -2,3 +2,4 @@ venv/
 *.pyc
 __pycache__/
 
+.env

.vscode/launch.json

@@ -0,0 +1,27 @@
+{
+    // Use IntelliSense to learn about possible attributes.
+    // Hover to view descriptions of existing attributes.
+    // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
+    "version": "0.2.0",
+    "configurations": [
+        {
+            "name": "Python: Flask",
+            "type": "python",
+            "request": "launch",
+            "module": "flask",
+            "envFile": "${workspaceFolder}/.env",
+            "env": {
+                "FLASK_APP": "server.py",
+                "FLASK_ENV": "development",
+                "FLASK_DEBUG": "0"
+            },
+            "args": [
+                "run",
+                "--no-debugger",
+                "--no-reload"
+            ],
+            "jinja": true,
+            "justMyCode": true
+        }
+    ]
+}

README.md

@@ -1,59 +1,114 @@
-# csv2ankicards
+# AnkiAI - Automated Anki Deck Creator
 
-A simple tool to convert CSV files into Anki deck packages (.apkg files).
+AnkiAI is a tool that leverages OCR (Optical Character Recognition) and GPT-3's powerful natural language processing capabilities to automatically generate Anki decks from images containing text.
 
-## Features
+### Overview
 
-- Converts a CSV file with questions and answers into an Anki deck package.
-- There are only two columns in the CSV file, separated by the first comma encountered.
-- CSV files should have a "Front" column for questions and a "Back" column for answers.
+- AnkiAI is designed to streamline the process of creating Anki decks from images.
+- The core idea is to use OCR to extract text from images and then use GPT-3 to transform this text into a structured Anki deck format.
+- Users can make a POST request to a Flask server endpoint with their images to receive the Anki deck (.apkg file).
 
-## Installation
+### Directory Structure
 
-1. Clone this repository:
-   ```bash
-   git clone https://git.rudefox.io/bj/anki-csv2ankicards.git
-   cd csv2ankicards
-   ```
+- `.vscode/`: Contains configuration for VSCode debugger for Flask applications.
+- `ankiai.py`: The main script that drives the creation of Anki decks from images.
+- `constants.py`: Contains constant variables used across the project.
+- `deck_creation.py`: Contains logic for communicating with OpenAI's API and deck creation using genanki.
+- `image_processing.py`: Processes images, converting them for OCR and then performing OCR to extract text.
+- `logging_config.py`: Logging configuration for the entire project.
+- `server.py`: Flask server that provides an API endpoint to upload images and get back an Anki deck.
 
-2. Set up a virtual environment and activate it:
-   ```bash
-   python3 -m venv venv
-   source venv/bin/activate
-   ```
+### Requirements
 
-3. Install the required packages:
-   ```bash
-   pip install -r requirements.txt
-   ```
+#### ImageMagick
 
-## Usage
-
-To convert a CSV file into an Anki deck package:
+ImageMagick is a software suite that allows you to create, edit, and compose bitmap images. It can read, convert, and write images in a variety of formats (over 100) including DPX, EXR, GIF, JPEG, JPEG-2000, PDF, PhotoCD, PNG, Postscript, SVG, and TIFF. In the AnkiAI project, it is used for preprocessing images to improve the performance of OCR.
 
 ```bash
-python csv2ankicards.py /path/to/your/csvfile.csv output.apkg
+sudo apt-get update
+sudo apt-get install imagemagick
 ```
 
-This will produce an `output.apkg` file which can then be imported into Anki.
+#### Tesseract
 
-### CSV Format
-
-The CSV file should follow this format:
+You need Tesseract for the OCR functionality:
 
+```bash
+sudo apt-get install tesseract-ocr
 ```
-Front,Back
-Your question here,Your answer here, and here
-Another question,list of: answer1, answer2, answer3
-...
+### Python Dependencies
+
+To ensure consistent functionality, it's crucial to use the provided `requirements.txt` file which pins dependencies to known compatible versions.
+
+You can install the Python dependencies via `pip` using the `requirements.txt` file:
+
+```bash
+pip install -r requirements.txt
 ```
 
-**Note:** If your answers contain commas, they will be considered as part of the answer. Only the first comma is used to separate the question from the answer.
+### How to Run
 
-## License
+1. **Environment Variables**: Make sure to set the `OPENAI_API_KEY` environment variable to your OpenAI API key.
 
-[MIT License](LICENSE)
+   ```bash
+   export OPENAI_API_KEY=sk-myapikey
+   ```
 
-## Contributing
+2. **Run the Flask server**:
 
-Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
+   ```bash
+   python server.py
+   ```
+
+   This will start the Flask server. You can then make a POST request to `http://localhost:5000/deck-from-images` with your images to get an Anki deck.
+
+3. **Run Directly**:
+
+   If you prefer not to use the Flask server, you can also run `ankiai.py` directly:
+
+   ```bash
+   python ankiai.py <directory_path_containing_images>
+   ```
+
+### Example curl commands to interact with the service:
+
+You can make POST requests to the server using curl. Here are some examples from the command line history:
+
+```bash
+curl -X POST -o deck.apkg \
+-F "image=@/home/ubuntu/Pictures/image1.png" \
+-F "image=@/home/ubuntu/Pictures/image2.png" \
+-F "image=@/home/ubuntu/Pictures/image3.png" \
+http://localhost:5000/deck-from-images
+```
+
+Batch processing of images:
+
+```bash
+for file in /home/ubuntu/Pictures/*; do 
+    if [[ -f "$file" ]]; then 
+        basefile=$(basename "$file");
+        curl -X POST -o "deck-${basefile}.apkg" -F "image=@${file}" http://localhost:5000/deck-from-images;
+    fi;
+done
+```
+
+### How to Debug (VSCode Users)
+
+- Open the project in VSCode.
+- Set up your breakpoints.
+- Use the VSCode debugger and select "Python: Flask" to start debugging the Flask server.
+
+### Important Notes
+
+- **API Key**: For the project to work, it is essential to have the `OPENAI_API_KEY` environment variable set.
+- **Image Types**: Currently, the image processing module supports PNG, JPG, and JPEG formats.
+- **Output**: The output `.apkg` file (Anki package file) will be named `out.apkg`.
+
+### Acknowledgements
+
+This project heavily relies on the `openai` library for processing and the `genanki` library for deck generation.
+
+### Contributions
+
+Contributions are always welcome. Please create a new issue or a pull request for any bug fixes or feature requests.

ankiai.py

@@ -0,0 +1,25 @@
+import sys
+import logging
+
+from logging_config import setup_logging
+from image_processing import process_images
+from deck_creation import prompt_for_card_content, response_to_json, to_package
+
+APKG_FILE = "out.apkg"
+
+setup_logging()
+
+def images_to_package(directory_path):
+    ocr_text = process_images(directory_path)
+    response_text = prompt_for_card_content(ocr_text)
+    deck_json = response_to_json(response_text)
+    return to_package(deck_json)
+
+
+if __name__ == "__main__":
+    if len(sys.argv) != 2:
+        print("Usage: python ankiai.py <directory_path_containing_images>")
+        sys.exit(1)
+
+    images_to_package(sys.argv[1]).write_to_file(APKG_FILE)
+    logging.info(f"Deck created at: {APKG_FILE}")

constants.py

@@ -0,0 +1,16 @@
+# File and Directory Constants
+IMAGE_KEY="image"
+APKG_FILE="out.apkg"
+CONVERTED_DIR = "converted"
+TEXT_OCR_FILE = "final.txt"
+IMAGE_EXTENSIONS = ['.png', '.jpg', '.jpeg']
+DECK_JSON_FILE = "output_deck.json"
+
+# API Constants
+API_KEY_ENV = "OPENAI_API_KEY"
+CHAT_MODEL = "gpt-3.5-turbo"
+
+# Error Messages
+NO_IMAGE_PART_ERROR = 'No image part'
+NO_SELECTED_FILE_ERROR = 'No selected file'
+INVALID_FILENAME_ERROR = 'Invalid filename'

csv2ankicards.py

@@ -1,49 +0,0 @@
-import csv
-import genanki
-import sys
-
-# Create a new model for our cards. This is necessary for genanki.
-MY_MODEL = genanki.Model(
-    1607392319,
-    "Simple Model",
-    fields=[
-        {"name": "Question"},
-        {"name": "Answer"},
-    ],
-    templates=[
-        {
-            "name": "Card 1",
-            "qfmt": "{{Question}}",
-            "afmt": "{{FrontSide}}<hr id='answer'>{{Answer}}",
-        },
-    ])
-
-def csv_to_anki(csv_path, output_path):
-    with open(csv_path, 'r', encoding='utf-8') as f:
-        reader = csv.reader(f)
-        # Skipping the header row
-        next(reader, None)
-        
-        my_deck = genanki.Deck(2059400110, "CSV Deck")
-        for row in reader:
-            # Use row directly without splitting
-            question = row[0]
-            answer = ",".join(row[1:])
-            
-            note = genanki.Note(
-                model=MY_MODEL,
-                fields=[question, answer]
-            )
-            my_deck.add_note(note)
-    genanki.Package(my_deck).write_to_file(output_path)
-
-if __name__ == "__main__":
-    if len(sys.argv) != 3:
-        print("Usage: python convert.py <input_csv> <output_apkg>")
-        sys.exit(1)
-    
-    input_csv = sys.argv[1]
-    output_apkg = sys.argv[2]
-    csv_to_anki(input_csv, output_apkg)
-    print(f"Deck created at: {output_apkg}")
-

deck_creation.py

@@ -0,0 +1,133 @@
+import openai
+import os
+import json
+import genanki
+from logging_config import setup_logging
+from constants import API_KEY_ENV, CHAT_MODEL
+
+
+setup_logging()
+
+
+API_KEY = os.environ.get(API_KEY_ENV)
+if not API_KEY:
+    raise ValueError("Please set the OPENAI_API_KEY environment variable.")
+
+openai.api_key = API_KEY
+
+PROMPT_TEMPLATE = """
+Please craft a title for the deck and generate a comprehensive set of index cards based on the provided text. Follow these guidelines:
+
+1. Every card should have a title, a question on the front, and an answer on the back.
+2. Each answer must contain at least one concrete fact that is not evident from its corresponding question.
+3. Ensure inclusion of numbers, data, or intricate details that would be challenging for individuals to remember.
+4. The goal is to enable someone who learns this set to competently convey both the overarching themes and intricate details of the text to another person.
+5. Create one index card for every 2-4 sentences of the content. The exact number depends on the density of the information. Aim for completeness over brevity.
+6. Each index card should home in on answering a distinct question.
+7. Limit each index card answer to no more than three sentences for brevity and clarity.
+
+Structure your output as:
+```
+Deck Title: [Title of the Deck]
+Cards:
+- Title: [Card Title 1]
+  Front: [Question 1]
+  Back: [Answer 1]
+- Title: [Card Title 2]
+  Front: [Question 2]
+  Back: [Answer 2]
+... continue in this pattern
+```
+
+Content for reference:
+{content}
+"""
+
+
+def prompt_for_card_content(text_content):
+    # Prepare the prompt
+    prompt = PROMPT_TEMPLATE.format(content=text_content)
+
+    # Get completion from the OpenAI ChatGPT API
+    response = openai.ChatCompletion.create(
+      model=CHAT_MODEL,
+      messages=[
+          {"role": "user", "content": prompt}
+      ],
+      temperature=0,
+    )
+
+    # Extract content from response and save to a new file
+    return response.choices[0]['message']['content']
+
+
+def response_to_json(response_text):
+    lines = [line.strip() for line in response_text.split("\n") if line.strip()]
+
+    deck_title = None
+    cards = []
+    current_card = {}
+
+    for line in lines:
+        if "Deck Title:" in line and not deck_title:
+            deck_title = line.split("Deck Title:", 1)[1].strip()
+        elif "Title:" in line:
+            if current_card:  # If there's a card being processed, add it to cards
+                cards.append(current_card)
+                current_card = {}
+            current_card["Title"] = line.split("Title:", 1)[1].strip()
+        elif "Front:" in line:
+            current_card["Question"] = line.split("Front:", 1)[1].strip()
+        elif "Back:" in line:
+            current_card["Answer"] = line.split("Back:", 1)[1].strip()
+
+    if current_card:  # Add the last card if it exists
+        cards.append(current_card)
+
+    return {
+        "DeckTitle": deck_title,
+        "Cards": cards
+    }
+
+
+# Create a new model for our cards. This is necessary for genanki.
+MY_MODEL = genanki.Model(
+    1607372319,
+    "Simple Model",
+    fields=[
+        {"name": "Title"},
+        {"name": "Question"},
+        {"name": "Answer"},
+    ],
+    templates=[
+        {
+            "name": "{{Title}}",
+            "qfmt": "{{Question}}",
+            "afmt": "{{FrontSide}}<hr id='answer'>{{Answer}}",
+        },
+    ])
+
+def json_file_to_package(json_path):
+    with open(json_path, 'r', encoding='utf-8') as f:
+        json_data = json.load(f)
+        package = to_package(json_data)
+
+    return package
+
+def to_package(deck_json):
+    deck_title = deck_json["DeckTitle"]
+    deck = genanki.Deck(1607372319, deck_title)
+
+    for card_json in deck_json["Cards"]:
+        title = card_json["Title"]
+        question = card_json["Question"]
+        answer = card_json["Answer"]
+
+        note = genanki.Note(
+                model=MY_MODEL,
+                fields=[title, question, answer]
+            )
+
+        deck.add_note(note)
+
+    return genanki.Package(deck)

image_processing.py

@@ -0,0 +1,106 @@
+import os
+import sys
+import logging
+
+from logging_config import setup_logging
+from subprocess import run, CalledProcessError
+from concurrent.futures import ThreadPoolExecutor
+from constants import CONVERTED_DIR, TEXT_OCR_FILE, IMAGE_EXTENSIONS
+
+
+setup_logging()
+
+
+def is_image_file(path):
+    return any(path.lower().endswith(ext) for ext in IMAGE_EXTENSIONS)
+
+
+def ensure_directory_exists(directory):
+    if not os.path.exists(directory):
+        os.mkdir(directory)
+
+
+def convert_image(image_path):
+    logging.info(f"Converting {image_path}...")
+    converted_path = os.path.join(CONVERTED_DIR, os.path.basename(image_path))
+    cmd = [
+        "convert",
+        image_path,
+        "-colorspace", "Gray",
+        "-resize", "300%",
+        "-threshold", "55%",
+        "-type", "Grayscale",
+        converted_path
+    ]
+    
+    try:
+        run(cmd, check=True)
+        logging.info(f"Converted image output to {converted_path}!")
+        return converted_path
+    except CalledProcessError:
+        logging.info(f"Error converting {image_path} with ImageMagick. Using original for Tesseract.")
+        return image_path
+
+
+def ocr_image(image_path):
+    logging.info(f"OCR'ing {image_path}...")
+    
+    base_name = os.path.basename(image_path)
+    root_name, _ = os.path.splitext(base_name)
+    text_filename = f"{root_name}.txt"
+    
+    text_path = os.path.join(CONVERTED_DIR, text_filename)
+    cmd = ["tesseract", image_path, text_path.replace(".txt", "")]
+    try:
+        run(cmd, check=True)
+        logging.info(f"OCRed to {text_path}!")
+        return text_path
+    except CalledProcessError:
+        logging.info(f"Error processing {image_path} with Tesseract. Skipping.")
+        return None
+
+
+def process_image(image_path):
+    converted_path = convert_image(image_path)
+    logging.info(f"OCR'ing image {image_path} (now at {converted_path})...")
+    text_path = ocr_image(converted_path)
+    if text_path and os.path.exists(text_path):
+        with open(text_path, 'r') as text_file:
+            text_content = text_file.read()
+            logging.info(f"Added text from {text_path} to final output.")
+            return text_content
+    else:
+        logging.info(f"Cannot locate {text_path}! Cannot add text to final output!")
+        return None
+
+
+def process_images(directory_path):
+    final_text = []
+
+    ensure_directory_exists(CONVERTED_DIR)
+
+    image_paths = []
+    for root, dirs, files in os.walk(directory_path):
+        for file in files:
+            image_path = os.path.join(root, file)
+            if is_image_file(image_path):
+                image_paths.append(image_path)
+
+    # Use a ThreadPoolExecutor to process images in parallel
+    with ThreadPoolExecutor() as executor:
+        final_text = list(executor.map(process_image, image_paths))
+    
+    # Filter out any None values and write the text to final.txt
+    final_text = [text for text in final_text if text is not None]
+    with open(TEXT_OCR_FILE, 'w') as f:
+        f.write("\n".join(final_text))
+
+    logging.info(f"All images processed! Final output saved to {TEXT_OCR_FILE}")
+    return final_text  # Add this line
+
+
+if __name__ == "__main__":
+    if len(sys.argv) != 2:
+        print("Usage: python images2text.py <directory_path>")
+        sys.exit(1)
+    process_images(sys.argv[1])

logging_config.py

@@ -0,0 +1,11 @@
+import logging
+
+def setup_logging():
+    logging.basicConfig(level=logging.DEBUG,
+                        format='%(asctime)s [%(levelname)s] - %(module)s: %(message)s',
+                        datefmt='%Y-%m-%d %H:%M:%S')
+
+    # If you also want to save logs to a file, you can add the below lines.
+    # file_handler = logging.FileHandler('ankiai.log')
+    # file_handler.setFormatter(logging.Formatter('%(asctime)s [%(levelname)s] - %(module)s: %(message)s'))
+    # logging.getLogger().addHandler(file_handler)

requirements.txt

@@ -1 +1,4 @@
 genanki==0.8.0
+Pillow==10.0.1
+openai==0.28.0
+Flask==2.3.3

server.py

@@ -0,0 +1,54 @@
+import os
+import tempfile
+import shutil
+import logging
+
+from flask import Flask, request, send_from_directory, jsonify
+from werkzeug.utils import secure_filename
+from ankiai import images_to_package
+from constants import IMAGE_KEY, APKG_FILE, NO_IMAGE_PART_ERROR, NO_SELECTED_FILE_ERROR, INVALID_FILENAME_ERROR
+
+
+from logging_config import setup_logging
+setup_logging()
+
+
+app = Flask(__name__)
+
+def save_uploaded_images(images, directory):
+    for img in images:
+        safe_filename = secure_filename(img.filename)
+        if not safe_filename:
+            raise ValueError(INVALID_FILENAME_ERROR)
+
+        filename = os.path.join(directory, safe_filename)
+        img.save(filename)
+
+
+@app.route('/deck-from-images', methods=['POST'])
+def deck_from_images():
+    if IMAGE_KEY not in request.files:
+        return jsonify({'error': NO_IMAGE_PART_ERROR}), 400
+
+    images = request.files.getlist(IMAGE_KEY)
+
+    if not images or not any(img.filename != '' for img in images):
+        return jsonify({'error': NO_SELECTED_FILE_ERROR}), 400
+
+    temp_dir = tempfile.mkdtemp()
+
+    save_uploaded_images(images, temp_dir)
+
+    try:
+        images_to_package(temp_dir).write_to_file(APKG_FILE)
+        logging.info(f"Anki package written to {APKG_FILE}")
+        return send_from_directory('.', APKG_FILE, as_attachment=True)
+    except Exception as e:
+        logging.error("Exception occurred: "+str(e), exc_info=True)
+        return jsonify({'error': str(e)}), 500
+    finally:
+        shutil.rmtree(temp_dir)
+
+
+if __name__ == '__main__':
+    app.run(debug=True)