feat: add production config and readiness checks

.env.example

@@ -0,0 +1,17 @@
+PROJECT_NAME="Support Docs Copilot"
+OLLAMA_BASE_URL=http://127.0.0.1:11434
+OLLAMA_MODEL=llama3
+QDRANT_URL=
+QDRANT_LOCATION=./qdrant_data
+COLLECTION_NAME=support_docs
+RETRIEVAL_MODE=hybrid
+RETRIEVAL_TOP_K=15
+RERANKER_TOP_N=3
+RERANKER_ENABLED=true
+CHUNK_SIZE=500
+CHUNK_OVERLAP=50
+MIN_RELEVANCE_SCORE=0.0
+MAX_CONTEXT_CHARS=12000
+ENABLE_GUARDRAILS=true
+ENABLE_RAG_EVAL=false
+MAX_QUERY_LENGTH=2000

.gitignore

@@ -15,7 +15,7 @@ frontend.log
 backend.log
 fastembed_cache/
 .cache/
-model files
-Hugging Face cache
-Ollama models
+reports/*.html
+reports/*.json
+data/document_registry.json
 

app/__init__.py

@@ -0,0 +1 @@
+

app/core/__init__.py

@@ -0,0 +1 @@
+

app/core/config.py

@@ -1,16 +1,15 @@
-import os
 from pydantic_settings import BaseSettings
 
 class Settings(BaseSettings):
     PROJECT_NAME: str = "Support Docs Copilot"
     
     # Ollama LLM Config
-    OLLAMA_BASE_URL: str = os.getenv("OLLAMA_BASE_URL", "http://ollama:11434")
-    OLLAMA_MODEL: str = os.getenv("OLLAMA_MODEL", "llama3")
+    OLLAMA_BASE_URL: str = "http://127.0.0.1:11434"
+    OLLAMA_MODEL: str = "llama3"
     
     # Qdrant Vector DB Config
-    QDRANT_URL: str = os.getenv("QDRANT_URL", "")
-    QDRANT_LOCATION: str = os.getenv("QDRANT_LOCATION", "./qdrant_data")
+    QDRANT_URL: str = ""
+    QDRANT_LOCATION: str = "./qdrant_data"
     COLLECTION_NAME: str = "support_docs"
     
     # Embeddings Config
@@ -18,4 +17,25 @@ class Settings(BaseSettings):
     SPARSE_EMBEDDING_MODEL: str = "Qdrant/bm25"
     RERANKER_MODEL: str = "BAAI/bge-reranker-base"
 
+    # Retrieval Config
+    RETRIEVAL_MODE: str = "hybrid"
+    RETRIEVAL_TOP_K: int = 15
+    RERANKER_TOP_N: int = 3
+    RERANKER_ENABLED: bool = True
+    MIN_RELEVANCE_SCORE: float = 0.0
+    MAX_CONTEXT_CHARS: int = 12000
+
+    # Chunking Config
+    CHUNK_SIZE: int = 500
+    CHUNK_OVERLAP: int = 50
+
+    # Runtime Safety
+    ENABLE_GUARDRAILS: bool = True
+    ENABLE_RAG_EVAL: bool = False
+    MAX_QUERY_LENGTH: int = 2000
+
+    class Config:
+        env_file = ".env"
+        extra = "ignore"
+
 settings = Settings()

app/core/dependencies.py

@@ -0,0 +1,34 @@
+from typing import Any
+
+import requests
+from qdrant_client import QdrantClient
+
+from app.core.config import settings
+
+
+def get_qdrant_client() -> QdrantClient:
+    if settings.QDRANT_URL:
+        return QdrantClient(url=settings.QDRANT_URL)
+    return QdrantClient(path=settings.QDRANT_LOCATION)
+
+
+def check_ollama() -> dict[str, Any]:
+    try:
+        response = requests.get(f"{settings.OLLAMA_BASE_URL}/api/tags", timeout=3)
+        return {"ok": response.ok, "status_code": response.status_code}
+    except requests.RequestException as exc:
+        return {"ok": False, "error": str(exc)}
+
+
+def check_qdrant() -> dict[str, Any]:
+    try:
+        client = get_qdrant_client()
+        collections = client.get_collections()
+        names = [collection.name for collection in collections.collections]
+        return {
+            "ok": settings.COLLECTION_NAME in names,
+            "collection": settings.COLLECTION_NAME,
+            "available_collections": names,
+        }
+    except Exception as exc:
+        return {"ok": False, "error": str(exc)}

app/core/errors.py

@@ -0,0 +1,7 @@
+class CopilotError(Exception):
+    """Base application error with a user-safe message."""
+
+    def __init__(self, message: str, status_code: int = 500):
+        self.message = message
+        self.status_code = status_code
+        super().__init__(message)

app/core/logging.py

@@ -0,0 +1,14 @@
+import logging
+import sys
+
+
+def configure_logging() -> None:
+    logging.basicConfig(
+        level=logging.INFO,
+        format="%(asctime)s %(levelname)s %(name)s %(message)s",
+        handlers=[logging.StreamHandler(sys.stdout)],
+        force=True,
+    )
+
+
+logger = logging.getLogger("support_docs_copilot")

app/engine/__init__.py

@@ -0,0 +1 @@
+

app/graph/__init__.py

@@ -0,0 +1 @@
+

app/guardrails/__init__.py

@@ -0,0 +1 @@
+

app/main.py

@@ -1,4 +1,5 @@
 import asyncio
+import time
 from fastapi import FastAPI, HTTPException
 from fastapi.responses import StreamingResponse
 from pydantic import BaseModel
@@ -7,9 +8,12 @@ from langchain_core.prompts import PromptTemplate
 from langchain_ollama import ChatOllama
 
 from app.core.config import settings
+from app.core.dependencies import check_ollama, check_qdrant
+from app.core.logging import configure_logging, logger
 from app.graph.workflow import compile_workflow
 from app.guardrails.validators import DetectPromptInjection
 
+configure_logging()
 app = FastAPI(title=settings.PROJECT_NAME)
 rag_agent = compile_workflow()
 input_guard = Guard().use(DetectPromptInjection, on_fail="exception")
@@ -21,12 +25,28 @@ class ChatResponse(BaseModel):
     query: str
     answer: str
 
+@app.get("/health")
+async def health_endpoint():
+    return {"status": "ok", "project": settings.PROJECT_NAME}
+
+@app.get("/ready")
+async def ready_endpoint():
+    ollama = check_ollama()
+    qdrant = check_qdrant()
+    return {
+        "ready": bool(ollama.get("ok") and qdrant.get("ok")),
+        "ollama": ollama,
+        "qdrant": qdrant,
+    }
+
 @app.post("/chat", response_model=ChatResponse)
 async def chat_endpoint(request: ChatRequest):
-    try:
-        input_guard.validate(request.query)
-    except Exception as e:
-        raise HTTPException(status_code=400, detail=str(getattr(e, "message", e)))
+    started_at = time.perf_counter()
+    if settings.ENABLE_GUARDRAILS:
+        try:
+            input_guard.validate(request.query)
+        except Exception as e:
+            raise HTTPException(status_code=400, detail=str(getattr(e, "message", e)))
 
     initial_state = {"question": request.query, "run_count": 0}
     try:
@@ -35,16 +55,19 @@ async def chat_endpoint(request: ChatRequest):
     except Exception as e:
         raise HTTPException(status_code=500, detail=str(e))
 
+    logger.info("chat completed latency_ms=%s", round((time.perf_counter() - started_at) * 1000, 2))
     return ChatResponse(query=request.query, answer=answer)
 
 @app.post("/chat/stream")
 async def chat_stream_endpoint(request: ChatRequest):
-    try:
-        input_guard.validate(request.query)
-    except Exception as e:
-        raise HTTPException(status_code=400, detail=str(getattr(e, "message", e)))
+    if settings.ENABLE_GUARDRAILS:
+        try:
+            input_guard.validate(request.query)
+        except Exception as e:
+            raise HTTPException(status_code=400, detail=str(getattr(e, "message", e)))
 
     async def token_generator():
+        started_at = time.perf_counter()
         initial_state = {"question": request.query, "run_count": 0}
         final_state = rag_agent.invoke(initial_state)
         documents = final_state.get("documents", [])
@@ -67,5 +90,6 @@ async def chat_stream_endpoint(request: ChatRequest):
             if chunk.content:
                 yield chunk.content
                 await asyncio.sleep(0.01)
+        logger.info("stream completed latency_ms=%s", round((time.perf_counter() - started_at) * 1000, 2))
 
     return StreamingResponse(token_generator(), media_type="text/event-stream")

app/tests/__init__.py

@@ -0,0 +1 @@
+

ui/app.py

@@ -1,9 +1,11 @@
-import streamlit as st
-import requests
 import os
 
-st.set_page_config(page_title="Support Docs Copilot", page_icon="🤖")
-st.title("🤖 Support Docs Copilot")
+import requests
+import streamlit as st
+
+
+st.set_page_config(page_title="Support Docs Copilot", page_icon="SD")
+st.title("Support Docs Copilot")
 
 BACKEND_URL = os.getenv("BACKEND_URL", "http://127.0.0.1:8000/chat/stream")
 
@@ -23,21 +25,21 @@ if user_query := st.chat_input("Ask a support question..."):
         response_placeholder = st.empty()
         full_response = ""
         try:
-            with requests.post(BACKEND_URL, json={"query": user_query}, stream=True) as response:
+            with requests.post(BACKEND_URL, json={"query": user_query}, stream=True, timeout=120) as response:
                 if response.status_code == 200:
                     for chunk in response.iter_content(chunk_size=None, decode_unicode=True):
                         if chunk:
                             full_response += chunk
-                            response_placeholder.markdown(full_response + "▌")
+                            response_placeholder.markdown(full_response + "...")
                     response_placeholder.markdown(full_response)
                 elif response.status_code == 400:
-                    full_response = f"⚠️ {response.json().get('detail', 'Violation.')}"
+                    full_response = f"Warning: {response.json().get('detail', 'Violation.')}"
                     response_placeholder.error(full_response)
                 else:
-                    full_response = "⚠️ Server communication error."
+                    full_response = "Warning: server communication error."
                     response_placeholder.error(full_response)
-        except:
-            full_response = "❌ Backend connection error."
+        except requests.RequestException:
+            full_response = "Backend connection error."
             response_placeholder.error(full_response)
 
     st.session_state.messages.append({"role": "assistant", "content": full_response})