changes

.gitignore

@@ -0,0 +1 @@
+.env

app.py

@@ -0,0 +1,47 @@
+"""
+FastAPI + Gradio combo app for Hugging Face Space
+"""
+import os, gradio as gr
+from fastapi import FastAPI, Header, HTTPException
+from pydantic import BaseModel
+from rag import retrieve_info     # your existing function
+
+# -------- 1)  Private key list ----------
+VALID_KEYS = set(os.getenv("RAG_KEYS", "alpha,beta").split(","))  # store secret in Space > Settings > Secrets
+
+# -------- 2)  FastAPI core app ----------
+app = FastAPI(title="Moffitt RAG API")
+
+class QueryIn(BaseModel):
+    query: str
+    k: int = 5
+
+@app.post("/v1/query")
+async def rag_query(body: QueryIn, x_api_key: str = Header(None)):
+    """
+    Secure JSON endpoint.
+    Caller must send:  X-API-Key: <one-of-valid-keys>
+    """
+    if x_api_key not in VALID_KEYS:
+        raise HTTPException(status_code=401, detail="Invalid or missing X-API-Key")
+
+    text = retrieve_info(body.query, body.k)
+    return {"answer": text}
+
+# -------- 3)  Public Gradio UI ----------
+def run(q, k):
+    return retrieve_info(q, int(k))
+
+demo = gr.Interface(
+    fn=run,
+    inputs=["text", gr.Number(label="k (Number of chunks to retrieve)")],
+    outputs=gr.Textbox(lines=25, label="Retrieved chunks"),
+    allow_flagging="never",
+    title="Moffitt RAG Demo",
+    description="Type a question; we search Chroma with E5 embeddings."
+)
+
+# Mount Gradio at ROOT path "/"
+if __name__ == "__main__":
+    demo.launch()
+

chroma_db/6eaba295-86d1-4c64-9b6e-f3f0d54f5cac/data_level0.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b97478e7c7ceec00d978669a0594a856756c91815c10abf5a9ec2269162a06a
+size 32120000

chroma_db/6eaba295-86d1-4c64-9b6e-f3f0d54f5cac/header.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9143bae7b81c127da7707ac7626f296457404dd366b7e0ce40d2f900ae21d8cb
+size 100

chroma_db/6eaba295-86d1-4c64-9b6e-f3f0d54f5cac/index_metadata.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af3caa649a549e8e441b1a32fb57b1616f5089ef44d3efc54e9a5737cdd33718
+size 125068

chroma_db/6eaba295-86d1-4c64-9b6e-f3f0d54f5cac/length.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bb6f471778ae0d114debd720ff7831af68fb2737ed858558f828e647981699e
+size 40000

chroma_db/6eaba295-86d1-4c64-9b6e-f3f0d54f5cac/link_lists.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1e1ce593ebbb113202ce280877fdb67ac5ba31a6ab3b82500da6bb417897260
+size 12164

chroma_db/chroma.sqlite3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:811181a70c31391199665bec074d9e108cde48a15c52dec83c6b78cd072e261e
+size 9785344

demo_client.py

@@ -0,0 +1,66 @@
+#!/usr/bin/env python3
+"""
+Simple CLI for the Moffitt RAG Space.
+
+Before running, run the following on your terminal:
+    pip install requests python-dotenv
+
+Then, in the same folder as this file demo_client.py, create a file named .env
+Add in it the following line
+
+RAG_API_KEY=password (replace the actual password here, no "" symbols needed)
+
+Usage:
+    python demo_client.py "What are the common types of cancer" -k 5
+
+Environment variables:
+    RAG_API_KEY   Your X-API-Key header value (e.g. "alpha")
+    RAG_API_URL   (optional) override the default Space URL
+"""
+import argparse
+import os
+import sys
+import requests
+from dotenv import load_dotenv
+
+load_dotenv()
+DEFAULT_URL = "https://buildng-moffitt-rag-demo.hf.space/v1/query"
+
+
+def call_rag_api(question: str, k: int = 5, url: str = DEFAULT_URL) -> str:
+    api_key = os.getenv("RAG_API_KEY")
+    if not api_key:
+        sys.exit("RAG_API_KEY not set in environment")
+
+    payload = {"query": question, "k": k}
+    headers = {
+        "Content-Type": "application/json",
+        "X-API-Key": api_key,
+    }
+
+    response = requests.post(url, json=payload, headers=headers, timeout=30)
+    response.raise_for_status()             # 4xx/5xx → raises HTTPError
+    return response.json()["answer"]
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description="Query the Moffitt RAG API")
+    parser.add_argument("question", help="Your natural-language question")
+    parser.add_argument("-k", type=int, default=5, help="Top-k passages to retrieve (default: 5)")
+    args = parser.parse_args()
+
+    answer = call_rag_api(args.question, args.k, DEFAULT_URL)
+    print("\n=== Answer ===\n")
+    print(answer)
+
+if __name__ == "__main__":
+    main()
+
+"""
+If you want to integrate this with your ChatGPT API,
+Reuse the same call_rag_api functions,
+and when you need it, simply call 
+call_rag_api(question, k, DEFAULT_URL)
+question is a string, your question, like "what is the risk of blood bone marrow transplant?"
+k is an integer. Keep it a small integer (<10)
+DEFAULT_URL is the variable above
+"""

rag.py

@@ -0,0 +1,46 @@
+import os
+import zipfile
+from huggingface_hub import hf_hub_download
+from sentence_transformers import SentenceTransformer
+from langchain_chroma import Chroma
+import torch
+import gradio as gr
+
+
+class SentenceTransformerWrapper:
+    def __init__(self, model_name):
+        self.model = SentenceTransformer(model_name)
+
+    def embed_documents(self, texts):
+        # Convert the list of texts to embeddings
+        return self.model.encode(texts, show_progress_bar=True).tolist()
+
+    def embed_query(self, text):
+        # Convert a single query to its embedding
+        return self.model.encode(text).tolist()
+    
+persist_directory = "chroma_db"
+embedding_model = SentenceTransformerWrapper("intfloat/e5-base-v2")
+
+vector_db = Chroma(
+    persist_directory=persist_directory,
+    embedding_function=embedding_model,
+)
+
+def retrieve_info(query, k=5):
+    results = vector_db.similarity_search(query, k)
+    output = ""
+    for i, doc in enumerate(results):
+        output += f"Result {i+1}:\n Metadata: {doc.metadata}\n Content: {doc.page_content[:1000]}\n\n"
+    return output
+
+
+demo = gr.Interface(
+    fn=retrieve_info,
+    inputs=["text", gr.Number(label="k (Number of chunks to retrieve)")],
+    outputs=[
+        gr.Textbox(label="Output from DenseRetriever", lines=25),
+    ],
+)
+
+    

requirements.txt

@@ -0,0 +1,8 @@
+fastapi
+uvicorn[standard]
+gradio
+sentence-transformers
+torch
+langchain-chroma
+chromadb
+python-dotenv

space.yaml

@@ -0,0 +1,4 @@
+# space.yaml  (must live at the root of the repo)
+sdk: fastapi            # tells Spaces to launch with uvicorn
+python_version: "3.10"  # optional; 3.10 by default
+entrypoint: app:app     # module:variable   (app.py defines 'app')