30 KiB
slsk-batchdl
A batch downloader for Soulseek built with Soulseek.NET. Accepts CSV files as well as Spotify and YouTube urls.
See the examples.
Index
- Options
- Input types
- Download modes
- Searching
- File conditions
- Name format
- Skip existing
- Configuration
- Examples
- Notes
- Docker
Options
Usage: sldl <input> [OPTIONS]
Required Arguments
<input> A url, search string, or path to a local CSV file.
Run --help "input" to view the accepted inputs.
Can also be passed with -i, --input <input>
--user <username> Soulseek username
--pass <password> Soulseek password
General Options
-p, --path <path> Download directory
-f, --folder <name> Subfolder name. Set to '.' to output directly to --path
--input-type <type> Force set input type, [csv|youtube|spotify|bandcamp|string]
--name-format <format> Name format for downloaded tracks. See --help name-format
-n, --number <maxtracks> Download the first n tracks of a playlist
-o, --offset <offset> Skip a specified number of tracks
-r, --reverse Download tracks in reverse order
-c, --config <path> Set config file location. Set to 'none' to ignore config
--profile <names> Configuration profile(s) to use. See --help "config".
--concurrent-downloads <num> Max concurrent downloads (default: 2)
--m3u <option> Create an m3u8 playlist file in the output directory
'none' (default for single inputs): Do not create
'index' (default): Write a line indexing all downloaded
files, required for skip-not-found or skip-existing=m3u
'all': Write the index and a list of paths and fails
-s, --skip-existing Skip if a track matching file conditions is found in the
output folder or your music library (if provided)
--skip-mode <mode> [name|tag|m3u|name-cond|tag-cond|m3u-cond]
See --help "skip-existing".
--music-dir <path> Specify to also skip downloading tracks found in a music
library. Use with --skip-existing
--skip-not-found Skip searching for tracks that weren't found on Soulseek
during the last run. Fails are read from the m3u file.
--skip-existing-pref-cond Use preferred instead of necessary conds for skip-existing
--display-mode <option> Changes how searches and downloads are displayed:
'single' (default): Show transfer state and percentage
'double': Transfer state and a large progress bar
'simple': No download bars or changing percentages
--print <option> Print tracks or search results instead of downloading:
'tracks': Print all tracks to be downloaded
'tracks-full': Print extended information about all tracks
'results': Print search results satisfying file conditions
'results-full': Print search results including full paths
--debug Print extra debug info
--listen-port <port> Port for incoming connections (default: 49998)
--on-complete <command> Run a command whenever a file is downloaded.
Available placeholders: {path} (local save path), {title},
{artist},{album},{uri},{length},{failure-reason},{state}.
Prepend a state number to only download in specific cases:
1:, 2:, 3:, 4: for the Downloaded, Failed, Exists, and
NotFoundLastTime states respectively.
E.g: '1:<cmd>' will only run the command if the file is
downloaded successfully. Prepend 's:' to use the system
shell to execute the command.
Searching
--fast-search Begin downloading as soon as a file satisfying the preferred
conditions is found. Only for normal download mode.
--remove-ft Remove 'feat.' and everything after before searching
--no-remove-special-chars Do not remove special characters before searching
--remove-brackets Remove square brackets and their contents before searching
--regex <regex> Remove a regexp from all track titles and artist names.
Optionally specify a replacement regex after a semicolon.
Add 'T:', 'A:' or 'L:' at the start to only apply this to
the track title, artist, or album respectively.
--artist-maybe-wrong Performs an additional search without the artist name.
Useful for sources like SoundCloud where the "artist"
could just be an uploader. Note that when downloading a
YouTube playlist via url, this option is set automatically
on a per-track basis, so it is best kept off in that case.
-d, --desperate Tries harder to find the desired track by searching for the
artist/album/title only, then filtering. (slower search)
--fails-to-downrank <num> Number of fails to downrank a user's shares (default: 1)
--fails-to-ignore <num> Number of fails to ban/ignore a user's shares (default: 2)
--yt-dlp Use yt-dlp to download tracks that weren't found on
Soulseek. yt-dlp must be available from the command line.
--yt-dlp-argument <str> The command line arguments when running yt-dlp. Default:
"{id}" -f bestaudio/best -cix -o "{savepath}.%(ext)s"
Available vars are: {id}, {savedir}, {savepath} (w/o ext).
Note that with -x, yt-dlp will download webms in case
ffmpeg is unavailable.
--search-timeout <ms> Max search time in ms (default: 6000)
--max-stale-time <ms> Max download time without progress in ms (default: 50000)
--searches-per-time <num> Max searches per time interval. Higher values may cause
30-minute bans, see --help "search". (default: 34)
--searches-renew-time <sec> Controls how often available searches are replenished.
See --help "search". (default: 220)
Spotify
--spotify-id <id> Spotify client ID
--spotify-secret <secret> Spotify client secret
--spotify-token <token> Spotify access token
--spotify-refresh <token> Spotify refresh token
--remove-from-source Remove downloaded tracks from source playlist
YouTube
--youtube-key <key> Youtube data API key
--get-deleted Attempt to retrieve titles of deleted videos from wayback
machine. Requires yt-dlp.
--deleted-only Only retrieve & download deleted music.
CSV Files
--artist-col Artist column name
--title-col Track title column name
--album-col Album column name
--length-col Track length column name
--album-track-count-col Album track count column name (sets --album-track-count)
--yt-desc-col Youtube description column (improves --yt-parse)
--yt-id-col Youtube video id column (improves --yt-parse)
--time-format <format> Time format in Length column of the csv file (e.g h:m:s.ms
for durations like 1:04:35.123). Default: s
--yt-parse Enable if the CSV contains YouTube video titles and channel
names; attempt to parse them into title and artist names.
--remove-from-source Remove downloaded tracks from source CSV file
File Conditions
--format <formats> Accepted file format(s), comma-separated, without periods
--length-tol <sec> Length tolerance in seconds
--min-bitrate <rate> Minimum file bitrate
--max-bitrate <rate> Maximum file bitrate
--min-samplerate <rate> Minimum file sample rate
--max-samplerate <rate> Maximum file sample rate
--min-bitdepth <depth> Minimum bit depth
--max-bitdepth <depth> Maximum bit depth
--banned-users <list> Comma-separated list of users to ignore
--pref-format <formats> Preferred file format(s), comma-separated (default: mp3)
--pref-length-tol <sec> Preferred length tolerance in seconds (default: 3)
--pref-min-bitrate <rate> Preferred minimum bitrate (default: 200)
--pref-max-bitrate <rate> Preferred maximum bitrate (default: 2500)
--pref-min-samplerate <rate> Preferred minimum sample rate
--pref-max-samplerate <rate> Preferred maximum sample rate (default: 48000)
--pref-min-bitdepth <depth> Preferred minimum bit depth
--pref-max-bitdepth <depth> Preferred maximum bit depth
--pref-banned-users <list> Comma-separated list of users to downrank
--strict-conditions Skip files with missing properties instead of accepting by
default; if --min-bitrate is set, ignores any files with
unknown bitrate.
Album Download
-a, --album Album download mode: Download a folder
-t, --interactive Interactive mode, allows to select the folder and images
--album-track-count <num> Specify the exact number of tracks in the album. Add a + or
- for inequalities, e.g '5+' for five or more tracks.
--album-ignore-fails Do not skip to the next source and do not delete all
successfully downloaded files if one of the files in the
folder fails to download
--album-art <option> Retrieve additional images after downloading the album:
'default': No additional images
'largest': Download from the folder with the largest image
'most': Download from the folder containing the most images
--album-art-only Only download album art for the provided album
--no-browse-folder Do not automatically browse user shares to get all files in
in the folder
Aggregate Download
-g, --aggregate Aggregate download mode: Find and download all distinct
songs associated with the provided artist, album, or title.
--min-shares-aggregate <num> Minimum number of shares of a track or album for it to be
downloaded in aggregate mode. (Default: 2)
--relax-filtering Slightly relax file filtering in aggregate mode to include
more results
Help
-h, --help [option] [all|input|download-modes|search|name-format|
file-conditions|skip-existing|config]
Notes
Acronyms of two- and --three-word-flags are also accepted, e.g. --twf. If the option
contains the word 'max' then the m should be uppercase. 'bitrate', 'sameplerate' and
'bitdepth' should be all treated as two separate words, e.g --Mbr for --max-bitrate.
Flags can be explicitly disabled by setting them to false, e.g '--interactive false'
Input types
The input type is usually determined automatically. To force a specific input type, set --input-type [spotify|youtube|csv|string|bandcamp]. The following input types are available:
CSV file
Path to a local CSV file: Use a csv file containing track info of the songs to download. The names of the columns should be Artist, Title, Album, Length, although alternative names are usually detected as well. Only the title or album column is required, but extra info may improve search results. Every row that does not have a title column text will be treated as an album download.
YouTube
A playlist url: Download songs from a youtube playlist. The default method to retrieve playlists doesn't always return all videos, especially not the ones which are unavailable. To get all video titles, you can use the official API by providing a key with --youtube-key. Get it here https://console.cloud.google.com. Create a new project, click "Enable Api" and search for "youtube data", then follow the prompts.
Tip: For playlists containing music videos, it may be better to remove all text in parentheses (to remove (Lyrics), (Official), etc) and disable song duration checking: --regex "[[(].*?[])]" --pref-length-tol -1
Spotify
A playlist/album url or 'spotify-likes': Download a spotify playlist, album, or your liked songs. Credentials are required when downloading a private playlist or liked music.
Using Credentials
Click to expand
Create a Spotify application at https://developer.spotify.com/dashboard/applications with a redirect url http://localhost:48721/callback. Obtain an application ID and secret from the created application dashboard.
Start sldl with the obtained credentials and an authorized action to trigger the Spotify app login flow:
sldl spotify-likes --spotify-id 123456 --spotify-secret 123456 -n 1 --print-tracks
sldl will try to open a browser automatically but will fallback to logging the login flow URL to output. After login flow is complete sldl will output a token and refresh token and finish running the current command.
To skip requiring login flow every time sldl is used the token and refresh token can be provided to sldl (hint: store this info in the config file to make commands less verbose):
sldl spotify-likes --spotify-id 123456 --spotify-secret 123456 --spotify-refresh 123456 --spotify-token 123456 -n 1 --pt
spotify-token access is only valid for 1 hour. spotify-refresh will enable sldl to renew access every time it is run (and can be used without including spotify-token)
Bandcamp
A bandcamp url: Download a single track, an album, or an artist's entire discography. Extracts the artist name, album name and sets --album-track-count="n+", where n is the number of visible tracks on the bandcamp page.
Search string
Name of the track, album, or artist to search for: Can either be any typical search string (like what you would enter into the soulseek search bar), or a comma-separated list of properties like 'title=Song Name, artist=Artist Name, length=215'.
The following properties are allowed:
title
artist
album
length (in seconds)
artist-maybe-wrong
Example inputs and their interpretations:
Input String | Artist | Title | Album | Length
---------------------------------------------------------------------------------
'Foo Bar' (without any hyphens) | | Foo Bar | |
'Foo - Bar' | Foo | Bar | |
'Foo - Bar' (with --album enabled) | Foo | | Bar |
'Artist - Title, length=42' | Artist | Title | | 42
'artist=AR, title=T, album=AL' | AR | T | AL |
Download modes
Normal
The program will download a single file for every input entry.
Album
sldl will search for the album and download an entire folder including non-audio files. Activated when the input is a link to a spotify or bandcamp album, when the input string or csv row has no track title, or when -a/--album is enabled.
Aggregate
With -g/--aggregate, sldl will first perform an ordinary search for the input, then attempt to
group the results into distinct songs and download one of each kind. A common use case is
finding all remixes of a song or printing all songs by an artist that are not your music dir.
Two files are considered equal if their inferred track title and artist name are equal
(ignoring case and some special characters), and their lengths are within --length-tol of each
other.
Note that this mode is not 100% reliable, which is why --min-shares-aggregate is set to 2 by
default, i.e. any song that is shared only once will be ignored.
Album Aggregate
Activated when --album and --aggregate are enabled, in this mode sldl searches for the query
and groups results into distinct albums. Two folders are considered same if they have the
same number of audio files, and the durations of the files are within --length-tol of each
other (or within 3 seconds if length-tol is not configured). If both folders have exactly one
audio file with similar lengths, also checks if the inferred title and artist name coincide.
More reliable than normal aggregate due to much simpler grouping logic.
Note that --min-shares-aggregate is 2 by default, which means that folders shared only once
will be ignored.
Searching
Search Query
The search query is determined as follows:
- For album downloads: If the album field is non-empty, search for 'Artist Album'. Otherwise, search for 'Artist Title'
- For all other download types: If the title field is non-empty, search for 'Artist Title'. Otherwise, search for 'Artist Album'
Soulseek's rate limits
The server will ban you for 30 minutes if too many searches are performed within a short timespan. The program has a search limiter which can be adjusted with --searches-per-time and --searches-renew-time (when the limit is reached, the status of the downloads will be "Waiting"). By default it is configured to allow up to 34 searches every 220 seconds. The default values were determined through experimentation, so they may be incorrect.
Quality vs Speed
The following options will make it go faster, but may decrease search result quality or cause instability:
- --fast-search skips waiting until the search completes and downloads as soon as a file matching the preferred conditions is found
- --concurrent-downloads - set it to 4 or more
- --max-stale-time is set to 50 seconds by default, so it will wait a long time before giving up on a file
- --searches-per-time increase at the risk of bans.
Quality vs Quantity
The options --strict-title, --strict-artist and --strict-album will filter any file that
does not contain the title/artist/album in the filename (ignoring case, bounded by boundary
chars).
Another way to prevent false downloads is to set --length-tol to 3 or less to make it ignore
any songs that differ from the input by more than 3 seconds. However, all 4 options are already
enabled as 'preferred' conditions by default, meaning that such files will only be downloaded
as a last resort anyways. Hence it is only recommended to enable them if you need to minimize
false downloads as much as possible.
File conditions
Files not satisfying the required conditions will not be downloaded. Files satisfying pref- conditions will be preferred; setting --pref-format "flac,wav" will make it download lossless files if available, and only download lossy files if there's nothing else.
There are no default required conditions. The default preferred conditions are:
pref-format = mp3
pref-length-tol = 3
pref-min-bitrate = 200
pref-max-bitrate = 2500
pref-max-samplerate = 48000
pref-strict-title = true
pref-strict-album = true
pref-accept-no-length = false
sldl will therefore prefer mp3 files with bitrate between 200 and 2500 kbps, and whose length
differs from the supplied length by no more than 3 seconds. It will also prefer files whose
paths contain the supplied title and album (ignoring case, and bounded by boundary characters)
and which have non-null length. Changing the last three preferred conditions is not recommended.
Note that files satisfying a subset of the preferred conditions will still be preferred over files
that don't satisfy any condition, but some conditions have precedence over others. For instance,
a file that only satisfies strict-title (if enabled) will always be preferred over a file that
only satisfies the format condition. Run with --print "results-full" to reveal the sorting logic.
Important note
Some info may be unavailable depending on the client used by the peer. For example, the standard
Soulseek client does not share the file bitrate. If (e.g) --min-bitrate is set, then sldl will
still accept any file with unknown bitrate. You can configure it to reject all files where one
or more of the checked properties is null (unknown) by enabling --strict-conditions.
As a consequence, if --min-bitrate is also set then any files shared by users with the default
client will be ignored. Also note that the default preferred conditions will already affect
ranking with this option due to the bitrate and samplerate checks.
Conditions can also be supplied as a semicolon-delimited string with --cond and --pref, e.g --cond "br>=320;f=mp3,ogg;sr<96000"
Name format
Variables enclosed in {} will be replaced by the corresponding file tag value. Name format supports subdirectories as well as conditional expressions like {tag1|tag2} - If tag1 is null, use tag2. String literals enclosed in parentheses are ignored in the null check.
Examples:
- "{artist} - {title}"
Always name it 'Artist - Title'. Because some files on Soulseek are untagged, the following is generally preferred: - "{artist( - )title|filename}"
If artist and title are not null, name it 'Artist - Title', otherwise use the original filename. - "{albumartist(/)album(/)track(. )title|(missing-tags/)filename}"
Sort files into artist/album folders if all tags are present, otherwise put them in the 'missing-tags' folder.
Available variables:
artist First artist (from the file tags)
sartist Source artist (as on CSV/Spotify/YouTube/etc)
artists Artists, joined with '&'
albumartist First album artist
albumartists Album artists, joined with '&'
title Track title
stitle Source track title
album Album name
salbum Source album name
year Track year or date
track Track number
disc Disc number
filename Soulseek filename without extension
foldername Soulseek folder name
default-foldername Default sldl folder name
extractor Name of the extractor used (CSV/Spotify/YouTube/etc)
Skip existing
sldl can skip downloads that exist in the output directory or a specified directory configured with --music-dir. The following modes are available for --skip-mode:
m3u
Default when checking in the output directory.
Checks whether the output m3u file contains the track in the '#SLDL' line. Does not check if
the audio file exists or satisfies the file conditions (use m3u-cond for that). m3u and
m3u-cond are the only modes that can skip album downloads.
name
Default when checking in the music directory.
Compares filenames to the track title and artist name to determine if a track already exists.
Specifically, a track will be skipped if there exists a file whose name contains the title
and whose full path contains the artist name.
tag
Compares file tags to the track title and artist name. A track is skipped if there is a file whose artist tag contains the track artist and whose title tag equals the track title (ignoring case and ws). Slower than name mode as it needs to read all file tags.
m3u-cond, name-cond, tag-cond
Same as the above modes but also checks whether the found file satisfies the configured conditions. Uses necessary conditions by default, run with --skip-existing-pref-cond to use preferred conditions instead. Equivalent to the above modes if no necessary conditions have been specified (except m3u-cond, which always checks if the file exists). May be slower and use a lot of memory for large libraries.
Configuration
Config Location:
sldl will look for a file named sldl.conf in the following locations:
~/AppData/Roaming/sldl/sldl.conf
~/.config/sldl/sldl.conf
as well as in the directory of the executable.
Syntax:
Example config file:
username = your-username
password = your-password
pref-format = flac
fast-search = true
Lines starting with hashtags (#) will be ignored. Tildes in paths are expanded as the user directory.
Configuration profiles:
Profiles are supported:
[lossless]
pref-format = flac,wav
To activate the above profile, run --profile "lossless". To list all available profiles,
run --profile "help".
Profiles can be activated automatically based on a few simple conditions:
[no-stale]
profile-cond = interactive && download-mode == "album"
max-stale-time = 999999
# album downloads will never be automatically cancelled in interactive mode
[youtube]
profile-cond = input-type == "youtube"
path = ~/downloads/sldl-youtube
# download to another location for youtube
The following operators are supported for use in profile-cond: &&, ||, ==, !=, !{bool}.
The following variables are available:
input-type ("youtube"|"csv"|"string"|"bandcamp"|"spotify")
download-mode ("normal"|"aggregate"|"album"|"album-aggregate")
interactive (bool)
Examples
Download tracks from a csv file:
sldl test.csv
Download spotify likes:
sldl spotify-likes
Download albums for every song in a spotify playlist:
sldl https://spotify/playlist/url --album --skip-existing
Retrieve deleted video names, then download from a youtube playlist with fallback to yt-dlp:
sldl "https://www.youtube.com/playlist?list=PLI_eFW8NAFzYAXZ5DrU6E6mQ_XfhaLBUX" --get-deleted --yt-dlp
Search & download a specific song, preferring lossless:
sldl "title=MC MENTAL @ HIS BEST,length=242" --pref-format "flac,wav"
Interactive album download:
sldl "album=Some Album" --interactive
Print all songs by an artist which are not in your library:
sldl "artist=MC MENTAL" --aggregate --skip-existing --music-dir "path/to/music" --print tracks-full
Download all albums by an artist that are on soulseek:
sldl "artist=MC MENTAL" --aggregate --album
Notes
- For macOS builds you can use publish.sh to build the app. Download dotnet from https://dotnet.microsoft.com/en-us/download/dotnet/6.0, then run
chmod +x publish.sh && sh publish.sh. For intel macs, uncomment the x64 and comment the arm64 section in publish.sh. --display singleand especiallydoublecan cause the printed lines to be duplicated or overwritten on some configurations. Usesimpleif that's an issue.
Docker
A docker container for running sldl can be built from this repository. The image supports linux x86/ARM.
To build and start container:
clone https://github.com/fiso64/slsk-batchdl
cd slsk-batchdl
docker compose up -d
exec into the container to start using sldl:
docker compose exec sldl sh
sldl --help
The compose stack mounts two directories relative to where docker-compose.yml is located which can be used for file management:
/config(at./configon host) - put yoursldl.confconfiguration in this directory and then usesldl -c /config ...to use your configuration in the container/data(at./dataon host) - use as the download directory IEsldl -p /data ...
File Permissions
If you are running Docker on a Linux Host you should specify user:group permissions of the user who owns the configuration and data directory on the host to avoid docker file permission problems. These can be specified using the environmental variables PUID and PGID.
To get the UID and GID for the current user run these commands from a terminal:
id -u-- prints UIDid -g-- prints GID
Replace these with the corresponding variable (PUID PGID) in docker-compose.yml.
Cron
One or more sldl commands can be run on a schedule using cron built into the container.
To create a schedule make a new file on the host ./config/crontabs/abc and use it with the standard crontab syntax.
Make sure to restart the container after any changes to the cron file are made.
Example => Run sldl every Sunday at 1am, search for missing tracks from the specified Spotify playlist
# min hour day month weekday command
0 1 * * 0 sldl https://open.spotify.com/playlist/6sf1WR5grXGJ6dET -c /config -p /data --music-dir /data --skip-existing"
crontab.guru could be used to help with the scheduling expression.