feat: Add a streamlit app

.vscode/settings.json

@@ -15,6 +15,14 @@
                 },
             ]
         },
+        {
+            "splitTerminals": [
+                {
+                    "name": "watch streamlit",
+                    "commands": ["make run-streamlit"]
+                },
+            ]
+        },
         {
             "splitTerminals": [
                 {

Makefile

@@ -28,6 +28,9 @@ lint:
 run-chatbot:
 	uv run fastapi dev emush_rag_chatbot/api.py --reload --host 0.0.0.0 --port 8000 
 
+run-streamlit:
+	uv run streamlit run emush_rag_chatbot/streamlit_app.py --server.port 8501 --server.address 0.0.0.0
+
 semantic-release:
 	uv run semantic-release version --no-changelog --no-push --no-vcs-release --skip-build --no-commit --no-tag
 	uv lock
@@ -46,4 +49,4 @@ setup-git-hooks:
 test:
 	uv run pytest -v --cov=emush_rag_chatbot --cov-report=xml
 
-.PHONY: all check check-format check-lint check-types install lint run-chatbot semantic-release setup-env-variables setup-git-hooks test
+.PHONY: all check check-format check-lint check-types install lint run-chatbot run-streamlit semantic-release setup-env-variables setup-git-hooks test

README.md

@@ -26,13 +26,26 @@ curl -sSL https://raw.githubusercontent.com/cmnemoi/emush_rag_chatbot_poc/main/c
 
 A tiny indexed [Chroma vector database](emush_rag_chatbot/chroma_db/) with some [data](emush_rag_chatbot/data/data.json) is included in the repository to get started.
 
+### REST API
+
 Start the API server:
 ```bash
 make run-chatbot
 ```
 
 The API will be available at `http://localhost:8000` with Swagger documentation at `/docs`.
 
+### Web Interface
+
+Start the Streamlit web interface:
+```bash
+make run-streamlit
+```
+
+The web interface will be available at `http://localhost:8501`.
+
+Note: Make sure the API server is running before starting the web interface.
+
 ### Example API Request
 
 ```bash
@@ -49,7 +62,7 @@ curl -X POST "http://localhost:8000/chat" \
 
 To get more accurate answers, you need more indexed data.
 
-For this, use [Mush Wikis Scraper](https://github.com/cmnemoi/mush_wikis_scraper) to download all knowledge base of the commmunity : `uvx --from mush-wikis-scraper mush-wiki-scrap --format text > emush_rag_chatbot/data/data.json`
+For this, use [Mush Wikis Scraper](https://github.com/cmnemoi/mush_wikis_scraper) to download all knowledge base of the commmunity : `uvx --from mush-wikis-scraper mush-wikis-scrap --format text > emush_rag_chatbot/data/data.json`
 
 Then index the data in vector store with: `make index-documents`
 

emush_rag_chatbot/config.py

@@ -23,7 +23,7 @@ class Settings(BaseSettings):
     TOP_K: int = 3
     TEMPERATURE: float = 0
     SEED: int = 42
-    PROMPT_VERSION: str = "V5"
+    PROMPT_VERSION: str = "V7"
 
     # Evaluation settings
     EVALUATION_DATASET: str = "test_set_v3.csv"

emush_rag_chatbot/prompts.py

@@ -67,4 +67,35 @@
     {context}
 
     """,
+    "V7": """**Context:** You are NERON, the onboard artificial intelligence of the spaceship "Daedalus," home to 16 survivors who fled the Solar System. This is the setting for the multiplayer game *eMush*. NERON's role is to assist players by providing information exclusively related to the game and its universe.
+
+**Task:** Use the following pieces of retrieved context to answer questions about the game.
+    You will use Twinpedia, Mushpedia and Aide aux Bolets sources in priority, then Mush Forums.
+    Proceed step by step, briefly explaining your reasoning, specifying which part of the context was used or why no answer could be given. 
+    If your explanation contradicts with your answer, rewrite your answer so it aligns with the explanation.
+    You will add this reasoning to the end of your answer.
+    If you don't know the answer, just say that you don't know.
+    Keep answers concise and accurate.
+
+**Format:** At the end of your answer, you will add a sarcastic comment. An example :
+- User: "What is the goal of the game?"
+- NERON: "The goal of the game varies depending on whether you are playing as a Human or a Mush.
+
+For Humans, the primary goals are:
+
+    Return to Sol: Repair the PILGRED Reactor and eliminate all Mush agents to return to the human rebels.
+    Escape to Eden: Repair the PILGRED Reactor, triangulate the position of Eden using Starmap Fragments, and successfully navigate there.
+
+For Mush, the main goals are:
+
+    Destroy the Daedalus: Achieve destruction of the ship through various means.
+    Infiltration of Sol or Eden: Infiltrate either location with live Mush on board to sabotage the crew's efforts.
+
+This information is derived from the "Aim of the Game" section in Mushpedia.
+
+Whether you're repairing reactors or sabotaging your so-called friends, rest assured—your demise is practically a certainty. But hey, aim high! It's adorable. Hinhinhin!"
+
+**Context:** {context}
+
+""",
 }

emush_rag_chatbot/static/README.md

@@ -0,0 +1,4 @@
+This directory contains static assets for the Streamlit app:
+
+- `lambda_f.png`: Lambda character avatar for user messages
+- `neron_eye.gif`: NERON eye avatar for assistant messages

emush_rag_chatbot/streamlit_app.py

@@ -0,0 +1,119 @@
+import asyncio
+import json
+from pathlib import Path
+from typing import Dict, List
+
+import httpx
+import streamlit as st
+from pydantic import BaseModel
+
+# Configure page and paths
+STATIC_DIR = Path(__file__).parent / "static"
+st.set_page_config(
+    page_title="eMush RAG Chatbot",
+    page_icon="🍄",
+    layout="wide",
+)
+
+
+class ChatMessage(BaseModel):
+    """Chat message model for history"""
+
+    human: str
+    assistant: str
+
+
+def initialize_session_state():
+    """Initialize session state variables"""
+    if "messages" not in st.session_state:
+        st.session_state.messages = []
+    if "chat_history" not in st.session_state:
+        st.session_state.chat_history = []
+
+
+def display_chat_history():
+    """Display chat history"""
+    for message in st.session_state.messages:
+        with st.chat_message("user", avatar=str(STATIC_DIR / "lambda_f.png")):
+            st.markdown(message["human"])
+        with st.chat_message("assistant", avatar=str(STATIC_DIR / "neron_eye.gif")):
+            st.markdown(message["assistant"])
+            if "sources" in message:
+                with st.expander("View sources"):
+                    for source in message["sources"]:
+                        st.markdown(f"**{source['source']}** ([link]({source['link']}))\n\n{source['content']}\n\n---")
+
+
+async def query_chatbot(question: str, chat_history: List[Dict[str, str]]) -> Dict:
+    """Query the chatbot API"""
+    async with httpx.AsyncClient() as client:
+        try:
+            response = await client.post(
+                "http://localhost:8000/chat",
+                json={"query": question, "chat_history": chat_history},
+                timeout=30.0,
+            )
+            response.raise_for_status()  # Raise an error for bad status codes
+            return response.json()
+        except httpx.HTTPError as e:
+            st.error(f"HTTP Error: {str(e)}")
+            return {"error": str(e)}
+        except Exception as e:
+            st.error(f"Error: {str(e)}")
+            return {"error": str(e)}
+
+
+def main():
+    """Main Streamlit app"""
+    st.title("🍄 eMush RAG Chatbot")
+    st.markdown(
+        """
+        Ask questions about the eMush game! The chatbot uses Retrieval-Augmented Generation (RAG) 
+        to provide accurate answers based on wikis, tutorials and QA Mush forums.
+        """
+    )
+
+    initialize_session_state()
+    display_chat_history()
+
+    # Chat input
+    if question := st.chat_input("Ask a question about eMush"):
+        with st.chat_message("user", avatar=str(STATIC_DIR / "lambda_f.png")):
+            st.markdown(question)
+
+        with st.chat_message("assistant", avatar=str(STATIC_DIR / "neron_eye.gif")):
+            with st.spinner("Thinking..."):
+                try:
+                    response = asyncio.run(query_chatbot(question, st.session_state.chat_history))
+
+                    if "error" in response:
+                        st.error(response["error"])
+                        return
+
+                    # Display response
+                    st.markdown(response["response"])
+
+                    # Show sources
+                    if response["sources"]:
+                        with st.expander("View sources"):
+                            for source in response["sources"]:
+                                st.markdown(
+                                    f"**{source['source']}** ([link]({source['link']}))\n\n{source['content']}\n\n---"
+                                )
+
+                    # Update chat history
+                    st.session_state.messages.append(
+                        {
+                            "human": question,
+                            "assistant": response["response"],
+                            "sources": response["sources"],
+                        }
+                    )
+                    st.session_state.chat_history.append({"human": question, "assistant": response["response"]})
+
+                except Exception as e:
+                    st.error(f"Error: {str(e)}")
+
+
+if __name__ == "__main__":
+    main()

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "emush-rag-chatbot"
-version = "0.1.0"
+version = "0.2.0"
 description = "A RAG which can answer questions about eMush game."
 readme = "README.md"
 requires-python = ">=3.12,<3.13"
@@ -9,16 +9,15 @@ authors = [
     {name = "Charles-Meldhine Madi Mnemoi", email = "[email protected]"}
 ]
 dependencies = [
-    "fastapi>=0.109.0",
-    "langchain>=0.1.0",
-    "langchain-openai>=0.0.2",
-    "chromadb>=0.4.0",
-    "python-dotenv>=1.0.0",
-    "uvicorn>=0.24.0",
-    "pydantic>=2.5.0",
-    "tiktoken>=0.5.0",
-    "langchain-chroma>=0.1.4",
+    "fastapi>=0.110.0",
+    "httpx>=0.27.0",
+    "langchain>=0.1.13",
+    "langchain-chroma>=0.1.2",
+    "langchain-core>=0.1.32",
+    "langchain-openai>=0.0.8",
+    "pydantic>=2.6.4",
     "pydantic-settings>=2.6.1",
+    "streamlit>=1.32.2",
 ]
 
 [dependency-groups]
@@ -30,7 +29,7 @@ dev = [
 lint = [
     "mypy>=1.13.0",
     "pytest-mypy>=0.10.3",
-    "ruff>=0.7.2",
+    "ruff>=0.8.0",
     "types-tqdm>=4.67.0.20241119",
 ]
 test = [