feat(model-server): afficher le téléchargement du modèle llamacpp au démarrage (#54)

Ajoute un handle du téléchargement des modèles lors du démarrage de
llamacpp : le domaine et l'application émettent la progression de
téléchargement du modèle, relayée en événement côté app-tauri, et l'UI
l'affiche via un badge de lancement et un overlay de cellule pendant que
le serveur de modèle démarre.

Backend (B1) : progression de téléchargement dans domain/application,
relais d'événement app-tauri, couverture de tests.
Frontend (F1) : modelServerLaunch, badge et overlay LayoutGrid, tests.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
2026-07-13 23:14:05 +02:00
parent c67ec4f7bd
commit fe7ed0aa20
12 changed files with 891 additions and 29 deletions

View File

@ -12,8 +12,9 @@ use application::{
};
use domain::events::DomainEvent;
use domain::model_server::{
ExecutablePath, LlamaCppOptions, LocalModelRef, LocalModelServerConfig, LocalModelServerKind,
ModelPath, ModelServerEndpoint, ModelServerStatus, ModelSource, StopPolicy,
ExecutablePath, HfModelRef, LlamaCppOptions, LocalModelRef, LocalModelServerConfig,
LocalModelServerKind, ModelPath, ModelServerEndpoint, ModelServerLifecycleStatus,
ModelServerStatus, ModelSource, StopPolicy,
};
use domain::ports::{
DirEntry, EventBus, EventStream, FileSystem, FsError, ManagedProcess, ManagedProcessHandle,
@ -56,6 +57,30 @@ fn config(
.unwrap()
}
fn hf_config(id: LocalModelServerId, port: u16, repo: &str) -> LocalModelServerConfig {
LocalModelServerConfig::new(
id,
LocalModelServerKind::LlamaCpp,
"llama.cpp",
ModelServerEndpoint::new(format!("http://localhost:{port}"), port).unwrap(),
LocalModelRef::new(
"qwen",
"Qwen",
Some(ModelSource::HuggingFace {
repo: HfModelRef::new(repo).unwrap(),
}),
"qwen3-coder-30b",
)
.unwrap(),
Some(ExecutablePath::new("llama-server").unwrap()),
LlamaCppOptions::default(),
Vec::new(),
true,
StopPolicy::StopOnAppExit,
)
.unwrap()
}
#[derive(Default)]
struct FakeRegistry(Mutex<HashMap<LocalModelServerId, LocalModelServerConfig>>);
@ -158,6 +183,7 @@ struct FakeProcess {
spawns: Mutex<Vec<SpawnSpec>>,
kills: Mutex<Vec<String>>,
statuses: Mutex<HashMap<String, ProcessStatus>>,
status_sequence: Mutex<VecDeque<ProcessStatus>>,
spawn_delay: Duration,
}
@ -187,6 +213,9 @@ impl ManagedProcess for FakeProcess {
&self,
handle: &ManagedProcessHandle,
) -> Result<ProcessStatus, ModelServerError> {
if let Some(status) = self.status_sequence.lock().unwrap().pop_front() {
return Ok(status);
}
Ok(*self
.statuses
.lock()
@ -203,25 +232,35 @@ impl ModelServerRuntime for FakeRuntime {
&self,
config: &LocalModelServerConfig,
) -> Result<ModelServerArgv, ModelServerError> {
let Some(ModelSource::LocalPath { path }) = config.model.source.as_ref() else {
return Err(ModelServerError::PathNotAccessible(
"model.source missing".to_owned(),
));
};
let mut args = Vec::new();
match config
.model
.source
.as_ref()
.ok_or_else(|| ModelServerError::PathNotAccessible("model.source missing".to_owned()))?
{
ModelSource::LocalPath { path } => {
args.push("--model".to_owned());
args.push(path.as_str().to_owned());
}
ModelSource::HuggingFace { repo } => {
args.push("-hf".to_owned());
args.push(repo.as_str().to_owned());
}
}
args.extend([
"--port".to_owned(),
config.endpoint.port.to_string(),
"--host".to_owned(),
config.options.host.clone(),
]);
Ok(ModelServerArgv {
command: config
.binary
.as_ref()
.map(|binary| binary.as_str().to_owned())
.unwrap_or_else(|| "llama-server".to_owned()),
args: vec![
"--model".to_owned(),
path.as_str().to_owned(),
"--port".to_owned(),
config.endpoint.port.to_string(),
"--host".to_owned(),
config.options.host.clone(),
],
args,
})
}
@ -303,6 +342,7 @@ fn ensure(
attempts: 2,
backoff: Duration::ZERO,
})
.with_hf_download_deadline(Duration::from_secs(5))
}
#[tokio::test]
@ -546,6 +586,158 @@ async fn readiness_timeout_kills_started_process() {
assert_eq!(process.kills.lock().unwrap().as_slice(), ["h1"]);
}
#[tokio::test]
async fn hf_unreachable_alive_publishes_downloading_without_short_timeout_kill() {
let registry = Arc::new(FakeRegistry::default());
registry
.save(hf_config(
sid(11),
8091,
"Qwen/Qwen3-Coder-30B-A3B-Instruct-GGUF",
))
.await
.unwrap();
let process = Arc::new(FakeProcess::default());
let events = Arc::new(FakeEvents::default());
let usecase = ensure(
Arc::clone(&registry),
Arc::new(FakeProbe::new(vec![
ModelServerStatus::Unreachable,
ModelServerStatus::Unreachable,
ModelServerStatus::Unreachable,
ModelServerStatus::ReadyReused,
])),
Arc::clone(&process),
Arc::new(FakeFs::default()),
Arc::clone(&events),
);
let out = usecase
.execute(EnsureLocalModelServerInput { server_id: sid(11) })
.await
.unwrap();
assert_eq!(out.ready.status, ModelServerStatus::ReadyStarted);
assert!(process.kills.lock().unwrap().is_empty());
assert!(events.0.lock().unwrap().iter().any(|event| {
matches!(
event,
DomainEvent::ModelServerStatusChanged {
server_id,
status: ModelServerLifecycleStatus::Downloading {
downloaded_bytes: None,
total_bytes: None,
percent: None,
source: Some(source),
},
} if *server_id == sid(11)
&& source == "Qwen/Qwen3-Coder-30B-A3B-Instruct-GGUF"
)
}));
}
#[tokio::test]
async fn hf_download_phase_reports_ready_when_probe_becomes_ok() {
let registry = Arc::new(FakeRegistry::default());
registry
.save(hf_config(
sid(12),
8092,
"Qwen/Qwen3-Coder-30B-A3B-Instruct-GGUF",
))
.await
.unwrap();
let process = Arc::new(FakeProcess::default());
let events = Arc::new(FakeEvents::default());
let usecase = ensure(
Arc::clone(&registry),
Arc::new(FakeProbe::new(vec![
ModelServerStatus::Unreachable,
ModelServerStatus::Unreachable,
ModelServerStatus::Unreachable,
ModelServerStatus::ReadyStarted,
])),
Arc::clone(&process),
Arc::new(FakeFs::default()),
Arc::clone(&events),
);
let out = usecase
.execute(EnsureLocalModelServerInput { server_id: sid(12) })
.await
.unwrap();
assert_eq!(out.ready.status, ModelServerStatus::ReadyStarted);
let statuses: Vec<ModelServerLifecycleStatus> = events
.0
.lock()
.unwrap()
.iter()
.filter_map(|event| match event {
DomainEvent::ModelServerStatusChanged { status, .. } => Some(status.clone()),
_ => None,
})
.collect();
assert!(matches!(
statuses.as_slice(),
[
ModelServerLifecycleStatus::Probing,
ModelServerLifecycleStatus::Starting,
ModelServerLifecycleStatus::Downloading { .. },
ModelServerLifecycleStatus::Ready { reused: false },
]
));
}
#[tokio::test]
async fn hf_download_phase_fails_when_process_exits() {
let registry = Arc::new(FakeRegistry::default());
registry
.save(hf_config(
sid(13),
8093,
"Qwen/Qwen3-Coder-30B-A3B-Instruct-GGUF",
))
.await
.unwrap();
let process = Arc::new(FakeProcess::default());
process
.status_sequence
.lock()
.unwrap()
.push_back(ProcessStatus::Exited { code: Some(42) });
let events = Arc::new(FakeEvents::default());
let usecase = ensure(
Arc::clone(&registry),
Arc::new(FakeProbe::new(vec![
ModelServerStatus::Unreachable,
ModelServerStatus::Unreachable,
ModelServerStatus::Unreachable,
])),
Arc::clone(&process),
Arc::new(FakeFs::default()),
Arc::clone(&events),
);
let err = usecase
.execute(EnsureLocalModelServerInput { server_id: sid(13) })
.await
.unwrap_err();
assert!(err.to_string().contains("process"));
assert!(err.to_string().contains("42"));
assert!(process.kills.lock().unwrap().is_empty());
assert!(events.0.lock().unwrap().iter().any(|event| {
matches!(
event,
DomainEvent::ModelServerStatusChanged {
status: ModelServerLifecycleStatus::Failed { code, .. },
..
} if code == "process"
)
}));
}
#[tokio::test]
async fn missing_registry_entry_is_model_server_not_configured() {
let usecase = ensure(