{"repo": "ashleshakadam/utd-jsom-chatbot", "files": {"app.py": {"24261162656c40a012f9dac41db0d2e460a55abb": {"t": 1785674364.311221, "n": 4180, "h": "900b821ca3bebb2f", "c": "import os\nimport streamlit as st\nimport requests\nfrom datetime import datetime\nfrom dotenv import load_dotenv\n\nfrom langchain_community.document_loaders import JSONLoader\nfrom langchain.text_splitter import CharacterTextSplitter\nfrom langchain_community.embeddings import HuggingFaceEmbeddings\nfrom langchain_community.vectorstores import FAISS\nfrom langchain.chains import RetrievalQA\n\n# ==== Load environment variables ====\nload_dotenv()\nOPENROUTER_API_KEY = os.getenv(\"OPENROUTER_API_KEY\")\n\n# ==== Load and preprocess website content ====\nloader = JSONLoader(file_path=\"data/full_site.json\", jq_schema=\".[]\", text_content=True)\ndocuments = loader.load()\n\ntext_splitter = CharacterTextSplitter(chunk_size=800, chunk_overlap=100)\ndocs = text_splitter.split_documents(documents)\n\n# ==== Build embeddings and retriever ====\nembeddings = HuggingFaceEmbeddings(model_name=\"all-MiniLM-L6-v2\")\nvectorstore = FAISS.from_documents(docs, embeddings)\nretriever = vectorstore.as_retriever(search_kwargs={\"k\": 3})\n\n# ==== OpenRouter query ====\ndef query_openrouter(prompt, api_key):\n    url = \"https://openrouter.ai/api/v1/chat/completions\"\n    headers = {\n        \"Authorization\": f\"Bearer {api_key}\",\n        \"Content-Type\": \"application/json\",\n        \"HTTP-Referer\": \"https://utd-jsom-chatbot.streamlit.app\"\n    }\n    payload = {\n        \"model\": \"mistralai/mistral-7b-instruct\",\n        \"messages\": [{\"role\": \"user\", \"content\": prompt}],\n    }\n    response = requests.post(url, headers=headers, json=payload)\n    response.raise_for_status()\n    return response.json()[\"choices\"][0][\"message\"][\"content\"]\n\n# ==== Streamlit Page Setup ====\nst.set_page_config(page_title=\"🤖 JSOM Chatbot\", layout=\"wide\")\n\n# ==== Custom CSS for UI polish ====\nst.markdown(\"\"\"\n    <style>\n        .title {\n            font-size: 2.8rem;\n            font-weight: 700;\n            color: #0d3c61;\n            text-align: center;\n            margin-top: 1rem;\n        }\n        .subtitle {\n            text-align: center;\n            font-size: 1.2rem;\n            color: #444;\n            margin-bottom: 2rem;\n        }\n        .answer-box {\n            background-color: #f0f6ff;\n            border-left: 5px solid #0078d4;\n            padding: 1rem;\n            border-radius: 8px;\n            margin-top: 1.5rem;\n            font-size: 1.1rem;\n        }\n        .chunk-box {\n            background-color: #fdfdfd;\n            border: 1px dashed #ccc;\n            border-radius: 5px;\n            padding: 0.75rem;\n            margin-bottom: 1rem;\n            font-size: 0.95rem;\n        }\n        .footer {\n            text-align: center;\n            color: #888;\n            font-size: 0.9rem;\n            margin-top: 2rem;\n        }\n        .stTextInput > div > div > input {\n            font-size: 1.05rem;\n        }\n    </style>\n\"\"\", unsafe_allow_html=True)\n\n# ==== Title & Subtitle ====\nst.markdown('<div class=\"title\">🤖 JSOM Chatbot</div>', unsafe_allow_html=True)\nst.markdown('<div class=\"subtitle\">Accurate answers from the official Jindal School of Management website</div>', unsafe_allow_html=True)\n\n# ==== Query Input ====\nquery = st.text_input(\"🔍 Ask me anything about JSOM:\")\n\nif query:\n    with st.spinner(\"🤖 Thinking...\"):\n        try:\n            relevant_docs = retriever.get_relevant_documents(query)\n            context = \"\\n\\n\".join(doc.page_content for doc in relevant_docs)\n\n            prompt = f\"Use the following JSOM information to answer:\\n\\n{context}\\n\\nQuestion: {query}\"\n            answer = query_openrouter(prompt, OPENROUTER_API_KEY)\n\n            st.markdown(f'<div class=\"answer-box\"><strong>Answer:</strong> {answer}</div>', unsafe_allow_html=True)\n\n            with st.expander(\"📄 View Sources Used\"):\n                for i, doc in enumerate(relevant_docs):\n                    st.markdown(f'<div class=\"chunk-box\"><strong>Snippet {i+1}:</strong> {doc.page_content.strip()}</div>', unsafe_allow_html=True)\n\n        except Exception as e:\n            st.error(f\"❌ Error: {e}\")\n\n# ==== Footer with timestamp ====\nlast_updated = datetime.now().strftime(\"%B %d, %Y at %I:%M %p\")\nst.markdown(f'<div class=\"footer\">📅 Last updated: {last_updated}</div>', unsafe_allow_html=True)\n"}}}}