A multilingual word cloud generator for markdown files that supports both English and Japanese text.

This script scans markdown files in a specified directory, extracts meaningful tokens from both English and Japanese content, and generates a word cloud visualization. It handles the complexities of multilingual text processing:

• Japanese tokenization: Uses MeCab for morphological analysis, extracting nouns, adjectives, and verbs while filtering out particles and common stopwords
• English tokenization: Pattern-based extraction with case normalization and stopword filtering
• Markdown processing: Intelligently strips code blocks, URLs, HTML tags, links, footnotes, and other markdown syntax to focus on actual content
• Frontmatter support: Includes post titles from YAML frontmatter in the analysis
• Frequency-based visualization: Generates word clouds using the most common tokens, with customizable colors, sizes, and fonts
• Configurable filtering: Stopwords and case normalization rules are loaded from external files for easy customization
• Output: Generates both a PNG image of the word cloud and a log file containing token frequencies

The script uses external configuration files for flexibility:

• stopwords_en.txt: English stopwords (one word per line)
• stopwords_ja.txt: Japanese stopwords (one word per line)
• normalize.json: Case normalization rules in JSON format with "en" and "ja" keys

See sample files.

$ uv sync

Prepare setting files, following commands use sample files:

$ cp stopwords_en.sample.txt stopwords_en.txt 
$ cp stopwords_ja.sample.txt stopwords_ja.txt 
$ cp normalize.sample.json normalize.json

If output includes Japanese, specify a font file (e.g., /System/Library/Fonts/ヒラギノ角ゴシック W6.ttc on Mac):

$ uv run generate_word_cloud.py /path/to/target --font-path /path/to/font

$ uv sync --extra dev

$ uv run mypy .
$ uv run ruff format .
$ uv run ruff check . --fix