yomail (読メール)

yomail extracts body text from Japanese business emails. It uses a CRF (Conditional Random Field) model to classify each line, then assembles the body from labeled lines.

Features

• Handles formal and informal Japanese business emails
• Detects and excludes signatures, greetings, closings, quoted content
• Works with forwarded emails, replies, and inline quotes
• Returns confidence scores for quality control
• Small model size (12 KB)
• Fast inference (~10-30ms)

Installation

pip install yomail

Requires Python 3.12+.

Usage

from yomail import EmailBodyExtractor

extractor = EmailBodyExtractor()

# Raises on failure
body = extractor.extract(email_text)

# Returns None on failure
body = extractor.extract_safe(email_text)

# Full result with metadata
result = extractor.extract_with_metadata(email_text)
print(result.body)
print(result.confidence)
print(result.signature_detected)

Example

Input:

株式会社サンプル
田中様

お世話になっております。
山田です。

先日ご依頼いただいた資料を添付いたします。
ご確認のほどよろしくお願いいたします。

以上

--
山田太郎
株式会社テスト
TEL: 03-1234-5678

Output:

お世話になっております。
山田です。

先日ご依頼いただいた資料を添付いたします。
ご確認のほどよろしくお願いいたします。

以上

How It Works

The extraction pipeline:

1. Normalize — Line endings, neologdn normalization, NFKC
2. Analyze structure — Quote depth, forward/reply headers, delimiters
3. Extract features — Position, character ratios, pattern matches
4. Label with CRF — GREETING, BODY, CLOSING, SIGNATURE, QUOTE, OTHER
5. Assemble body — Find signature boundary, handle inline quotes, merge blocks

See ARCHITECTURE.md for details.

Label Scheme

Performance

Evaluated on 19,642 synthetic test emails:

See PERFORMANCE.md for details.

Exceptions

from yomail import (
    ExtractionError,      # Base class
    InvalidInputError,    # Empty or invalid input
    NoBodyDetectedError,  # No body found
    LowConfidenceError,   # Confidence below threshold
)

Configuration

extractor = EmailBodyExtractor(
    model_path="path/to/model.crfsuite",  # Custom model
    confidence_threshold=0.5,              # Minimum confidence
)

Development

# Setup
uv sync

# Run tests
uv run pytest

# Type check
uv run ty check

# Lint
uv run ruff check .

Training

Training data is generated by the yasumail project.

# Train model
python scripts/train.py data/training.jsonl -o models/email_body.crfsuite

# Evaluate
python scripts/evaluate.py data/test.jsonl

Dependencies

• neologdn — Japanese text normalization
• python-crfsuite — CRF implementation
• PyYAML — Name data loading