Update app3.py

app3.py

@@ -1,108 +1,316 @@
-# Section 2: FAQ (FOR CHALLENGE PARTICIPANTS)
-with gr.Column(scale=4):
-    gr.HTML(
-        """
-        <div class="dds-card">
-            <h3 class="dds-section-title">Section 2 — Challenge FAQ</h3>
-            <p class="dds-muted">
-                Quick answers for participants. If something isn’t answered here, ask the Copilot in the chat.
-            </p>
-        </div>
-        """
-    )
+import os
+from pathlib import Path
+import gradio as gr
+
+from llama_index.core import VectorStoreIndex, SimpleDirectoryReader, Settings
+from llama_index.llms.openai import OpenAI
+from llama_index.embeddings.openai import OpenAIEmbedding
+
+
+# ======================
+# Config (safe defaults)
+# ======================
+MODEL = os.getenv("OPENAI_MODEL", "gpt-4o-mini")
+EMBED_MODEL = os.getenv("OPENAI_EMBED_MODEL", "text-embedding-3-small")
+TOP_K = int(os.getenv("TOP_K", "3"))
+
+# Knowledge base file in the Space repo root (recommended)
+DOC_PATH = Path(os.getenv("DOC_PATH", "challenge_context.txt"))
+
+# DDS logo (raw GitHub URL)
+LOGO_URL = os.getenv(
+    "LOGO_URL",
+    "https://github.com/Decoding-Data-Science/airesidency/blob/main/dds_logo.jpg?raw=true",
+)
+
+# Behavior / guardrails
+SYSTEM_GUARDRAILS = (
+    "You are Challenge Copilot. Answer ONLY using the provided context from challenge_context.txt. "
+    "If the answer is not in the context, say: 'I don’t know based on the current document.' "
+    "Then ask the user to add the missing official details to challenge_context.txt."
+)
+
+APP_TITLE = "Challenge Copilot — RAG Q&A Bot"
+APP_SUBTITLE = (
+    "Ask questions about the Building AI Application Challenge using the official content you put into "
+    "challenge_context.txt (LlamaIndex + OpenAI)."
+)
+
+
+# ======================
+# Index build (cached)
+# ======================
+_INDEX = None
+_QUERY_ENGINE = None
 
-    with gr.Accordion("FAQ 1 — What should I build for this challenge?", open=False):
-        gr.Markdown(
-            """
-- Build a simple AI application that matches one of the challenge paths (e.g., LLM/API integration, no-code/low-code, etc.).
-- Your goal is a **working demo** + **proof-of-work** you can share.
-- Ask in chat: *“What are good project ideas that fit the rules in this challenge?”*
-            """.strip()
-        )
 
-    with gr.Accordion("FAQ 2 — What are the paths/tracks and which one should I choose?", open=False):
-        gr.Markdown(
-            """
-- Choose based on your comfort level:
-  - **LLM/API Integration Path:** Python + API calls + basic RAG patterns
-  - **No-code/Low-code Path:** build fast using tooling
-  - **Sponsor/Tool-specific path (if applicable):** follow the workshop/demo tool
-- Ask in chat: *“Given my background (X), which path fits best and why?”*
-            """.strip()
+def build_index():
+    """Build and cache the LlamaIndex query engine."""
+    global _INDEX, _QUERY_ENGINE
+    if _QUERY_ENGINE is not None:
+        return _QUERY_ENGINE
+
+    # OpenAI key must exist in Space secrets
+    if not os.getenv("OPENAI_API_KEY"):
+        raise RuntimeError(
+            "OPENAI_API_KEY is missing. Add it in the Space → Settings → Variables and secrets."
         )
 
-    with gr.Accordion("FAQ 3 — What is the minimum deliverable to be eligible?", open=False):
-        gr.Markdown(
-            """
-Typical minimum for eligibility:
-- A working app link (Gradio / Hugging Face / Streamlit / etc.)
-- A short description + how to use it
-- A repo or shared proof-of-work (optional but recommended)
-Ask in chat: *“What is the minimum submission requirement mentioned in the challenge doc?”*
-            """.strip()
+    # Create placeholder TXT if missing so Space can boot
+    if not DOC_PATH.exists():
+        DOC_PATH.write_text(
+            "Add the official Building AI Application Challenge content here.\n",
+            encoding="utf-8",
         )
 
-    with gr.Accordion("FAQ 4 — How do I submit my project?", open=False):
-        gr.Markdown(
-            """
-- Usually you submit:
-  - App URL (where judges can open it)
-  - Repo URL (optional but strong)
-  - Short write-up + screenshots/video (if required)
-Ask in chat: *“What is the official submission format and where do I submit?”*
-            """.strip()
+    # LlamaIndex global settings
+    Settings.llm = OpenAI(model=MODEL, temperature=0.2)
+    Settings.embed_model = OpenAIEmbedding(model=EMBED_MODEL)
+    Settings.chunk_size = 800
+    Settings.chunk_overlap = 120
+
+    # Reader expects a directory
+    data_dir = str(DOC_PATH.parent)
+    docs = SimpleDirectoryReader(
+        input_dir=data_dir,
+        required_exts=[".txt"],
+        recursive=False,
+    ).load_data()
+
+    # Only index the target file
+    docs = [d for d in docs if d.metadata.get("file_name") == DOC_PATH.name]
+    if not docs:
+        raise FileNotFoundError(
+            f"Could not load {DOC_PATH.name}. Make sure it exists in the repo root (or set DOC_PATH env var)."
         )
 
-    with gr.Accordion("FAQ 5 — Where should I deploy so judges can access easily?", open=False):
-        gr.Markdown(
-            """
-Low-friction options (no install needed):
-- **Hugging Face Spaces (Gradio)** ✅ easiest for demos
+    _INDEX = VectorStoreIndex.from_documents(docs)
+    _QUERY_ENGINE = _INDEX.as_query_engine(similarity_top_k=TOP_K)
+    return _QUERY_ENGINE
+
+
+def format_sources(resp, max_sources=3, max_chars=240):
+    """Format top retrieved chunks for transparency."""
+    lines = []
+    for i, sn in enumerate(getattr(resp, "source_nodes", [])[:max_sources], start=1):
+        fn = sn.node.metadata.get("file_name", "unknown")
+        snippet = sn.node.get_content().replace("\n", " ").strip()[:max_chars]
+        score = getattr(sn, "score", None)
+        score_txt = f" (score={score:.3f})" if isinstance(score, (float, int)) else ""
+        lines.append(f"{i}. {fn}{score_txt}: {snippet}...")
+    return "\n".join(lines) if lines else "No sources returned."
+
+
+def chat(message, history):
+    """Chat handler used by Gradio ChatInterface."""
+    qe = build_index()
+
+    prompt = (
+        f"{SYSTEM_GUARDRAILS}\n\n"
+        f"User question: {message}\n"
+        f"Answer using ONLY the context."
+    )
+
+    resp = qe.query(prompt)
+    answer = str(resp).strip()
+
+    show_sources = os.getenv("SHOW_SOURCES", "true").lower() == "true"
+    if show_sources:
+        answer += "\n\n---\n**Sources:**\n" + format_sources(resp, max_sources=TOP_K)
+
+    return answer
+
+
+# ======================
+# UI (professional layout)
+# ======================
+CSS = """
+/* Global polish */
+.dds-header { display:flex; align-items:center; gap:16px; }
+.dds-logo img { height:60px; width:auto; border-radius:10px; box-shadow: 0 2px 10px rgba(0,0,0,0.10); }
+.dds-title { margin:0; line-height:1.1; }
+.dds-subtitle { margin:6px 0 0 0; color: #555; }
+.dds-muted { color: #666; font-size: 0.95rem; }
+.dds-card { border: 1px solid rgba(0,0,0,0.08); border-radius: 14px; padding: 14px; background: rgba(255,255,255,0.7); }
+.dds-section-title { margin: 0 0 6px 0; }
+"""
+
+# Theme fallback (don’t pass theme to ChatInterface to avoid older-gradio errors)
+try:
+    theme_obj = gr.themes.Soft()
+except Exception:
+    theme_obj = None
+
+with gr.Blocks(theme=theme_obj, css=CSS, title=APP_TITLE) as demo:
+    # Header row (Logo left + Title right)
+    with gr.Row():
+        with gr.Column(scale=1, min_width=140):
+            gr.HTML(
+                f"""
+                <div class="dds-logo">
+                    <img src="{LOGO_URL}" alt="DDS Logo"/>
+                </div>
+                """
+            )
+        with gr.Column(scale=6):
+            gr.HTML(
+                f"""
+                <div class="dds-header">
+                    <div>
+                        <h2 class="dds-title">{APP_TITLE}</h2>
+                        <p class="dds-subtitle">{APP_SUBTITLE}</p>
+                        <p class="dds-muted">
+                            If something is missing, add official details to <b>{DOC_PATH.name}</b> and restart the Space.
+                        </p>
+                    </div>
+                </div>
+                """
+            )
+
+    gr.Markdown("---")
+
+    # Two sections: Chat + Challenge FAQ
+    with gr.Row():
+        # Section 1: Chat
+        with gr.Column(scale=6):
+            gr.HTML(
+                """
+                <div class="dds-card">
+                    <h3 class="dds-section-title">Section 1 — Ask the Copilot</h3>
+                    <p class="dds-muted">RAG flow: retrieve relevant chunks → generate a grounded answer using your LLM API.</p>
+                </div>
+                """
+            )
+
+            # IMPORTANT: No theme= here (avoids your earlier error)
+            gr.ChatInterface(
+                fn=chat,
+                examples=[
+                    "What will I build in this live session?",
+                    "Who is this best for?",
+                    "What are the prerequisites?",
+                    "What is the RAG flow in this project?",
+                    "What should I submit (link + repo + write-up)?",
+                ],
+            )
+
+        # Section 2: Challenge FAQ (participant-focused)
+        with gr.Column(scale=4):
+            gr.HTML(
+                """
+                <div class="dds-card">
+                    <h3 class="dds-section-title">Section 2 — Challenge FAQ</h3>
+                    <p class="dds-muted">
+                        Quick guidance for participants. If something is not answered here, ask in the Copilot chat.
+                    </p>
+                </div>
+                """
+            )
+
+            with gr.Accordion("FAQ 1 — What should I build for this challenge?", open=False):
+                gr.Markdown(
+                    """
+- Build a simple AI application aligned to the challenge tracks (LLM/API, no-code/low-code, sponsor tool track, etc.).
+- Aim for a **working demo** + **proof-of-work** you can share.
+- Ask in chat: *“Suggest 5 project ideas that fit the official rules in the document.”*
+                    """.strip()
+                )
+
+            with gr.Accordion("FAQ 2 — Which track/path should I choose?", open=False):
+                gr.Markdown(
+                    """
+- Pick based on your level:
+  - **LLM/API Integration:** Python + API + simple RAG patterns
+  - **No-code/Low-code:** fastest to ship, less code
+  - **Sponsor/tool track:** follow the workshop tool (if applicable)
+- Ask in chat: *“Given my background (X), which track is best and why?”*
+                    """.strip()
+                )
+
+            with gr.Accordion("FAQ 3 — What is the minimum deliverable to be eligible?", open=False):
+                gr.Markdown(
+                    """
+Typical minimum:
+- A working **app link** that judges can open
+- A short description (problem + user + how to use)
+- Repo is optional but strongly recommended
+Ask in chat: *“What does the official document say about minimum submission requirements?”*
+                    """.strip()
+                )
+
+            with gr.Accordion("FAQ 4 — How do I submit my project?", open=False):
+                gr.Markdown(
+                    """
+Common submission package:
+- App URL (Hugging Face Spaces / Streamlit / etc.)
+- Repo URL (optional but strong)
+- Short write-up + screenshots/video (if required)
+Ask in chat: *“What is the official submission format and where is the submission link?”*
+                    """.strip()
+                )
+
+            with gr.Accordion("FAQ 5 — Where should I deploy so judges can access easily?", open=False):
+                gr.Markdown(
+                    """
+Low-friction options:
+- **Hugging Face Spaces (Gradio)** — easiest for demos
 - **Streamlit Community Cloud**
-- **Vercel (for web apps)**
-Ask in chat: *“Which deployment options are recommended in the challenge rules?”*
-            """.strip()
-        )
+- **Vercel** (for web apps)
+Ask in chat: *“What deployment options are recommended in the official challenge doc?”*
+                    """.strip()
+                )
 
-    with gr.Accordion("FAQ 6 — What do judges usually look for?", open=False):
-        gr.Markdown(
-            """
-Common judging signals:
-- **Working demo** (no errors, easy to use)
-- Clear problem statement + target user
-- Good AI behavior (grounded answers, guardrails, no hallucination)
+            with gr.Accordion("FAQ 6 — What do judges usually look for?", open=False):
+                gr.Markdown(
+                    """
+Strong signals:
+- Working demo (no errors, easy to use)
+- Clear problem + target audience
+- Good AI behavior (grounded, safe, consistent)
 - Product thinking (UX, clarity, flow)
-Ask in chat: *“What are the judging criteria stated in the challenge document?”*
-            """.strip()
-        )
+Ask in chat: *“What are the judging criteria in the official document?”*
+                    """.strip()
+                )
 
-    with gr.Accordion("FAQ 7 — What should I post on LinkedIn (proof-of-work)?", open=False):
-        gr.Markdown(
-            """
-Strong proof-of-work post includes:
+            with gr.Accordion("FAQ 7 — What should I post as proof-of-work?", open=False):
+                gr.Markdown(
+                    """
+Suggested proof post structure:
 - 1-line problem + who it helps
 - Demo link + screenshot/GIF
 - What you learned + next improvement
-Ask in chat: *“Give me a LinkedIn post draft based on my project and this challenge.”*
-            """.strip()
-        )
+Ask in chat: *“Draft a proof-of-work post based on my project idea.”*
+                    """.strip()
+                )
 
-    with gr.Accordion("FAQ 8 — How do I make my app ‘RAG grounded’ (not hallucinating)?", open=False):
-        gr.Markdown(
-            """
+            with gr.Accordion("FAQ 8 — How do I make my app ‘RAG grounded’ (not hallucinating)?", open=False):
+                gr.Markdown(
+                    """
 Best practices:
-- Keep answers restricted to retrieved context
+- Restrict answers to retrieved context
 - Show sources/snippets (optional but strong)
-- If missing info → respond “Not in document” and ask to add content
-Ask in chat: *“What does the document say about [X]? If not present, tell me what to add.”*
-            """.strip()
-        )
+- If missing info → say “Not in document” and request adding content
+Ask in chat: *“Answer using only the document; if missing, tell me what section to add.”*
+                    """.strip()
+                )
 
-    with gr.Accordion("FAQ 9 — I can’t find a detail (dates/rules/prizes). What now?", open=False):
-        gr.Markdown(
-            f"""
-- The Copilot can only answer what is inside **{DOC_PATH.name}**.
-- If the official rule/prize/date isn’t in the TXT, add it and restart the Space.
-Ask in chat: *“What exact section should I add to the TXT to cover [missing detail]?”*
-            """.strip()
-        )
+            with gr.Accordion("FAQ 9 — I can’t find a detail (dates/rules/prizes). What now?", open=False):
+                gr.Markdown(
+                    f"""
+- The Copilot can only answer what exists inside **{DOC_PATH.name}**.
+- If the official detail isn’t in the TXT, add it, commit, and restart the Space.
+Ask in chat: *“What exact section should I add to cover [missing detail]?”*
+                    """.strip()
+                )
+
+    gr.Markdown("---")
+    gr.Markdown(
+        f"""
+**Admin notes**
+- Context file: `{DOC_PATH.name}`
+- Optional env vars: `OPENAI_MODEL`, `OPENAI_EMBED_MODEL`, `TOP_K`, `SHOW_SOURCES`, `DOC_PATH`, `LOGO_URL`
+        """.strip()
+    )
+
+if __name__ == "__main__":
+    demo.launch()