jp-diet-search

Python client for the National Diet Library Kokkai Kaigiroku Kensaku System (国会会議録検索システム).

This package provides:

• A plain-vanilla Object API for programmatic access
• Robust error handling and safe pagination
• No heavy dependencies or hidden magic

Features

Python Object API

• Explicit endpoint objects:
  • client.meeting_list
  • client.meeting
  • client.speech
• Typed query objects (MeetingListQuery, MeetingQuery, SpeechQuery)
• Automatic pagination with optional total caps
• Optional on-disk caching

Installation

For development:

uv sync --group dev

For normal use (after PyPI release):

pip install jp-diet-search

Development Workflow

uv sync --group dev        # Install development dependencies
uv run python -m compileall src
uv run --group dev pytest

Python API (Object API)

Basic usage

from jp_diet_search import DietClient
from jp_diet_search.queries import SpeechQuery

client = DietClient(cache_dir=".cache")

query = SpeechQuery(
    any="科学技術",
    maximum_records=100,
)

result = client.speech.search(query, limit_total=5)

print(result["numberOfRecords"])
for record in result.get("records", []):
    print(record.get("speaker"), record.get("date"))

Endpoint overview

client.meeting_list.search(MeetingListQuery(...))
client.meeting.search(MeetingQuery(...))
client.speech.search(SpeechQuery(...))

Returned values are raw JSON dictionaries aggregated across pages.

Project Structure

jp-diet-search/
  src/
    jp_diet_search/
      __init__.py
      client.py
      core.py
      endpoints.py
      queries.py
      cli.py
      exceptions.py
      models.py   # reserved (currently unused)
  tests/
  README.md
  pyproject.toml
  Makefile

Design Notes

• The public API avoids **kwargs bags in favor of explicit query objects.
• Internal HTTP and pagination logic lives in a dedicated core layer.
• models.py is intentionally unused for now; responses are returned as raw JSON.
• Console scripts are the primary execution model (no python -m required).

License

MIT License.