Re-enable VectorRAG init with lazy retry

Personal Docs (POST /api/personal/add_directory and friends) currently
returns HTTP 503 'RAG system is not available' for every request,
because get_rag_manager() and rag_manager are both hardcoded off. The
disablement was added when chromadb 1.4.1 / pydantic 2.12 were mutually
incompatible at the client init layer.

That compat issue is fixed in the current pins (chromadb 1.5.x +
pydantic 2.13.x). Verified by calling the original lazy initializer
against a running chroma server — VectorRAG instantiates, reports
healthy=True, and indexes successfully.

This change:

1. src/rag_singleton.py — replace the hardcoded `return None` in
   get_rag_manager() with the original lazy init body. Keeps the
   30s retry-throttle so a missing chroma server doesn't busy-retry
   on every request.

2. app.py — replace the parallel `rag_manager = None` /
   `rag_available = False` hardcoding with a get_rag_manager() call.
   Logs the resolved state at startup. If chroma isn't reachable yet,
   rag_manager stays None and personal-doc routes still return 503,
   but the *next* request will hit the retry-throttle path in
   get_rag_manager() and try to init again.

Doesn't touch requirements.txt. Repos using docker-compose get chroma
automatically; manual installs that want Personal Docs to work still
need to either pip install chromadb (full package) and run `chroma run`
or point at an external chroma instance via env. That can be a
follow-up README / requirements-optional note.

app.py

@@ -355,15 +355,26 @@ async def serve_generated_image(filename: str, request: Request):
 from services.youtube import init_youtube
 init_youtube()
 
-# ========= RAG (vector document RAG — DISABLED) =========
+# ========= RAG (vector document RAG) =========
-# VectorRAG (ChromaDB-backed personal-document semantic search) is unused
+# VectorRAG (ChromaDB-backed personal-document semantic search). Initialized
-# (0 directories ever indexed) and its chromadb 1.4.1 / pydantic 2.12 client
+# lazily via get_rag_manager() — returns None if ChromaDB isn't reachable
-# can't even instantiate — it threw at init and cost ~30s of startup waiting on
+# (no server running on the configured host:port), in which case personal-doc
-# the embedding probe. Disabled. All callers already guard on rag_available /
+# routes return a clean 503 instead of busy-retrying every request.
-# `if rag_manager`, so personal-doc routes degrade cleanly.
+#
-rag_manager = None
+# Note: this was previously hardcoded off because chromadb 1.4.1 / pydantic
-rag_available = False
+# 2.12 were mutually incompatible at the time. With the current pins
-logger.info("Vector document RAG disabled (unused)")
+# (chromadb 1.5.x + pydantic 2.13.x) the init works and Personal Docs
+# (POST /api/personal/add_directory etc.) is functional again.
+from src.rag_singleton import get_rag_manager
+rag_manager = get_rag_manager()
+rag_available = rag_manager is not None
+if rag_available:
+    logger.info("Vector document RAG initialized")
+else:
+    logger.info(
+        "Vector document RAG not available at startup "
+        "(ChromaDB may not be reachable yet — routes will retry lazily)"
+    )
 
 # ========= IMPORT CONFIG =========
 from src.config import config

src/rag_singleton.py

@@ -12,16 +12,21 @@ rag_instance = None
 _last_attempt = 0.0
 _RETRY_INTERVAL = 30  # seconds between re-init attempts
 
+
 def get_rag_manager():
-    """Disabled: vector document RAG (VectorRAG/ChromaDB) is unused and its
+    """Lazy ChromaDB-backed VectorRAG initializer.
-    client is incompatible with the installed pydantic. Return None so personal-
-    doc routes fall back to non-vector behavior instead of re-attempting (and
-    re-hanging on) a broken ChromaDB init every 30s."""
-    return None
 
+    Returns the VectorRAG instance on first successful init, None if ChromaDB
+    isn't reachable / available. Failed init attempts are throttled to once
+    per _RETRY_INTERVAL seconds so a missing ChromaDB doesn't busy-retry on
+    every request — callers (personal-doc routes etc.) get None back and
+    return a clean 503 to the user instead.
 
-def _get_rag_manager_legacy():
+    Historical note: this used to be hardcoded to ``return None`` with a
-    """Original lazy initializer, kept for reference / easy re-enable."""
+    comment about chromadb 1.4.1 / pydantic 2.12 being mutually incompatible.
+    That compat issue is resolved in current pinned versions
+    (chromadb 1.5.x + pydantic 2.13.x), so the real initializer is back.
+    """
     global rag_instance, _last_attempt
 
     if rag_instance is not None:
@@ -29,7 +34,7 @@ def _get_rag_manager_legacy():
 
     now = time.monotonic()
     if now - _last_attempt < _RETRY_INTERVAL:
-        return None  # too soon to retry
+        return None  # too soon to retry — last attempt failed
 
     _last_attempt = now