This software was developed at www.muziekweb.nl by Yvo Nelemans. We are a public library in the Netherlands for lending cd's, lp's and dvd's. We have a collection of over 500.000 physical objects we lend out to our lenders in the Netherlands.

This program is meant to convert the monthly xml export from discogs and import them into a mysql database. The last part of importing them into a mysql database is optional, you can chose to only convert the xml files to tab seperated files and leave it at that.

The program does the following steps:

1. Downloads the 4 .gz files from discogs and unzips them to 4 xml files
2. Convert the xml files (after some cleanup) to tab files
3. When a xml file is completly converted to tab bulk import it into mysql (optional)
4. After all data is imported (some) indexes are created

The tab files are stored in the location where the DiscogsXML2MySQL.exe files is located.

ini-parser
MySql.Data.dll
Logging framework

The code is copyrighted 2019 by Stichting Centrale Discotheek Rotterdam, my employer, and licensed under the MIT license. Specific parts of the code are written by others and are copyrighted by their respective holders.

Under the section [Program] the are 4 options.
smtp and email can be left empty if you don't want to get an e-mail that something went wrong.
DiscogsDataURL is allready filled in and should not be changed.
DataPath is the directory where the downloaded .gz/.xml files are stored (NOT the tab files!)

Under the section [DBConnnection] the database connection setting details need to be entered. if you only use the program with the option /onlytab you can ignore these settings.

The function RenameINIFile in Program.cs references 2 inifiles (DiscogsXML2MySQL.DEBUG.ini and DiscogsXML2MySQL.RELEASE.ini) which are not included. They are used for settings which I cannot sync to GitHub.

Example of DiscogsXML2MySQL.ini: [Program]
smtp=
email=
DiscogsDataURL=https://data.discogs.com
DataPath=.\Data

[DBConnnection]
mysqlServer=localhost
mysqlServerPort=3306
mysqlDB=discogs
mysqlUser=root
mysqlPassword=123456

/forcedownload
Always (re)download the last discogs gz files and unzip them.
/forceupdate
The date of the discogs export files is stored in the table 'SETTINGS', when this is the same as the file date, the data is not imported. This flag forces the program to ignore the date.
/onlytab
This flag will only create the TAB files (and not throw them away when importing them in the database). Below you can find the reference for the layout of the fields per file. /useexistingfiles
This flag stops the program from downloading the latest files from discogs and uses the files in the data directory instead (see ini file where this is defined).

Most of the output from DiscogsXML2MySQL is written to a logfile with the name DiscogsXML2MySQL-yyyy-mm.log this is done because most of the time this program is run unattended and when something goes wrong it makes the job of debug the problem a lot easier.
Another log file DiscogsXML2MySQL.log is also created and mostly contains the start timestamps. Exceptions created by the program are also logged in this file.

I've only tested the import into mysql with some settings which could break it on other systems if they are not set (not tested so I don't know).
lower_case_table_names=1
We have this set on, because in the past we had a mix of windows and linux servers and name casing could lead to trouble on one server and not on the other. This forces them to always convert to lower case.

The xml files contain all kind of characters. utf8 is needed. On mysql this means the database, tables and fields must use utf8mb4 as character set (not utf8!). The program checks if the database is created with this option and stops if not!

The company I work for wants/wanted to match our music data with discogs. To do this would mean a lot of web service request, which are limited to 1 per second. This would be to slow for us. We then looked at using the discogs monthly export. We allready use mysql for our own website and wanted to use this database for matching. After some searching I couldn't find a good conversion program for the discogs data so that's why we created our own.
The program is written in C# and we asume you run it on a windows pc (haven't tested it with mono). The mysql part can of course be on a linux machine.

Download from discogs and unzipping the files takes about 1 hour and 21 minutes.
Converting the xml files and importing them to mysql takes about 3 hours on a vmware server with SSD and 4 hours was on a bare bone server with five disk in raid5.
Converting the xml files only to tab files take about 2 hour and 10 minutes on a desktop machine with SSD.

Some tab files contain ID's and some not, the decision when or not to include them is based on the datamodel I used.

discogs_yyyymmdd_artists.xml

• ARTIST.TAB
  • Fields: ARTIST_ID, NAME, REALNAME, PROFILE, DATA_QUALITY
• NAMEVARIATION.TAB
  • Fields: ARTIST_ID, NAME
• ALIAS.TAB
  • Fields: MAIN_ARTIST_ID, ALIAS_ARTIST_ID
• MEMBER.TAB
  • Fields: MAIN_ARTIST_ID, MEMBER_ARTIST_ID
• GROUP.TAB
  • Fields: MAIN_ARTIST_ID, GROUP_ARTIST_ID
• IMAGE.TAB
  • Fields: ARTIST_ID, HEIGHT, WIDTH, TYPE, URI, URI150
• URL.TAB
  • Fields: ARTIST_ID, URL

discogs_yyyymmdd_labels.xml

• LABEL.TAB
  • Fields: LABEL_ID, NAME, CONTACTINFO, PROFILE, DATA_QUALITY, PARENT_LABEL_ID
• LABEL-IMAGE.TAB
  • Fields: LABEL_ID, HEIGHT, WIDTH, TYPE, URI, URI150
• LABEL-URL.TAB
  • Fields: LABEL_ID, URL
• SUBLABEL.TAB
  • Fields: MAIN_LABEL_ID, CHILD_LABEL_ID

discogs_yyyymmdd_releases.xml

• RELEASE.TAB
  • Fields: RELEASE_ID, MASTER_ID, STATUS, TITLE, COUNTRY, RELEASED, NOTES, DATA_QUALITY, IS_MAIN_RELEASE
• RELEASE-IMAGE.TAB
  • Fields: RELEASE_ID, HEIGHT, WIDTH, TYPE, URI, URI150
• RELEASE-ARTIST.TAB
  • Fields: RELEASE_ID, ARTIST_ID, ANV, JOIN, ROLE, NAME, EXTRA_ARTIST
• RELEASE-GENRE.TAB
  • Fields: RELEASE_ID, GENRE_NAME
• RELEASE-STYLE.TAB
  • Fields: RELEASE_ID, STYLE_NAME
• FORMAT.TAB
  • Fields: FORMAT_ID, RELEASE_ID, FORMAT_NAME, FORMAT_TEXT, QUANTITY
• FORMAT-DESCRIPTION.TAB
  • Fields: FORMAT_ID, DESCRIPTION, DESCRIPTION_ORDER
• RELEASE-LABEL.TAB
  • Fields: RELEASE_ID, LABEL_ID, CATNO
• TRACK.TAB
  • Fields: TRACK_ID, RELEASE_ID, MAIN_TRACK_ID, HAS_SUBTRACKS, IS_SUBTRACK, TRACKNUMBER, TITLE, SUBTRACK_TITLE, POSITION, DURATION_IN_SEC
• TRACK-ARTIST.TAB
  • Fields: TRACK_ID, ARTIST_ID, ANV, JOIN, ROLE, NAME, EXTRA_ARTIST
• IDENTIFIER.TAB
  • Fields: RELEASE_ID, DESCRIPTION, TYPE, VALUE
• RELEASE-VIDEO.TAB
  • Fields: RELEASE_ID, EMBED, DURATION_IN_SEC, SRC, TITLE, DESCRIPTION
• COMPANY.TAB
  • Fields: COMPANY_ID, RELEASE_ID, NAME, CATNO, ENTITY_TYPE, ENTITY_TYPE_NAME, RESOURCE_URL

discogs_yyyymmdd_masters.xml

• MASTER.TAB
  • Fields: MASTER_ID, MAIN_RELEASE_ID, TITLE, RELEASED, NOTES, DATA_QUALITY
• MASTER-IMAGE.TAB
  • Fields: MASTER_ID, HEIGHT, WIDTH, TYPE, URI, URI150
• MASTER-ARTIST.TAB
  • Fields: MASTER_ID, ARTIST_ID, ANV, JOIN, NAME, ROLE
• MASTER-GENRE.TAB
  • Fields: MASTER_ID, GENRE_NAME
• MASTER-STYLE.TAB
  • Fields: MASTER_ID, STYLE_NAME
• MASTER-VIDEO.TAB
  • Fields: MASTER_ID, EMBED, DURATION_IN_SEC, SRC, TITLE, DESCRIPTION

The datamodel is made in DeZign for Databases. I've also included the .dez file and a .emf image file for reference.

You can contact me (Yvo Nelemans) at y.nelemans@muziekweb.nl