108 lines
5.3 KiB
Markdown
108 lines
5.3 KiB
Markdown
# dam
|
|
### Digital Audio Manager
|
|
|
|
© 2017 Antoine Martin \<antoine.martin@protonmail.com\>, provided under GNU General Public License v3.0.
|
|
|
|
Utility used for deployment of music collection organized by CD images, delimited by CUE sheet files,
|
|
tagged by JSON file based on MTAG specification and covered by JPG file.
|
|
|
|
|
|
## dam --help
|
|
Usage: dam [--help] [--info] [<options>] <command> </path/to/target> [<args>]
|
|
|
|
Options
|
|
--git-dir=</path/to/git/dir>
|
|
Defines path to git directory that contains the music collection. Defaults to current directory when not set.
|
|
|
|
Commands
|
|
update
|
|
Populates database present in target folder with new images.
|
|
deploy
|
|
Deploys tracks to target folder, with applies metadata and cover image
|
|
exclude <condition> [<additional condition>]
|
|
Exclude tracks based on conditions (see dam --help conditions for more information)
|
|
include <condition> [<additional condition>]
|
|
Include tracks based on conditions (see dam --help conditions for more information)
|
|
|
|
## dam --help condition
|
|
Conditions
|
|
|
|
Note
|
|
Conditions are defined using the following format:
|
|
<metadatafield>=<value>
|
|
The metadta field can be anything that can be injected as metadta in a Vorbis tag.
|
|
|
|
Example
|
|
You want to exclude everything from "GENRE" "Pop":
|
|
dam exclude "GENRE=Pop"
|
|
|
|
...but want to keep Lady Gaga, just because
|
|
dam include "ARTIST=Lady Gaga"
|
|
|
|
...but not THAT album
|
|
dam exclude "ALBUM=THAT"
|
|
|
|
## A note on music library format
|
|
The music library (git directory) is organized by album images. When the audio is ripped from cd, the cuetools
|
|
image is kept as is. When the source is digital (i.e. Bandcamp), the album is conjoined into one .flac file to
|
|
be treated as if it was a CD image, with its own cue sheet, and accompanying files.
|
|
|
|
### Files
|
|
They are four mandatory files per image:
|
|
* a FLAC file (the image file compressed to FLAC)
|
|
* a CUE file (either generated from cuetools, or generated using import function from track files)
|
|
* a MTAG file (metadata set using the MTAG specification)
|
|
* a JPG file (480x480 resolution cover to be imported into the deployed FLAC files)
|
|
* (Optionally) a PNG file that contains a larger resolution version of the JPG cover file
|
|
* (Optionally ACCURIP and LOG file generated by cueripper.
|
|
|
|
### File name format
|
|
It is assumed that they are all named the same, only differentiated by their extension. The format used by
|
|
yours truly is the following:
|
|
SHA256-<ctdb id>-<sha256 sum of FLAC file>.<ext>
|
|
|
|
In the event that the source is not a CD, the ctdb id is then replaced with the source info. in my case, <FLAC>
|
|
means a lossless source and <MP3> means a lossy source. A few examples:
|
|
|
|
* The Alan Parsons Project's I Robot
|
|
SHA256-sGYFy0TxXWucFAm.WYiPeQCSuv8--edba73bda9a2dc829405f979956e6dbf748896b48f0604bee2a5249e9dfb7eb0.<ext>
|
|
|
|
* Antoine Corriveau's Cette chose qui cognait au creux de sa poitrine sans vouloir s'arreter
|
|
SHA256-FLAC--0f25f26caa8b321dfc8e84d068f7a1225d1c7893f66f225b8e04bb51bbaf56e9.<ext>
|
|
|
|
### Git annex
|
|
The music collection is handled using git annex, which is why this was done (more on this later).
|
|
The version used is v6, which allows some files to be locked and serviced by git annex, and other files
|
|
to be serviced by git directory. In my case, I wanted the metadata to be services directly by git, and thus
|
|
the CUE file, MTAG file and JPG file are imported directly in the branch, while the FLAC file, PNg file are
|
|
handled by git-annex as they are larger.
|
|
|
|
Thus, when one clones the music repository, the metadata is downloaded without the data, allowing the management
|
|
of a really large collection to be much easier. It is then as easy as doing "git annex get <file>.flac to download
|
|
the image from the server. If a change is made (i.e., an image gets added, or metadata changed), the changes can get
|
|
propagated.
|
|
|
|
# Why?
|
|
One thing I wanted to do was use git-annex to manage my large music collection (1194 albums, 11914 tracks, ~400 GB).
|
|
What really sucks with metadata is how its embeded in the audio file. Thus, when a small metadata change occurs, it
|
|
was necessary to reupload (and redownload) the whole FLAC file. Not ideal.
|
|
|
|
Thus, came the idea of using CUE sheet files as a way to decouple metadata from data. Unfortunately, the CUE sheet
|
|
specification is very limited. I have too many classical discs to accept such a thing. Thus, I went searching and
|
|
found the MTAG specification that basically sets the metadata using a JSON text file. Unfortunately, nothing supports
|
|
this specification except foobar2000.
|
|
|
|
No matter, when nothing supports it, just code a program that splits the image files in seperate tracks using the
|
|
cue sheet, then apply the metadata by parsing the MTAG file, then copy the file to a target directory (i.e DAP),
|
|
then keep track of where everything is using a database file.
|
|
|
|
In the event of a small metadata change, the program is smart enough to apply the new metadata without having to
|
|
retransfer the same data.
|
|
|
|
# Why bash?
|
|
Because I was lazy and didn't feel like teaching myself Python. God knows I should've. Hell, I will probably port
|
|
this script to python as a way to get out of my comfort zone, and a way to have this be more portable. I would
|
|
definitely enjoy an ncurse interface for this. Alas, for when life gives me more time to waste on my OCD and
|
|
music collection.
|
|
|
|
Hope someone finds inspiration in my perfectionism..
|