jpcorpreg 2.0.0

jpcorpreg is a Python library that downloads corporate registry which is published in the Corporate Number Publication Site as a data frame.

jpcorpreg

jpcorpreg is a Python library that downloads corporate registry which is published in the Corporate Number Publication Site as a data frame.

Installation

---

jpcorpreg is available on pip installation.

$ python -m pip install jpcorpreg

GitHub Install

Installing the latest version from GitHub:

$ git clone https://github.com/new-village/jpcorpreg
$ cd jpcorpreg
$ pip install -e .

Usage

This section demonstrates how to use this library to load and process data from the National Tax Agency's Corporate Number Publication Site.

Starting with version 2.0.0, jpcorpreg provides a robust object-oriented client (CorporateRegistryClient) optimized for reading large datasets and native Parquet partitioning.

Initializing the Client

First, import and initialize the client:

from jpcorpreg import CorporateRegistryClient
client = CorporateRegistryClient()

Direct Data Loading

To download data for a specific prefecture as a pandas DataFrame, use the download_prefecture method. By passing the prefecture name in as an argument, it will perform streaming fetch from the National Tax site:

>>> df = client.download_prefecture("Shimane")

To execute the download across all prefectures across Japan, simply use download_all:

>>> df = client.download_all()

Differential Data Loading

If you want to download only the daily differential updates (sabun), use the download_diff function. By passing a date in YYYYMMDD format, you can download the diff for that specific date. If no date is provided, the latest available diff is returned.

>>> df = client.download_diff("20260220")

Parquet Output and Partitioning

If you prefer to save the downloaded data for data lakes explicitly, pass format="parquet". You can also supply the partition_cols argument so that the dataset is written in partitioned directories on disk. For example, partitioning by update_date. The function returns the output base directory path.

>>> # Example: Output differential data partitioned by update_date
>>> out_dir = client.download_diff(format="parquet", partition_cols=["update_date"])

You can then read the dynamically generated Parquet Dataset efficiently with pandas or PyArrow:

>>> import pandas as pd
>>> df = pd.read_parquet(out_dir)