feat(memory): embedders vectoriels réels HTTP + ONNX derrière features (LOT C1)

Remplace les StubEmbedder pour les stratégies localServer/api/localOnnx par de
vrais moteurs, chacun derrière une feature cargo off-by-default — la posture
fondatrice « rien d'imposé, zéro dépendance » (défaut none → rappel naïf) reste
byte-for-byte inchangée.

C1a (feature vector-http, reqwest rustls optional):
- HttpEmbedder couvrant localServer (Ollama/llama.cpp) et api (OpenAI/Voyage…),
  payload OpenAI-compatible /v1/embeddings, ordre restauré par index, bearer
  token lu via env var (jamais en clair), timeout client 30s.
- detect_ollama() pour la détection de l'existant (C3).

C1b (feature vector-onnx, fastembed v5 optional):
- OnnxEmbedder en-process (e5-small, dim 384), init paresseuse + spawn_blocking,
  cache modèle sous <app_data>/embedders/onnx — aucun download au build ni au
  first-run, uniquement à la demande au 1er embed.
- Catalogue RECOMMENDED_ONNX_MODELS + ONNX_CACHE_SUBDIR + onnx_model_is_cached
  exposés (sans feature) pour la config (C2) et la popup (C3).

embedder_from_profile(profile, onnx_cache_dir) dispatche feature-gated ; sans la
feature, retombe sur StubEmbedder (Unsupported) → fallback naïf via
AdaptiveMemoryRecall. Composition root (build_memory_recall) propage le cache dir.

Tests: 10 HTTP + 6 ONNX (dont 2 #[ignore] download réel) + 26 vectoriels, verts
en défaut, --features vector-http et --features vector-onnx.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
2026-06-08 19:00:40 +02:00
parent b39c11a64d
commit 32398827fb
10 changed files with 2012 additions and 50 deletions

View File

@ -31,6 +31,12 @@ serde_json = { workspace = true }
thiserror = { workspace = true }
uuid = { workspace = true }
[features]
# Passthrough toggles to enable the real embedders in an IDE build. OFF by default
# (founding posture: `none` ⇒ naïve recall, zero dependency).
vector-http = ["infrastructure/vector-http"]
vector-onnx = ["infrastructure/vector-onnx"]
[dev-dependencies]
uuid = { workspace = true }
async-trait = { workspace = true }

View File

@ -35,7 +35,7 @@ use infrastructure::{
FsEmbedderProfileStore, FsMemoryStore, FsOrchestratorWatcher, FsProfileStore, FsProjectStore,
FsSkillStore, FsTemplateStore, Git2Repository, IdeaiContextStore, LocalFileSystem,
LocalProcessSpawner, NaiveMemoryRecall, OrchestratorWatchHandle, PortablePtyAdapter,
SystemClock, TokioBroadcastEventBus, UuidGenerator, VectorMemoryRecall,
SystemClock, TokioBroadcastEventBus, UuidGenerator, VectorMemoryRecall, ONNX_CACHE_SUBDIR,
};
use crate::pty::PtyBridge;
@ -393,10 +393,12 @@ impl AppState {
.flatten()
})
.unwrap_or_else(EmbedderProfile::none);
let onnx_cache_dir = app_data_dir.join(ONNX_CACHE_SUBDIR);
let memory_recall_port = build_memory_recall(
Arc::clone(&fs_port),
Arc::clone(&memory_store_port),
&embedder_profile,
&onnx_cache_dir,
);
let create_agent = Arc::new(CreateAgentFromScratch::new(
@ -711,9 +713,10 @@ pub(crate) fn build_memory_recall(
fs: Arc<dyn FileSystem>,
store: Arc<dyn MemoryStore>,
profile: &EmbedderProfile,
onnx_cache_dir: &std::path::Path,
) -> Arc<dyn MemoryRecall> {
let naive = Arc::new(NaiveMemoryRecall::new(Arc::clone(&store))) as Arc<dyn MemoryRecall>;
match embedder_from_profile(profile) {
match embedder_from_profile(profile, onnx_cache_dir) {
None => naive,
Some(embedder) => {
let embedder: Arc<dyn Embedder> = Arc::from(embedder);
@ -839,7 +842,12 @@ mod tests {
async fn build_memory_recall_none_profile_matches_naive_recall() {
let (fs, store) = seed_store().await;
let wired = build_memory_recall(Arc::clone(&fs), Arc::clone(&store), &EmbedderProfile::none());
let wired = build_memory_recall(
Arc::clone(&fs),
Arc::clone(&store),
&EmbedderProfile::none(),
std::path::Path::new("/unused-onnx-cache"),
);
let naive = NaiveMemoryRecall::new(Arc::clone(&store));
// 0 ⇒ empty; mid budgets ⇒ partial truncation; huge ⇒ full index order.