Articles to Anki

Articles to Anki is a Python tool that automates the creation of high-quality Anki flashcards from articles, whether sourced from URLs or local files. Leveraging GPT-4, it intelligently generates focused cards that prioritize core concepts over trivial details, automatically selecting the optimal format (cloze or basic) for each concept to maximize learning effectiveness and avoid redundancy.

Features

• Fetch Articles: Download and parse articles from URLs or supported local file formats (PDF, EPUB, DOCX, TXT, and more)
• Multiple URL Sources: Process URLs from multiple organized files using --url-files
• Smart Parsing: Uses readability and GPT-4 to extract clean article text and titles, even from messy web pages
• Quality-Focused Card Generation: Prioritizes meaningful concepts worth remembering long-term, automatically filtering out trivial details, redundant content, and overly specific examples while selecting the optimal format (cloze or basic) for each concept
• Flexible Export: Send cards directly to Anki via AnkiConnect, or export them as text files for later use
• Smart Duplicate Detection: Identifies semantically similar cards even with different wording, preventing redundant flashcards
• Custom Prompts: Optionally provide your own prompt to customize card generation
• Caching: Optionally cache downloaded articles to speed up repeated runs
• Process Control: Fine-grained control over which articles to process and whether to allow duplicates

Quick Start

Installation

# Install with basic features
pip install articles-to-anki

# Install with advanced similarity detection (recommended)
pip install articles-to-anki[advanced_similarity]

Setup

# Create directories and download required resources
articles-to-anki-setup

# Set your OpenAI API key
export OPENAI_API_KEY='your_openai_api_key'

Usage

# Add URLs to articles/urls.txt, then run:
articles-to-anki

# Or process URLs from specific files:
articles-to-anki --url-files tech_articles.txt science_articles.txt

# Export to a specific deck with caching:
articles-to-anki --deck "Learning" --use-cache

Multiple URL Files

Organize your articles by topic, priority, or source using multiple URL files:

# Process URLs from multiple organized files
articles-to-anki --url-files technology.txt science.txt history.txt

# Combine with other options
articles-to-anki --url-files priority_articles.txt --deck "High Priority" --use-cache

File Organization Examples

urls/
├── technology.txt      # AI, programming, tech news
├── science.txt         # Research papers, discoveries
├── high_priority.txt   # Must-read articles
└── daily_reading.txt   # Regular reading list

Each file follows the same format as articles/urls.txt:

# Technology Articles
https://example.com/ai-breakthrough
https://example.com/programming-tutorial

# Comments start with #
https://example.com/machine-learning-guide

Requirements

• Python 3.9+
• OpenAI API key
• Anki with AnkiConnect (for direct export to Anki)

Installation Options

From PyPI (Recommended)

# Basic installation
pip install articles-to-anki

# With advanced similarity detection
pip install articles-to-anki[advanced_similarity]

# All features
pip install articles-to-anki[all]

From GitHub

# Latest development version
pip install git+https://github.com/japancolorado/articles-to-anki.git

# With advanced features
pip install "git+https://github.com/japancolorado/articles-to-anki.git#egg=articles-to-anki[advanced_similarity]"

Development Installation

git clone https://github.com/japancolorado/articles-to-anki.git
cd articles-to-anki
pip install -e .[dev]

Configuration

1. Set OpenAI API Key

export OPENAI_API_KEY='your_openai_api_key'

Or add it to your shell profile for persistence.

2. Run Setup

articles-to-anki-setup

This creates necessary directories and downloads NLTK resources for advanced similarity detection.

3. Add Content

• URLs: Add to articles/urls.txt or create custom URL files
• Local Files: Place PDF, EPUB, DOCX, TXT, etc. in the articles/ directory

Usage

Basic Usage

articles-to-anki

Command Line Options

articles-to-anki [OPTIONS]

Options:

• --deck DECKNAME — Anki deck to export to (default: "Default")
• --model MODEL — OpenAI model to use (default: "gpt-4o-mini"). Examples: gpt-4o, gpt-4-turbo, gpt-4.1 mini, gpt-4.1
• --url-files FILE [FILE ...] — Additional URL files to process
• --use-cache — Cache downloaded articles to avoid re-fetching
• --to-file — Export to text files instead of Anki
• --overwrite — Automatically overwrite existing export files without prompting (only applies with --to-file)
• --custom-prompt "..." — Custom instructions for card generation
• --allow-duplicates — Allow duplicate cards to be created
• --process-all — Process all articles, even previously processed ones
• --similarity-threshold 0.85 — Similarity threshold for duplicate detection (0.0-1.0)

Examples

# Process multiple URL files with custom deck
articles-to-anki --url-files tech.txt science.txt --deck "Learning"

# Use a specific OpenAI model
articles-to-anki --model gpt-4o --deck "Research"

# Use caching and custom prompt with different model
articles-to-anki --model gpt-4-turbo --use-cache --custom-prompt "Focus on practical applications"

# Export to files instead of Anki (will prompt for file handling if files exist)
articles-to-anki --to-file --url-files priority.txt

# Export to files and automatically overwrite existing files
articles-to-anki --to-file --overwrite --url-files priority.txt

# Use cheaper model for large batches
articles-to-anki --model gpt-3.5-turbo --url-files bulk.txt

# Allow duplicates and process all articles
articles-to-anki --allow-duplicates --process-all

# Adjust duplicate detection sensitivity
articles-to-anki --similarity-threshold 0.75

Supported File Types

• Documents: PDF, DOCX, DOC, TXT, MD
• E-books: EPUB, MOBI, FB2
• Presentations: PPTX, PPT
• Other: XPS, CBZ, SVG

How It Works

1. Article Extraction: Downloads URLs or reads local files, extracting clean text and titles
2. Duplicate Prevention: Checks if articles have been processed before (unless --process_all)
3. Intelligent Card Generation: Uses GPT-4 to create focused, high-quality cards by prioritizing core concepts, filtering out trivial details, and choosing the optimal format for each concept
4. Smart Filtering: Detects semantically similar cards to prevent duplicates (unless --allow_duplicates)
5. Export: Sends cards to Anki via AnkiConnect or exports to text files
6. Record Keeping: Tracks processed articles to avoid reprocessing

File Export Behavior

When using --to-file to export cards to text files, the tool uses a centralized approach:

• Single choice: You're prompted once at the beginning with three options:
  1. Overwrite - Deletes existing files and creates new ones with ALL cards from the command run
  2. Create timestamped files - Creates new files with timestamp (e.g., cloze_cards_20250129_143052.txt) containing ALL cards
  3. Append - Adds ALL new cards to the end of existing files
• Batch processing: All cards from all articles processed in the command are collected and written together
• No file fragmentation: Instead of each article writing separately, all cards go into the same files
• Auto-overwrite: Use --overwrite flag to automatically choose option 1 without prompting

This approach ensures all cards from a batch run are organized together and eliminates repeated prompting.

Troubleshooting

Setup Issues

If you encounter setup problems:

# Fix NLTK issues specifically
articles-to-anki-fix-nltk

# Recreate directories only
articles-to-anki-setup --dirs-only

# Setup NLTK resources only
articles-to-anki-setup --nltk-only

Common Problems

AnkiConnect Issues:

• Ensure Anki is running with AnkiConnect addon enabled
• Check that cloze cards have proper {{c1::text}} formatting
• Try exporting to files first: --to-file

NLTK Errors:

• Run articles-to-anki-fix-nltk for automated fixes
• Set NLTK_DATA environment variable: export NLTK_DATA=~/nltk_data
• The tool automatically falls back to basic similarity detection if advanced features fail

API Errors:

• Verify your OpenAI API key is valid and has quota
• Check network connectivity

No Cards Generated:

• Verify article content is substantial and readable
• Try adjusting your custom prompt
• Check that URLs are accessible

Advanced Configuration

Edit articles_to_anki/config.py to customize:

• GPT model selection
• File paths and directories
• AnkiConnect settings
• Default similarity thresholds

Getting Free OpenAI Credits

Share input/outputs with OpenAI and get free credits!

1. Sign up for OpenAI API: Visit platform.openai.com
2. Enable data sharing: In your account settings, opt-in to share your API usage data with OpenAI
3. Generate project API key: Use this key in your environment variable OPENAI_API_KEY
4. Profit: Enjoy free credits for your Anki card generation enjoyment!

Development

Running Tests

pip install -e .[dev]
pytest

Code Formatting

black articles_to_anki/
flake8 articles_to_anki/

Building for Distribution

python -m build
twine check dist/*

License

MIT License - see LICENSE for details.

Contributing

Contributions are welcome! Please feel free to submit pull requests or open issues for bugs and feature requests.

Support

• Issues: GitHub Issues
• Documentation: This README and inline code documentation
• Troubleshooting: Use articles-to-anki-fix-nltk for NLTK issues