bb16cd6927
refactor: extract eviction engine into eviction.py (rag.py 303→109 lines)
...
- Move all eviction logic (evict_batch, maybe_evict, EVICTION_LOG,
get_collection_count/stats, get_rag_operational_stats) into eviction.py
- Move QDRANT_URL, RAG_COLLECTION into config.py to break circular dep
- rag.py re-exports eviction symbols for backward compatibility
- Router imports updated to use eviction module directly
- All 130 tests pass
2026-07-06 08:00:26 -07:00
8072fb3dd0
feat: Roadmap K — RAG corpus management with score-based eviction (v0.13.0)
...
- Config: RAG_MAX_VECTORS, high/low water marks, grace period, weights
- rag.py: get_collection_count, evict_batch, maybe_evict (asyncio.Lock),
get_rag_operational_stats, EVICTION_LOG, retrieval_count tracking
- routers/rag_admin.py: GET /api/rag/stats, POST /api/rag/flush (admin)
- Wire maybe_evict() into upload.py and ingest.py after Qdrant upsert
- 16 tests: collection stats, eviction scoring, pinned/grace/batch guards,
endpoint auth, race lock, flush, operational stats shape
- Bump to v0.13.0
2026-07-06 07:56:09 -07:00
4a891c8435
v1.8.9 -> v1.9.0: file upload backend (PDF/text, Qdrant ingest, SQLite context)
2026-07-01 18:15:23 -07:00
66b086c3f3
fix: restore EMBED_URL pointing to ollama on 192.168.50.210:11434
2026-06-27 16:03:19 -07:00
4b36fd315a
fix: replace hardcoded EMBED_URL with LLAMA_SERVER_BASE from config
...
EMBED_URL in rag.py hardcoded the IP and port instead of using
LLAMA_SERVER_BASE, so the env var JARVISCHAT_LLAMA_SERVER_BASE
was ignored for embedding requests.
2026-06-27 15:59:43 -07:00
cc1efa7a21
fix: resolve all critical runtime errors and bugs from audit
...
- Add COMPLETIONS_API_KEY to config.py (env var + auto-generated fallback)
- Fix perplexity auto-search: upstream sends logprobs=true, parse_llama_stream_chunk
extracts per-token logprobs, all_logprobs populated during streaming
- Fix all /api/models endpoints to target LLAMA_SERVER_BASE (port 8081) not OLLAMA_BASE
- Fix RAG embedding endpoint URL from port 11434 (Ollama) to 8081 (llama-server)
- Correct misleading error messages: 'inference server' not 'Ollama'
- Remove raw_results leak from SSE event stream in /api/search
- Fix weather query extractor: pattern-match instead of unconditional suffix append
- Escape FTS5 operator keywords (AND/OR/NOT/NEAR) in memory search
- Move auth.py BODY_LIMIT_DEFAULT_BYTES imports to module level
- Change RAG injection log level from warning to info
- Fix all 8 test files after modular refactor (rewire imports from correct modules)
- Update AGENTS.md and README.md to reflect v1.8.0 changes
2026-06-27 15:12:18 -07:00
gramps
d01dd3b761
refactor(arch): modular package structure — split monolithic app.py into config/db/auth/memory/search/rag/gpu + routers/
...
- config.py: all constants, env vars, limits, skill registry, profiles
- db.py: schema init, connection factory, skill state helpers
- security.py: PIN hashing, audit logging, rate limiting, CSRF, request helpers
- auth.py: session management, PIN verify, auth routes
- memory.py: FTS5 CRUD + remember/forget command processing
- search.py: SearXNG integration, perplexity scoring, refusal/hedge detection
- gpu.py: rocm-smi stats
- rag.py: Qdrant vector search + system prompt assembly
- routers/: conversations, memories, models, presets, profile, settings, skills, chat, search
- app.py: slim entry point, middleware, router registration only
Bumps to v1.9.0
2026-06-16 08:17:46 -07:00
gramps
gramps