Create app.py

app.py

@@ -0,0 +1,706 @@
+import os
+import re
+import json
+import pickle
+from urllib.parse import quote
+
+import numpy as np
+import gradio as gr
+from rank_bm25 import BM25Okapi
+from sentence_transformers import SentenceTransformer
+from openai import OpenAI
+
+# ============================================================
+# BrainChat: High-contrast phone UI + evaluated quizzes (RAG)
+# ============================================================
+
+# --------- Build artifacts (RAG) ---------
+BUILD_DIR = "brainchat_build"
+CHUNKS_PATH = os.path.join(BUILD_DIR, "chunks.pkl")
+TOKENS_PATH = os.path.join(BUILD_DIR, "tokenized_chunks.pkl")
+EMBED_PATH = os.path.join(BUILD_DIR, "embeddings.npy")
+CONFIG_PATH = os.path.join(BUILD_DIR, "config.json")
+
+# --------- Logo auto-detection ---------
+LOGO_CANDIDATES = [
+    "brainchat_logo.png.png",
+    "Brain chat-09.png",
+    "Brain Chat Imagen.svg",
+    "ebcbb9f5-022f-473a-bf51-7e7974f794b4.png",
+    "logo.png",
+    "logo.svg",
+]
+
+OPENAI_MODEL = os.getenv("OPENAI_MODEL", "gpt-4o-mini")
+
+# --------- Globals (lazy loaded) ---------
+BM25 = None
+CHUNKS = None
+EMBEDDINGS = None
+EMBED_MODEL = None
+CLIENT = None
+
+
+# ============================================================
+# Retrieval
+# ============================================================
+def tokenize(text: str):
+    return re.findall(r"\w+", text.lower(), flags=re.UNICODE)
+
+
+def ensure_loaded():
+    global BM25, CHUNKS, EMBEDDINGS, EMBED_MODEL, CLIENT
+
+    if CHUNKS is None:
+        missing = [p for p in [CHUNKS_PATH, TOKENS_PATH, EMBED_PATH, CONFIG_PATH] if not os.path.exists(p)]
+        if missing:
+            raise FileNotFoundError(
+                "Missing build files. Ensure brainchat_build/ exists and contains:\n"
+                "chunks.pkl, tokenized_chunks.pkl, embeddings.npy, config.json\n\nMissing:\n"
+                + "\n".join(missing)
+            )
+
+        with open(CHUNKS_PATH, "rb") as f:
+            CHUNKS = pickle.load(f)
+
+        with open(TOKENS_PATH, "rb") as f:
+            tokenized_chunks = pickle.load(f)
+
+        EMBEDDINGS = np.load(EMBED_PATH)
+
+        with open(CONFIG_PATH, "r", encoding="utf-8") as f:
+            cfg = json.load(f)
+
+        BM25 = BM25Okapi(tokenized_chunks)
+        EMBED_MODEL = SentenceTransformer(cfg["embedding_model"])
+
+    if CLIENT is None:
+        api_key = os.getenv("OPENAI_API_KEY")
+        if not api_key:
+            raise ValueError("OPENAI_API_KEY missing. Add it in your Space Secrets.")
+        CLIENT = OpenAI(api_key=api_key)
+
+
+def search_hybrid(query: str, shortlist_k: int = 30, final_k: int = 5):
+    """
+    Hybrid retrieval:
+      1) BM25 shortlist
+      2) Dense cosine re-rank
+    """
+    ensure_loaded()
+    q_tokens = tokenize(query)
+    bm25_scores = BM25.get_scores(q_tokens)
+    shortlist_idx = np.argsort(bm25_scores)[::-1][:shortlist_k]
+
+    qvec = EMBED_MODEL.encode([query], normalize_embeddings=True).astype("float32")[0]
+    shortlist_emb = EMBEDDINGS[shortlist_idx]
+    dense_scores = shortlist_emb @ qvec
+
+    rerank = np.argsort(dense_scores)[::-1][:final_k]
+    final_idx = shortlist_idx[rerank]
+    return [CHUNKS[int(i)] for i in final_idx]
+
+
+def build_context(records):
+    blocks = []
+    for i, r in enumerate(records, start=1):
+        blocks.append(
+            f"""[Source {i}]
+Book: {r.get('book','')}
+Section: {r.get('section_title','')}
+Pages: {r.get('page_start','')}-{r.get('page_end','')}
+Text:
+{r.get('text','')}"""
+        )
+    return "\n\n".join(blocks)
+
+
+def make_sources(records):
+    seen = set()
+    lines = []
+    for r in records:
+        key = (r.get("book"), r.get("section_title"), r.get("page_start"), r.get("page_end"))
+        if key in seen:
+            continue
+        seen.add(key)
+        lines.append(
+            f"• {r.get('book','')} | {r.get('section_title','')} | pp. {r.get('page_start','')}-{r.get('page_end','')}"
+        )
+    return "\n".join(lines)
+
+
+# ============================================================
+# Prompts
+# ============================================================
+def language_instruction(language_mode: str) -> str:
+    if language_mode == "English":
+        return "Answer only in English."
+    if language_mode == "Spanish":
+        return "Answer only in Spanish."
+    if language_mode == "Bilingual":
+        return "Answer first in English, then provide a Spanish version under the heading 'Español:'."
+    return "If the user's message is in Spanish, answer in Spanish; otherwise answer in English."
+
+
+def choose_quiz_count(user_text: str, selector: str) -> int:
+    if selector in {"3", "5", "7"}:
+        return int(selector)
+
+    t = user_text.lower()
+    if any(k in t for k in ["mock test", "final exam", "exam practice", "full test"]):
+        return 7
+    if any(k in t for k in ["detailed", "revision", "comprehensive", "study"]):
+        return 5
+    return 3
+
+
+def build_tutor_prompt(mode: str, language_mode: str, question: str, context: str) -> str:
+    style = {
+        "Explain": "Explain clearly like a friendly tutor using simple language and short headings.",
+        "Detailed": "Give a detailed explanation with key points and clinical relevance only when supported by context.",
+        "Short Notes": "Write concise revision notes using bullet points.",
+        "Flashcards": "Create 6 flashcards in Q/A format.",
+        "Case-Based": "Create a short clinical scenario (2–4 lines) and then explain the concept using the context.",
+    }.get(mode, "Explain clearly like a friendly tutor.")
+
+    return f"""
+You are BrainChat, an interactive neurology and neuroanatomy tutor.
+
+Rules:
+- Use ONLY the provided context from the books.
+- If the answer is not supported by the context, say exactly:
+  Not found in the course material.
+- Do not invent facts outside the context.
+- {language_instruction(language_mode)}
+
+Style:
+{style}
+
+Context:
+{context}
+
+Student question:
+{question}
+""".strip()
+
+
+def build_quiz_generation_prompt(language_mode: str, topic: str, context: str, n_questions: int) -> str:
+    return f"""
+You are BrainChat, an interactive tutor.
+
+Rules:
+- Use ONLY the provided context.
+- Create exactly {n_questions} questions.
+- Provide a short answer key for each question.
+- Keep answers short (1–2 lines).
+- Return VALID JSON only.
+- {language_instruction(language_mode)}
+
+Return JSON in this schema:
+{{
+  "title": "short quiz title",
+  "questions": [
+    {{"q": "question 1", "answer_key": "expected short answer"}},
+    {{"q": "question 2", "answer_key": "expected short answer"}}
+  ]
+}}
+
+Context:
+{context}
+
+Topic:
+{topic}
+""".strip()
+
+
+def build_quiz_eval_prompt(language_mode: str, quiz_data: dict, user_answers: str) -> str:
+    quiz_json = json.dumps(quiz_data, ensure_ascii=False)
+    return f"""
+You are BrainChat, an interactive tutor.
+
+Evaluate the student's answers fairly using the answer keys.
+Be lenient for paraphrases that are semantically correct.
+
+Return VALID JSON only.
+
+Return JSON in this schema:
+{{
+  "score_obtained": 0,
+  "score_total": 0,
+  "summary": "short overall feedback",
+  "results": [
+    {{
+      "question": "question text",
+      "answer_key": "expected short answer",
+      "student_answer": "student answer",
+      "result": "Correct / Partially Correct / Incorrect",
+      "feedback": "short explanation"
+    }}
+  ],
+  "improvement_tip": "one short study suggestion"
+}}
+
+Quiz:
+{quiz_json}
+
+Student answers:
+{user_answers}
+
+Language:
+{language_instruction(language_mode)}
+""".strip()
+
+
+# ============================================================
+# OpenAI helpers (Chat Completions)
+# ============================================================
+def oai_text(prompt: str) -> str:
+    ensure_loaded()
+    resp = CLIENT.chat.completions.create(
+        model=OPENAI_MODEL,
+        temperature=0.2,
+        messages=[
+            {"role": "system", "content": "You are a helpful educational assistant."},
+            {"role": "user", "content": prompt},
+        ],
+    )
+    return resp.choices[0].message.content.strip()
+
+
+def oai_json(prompt: str) -> dict:
+    ensure_loaded()
+    resp = CLIENT.chat.completions.create(
+        model=OPENAI_MODEL,
+        temperature=0.2,
+        response_format={"type": "json_object"},
+        messages=[
+            {"role": "system", "content": "Return only valid JSON."},
+            {"role": "user", "content": prompt},
+        ],
+    )
+    return json.loads(resp.choices[0].message.content)
+
+
+# ============================================================
+# Logo
+# ============================================================
+def find_logo_filename():
+    for name in LOGO_CANDIDATES:
+        if os.path.exists(name):
+            return name
+    return None
+
+
+def logo_url():
+    f = find_logo_filename()
+    if not f:
+        return None
+    return f"/gradio_api/file={quote(f)}"
+
+
+def phone_logo_html():
+    url = logo_url()
+    if url:
+        return f'<img class="bc-logo-img" src="{url}" alt="BrainChat logo">'
+    return '<div class="bc-logo-fallback">BRAIN<br/>CHAT</div>'
+
+
+# ============================================================
+# Chat logic
+# ============================================================
+def respond(user_msg, history, mode, language_mode, quiz_count_mode, show_sources, quiz_state):
+    history = history or []
+    quiz_state = quiz_state or {"active": False, "quiz_data": None, "language_mode": "Auto"}
+
+    text = (user_msg or "").strip()
+    if not text:
+        return "", history, quiz_state
+
+    try:
+        history = history + [{"role": "user", "content": text}]
+
+        # If quiz is active, treat this message as the answer sheet and evaluate.
+        if quiz_state.get("active", False):
+            evaluation = oai_json(build_quiz_eval_prompt(
+                quiz_state.get("language_mode", language_mode),
+                quiz_state.get("quiz_data", {}),
+                text
+            ))
+
+            lines = []
+            lines.append(f"**Score:** {evaluation.get('score_obtained', 0)}/{evaluation.get('score_total', 0)}")
+            if evaluation.get("summary"):
+                lines.append(f"\n**Overall:** {evaluation['summary']}")
+            if evaluation.get("improvement_tip"):
+                lines.append(f"\n**Tip:** {evaluation['improvement_tip']}\n")
+
+            results = evaluation.get("results", [])
+            if results:
+                lines.append("**Question-wise feedback:**")
+                for item in results:
+                    lines.append("")
+                    lines.append(f"**Q:** {item.get('question','')}")
+                    lines.append(f"**Your answer:** {item.get('student_answer','')}")
+                    lines.append(f"**Expected:** {item.get('answer_key','')}")
+                    lines.append(f"**Result:** {item.get('result','')}")
+                    lines.append(f"**Feedback:** {item.get('feedback','')}")
+
+            history = history + [{"role": "assistant", "content": "\n".join(lines).strip()}]
+            quiz_state = {"active": False, "quiz_data": None, "language_mode": language_mode}
+            return "", history, quiz_state
+
+        # Normal retrieval
+        records = search_hybrid(text, shortlist_k=30, final_k=5)
+        context = build_context(records)
+
+        # Quiz generation
+        if mode == "Quiz Me":
+            n_questions = choose_quiz_count(text, quiz_count_mode)
+            quiz_data = oai_json(build_quiz_generation_prompt(language_mode, text, context, n_questions))
+
+            lines = []
+            lines.append(f"**{quiz_data.get('title','Quiz')}**")
+            lines.append(f"\n**Total questions:** {len(quiz_data.get('questions', []))}\n")
+            lines.append("Reply in ONE message with numbered answers (e.g., `1. ... 2. ...`).\n")
+            for i, q in enumerate(quiz_data.get("questions", []), start=1):
+                lines.append(f"**Q{i}.** {q.get('q','')}")
+
+            if show_sources:
+                lines.append("\n\n**Sources used to create this quiz:**")
+                lines.append(make_sources(records))
+
+            history = history + [{"role": "assistant", "content": "\n".join(lines).strip()}]
+            quiz_state = {"active": True, "quiz_data": quiz_data, "language_mode": language_mode}
+            return "", history, quiz_state
+
+        # Tutor modes
+        answer = oai_text(build_tutor_prompt(mode, language_mode, text, context))
+        if show_sources:
+            answer = answer.strip() + "\n\n**Sources:**\n" + make_sources(records)
+
+        history = history + [{"role": "assistant", "content": answer.strip()}]
+        return "", history, quiz_state
+
+    except Exception as e:
+        history = history + [{"role": "assistant", "content": f"Error: {str(e)}"}]
+        quiz_state = {"active": False, "quiz_data": None, "language_mode": language_mode}
+        return "", history, quiz_state
+
+
+def clear_all():
+    return "", [], {"active": False, "quiz_data": None, "language_mode": "Auto"}
+
+
+# ============================================================
+# CSS: High-contrast palette + phone UI
+# Note: Custom CSS selectors may need updates across Gradio versions.
+# Gradio explicitly warns DOM query selectors may not be stable. (See Change log.)
+# ============================================================
+
+CSS = r"""
+:root{
+  --bc-page-bg: #DCDCDC;
+
+  --bc-grad-top: #E8C7D4;
+  --bc-grad-mid: #7D4579;
+  --bc-grad-bot: #2B0C46;
+
+  --bc-ink: #141414;
+  --bc-ink-muted: #3A3A3A;
+  --bc-ink-inverse: #FFFFFF;
+
+  --bc-accent: #FEF24A;
+
+  --bc-bubble-user: #FFFFFF;
+  --bc-bubble-bot: #F7F3B0;
+
+  --bc-input-bg: #FFFCE6;
+  --bc-btn-bg: #FEF24A;
+  --bc-btn-ink: #2B0C46;
+
+  --bc-shadow: rgba(0,0,0,.18);
+}
+
+body, .gradio-container{
+  background: var(--bc-page-bg) !important;
+  font-family: ui-sans-serif, system-ui, -apple-system, Segoe UI, Roboto, Helvetica, Arial;
+}
+footer{ display:none !important; }
+
+#bc_settings{
+  max-width: 980px;
+  margin: 12px auto 10px auto;
+}
+
+#bc_phone{
+  max-width: 420px;
+  margin: 0 auto 0 auto;
+  border-radius: 40px;
+  background: linear-gradient(180deg, var(--bc-grad-top) 0%, var(--bc-grad-mid) 48%, var(--bc-grad-bot) 100%);
+  box-shadow: 0 18px 40px var(--bc-shadow);
+  border: 1px solid rgba(255,255,255,.22);
+  padding: 14px 14px 12px 14px;
+  position: relative;
+}
+
+#bc_phone_logo{
+  position: absolute;
+  top: 12px;
+  left: 50%;
+  transform: translateX(-50%);
+  z-index: 10;
+}
+
+.bc-logo-wrap{
+  width: 96px;
+  height: 96px;
+  border-radius: 999px;
+  background: var(--bc-accent);
+  display:flex;
+  align-items:center;
+  justify-content:center;
+  box-shadow: 0 10px 22px rgba(0,0,0,.20);
+}
+
+.bc-logo-img{
+  width: 84px;
+  height: 84px;
+  object-fit: contain;
+  display:block;
+}
+
+.bc-logo-fallback{
+  width: 84px;
+  height: 84px;
+  border-radius: 999px;
+  display:flex;
+  align-items:center;
+  justify-content:center;
+  font-weight: 900;
+  color: var(--bc-btn-ink);
+  text-align:center;
+  font-size: 12px;
+  line-height: 1.05;
+  background: rgba(255,255,255,.35);
+}
+
+#bc_chatbot{
+  margin-top: 96px;
+}
+
+#bc_chatbot, #bc_chatbot > div{
+  background: transparent !important;
+  border: none !important;
+  box-shadow: none !important;
+}
+
+#bc_chatbot .toolbar{ display:none !important; }
+
+/* ============================================================
+   BC_DOM_SELECTORS_START
+   If Gradio updates internal DOM, update selectors here.
+   ============================================================ */
+#bc_chatbot [data-testid="user"],
+#bc_chatbot [data-testid="bot"],
+#bc_chatbot [data-testid="assistant"]{
+  max-width: 82%;
+  border-radius: 18px !important;
+  padding: 12px 14px !important;
+  line-height: 1.35;
+  font-size: 14px;
+  color: var(--bc-ink) !important;
+  border: 0 !important;
+  box-shadow: 0 8px 18px rgba(0,0,0,.10);
+  position: relative;
+}
+
+#bc_chatbot [data-testid="user"]{
+  background: var(--bc-bubble-user) !important;
+}
+#bc_chatbot [data-testid="bot"],
+#bc_chatbot [data-testid="assistant"]{
+  background: var(--bc-bubble-bot) !important;
+}
+
+/* Tails */
+#bc_chatbot [data-testid="user"]::after{
+  content:"";
+  position:absolute;
+  right:-7px;
+  bottom: 12px;
+  width:0; height:0;
+  border-left: 10px solid var(--bc-bubble-user);
+  border-top: 8px solid transparent;
+  border-bottom: 8px solid transparent;
+}
+#bc_chatbot [data-testid="bot"]::before,
+#bc_chatbot [data-testid="assistant"]::before{
+  content:"";
+  position:absolute;
+  left:-7px;
+  bottom: 12px;
+  width:0; height:0;
+  border-right: 10px solid var(--bc-bubble-bot);
+  border-top: 8px solid transparent;
+  border-bottom: 8px solid transparent;
+}
+/* ============================================================
+   BC_DOM_SELECTORS_END
+   ============================================================ */
+
+#bc_input_row{
+  margin-top: 10px;
+  background: rgba(254,242,74,.96);
+  border-radius: 999px;
+  padding: 10px 10px;
+  box-shadow: 0 10px 22px rgba(0,0,0,.14);
+  align-items: center;
+}
+
+#bc_plus{
+  width: 34px;
+  height: 34px;
+  border-radius: 999px;
+  display:flex;
+  align-items:center;
+  justify-content:center;
+  font-weight: 900;
+  color: var(--bc-btn-ink);
+  background: rgba(255,255,255,.35);
+  user-select: none;
+}
+
+#bc_msg textarea{
+  background: rgba(255,255,255,.40) !important;
+  border-radius: 999px !important;
+  border: none !important;
+  padding: 10px 12px !important;
+  color: var(--bc-btn-ink) !important;
+  box-shadow: none !important;
+}
+#bc_msg textarea::placeholder{
+  color: rgba(43,12,70,.75) !important;
+}
+
+#bc_send{
+  min-width: 42px !important;
+  height: 38px !important;
+  border-radius: 999px !important;
+  border: none !important;
+  background: rgba(255,255,255,.35) !important;
+  color: var(--bc-btn-ink) !important;
+  font-size: 18px !important;
+  font-weight: 900 !important;
+}
+#bc_send:hover{ background: rgba(255,255,255,.55) !important; }
+
+#bc_clear{
+  max-width: 420px;
+  margin: 10px auto 0 auto;
+  border-radius: 14px !important;
+}
+
+@media (max-width: 480px){
+  #bc_phone{ max-width: 95vw; }
+  #bc_chatbot [data-testid="user"],
+  #bc_chatbot [data-testid="bot"],
+  #bc_chatbot [data-testid="assistant"]{
+    max-width: 88%;
+    font-size: 14px;
+  }
+}
+"""
+
+
+# ============================================================
+# UI
+# ============================================================
+def make_chatbot():
+    """
+    Compatibility shim:
+    - Some Gradio versions expose type="messages" (older docs mention it).
+    - Some Gradio versions default to messages and may not accept type.
+    See migration guide + chatbot docs. citeturn0search1turn39view0
+    """
+    base_kwargs = dict(
+        value=[],
+        elem_id="bc_chatbot",
+        height=560,
+        layout="bubble",
+        container=False,
+        show_label=False,
+        autoscroll=True,
+        buttons=[],
+        group_consecutive_messages=False,
+        placeholder="Ask a question, or start a quiz…",
+    )
+    try:
+        return gr.Chatbot(type="messages", **base_kwargs)
+    except TypeError:
+        # Gradio 6 may not accept `type`, but still expects messages-format values.
+        return gr.Chatbot(**base_kwargs)
+
+
+with gr.Blocks() as demo:
+    quiz_state = gr.State({"active": False, "quiz_data": None, "language_mode": "Auto"})
+
+    with gr.Accordion("Settings", open=False, elem_id="bc_settings"):
+        mode = gr.Dropdown(
+            choices=["Explain", "Detailed", "Short Notes", "Flashcards", "Case-Based", "Quiz Me"],
+            value="Explain",
+            label="Tutor Mode",
+        )
+        language_mode = gr.Dropdown(
+            choices=["Auto", "English", "Spanish", "Bilingual"],
+            value="Auto",
+            label="Answer Language",
+        )
+        quiz_count_mode = gr.Dropdown(
+            choices=["Auto", "3", "5", "7"],
+            value="Auto",
+            label="Quiz Questions",
+        )
+        show_sources = gr.Checkbox(value=True, label="Show Sources")
+        gr.Markdown(
+            "Tip: Choose **Quiz Me** and type a topic (e.g., `cranial nerves`). "
+            "Your next message will be evaluated automatically."
+        )
+
+    with gr.Group(elem_id="bc_phone"):
+        gr.HTML(f'<div class="bc-logo-wrap">{phone_logo_html()}</div>', elem_id="bc_phone_logo")
+
+        chatbot = make_chatbot()
+
+        with gr.Row(elem_id="bc_input_row"):
+            gr.HTML("<div>+</div>", elem_id="bc_plus")
+            msg = gr.Textbox(
+                placeholder="Type a message…",
+                show_label=False,
+                container=False,
+                scale=8,
+                elem_id="bc_msg",
+            )
+            send_btn = gr.Button("➤", elem_id="bc_send", scale=1)
+
+    clear_btn = gr.Button("Clear chat", elem_id="bc_clear")
+
+    msg.submit(
+        respond,
+        inputs=[msg, chatbot, mode, language_mode, quiz_count_mode, show_sources, quiz_state],
+        outputs=[msg, chatbot, quiz_state],
+    )
+    send_btn.click(
+        respond,
+        inputs=[msg, chatbot, mode, language_mode, quiz_count_mode, show_sources, quiz_state],
+        outputs=[msg, chatbot, quiz_state],
+    )
+    clear_btn.click(
+        clear_all,
+        inputs=None,
+        outputs=[msg, chatbot, quiz_state],
+        queue=False,
+    )
+
+if __name__ == "__main__":
+    demo.queue()
+    demo.launch(css=CSS)