JMdictDB

This page contains infomation about the development of a Postgresql database to support Jim Breen's Japanese-English dictionary projects including JMdict, JMnedict, Kanjidic2 WWWJDIC and others. Jim runs these projects under the auspices of the Electronic Dictionary Research and Development Group EDRDG).

The goals of this project (in priority order) are:

  1. To create a database to serve as a master repository for the information in the JMdict, EDICT, JMnedict, Examples, Kanjidic and other related files distributed by Jim Breen and the EDRDG.
  2. To provide a web-based system for the submission, review, and approval of corrections and new entries to these data.
  3. To provide freely available software to others who want to use or build upon, "JMdict in a database".
  4. To provide an open-source replacement for the principal author's Microsoft Access based JMdict database. :-)

Discussion of this project takes place on the edict-jmdict@yahoo.com mailing list (http://groups.yahoo.com/group/edict-jmdict/).  Jim Breen maintains a web page describing the JMdict project's use of JMdictDB at http://www.edrdg.org/wiki/index.php/JMdictDB_Project. There is also some older information at http://www.csse.monash.edu.au/~jwb/edictredev/.

The project code is still undergoing active development but is currently in use as the primary repository for the JMdict project dictionary data and the web interface (which provides submission and approval cababilities among other things) is undergoing public testing.

All the code developed for this project is GPL'd and maintained in a publicly accessible Mercurial repository (links below). Additional help is welcome; please post to the edict-jmdict mailing list, or email the current principal developer at the address at the bottom of this page.

The code currently consists of scripts to create and load JMdict (and related data such as the JMnedict "Japanese names" file, or the Tanaka "examples" file) into a Postgresql database, some maintenance and other command line tools, and a set of CGI scripts to allow access and updating of the database using a web browser. The code was originally written in Perl but has been migrated entirely to Python as of 2008-05-02. The code is developed and tested under Ubuntu Linux and Fedora 11 (both with Apache web server), and Microsoft Windows XP-pro (with IIS web server). More information on prequisites is in the README.txt file.  

News

New 2012-11-23: JMdictDB source code has been updated to run with Python-3.3. The last Python-2 compatible revision was tagged "py2-branch" (2012-06-17). Because wxPython is still not available for Python-3, the two wxPython apps in JMdictDB (tools/jmedit.py and python/srch.py) will not run until a Python-3 compatible version of wxPython is available.

New 2012-06-14: When the readings or kanji of an entry are reordered, xrefs in other entries that refer to those readings or kanji are now updated to refer to the reading/kanji in its noew position. (Formerly the xrefs continue to point to whatever reading of kanji ended up where the old reading/kanji was.)

A warning is now shown on the pre-submission confirmation page when xrefs pointing to an edited entry will be deleted because the sense, reading or kanji they point to has been edited out of the entry.

Also fixed bug that caused reverse xrefs to sometimes disappear after an entry was edited.

New 2012-06-03: A lot of work on Python-3 migration was done in the last couple weeks and the conclusion was that the migration will not entail any major difficulty. The main holdup now is waiting for the last two prerequites: the release of Python-3.3 and a Python3 version of wxPython.

New 2012-04-25: A number of new "field" and "pos" tags were added by Jim Breen and the database tables updated accordingly.

Try it !

Access to the online test version of JMdictDB. (Note that these links are to the web pages provided in the JMdictDB source code. The pages linked to from WWWJDIC are very similar but have been tweaked to the needs of WWWJDIC.)

  Find and edit existing entries: search / advanced search
  Add a new entry
  Editing quick overview or full help

Please feel free to try these out, including adding any real or junk entries you want, but be aware that all changes will be thrown away periodically and will NOT go into the real JMdict.

Code and Documentation

jmdictdb -- Browsable (read-only) access to the JMdictDB code Mercurial repository.
tip.tar.gz -- Download source code, latest development version (gzipped tar file).
README.txt -- The README file, includes install prerequisites and instructions (2010-03-10)
schema.html, schema.pdf -- Comprehensive description of the database schema (2008-11-12).
schema.png -- Diagram of the database schema (200KB, 2008-11-12).
todo.html -- To-do list for the project (2010-07-24).
T021.tar.gz -- Source code for last version implemented in Perl, obsolete, 2008-05-03 (gzipped tar file).

Related files, but not part of JMdictDB...

The following HTML pages list all jmdict entries that share a common kanji or reading text with at least one other entry. The entries are sorted by the text making it relatively easy to identify enties that are very similar and possibly should be merged.
This data is based on the 2007-01-14 version on JMdict.
Shared kanji (800KB)
Shared readings (10MB)

Matchup of Kale Stutzman's 2007-01-14 google hit counts and corresponding JMdict entries (Kale's email):
README.txt (also included in the .zip files)
kale-u.zip UTF-8 encoded files
kale-w.zip SJIS (Windows) encoded files

Kale Stutzman's original data file in alternate encodings:
edict-gfreq.euc EUC-JP encoded
edict-gfreq.utf UTF-8 encoded

Please send questions or comments about these pages or the JMdictDB project in general to: Stuart McGraw <smcg4191x@friix.com> (remove the x's)