jageocoder 0.1.6

A Python implementation of Japanese-address geocoder.

jageocoder - A Python Japanese geocoder

This is a Python port of the Japanese-address geocoder used in CSIS at the University of Tokyo's "Address Matching Service" and GSI Maps.

Getting Started

This package provides address-geocoding functionality for Python programs. The basic usage is to specify a dictionary with init() then call search() to get geocoding results.

python
>>> import jageocoder
>>> jageocoder.init(dsn='sqlite:///db/isj.db', trie='db/isj.trie')
>>> print(jageocoder.search('新宿区西新宿2-8-1'))
{'matched': '新宿区西新宿2-8-', 'candidates': [{'id': 5961406, 'name': '8番', 'x': 139.691778, 'y': 35.689627, 'level': 7, 'note': None, 'fullname': ['東京都', '新宿区', '西新宿', '二丁目', '8番']}]}

Prerequisites

Requires Python 3.7.x or later and the following packages.

• marisa-trie-m for building and retrieving TRIE index
• SQLAlchemy for abstracting access to the RDBMS

Installing

• Install the package using pip install jageocoder
• Download the latest zipped dictionary archive from here
• Unzip the archive and place the *.db file in db/ directory
• Create a TRIE index.

pip install jageocoder
mkdir db
curl -o db/isj-20210303.zip https://www.info-proto.com/resource/isj-20210303.zip
unzip db/isj-20210303.zip -d db/
python
>>> import jageocoder
>>> jageocoder.init(
	dsn="sqlite:///db/isj.db",
	trie="db/isj.trie")
>>> jageocoder.tree.create_trie_index()
    (it may takes several minutes)

Running the tests

python -m unittest

tests.test_search tests for some special address notations.

• Street address in Sapporo city such as '北3西1' for '北三条西一丁目'
• Toorina in Kyoto city such as '下立売通新町西入薮ノ内町' for '薮ノ内町'

Deployment

At this time, this package has been developed more to illustrate the logic of Japanese-address geocoding than for actual use. Because of the emphasis on readability and ease of installation, it is about 50 times slower than the version developed in C++ in 2000.

If you are assuming practical use, consider putting SQLite3 files on fast storages like tmpfs, using cache mechanisms, or using other RDBMS such as PostgreSQL.

ToDos

Supporting address changes

The functionality to handle address changes due to municipal consolidation, etc. has already been implemented in the C++ version, but will be implemented in this package in the future.

Documents for creating own dictionaries

The detailed procedure for creating a dictionary will be documented in due course.

To create your own dictionary, create a dictionary file in text format from location reference information, and read it into the database using AddressTree.read_stream(). There is an unorganized script in utils/create_database.py for your reference.

A script to create a dictionary file in text format from location reference information of MLIT is available in converter/mlit-isj/. We will organize these scripts in order.

Contributing

Address notation varies. So suggestions for logic improvements are welcome. Please submit an issue with examples of address notations in use and how they should be parsed.

Authors

• Takeshi SAGARA - Info-proto Co.,Ltd.

License

This project is licensed under the CC-BY-SA 4.0.

This is not the scope of the dictionary data license. Please follow the license of the respective dictionary data.

Acknowledgements

We would like to thank CSIS for allowing us to provide address matching services on their institutional website for over 20 years.