Compare commits

...

5 commits

Author SHA1 Message Date
Henri Bourcereau 575e89f34b python bindings 2026-01-15 18:37:55 +01:00
Henri Bourcereau daf84f29ab doc 2026-01-15 18:31:12 +01:00
Henri Bourcereau e4bd20578a Merge tag 'v0.1.1' into develop
v0.1.1
2026-01-15 17:14:44 +01:00
Henri Bourcereau 228dc5d50c Merge branch 'release/v0.1.1' 2026-01-15 17:14:40 +01:00
Henri Bourcereau de303ad574 refact: docs 2026-01-15 17:13:07 +01:00
24 changed files with 550 additions and 37 deletions

79
Cargo.lock generated
View file

@ -3460,6 +3460,15 @@ dependencies = [
"stable_deref_trait",
]
[[package]]
name = "memoffset"
version = "0.9.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "488016bfae457b036d996092f6cb448677611ce4449e970ceaf42695203f218a"
dependencies = [
"autocfg",
]
[[package]]
name = "merge"
version = "0.1.0"
@ -4210,6 +4219,69 @@ dependencies = [
"num-traits",
]
[[package]]
name = "pyo3"
version = "0.23.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7778bffd85cf38175ac1f545509665d0b9b92a198ca7941f131f85f7a4f9a872"
dependencies = [
"cfg-if",
"indoc",
"libc",
"memoffset",
"once_cell",
"portable-atomic",
"pyo3-build-config",
"pyo3-ffi",
"pyo3-macros",
"unindent",
]
[[package]]
name = "pyo3-build-config"
version = "0.23.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "94f6cbe86ef3bf18998d9df6e0f3fc1050a8c5efa409bf712e661a4366e010fb"
dependencies = [
"once_cell",
"target-lexicon",
]
[[package]]
name = "pyo3-ffi"
version = "0.23.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e9f1b4c431c0bb1c8fb0a338709859eed0d030ff6daa34368d3b152a63dfdd8d"
dependencies = [
"libc",
"pyo3-build-config",
]
[[package]]
name = "pyo3-macros"
version = "0.23.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fbc2201328f63c4710f68abdf653c89d8dbc2858b88c5d88b0ff38a75288a9da"
dependencies = [
"proc-macro2",
"pyo3-macros-backend",
"quote",
"syn 2.0.106",
]
[[package]]
name = "pyo3-macros-backend"
version = "0.23.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fca6726ad0f3da9c9de093d6f116a93c1a38e417ed73bf138472cf4064f72028"
dependencies = [
"heck",
"proc-macro2",
"pyo3-build-config",
"quote",
"syn 2.0.106",
]
[[package]]
name = "qoi"
version = "0.4.1"
@ -5154,6 +5226,7 @@ dependencies = [
"base64 0.21.7",
"log",
"merge",
"pyo3",
"rand 0.8.5",
"serde",
"transpose",
@ -5892,6 +5965,12 @@ version = "0.2.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ebc1c04c71510c7f702b52b7c350734c9ff1295c464a03335b00bb84fc54f853"
[[package]]
name = "unindent"
version = "0.2.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7264e107f553ccae879d21fbea1d6724ac785e8c3bfc762137959b5802826ef3"
[[package]]
name = "universal-hash"
version = "0.5.1"

View file

@ -1,7 +1,41 @@
# Trictrac
Game of [Trictrac](https://en.wikipedia.org/wiki/Trictrac) in rust.
This is a game of [Trictrac](https://en.wikipedia.org/wiki/Trictrac) rust implementation.
wip
The project is on its early stages.
Rules (without "schools") are implemented, as well as a rudimentary terminal interface which allow you to play against a bot which plays randomly.
Training of AI bots is the work in progress.
## Usage
`cargo run --bin=client_cli -- --bot random`
## Roadmap
- [x] rules
- [x] command line interface
- [x] basic bot (random play)
- [ ] AI bot
- [ ] network game
- [ ] web client
## Code structure
- game rules and game state are implemented in the _store/_ folder.
- the command-line application is implemented in _client_cli/_; it allows you to play against a bot, or to have two bots play against each other
- the bots algorithms and the training of their models are implemented in the _bot/_ folder
### _store_ package
The game state is defined by the `GameState` struct in _store/src/game.rs_. The `to_string_id()` method allows this state to be encoded compactly in a string (without the played moves history). For a more readable textual representation, the `fmt::Display` trait is implemented.
### _client_cli_ package
`client_cli/src/game_runner.rs` contains the logic to make two bots play against each other.
### _bot_ package
- `bot/src/strategy/default.rs` contains the code for a basic bot strategy: it determines the list of valid moves (using the `get_possible_moves_sequences` method of `store::MoveRules`) and simply executes the first move in the list.
- `bot/src/strategy/dqnburn.rs` is another bot strategy that uses a reinforcement learning trained model with the DQN algorithm via the burn library (<https://burn.dev/>).
- `bot/scripts/trains.sh` allows you to train agents using different algorithms (DQN, PPO, SAC).

9
bot/pyproject.toml Normal file
View file

@ -0,0 +1,9 @@
[build-system]
requires = ["maturin>=1.0,<2.0"]
build-backend = "maturin"
[tool.maturin]
# "extension-module" tells pyo3 we want to build an extension module (skips linking against libpython.so)
features = ["pyo3/extension-module"]
# python-source = "python"
# module-name = "trictrac.game"

4
bot/python/test.py Normal file
View file

@ -0,0 +1,4 @@
import store
game = store.TricTrac()
print(game.get_state_dict())

View file

@ -237,7 +237,7 @@ impl TrictracEnvironment {
// Mapper l'index d'action sur une action valide
let action_index = (action.index as usize) % valid_actions.len();
Some(valid_actions[action_index].clone())
Some(valid_actions[action_index])
}
/// Exécute une action Trictrac dans le jeu

View file

@ -6,8 +6,8 @@ use std::fmt::{Debug, Display, Formatter};
use serde::{Deserialize, Serialize};
use store::{CheckerMove, GameEvent, GameState};
// 1 (Roll) + 1 (Go) + mouvements possibles
// Pour les mouvements : 2*16*16 = 514 (choix du dé + choix de la dame 0-15 pour chaque from)
// 1 (Roll) + 1 (Go) + 512 (mouvements possibles)
// avec 512 = 2 (choix du dé) * 16 * 16 (choix de la dame 0-15 pour chaque from)
pub const ACTION_SPACE_SIZE: usize = 514;
/// Types d'actions possibles dans le jeu

View file

@ -3,10 +3,10 @@
"devenv": {
"locked": {
"dir": "src/modules",
"lastModified": 1753667201,
"lastModified": 1768056019,
"owner": "cachix",
"repo": "devenv",
"rev": "4d584d7686a50387f975879788043e55af9f0ad4",
"rev": "9bfc4a64c3a798ed8fa6cee3a519a9eac5e73cb5",
"type": "github"
},
"original": {
@ -19,14 +19,14 @@
"flake-compat": {
"flake": false,
"locked": {
"lastModified": 1747046372,
"owner": "edolstra",
"lastModified": 1767039857,
"owner": "NixOS",
"repo": "flake-compat",
"rev": "9100a0f413b0c601e0533d1d94ffd501ce2e7885",
"rev": "5edf11c44bc78a0d334f6334cdaf7d60d732daab",
"type": "github"
},
"original": {
"owner": "edolstra",
"owner": "NixOS",
"repo": "flake-compat",
"type": "github"
}
@ -40,10 +40,10 @@
]
},
"locked": {
"lastModified": 1750779888,
"lastModified": 1767281941,
"owner": "cachix",
"repo": "git-hooks.nix",
"rev": "16ec914f6fb6f599ce988427d9d94efddf25fe6d",
"rev": "f0927703b7b1c8d97511c4116eb9b4ec6645a0fa",
"type": "github"
},
"original": {
@ -60,10 +60,10 @@
]
},
"locked": {
"lastModified": 1709087332,
"lastModified": 1762808025,
"owner": "hercules-ci",
"repo": "gitignore.nix",
"rev": "637db329424fd7e46cf4185293b9cc8c88c95394",
"rev": "cb5e3fdca1de58ccbc3ef53de65bd372b48f567c",
"type": "github"
},
"original": {
@ -74,10 +74,10 @@
},
"nixpkgs": {
"locked": {
"lastModified": 1753432016,
"lastModified": 1767995494,
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "6027c30c8e9810896b92429f0092f624f7b1aace",
"rev": "45a1530683263666f42d1de4cdda328109d5a676",
"type": "github"
},
"original": {

View file

@ -15,6 +15,12 @@
pkgs.samply # code profiler
pkgs.feedgnuplot # to visualize bots training results
# --- AI training with python ---
# generate python classes from rust code
pkgs.maturin
# required by python numpy
pkgs.libz
# for bevy
pkgs.alsa-lib
pkgs.udev
@ -47,6 +53,25 @@
# https://devenv.sh/languages/
languages.rust.enable = true;
# AI training with python
enterShell = ''
PYTHONPATH=$PYTHONPATH:$PWD/.devenv/state/venv/lib/python3/site-packages
'';
languages.python = {
enable = true;
uv.enable = true;
venv.enable = true;
venv.requirements = "
pip
gymnasium
numpy
stable-baselines3
shimmy
";
};
# https://devenv.sh/scripts/
# scripts.hello.exec = "echo hello from $GREET";

View file

@ -53,6 +53,10 @@ Client
### Epic : Bot
- PGX
- https://joe-antognini.github.io/ml/jax-tic-tac-toe
- https://www.sotets.uk/pgx/api_usage/
- OpenAi gym
- doc gymnasium <https://gymnasium.farama.org/introduction/basic_usage/>
- Rust implementation for OpenAi gym <https://github.com/MathisWellmann/gym-rs>

31
doc/python.md Normal file
View file

@ -0,0 +1,31 @@
# Python bindings
## Génération bindings
```sh
# Generate trictrac python lib as a wheel
maturin build -m store/Cargo.toml --release
# Install wheel in local python env
pip install --no-deps --force-reinstall --prefix .devenv/state/venv target/wheels/*.whl
```
## Usage
Pour vérifier l'accès à la lib : lancer le shell interactif `python`
```python
Python 3.13.11 (main, Dec 5 2025, 16:06:33) [GCC 15.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import store
>>> game = store.TricTrac()
>>> game.get_active_player_id()
1
```
### Appels depuis python
`python bot/python/test.py`
## Interfaces
## Entraînement

View file

@ -1,19 +0,0 @@
# Description du projet
Je développe un jeu de TricTrac (<https://fr.wikipedia.org/wiki/Trictrac>) dans le langage rust.
Pour le moment je me concentre sur l'application en ligne de commande simple, donc ne t'occupe pas des dossiers 'client_bevy', 'client_tui', et 'server' qui ne seront utilisés que pour de prochaines évolutions.
Les règles du jeu et l'état d'une partie sont implémentées dans 'store', l'application ligne de commande est implémentée dans 'client_cli', elle permet déjà de jouer contre un bot, ou de faire jouer deux bots l'un contre l'autre.
Les stratégies de bots sont implémentées dans le dossier 'bot'.
Plus précisément, l'état du jeu est défini par le struct GameState dans store/src/game.rs, la méthode to_string_id() permet de coder cet état de manière compacte dans une chaîne de caractères, mais il n'y a pas l'historique des coups joués. Il y a aussi fmt::Display d'implémenté pour une representation textuelle plus lisible.
'client_cli/src/game_runner.rs' contient la logique permettant de faire jouer deux bots l'un contre l'autre.
'bot/src/strategy/default.rs' contient le code d'une stratégie de bot basique : il détermine la liste des mouvements valides (avec la méthode get_possible_moves_sequences de store::MoveRules) et joue simplement le premier de la liste.
Je cherche maintenant à ajouter des stratégies de bot plus fortes en entrainant un agent/bot par reinforcement learning.
J'utilise la bibliothèque burn (<https://burn.dev/>).
Une version utilisant l'algorithme DQN peut être lancée avec `cargo run --bin=burn_train -- dqn`). Elle effectue un entraînement, sauvegarde les données du modèle obtenu puis recharge le modèle depuis le disque pour tester l'agent. L'entraînement est fait dans la fonction 'run' du fichier bot/src/burnrl/dqn_model.rs, la sauvegarde du modèle dans la fonction 'save_model' et le chargement dans la fonction 'load_model'.
J'essaie de faire l'équivalent avec les algorithmes PPO (fichier bot/src/burnrl/ppo_model.rs) et SAC (fichier bot/src/burnrl/sac_model.rs) : les fonctions 'run' sont implémentées mais pas les fonctions 'save_model' et 'load_model'. Peux-tu les implémenter ?

View file

@ -20,6 +20,7 @@ profile:
cargo build --profile profiling
samply record ./target/profiling/client_cli --bot dummy,dummy
pythonlib:
rm -rf target/wheels
maturin build -m store/Cargo.toml --release
pip install --no-deps --force-reinstall --prefix .devenv/state/venv target/wheels/*.whl
trainbot algo:

View file

@ -7,14 +7,17 @@ edition = "2021"
[lib]
name = "store"
# "cdylib" is necessary to produce a shared library for Python to import from.
# Only "rlib" is needed for other Rust crates to use this library
crate-type = ["rlib"]
crate-type = ["cdylib", "rlib"]
[dependencies]
base64 = "0.21.7"
# provides macros for creating log messages to be used by a logger (for example env_logger)
log = "0.4.20"
merge = "0.1.0"
# generate python lib (with maturin) to be used in AI training
pyo3 = { version = "0.23", features = ["extension-module", "abi3-py38"] }
rand = "0.8.5"
serde = { version = "1.0", features = ["derive"] }
transpose = "0.2.2"

View file

@ -16,3 +16,6 @@ pub use board::CheckerMove;
mod dice;
pub use dice::{Dice, DiceRoller};
// python interface "trictrac_engine" (for AI training..)
mod pyengine;

View file

@ -1,9 +1,11 @@
use pyo3::prelude::*;
use serde::{Deserialize, Serialize};
use std::fmt;
// This just makes it easier to dissern between a player id and any ol' u64
pub type PlayerId = u64;
#[pyclass(eq, eq_int)]
#[derive(Copy, Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub enum Color {
White,

337
store/src/pyengine.rs Normal file
View file

@ -0,0 +1,337 @@
//! # Expose trictrac game state and rules in a python module
use pyo3::prelude::*;
use pyo3::types::PyDict;
use crate::board::CheckerMove;
use crate::dice::Dice;
use crate::game::{GameEvent, GameState, Stage, TurnStage};
use crate::game_rules_moves::MoveRules;
use crate::game_rules_points::PointsRules;
use crate::player::{Color, PlayerId};
#[pyclass]
struct TricTrac {
game_state: GameState,
dice_roll_sequence: Vec<(u8, u8)>,
current_dice_index: usize,
}
#[pymethods]
impl TricTrac {
#[new]
fn new() -> Self {
let mut game_state = GameState::new(false); // schools_enabled = false
// Initialiser 2 joueurs
game_state.init_player("player1");
game_state.init_player("bot");
// Commencer la partie avec le joueur 1
game_state.consume(&GameEvent::BeginGame { goes_first: 1 });
TricTrac {
game_state,
dice_roll_sequence: Vec::new(),
current_dice_index: 0,
}
}
/// Obtenir l'état du jeu sous forme de chaîne de caractères compacte
fn get_state_id(&self) -> String {
self.game_state.to_string_id()
}
/// Obtenir l'état du jeu sous forme de dictionnaire pour faciliter l'entrainement
fn get_state_dict(&self) -> PyResult<Py<PyDict>> {
Python::with_gil(|py| {
let state_dict = PyDict::new(py);
// Informations essentielles sur l'état du jeu
state_dict.set_item("active_player", self.game_state.active_player_id)?;
state_dict.set_item("stage", format!("{:?}", self.game_state.stage))?;
state_dict.set_item("turn_stage", format!("{:?}", self.game_state.turn_stage))?;
// Dés
let (dice1, dice2) = self.game_state.dice.values;
state_dict.set_item("dice", (dice1, dice2))?;
// Points des joueurs
if let Some(white_player) = self.game_state.get_white_player() {
state_dict.set_item("white_points", white_player.points)?;
state_dict.set_item("white_holes", white_player.holes)?;
}
if let Some(black_player) = self.game_state.get_black_player() {
state_dict.set_item("black_points", black_player.points)?;
state_dict.set_item("black_holes", black_player.holes)?;
}
// Positions des pièces
let white_positions = self.get_checker_positions(Color::White);
let black_positions = self.get_checker_positions(Color::Black);
state_dict.set_item("white_positions", white_positions)?;
state_dict.set_item("black_positions", black_positions)?;
// État compact pour la comparaison d'états
state_dict.set_item("state_id", self.game_state.to_string_id())?;
Ok(state_dict.into())
})
}
/// Renvoie les positions des pièces pour un joueur spécifique
fn get_checker_positions(&self, color: Color) -> Vec<(usize, i8)> {
self.game_state.board.get_color_fields(color)
}
/// Obtenir la liste des mouvements légaux sous forme de paires (from, to)
fn get_available_moves(&self) -> Vec<((usize, usize), (usize, usize))> {
// L'agent joue toujours le joueur actif
let color = self
.game_state
.player_color_by_id(&self.game_state.active_player_id)
.unwrap_or(Color::White);
// Si ce n'est pas le moment de déplacer les pièces, retourner une liste vide
if self.game_state.turn_stage != TurnStage::Move
&& self.game_state.turn_stage != TurnStage::HoldOrGoChoice
{
return vec![];
}
let rules = MoveRules::new(&color, &self.game_state.board, self.game_state.dice);
let possible_moves = rules.get_possible_moves_sequences(true, vec![]);
// Convertir les mouvements CheckerMove en tuples (from, to) pour Python
possible_moves
.into_iter()
.map(|(move1, move2)| {
(
(move1.get_from(), move1.get_to()),
(move2.get_from(), move2.get_to()),
)
})
.collect()
}
/// Jouer un coup ((from1, to1), (from2, to2))
fn play_move(&mut self, moves: ((usize, usize), (usize, usize))) -> bool {
let ((from1, to1), (from2, to2)) = moves;
// Vérifier que c'est au tour du joueur de jouer
if self.game_state.turn_stage != TurnStage::Move
&& self.game_state.turn_stage != TurnStage::HoldOrGoChoice
{
return false;
}
let move1 = CheckerMove::new(from1, to1).unwrap_or_default();
let move2 = CheckerMove::new(from2, to2).unwrap_or_default();
let event = GameEvent::Move {
player_id: self.game_state.active_player_id,
moves: (move1, move2),
};
// Vérifier si le mouvement est valide
if !self.game_state.validate(&event) {
return false;
}
// Exécuter le mouvement
self.game_state.consume(&event);
// Si l'autre joueur doit lancer les dés maintenant, simuler ce lancement
if self.game_state.turn_stage == TurnStage::RollDice {
self.roll_dice();
}
true
}
/// Lancer les dés (soit aléatoirement, soit en utilisant une séquence prédéfinie)
fn roll_dice(&mut self) -> (u8, u8) {
// Vérifier que c'est au bon moment pour lancer les dés
if self.game_state.turn_stage != TurnStage::RollDice
&& self.game_state.turn_stage != TurnStage::RollWaiting
{
return self.game_state.dice.values;
}
// Simuler un lancer de dés
let dice_values = if !self.dice_roll_sequence.is_empty()
&& self.current_dice_index < self.dice_roll_sequence.len()
{
// Utiliser la séquence prédéfinie
let dice = self.dice_roll_sequence[self.current_dice_index];
self.current_dice_index += 1;
dice
} else {
// Générer aléatoirement
(
(1 + (rand::random::<u8>() % 6)),
(1 + (rand::random::<u8>() % 6)),
)
};
// Envoyer les événements appropriés
let roll_event = GameEvent::Roll {
player_id: self.game_state.active_player_id,
};
if self.game_state.validate(&roll_event) {
self.game_state.consume(&roll_event);
}
let roll_result_event = GameEvent::RollResult {
player_id: self.game_state.active_player_id,
dice: Dice {
values: dice_values,
},
};
if self.game_state.validate(&roll_result_event) {
self.game_state.consume(&roll_result_event);
}
dice_values
}
/// Marquer des points
fn mark_points(&mut self, points: u8) -> bool {
// Vérifier que c'est au bon moment pour marquer des points
if self.game_state.turn_stage != TurnStage::MarkPoints
&& self.game_state.turn_stage != TurnStage::MarkAdvPoints
{
return false;
}
let event = GameEvent::Mark {
player_id: self.game_state.active_player_id,
points,
};
// Vérifier si l'événement est valide
if !self.game_state.validate(&event) {
return false;
}
// Exécuter l'événement
self.game_state.consume(&event);
// Si l'autre joueur doit lancer les dés maintenant, simuler ce lancement
if self.game_state.turn_stage == TurnStage::RollDice {
self.roll_dice();
}
true
}
/// Choisir de "continuer" (Go) après avoir gagné un trou
fn choose_go(&mut self) -> bool {
// Vérifier que c'est au bon moment pour choisir de continuer
if self.game_state.turn_stage != TurnStage::HoldOrGoChoice {
return false;
}
let event = GameEvent::Go {
player_id: self.game_state.active_player_id,
};
// Vérifier si l'événement est valide
if !self.game_state.validate(&event) {
return false;
}
// Exécuter l'événement
self.game_state.consume(&event);
// Simuler le lancer de dés pour le prochain tour
self.roll_dice();
true
}
/// Calcule les points maximaux que le joueur actif peut obtenir avec les dés actuels
fn calculate_points(&self) -> u8 {
let active_player = self
.game_state
.players
.get(&self.game_state.active_player_id);
if let Some(player) = active_player {
let dice_roll_count = player.dice_roll_count;
let color = player.color;
let points_rules =
PointsRules::new(&color, &self.game_state.board, self.game_state.dice);
let (points, _) = points_rules.get_points(dice_roll_count);
points
} else {
0
}
}
/// Réinitialise la partie
fn reset(&mut self) {
self.game_state = GameState::new(false);
// Initialiser 2 joueurs
self.game_state.init_player("player1");
self.game_state.init_player("bot");
// Commencer la partie avec le joueur 1
self.game_state
.consume(&GameEvent::BeginGame { goes_first: 1 });
// Réinitialiser l'index de la séquence de dés
self.current_dice_index = 0;
}
/// Vérifie si la partie est terminée
fn is_done(&self) -> bool {
self.game_state.stage == Stage::Ended || self.game_state.determine_winner().is_some()
}
/// Obtenir le gagnant de la partie
fn get_winner(&self) -> Option<PlayerId> {
self.game_state.determine_winner()
}
/// Obtenir le score du joueur actif (nombre de trous)
fn get_score(&self, player_id: PlayerId) -> i32 {
if let Some(player) = self.game_state.players.get(&player_id) {
player.holes as i32
} else {
-1
}
}
/// Obtenir l'ID du joueur actif
fn get_active_player_id(&self) -> PlayerId {
self.game_state.active_player_id
}
/// Définir une séquence de dés à utiliser (pour la reproductibilité)
fn set_dice_sequence(&mut self, sequence: Vec<(u8, u8)>) {
self.dice_roll_sequence = sequence;
self.current_dice_index = 0;
}
/// Afficher l'état du jeu (pour le débogage)
fn __str__(&self) -> String {
format!("{}", self.game_state)
}
}
/// A Python module implemented in Rust. The name of this function must match
/// the `lib.name` setting in the `Cargo.toml`, else Python will not be able to
/// import the module.
#[pymodule]
fn store(m: &Bound<'_, PyModule>) -> PyResult<()> {
m.add_class::<TricTrac>()?;
Ok(())
}