A Java library for fetching and parsing rekordbox media exports and track analysis files.
This project uses the Kaitai Struct compiler with the help of a Maven plugin to create classes that can parse and output binary data structures in a convenient and efficient way. It uses them to create Java classes, because it was created to support Beat Link, but other projects can use them to output code for other languages.
It also uses the
jrpcgen tool (which is part of the
Remote Tea project),
via another plugin, to generate classes that know how to talk to the
non-standard NFSv2 file servers running in Link-capable players, so
the rekordbox data can be reliably obtained even during big shows
where all four players are in use.
the structure definitions needed to parse exported rekordbox databases
||Huge thanks to Fabian Lesniak for figuring out the details of how to interpret these files in his python-prodj-link project and Mikhail Yakshin for helping me quickly learn the more subtle aspects of Kaitai Struct. And this was all started by a question Evan Purkhiser posted on Stack Exchange.|
There is an Export Structure Analysis site describing the details of what we have learned about these file formats. Reading that will help make sense of the exploration tools and the objects returned by this library.
Exploring the Analysis
One of the amazingly cool things about Kaitai Struct is that you can
use its Web IDE to see how the structure
definitions work and visually explore the contents of files you are
analyzing. This also means you can look inside your own
and check my work, or get a better understanding of how to use the
generated parsers. To do that, simply upload the
.pdb file you want
to examine to the Web IDE (it doesn’t actually go to the web, it just
gets put in your local browser storage), then also upload my
rekordbox_pdb.ksy file, and
the Web IDE will parse the exported database, letting you explore the
structures in the tree view, and see the corresponding raw bytes in
the hex viewer.
You can find the
Each track in a rekordbox database also has
ANLZnnnn.EXT files associated with it, containing the beat grid, an
index allowing rapid seeking to any time in variable-bit rate audio
files, the waveforms, memory cues and loop points. The paths to these
files are found inside the corresponding track record.
Using the Library
Crate Digger is available through Maven Central, so to use it in your Maven project, all you need is to include the appropriate dependency.
Click the maven central badge above to view the repository entry for crate-digger. The proper format for including the latest release as a dependency in a variety of tools, including Leiningen if you are using beat-link from Clojure, can be found in the Dependency Information section.
There are two halves to what Crate Digger offers. The first is an ability to talk to the nonstandard Network File System servers that are running in Pioneer players, and ask them to deliver the rekordbox data export and track analysis files (programs like Beat Link need these files to provide smooth integrations with the music being performed). The second half is the ability to parse the contents of those files, as described above.
org.deepsymmetry.cratedigger.FileFetcher is a singleton,
so to work with it you will start by calling
getInstance(), as is
customary. Retrieving a file is then as simple as this:
FileFetcher fetcher = FileFetcher.getInstance(); fetcher.fetch(playerAddress, mountPath, sourcePath, destination);
playerAddress is an
InetAddress object holding the address of the
player from which you want to download a file.
the media slot you want to get information from, as shown in the table
sourcePath is the path to the specific file you want within
the mounted media, and
destination is a
File object identifying
where you want the downloaded data to be stored.
org.deepsymmetry.cratedigger.Database provides support for
accessing the contents of rekordbox database export files. You can create
an instance to wrap a
File instance that contains such an export (for
example one that you downloaded using the
fetch method above). Then
you can query it for track and other information:
Database database = new Database(downloadedFile); RekordboxPdb.TrackRow track = database.findTrack(1); System.out.println(database.getText(track.title()));
Strings (like titles, artist names, etc.) are represented by a variety
of structures with different encodings, so a
getText() method is
provided to convert them into ordinary Java strings.
Crate Digger uses slf4j to allow you to integrate it with whatever Java logging framework your project is using, so you will need to include the appropriate slf4j binding on your class path.
There are still more tables to be figured out.
Columnslooks like the list of things that can be searched by, so perhaps it will hold some clues for how to find and use the index tables, which must exist because it would be horribly slow for the players to do a linear scan through the main sparse tables whenever they wanted a record.
If we could figure out how to use the indices ourselves, we could avoid having to load the whole file and index it ourselves.
Building the source
As noted above, he Maven project uses a plugin to run the the
jrpcgen tool (which is part of the
project) to generate Java classes to implement the
ONC RPC specifications found in
(These are used for communicating with the NFS servers
in CDJs.) It also uses the Kaitai Struct Compiler
through another plugin to generate Java classes that can parse the
rekordbox databases it downloads from the players, based on the
specifications found in
These things happen for you automatically during the code generation phase of the Maven build. If you want to use something other than Maven, you will need to figure out how to configure and run the tools yourself.
Building the Structure Analysis
I started out using pdfLaTeX to write and format the document, but then, at the recommendation of one of the Kaitai Struct developers, switched to XeLaTeX in order to take advantage of newer features. But over time some of the packages I was using, especially for tables, became unsupported and started having issues. So this and the dysentery project’s protocol analysis document have been ported to more modern Asciidoc source in the form of Antora sites.
To re-create (and even improve on) the byte field diagrams I was able to achieve in LaTeX, I ended up writing my own diagram generator, bytefied-svg, which runs as an Antora plugin with the help of David Jencks' generic-svg-extension.
This documentation site can be built alongside the dysentery project’s protocol analysis, by following the directions in that project.
If you have ideas, discoveries, or even code you’d like to share, that’s fantastic! Please take a look at the guidelines and get in touch!
Copyright © 2018–2021 Deep Symmetry, LLC
Distributed under the Eclipse Public License 1.0. By using this software in any fashion, you are agreeing to be bound by the terms of this license. You must not remove this notice, or any other, from this software. A copy of the license can be found in LICENSE within this project.