NodMOD

NodMOD is a Python library for reading, editing, and writing tracker modules.

It currently focuses on three classic formats:

• MOD
• XM
• S3M

The project is built around direct programmatic editing. You load or create a song, manipulate patterns, notes, samples, instruments, and effects in Python, then save the result back to disk.

What It Does

• Load MOD, XM, and S3M files
• Save edited or newly created MOD, XM, and S3M files
• Create, duplicate, resize, clear, and reorder patterns
• Edit notes, effects, rows, channels, samples, and XM instruments
• Import WAV audio into MOD samples or XM instrument samples
• Export human-readable ASCII dumps for inspection
• Render modules to WAV through external tools when available
• Validate/sanitize MOD sample-loop metadata before strict saves

Installation

From PyPI (recommended):

uv add nodmod

or:

pip install nodmod

Verify with:

source .venv/bin/activate  # macOS/Linux
# OR
.venv\Scripts\activate     # Windows

python -c "import nodmod; print('nodmod installed successfully!')"

For development from source:

git clone https://github.com/erodola/nodmod.git
cd nodmod
uv venv
uv pip install -e .

Quick Start

from nodmod import MODSong

song = MODSong()
song.load("music/spice_it_up.mod")

# MOD channels are 0-based (0-3)
song.mute_channel(2)
song.mute_channel(3)

song.save("music/ch1_2.mod")

from nodmod import XMSong
from nodmod.types import XMSample

song = XMSong()
song.set_n_channels(4)

inst = song.new_instrument("Lead")
smp = XMSample()
import array

smp.waveform = array.array('b', [0, 10, -10, 0])
song.add_sample(inst, smp)
song.set_sample_map_all(inst, 1)

song.set_note(0, 0, 0, inst, "C-4", "")
song.set_sample_panning(inst, 1, 192)
song.set_global_volume(0, 0, 0, 64)

song.save("music/lead.xm")

from nodmod import S3MSong

song = S3MSong()
song.set_n_channels(8)

song.set_note(0, 0, 0, 1, "C-4", effect="A03", volume=32)
song.set_note(0, 1, 4, 2, "G-4", effect="T96", volume=40)

song.save("music/sketch.s3m")

Core Model

At a high level, the library exposes a small set of central objects:

• MODSong for MOD modules
• XMSong for XM modules
• S3MSong for S3M modules
• Pattern for pattern data
• Note, XMNote, and S3MNote for note cells
• Sample, XMSample, and S3MSample for waveform data
• Instrument for XM instruments and sample maps

The API is intentionally close to tracker structure rather than trying to hide it behind a higher-level composition DSL.

Format Notes

• MOD notes reference samples directly.
• MOD read APIs (get_note, iter_cells, iter_rows, get_used_samples) resolve sample-memory semantics by default: note rows with sample 00 inherit the last latched sample in the same channel.
• In MOD, sample-only rows (sample set with empty note) are valid and update that channel's latched sample for later note rows.
• MOD sample slots expose loop safety helpers (Sample.validate_loop, Sample.sanitize_loop, MODSong.validate_samples, MODSong.sanitize_samples) and save(..., validate_samples=True) for strict pre-save validation.
• Use get_note_raw(...) (and resolved=False where available) when you need exact raw MOD cell sample nibbles.
• XM notes reference instruments, and instruments contain samples.
• S3M notes reference sample / instrument slots directly for PCM modules.
• MOD sample slots are 1-based in the public API.
• XM instrument indices and XM sample indices are 1-based in the public API.
• S3M sample slots are 1-based in the public API.
• Pattern order and pattern storage are separate concepts, as they are in tracker files.

Current S3M scope:

• PCM S3M modules are supported for load, edit, save, and round-trip tests.
• Adlib / OPL S3M instruments are detected and rejected explicitly; they are not supported yet.

For most day-to-day use, the practical rule is simple: sequence positions, rows, and channels behave like normal Python indices, while tracker sample and instrument slots follow tracker conventions.

Extras

ASCII dumps are available for both formats:

song.save_ascii("music/debug.txt")

Rendering can target mono or multi-channel output when openmpt123 or a compatible ffmpeg build is available:

song.render("music/render.wav", channels=2)

New API Additions

Recent enhancements add inspection-focused, additive APIs without breaking existing method signatures. The v1.0.3 release hardens MOD/XM/S3M edge-case behavior, including fixed MOD channel invariants, XM serialization correctness, and timing/documentation consistency improvements.

from nodmod import MODSong

song = MODSong()
song.set_restart_position(3)            # normalized view
raw = song.get_restart_position(raw=True)  # exact MOD header byte

# Canonical coordinate order: (sequence_idx, row, channel, ...)
song.set_note_rc(0, 8, 1, 4, "C-4", "F06")
song.set_effect_rc(0, 8, 1, "B01")
cell = song.get_note_rc(0, 8, 1)

from nodmod import decode_mod_effect, encode_mod_effect

info = decode_mod_effect("E6F")
assert info.extended_cmd == "E6"
assert encode_mod_effect("F", 125) == "F7D"

# Immutable snapshots for read-only analysis
summary = song.view()
cells = list(song.iter_cells(sequence_only=True))
rows = list(song.iter_rows(sequence_only=True))
effects = list(song.iter_effects(sequence_only=True, include_empty=False))
samples = list(song.iter_samples(include_empty=False))

# MOD raw signed 8-bit PCM helpers
pcm = song.get_sample_pcm_i8(1)
song.set_sample_pcm_i8(1, pcm, reset_meta=False)
song.set_sample_loop_bytes(1, start_byte=128, length_byte=256)

from nodmod import probe_file

probe = probe_file("music/demo.mod")
print(probe.detected_format, probe.supported, probe.metadata)

# In-memory ASCII dump (no temp files needed)
text = song.to_ascii(sequence_only=True, include_headers=False)

# Playback-aware row timeline with source coordinates
playback = list(song.iter_playback_rows(max_steps=250_000))
first = playback[0]
print(first.sequence_idx, first.pattern_idx, first.row, first.start_sec, first.end_sec)

# Reachability-aware used-resource scans
mod_used = song.get_used_samples(scope="reachable", order="first_use")
xm_insts = xm_song.get_used_instruments(scope="sequence", order="sorted")
xm_samples = xm_song.get_used_samples(scope="reachable", order="sorted")
s3m_used = s3m_song.get_used_samples(scope="reachable", order="sorted")

# One-call loading with format dispatch
from nodmod import load_song

song = load_song("music/demo.mod")

Scope semantics:

• scope="sequence" inspects every row in sequence-referenced patterns.
• scope="reachable" inspects rows actually visited during playback flow.

Requirements

• Python 3.10, 3.11, 3.12, 3.13, or 3.14
• pydub 0.25.1+
• audioop-lts is pulled automatically on Python 3.13+
• Optional: openmpt123 or ffmpeg for WAV rendering

Project Status

This is an editing-focused library, not yet a full tracker toolkit. The codebase is strongest in direct manipulation of existing songs and in generating small-to-medium scripted edits. The public API is still evolving.

Contributing

Useful contributions include:

• bug fixes and behavioral cleanup
• stronger round-trip coverage for MOD, XM, and S3M files
• better public examples
• support for more tracker operations and formats

Reference Material

Large collections of legal-to-study modules can be found at The Mod Archive and Amiga Music Preservation.

Good players and editors for checking output include XMPlay, Qmmp, and MilkyTracker.

License

MIT License. See LICENSE.