Update app.py

app.py

@@ -1,8 +1,9 @@
 from __future__ import annotations
 
-import os
-import re
+import base64
 import html
+import mimetypes
+import re
 from pathlib import Path
 from typing import List, Tuple, Dict, Optional
 
@@ -34,6 +35,7 @@ SEARCH_DIRS = [
 ]
 
 SUPPORTED_EXTENSIONS = {".txt", ".md", ".pdf", ".docx"}
+
 STOPWORDS = {
     "the", "is", "am", "are", "was", "were", "be", "been", "being",
     "a", "an", "and", "or", "of", "to", "in", "on", "for", "with",
@@ -58,6 +60,14 @@ def tokenize(text: str) -> List[str]:
     return [w for w in words if w not in STOPWORDS and len(w) > 1]
 
 
+def split_sentences(text: str) -> List[str]:
+    text = normalize_spaces(text)
+    if not text:
+        return []
+    parts = re.split(r"(?<=[.!?])\s+", text)
+    return [p.strip() for p in parts if p.strip()]
+
+
 def chunk_text(text: str, chunk_size: int = 900, overlap: int = 150) -> List[str]:
     text = normalize_spaces(text)
     if not text:
@@ -135,7 +145,7 @@ def extract_text_from_file(path: Path) -> str:
     return ""
 
 
-def find_asset(possible_names: List[str]) -> Optional[str]:
+def find_asset(possible_names: List[str]) -> Optional[Path]:
     lowered = [x.lower() for x in possible_names]
 
     for d in SEARCH_DIRS:
@@ -143,17 +153,35 @@ def find_asset(possible_names: List[str]) -> Optional[str]:
             for name in possible_names:
                 p = d / name
                 if p.exists() and p.is_file():
-                    return str(p)
+                    return p
 
     for d in SEARCH_DIRS:
         if d.exists():
             for p in d.rglob("*"):
                 if p.is_file() and p.name.lower() in lowered:
-                    return str(p)
+                    return p
     return None
 
 
-LOGO_PATH = find_asset([
+def file_to_data_uri(path: Optional[Path]) -> Optional[str]:
+    if not path or not path.exists():
+        return None
+    try:
+        data = path.read_bytes()
+        mime, _ = mimetypes.guess_type(str(path))
+        if mime is None:
+            if path.suffix.lower() == ".svg":
+                mime = "image/svg+xml"
+            else:
+                mime = "image/png"
+        encoded = base64.b64encode(data).decode("utf-8")
+        return f"data:{mime};base64,{encoded}"
+    except Exception:
+        return None
+
+
+LOGO_FILE = find_asset([
+    "Brain Chat Imagen.svg",
     "Brain chat-09.png",
     "BrainChat-09.png",
     "brain chat-09.png",
@@ -162,8 +190,12 @@ LOGO_PATH = find_asset([
     "brainchat_logo.png",
     "logo.png",
     "Logo.png",
+    "logo.svg",
+    "Logo.svg",
 ])
 
+LOGO_URI = file_to_data_uri(LOGO_FILE)
+
 
 # =========================================================
 # KNOWLEDGE BASE
@@ -225,7 +257,6 @@ class LocalKnowledgeBase:
             return []
 
         scored = []
-
         for item in self.chunks:
             overlap = len(q_tokens.intersection(item["tokens"]))
             if overlap == 0:
@@ -237,12 +268,12 @@ class LocalKnowledgeBase:
         scored.sort(key=lambda x: x[0], reverse=True)
 
         unique = []
-        seen_text = set()
+        seen_keys = set()
         for score, item in scored:
             key = (item["source"], item["chunk_id"])
-            if key in seen_text:
+            if key in seen_keys:
                 continue
-            seen_text.add(key)
+            seen_keys.add(key)
 
             result = dict(item)
             result["score"] = score
@@ -259,41 +290,204 @@ KB.load_from_directories()
 
 
 # =========================================================
-# RESPONSE LOGIC
+# CONTENT BUILDERS
 # =========================================================
-def build_answer_from_hits(query: str, hits: List[Dict], tutor_mode: str) -> Tuple[str, List[str]]:
+def collect_context(hits: List[Dict], max_chars: int = 2000) -> Tuple[str, List[str]]:
     if not hits:
-        return NOT_FOUND_TEXT, []
+        return "", []
 
-    mode = (tutor_mode or "Detailed").lower()
-    max_snippets = 2 if mode == "brief" else 4
-
-    selected = hits[:max_snippets]
-    snippets = []
+    chunks = []
     sources = []
-
     seen_sources = set()
-    for h in selected:
-        snippet = h["text"].strip()
-        if len(snippet) > 420 and mode == "brief":
-            snippet = snippet[:420].rsplit(" ", 1)[0] + "..."
-        elif len(snippet) > 750 and mode != "brief":
-            snippet = snippet[:750].rsplit(" ", 1)[0] + "..."
-
-        snippets.append(snippet)
-
-        source_label = h["source"]
-        if source_label not in seen_sources:
-            seen_sources.add(source_label)
-            sources.append(source_label)
-
-    if mode == "brief":
-        answer = "\n\n".join(snippets[:2])
+    total = 0
+
+    for h in hits:
+        txt = normalize_spaces(h["text"])
+        if not txt:
+            continue
+
+        remaining = max_chars - total
+        if remaining <= 0:
+            break
+
+        piece = txt[:remaining]
+        chunks.append(piece)
+        total += len(piece)
+
+        src = h["source"]
+        if src not in seen_sources:
+            seen_sources.add(src)
+            sources.append(src)
+
+    return "\n\n".join(chunks), sources
+
+
+def simple_definition_style(text: str, short: bool = False) -> str:
+    sentences = split_sentences(text)
+    if not sentences:
+        return NOT_FOUND_TEXT
+
+    if short:
+        chosen = sentences[:3]
+        return "\n\n".join(chosen)
+
+    chosen = sentences[:6]
+    return "\n\n".join(chosen)
+
+
+def detailed_teaching_style(query: str, text: str) -> str:
+    sentences = split_sentences(text)
+    if not sentences:
+        return NOT_FOUND_TEXT
+
+    first = sentences[:2]
+    rest = sentences[2:6]
+
+    out = [f"**Topic:** {query.strip()}"]
+
+    if first:
+        out.append("\n**Simple explanation:**")
+        out.append(" ".join(first))
+
+    if rest:
+        out.append("\n**More detail:**")
+        for s in rest:
+            out.append(f"- {s}")
+
+    out.append("\n**Why this matters:**")
+    out.append("This concept is important because it helps students connect theory with brain structure, function, and clinical understanding.")
+
+    return "\n".join(out)
+
+
+def build_flashcards(query: str, text: str, count: int = 5) -> str:
+    sentences = split_sentences(text)
+    words = [w for w in re.findall(r"[A-Za-z][A-Za-z\-]+", text) if len(w) > 5]
+
+    terms = []
+    seen = set()
+    for w in words:
+        wl = w.lower()
+        if wl not in seen:
+            seen.add(wl)
+            terms.append(w)
+        if len(terms) >= count:
+            break
+
+    if not terms:
+        terms = [query.title()]
+
+    out = ["**Flash Cards**"]
+    for i, term in enumerate(terms, 1):
+        meaning = sentences[min(i - 1, len(sentences) - 1)] if sentences else f"{term} is an important concept from the material."
+        out.append(f"\n**Card {i}**")
+        out.append(f"- **Term:** {term}")
+        out.append(f"- **Meaning:** {meaning}")
+        out.append(f"- **Remember:** Know the role, location, or function of {term.lower()}.")
+
+    return "\n".join(out)
+
+
+def build_case_study(query: str, text: str) -> str:
+    sentences = split_sentences(text)
+    base = " ".join(sentences[:3]) if sentences else f"This topic is related to {query}."
+
+    return (
+        "**Case Study**\n\n"
+        f"A student is reviewing a patient-related topic connected to **{query}**. "
+        f"While studying, the student learns that: {base}\n\n"
+        "**Scenario:**\n"
+        "A patient comes with symptoms that may involve this part of the nervous system. "
+        "The student must explain what structure or concept is involved, what it normally does, "
+        "and what may happen when it is affected.\n\n"
+        "**Questions to think about:**\n"
+        "1. What is the main structure or idea here?\n"
+        "2. What is its normal function?\n"
+        "3. What symptoms may appear if it is damaged or disturbed?\n"
+        "4. Why is this topic important in neuroanatomy or neurology?\n\n"
+        "**Teacher-style note:**\n"
+        "This case study is for understanding and classroom learning, not for real medical diagnosis."
+    )
+
+
+def build_quiz(query: str, text: str, n_questions: int = 5) -> str:
+    words = [w for w in re.findall(r"[A-Za-z][A-Za-z\-]+", text) if len(w) > 5]
+    unique_words = []
+    seen = set()
+
+    for w in words:
+        wl = w.lower()
+        if wl not in seen:
+            seen.add(wl)
+            unique_words.append(w)
+        if len(unique_words) >= n_questions:
+            break
+
+    if not unique_words:
+        unique_words = [query.title()] * n_questions
+
+    out = [f"**Quiz: {query.strip()}**"]
+    for i, w in enumerate(unique_words[:n_questions], 1):
+        out.append(f"{i}. What is **{w}**?")
+        out.append(f"{i}.a Why is **{w}** important?")
+    return "\n".join(out)
+
+
+def build_revision_questions(query: str, text: str, n_questions: int = 5) -> str:
+    sentences = split_sentences(text)
+    out = [f"**Revision Questions: {query.strip()}**"]
+
+    for i in range(1, n_questions + 1):
+        if i == 1:
+            out.append(f"{i}. Define {query.strip()} in simple words.")
+        elif i == 2:
+            out.append(f"{i}. Explain the main function or role of {query.strip()}.")
+        elif i == 3:
+            out.append(f"{i}. Why is {query.strip()} important in neurology or neuroanatomy?")
+        elif i == 4:
+            out.append(f"{i}. Give one example related to {query.strip()}.")
+        else:
+            if sentences:
+                out.append(f"{i}. Based on the material, explain this statement: “{sentences[min(i-5, len(sentences)-1)]}”")
+            else:
+                out.append(f"{i}. Write short notes on {query.strip()}.")
+
+    return "\n".join(out)
+
+
+def make_learning_output(
+    query: str,
+    learning_mode: str,
+    tutor_mode: str,
+    quiz_questions: str,
+    hits: List[Dict]
+) -> Tuple[str, List[str]]:
+    if not hits:
+        return NOT_FOUND_TEXT, []
+
+    context, sources = collect_context(hits, max_chars=2400)
+    mode = (learning_mode or "Detailed Explanation").strip().lower()
+
+    qn = 5 if str(quiz_questions).lower() == "auto" else int(quiz_questions)
+
+    if mode == "normal answer":
+        answer = simple_definition_style(context, short=False)
+    elif mode == "short explanation":
+        answer = simple_definition_style(context, short=True)
+    elif mode == "detailed explanation":
+        answer = detailed_teaching_style(query, context)
+    elif mode == "flash cards":
+        answer = build_flashcards(query, context, count=min(max(qn, 3), 8))
+    elif mode == "case study":
+        answer = build_case_study(query, context)
+    elif mode == "quiz":
+        answer = build_quiz(query, context, n_questions=qn)
+    elif mode == "revision questions":
+        answer = build_revision_questions(query, context, n_questions=qn)
     else:
-        intro = f"Here is what I found related to: “{query.strip()}”\n\n"
-        answer = intro + "\n\n".join(snippets)
+        answer = detailed_teaching_style(query, context)
 
-    return answer.strip(), sources
+    return answer, sources
 
 
 def format_answer(answer_text: str, sources: List[str], show_sources: bool) -> str:
@@ -312,6 +506,7 @@ def format_answer(answer_text: str, sources: List[str], show_sources: bool) -> s
 
 def get_answer_and_sources(
     message: str,
+    learning_mode: str,
     tutor_mode: str,
     answer_language: str,
     quiz_questions: str
@@ -324,28 +519,13 @@ def get_answer_and_sources(
 
     if lower_msg in {"hi", "hello", "hey"}:
         return (
-            "Hello. Ask me anything from your uploaded neurology or neuroanatomy material.",
+            "Hello. Ask me anything from your neurology or neuroanatomy material. "
+            "You can also choose a learning mode such as flash cards, case study, quiz, or detailed explanation.",
             []
         )
 
-    if "quiz" in lower_msg:
-        hits = KB.search(msg, top_k=5)
-        if not hits:
-            return NOT_FOUND_TEXT, []
-
-        qn = 5 if str(quiz_questions).lower() == "auto" else int(quiz_questions)
-        base_text = hits[0]["text"]
-
-        quiz = [f"**Mini Quiz ({qn} questions)**"]
-        words = [w for w in re.findall(r"[A-Za-z][A-Za-z\-]+", base_text) if len(w) > 5][:qn]
-
-        for i, w in enumerate(words[:qn], 1):
-            quiz.append(f"{i}. Explain the term **{w}** in simple words.")
-
-        return "\n".join(quiz), [hits[0]["source"]]
-
-    hits = KB.search(msg, top_k=5)
-    return build_answer_from_hits(msg, hits, tutor_mode)
+    hits = KB.search(msg, top_k=6)
+    return make_learning_output(msg, learning_mode, tutor_mode, quiz_questions, hits)
 
 
 # =========================================================
@@ -353,66 +533,68 @@ def get_answer_and_sources(
 # =========================================================
 CUSTOM_CSS = """
 :root{
-    --bg-main: #f6f7ff;
+    --bg1: #fffdf3;
+    --bg2: #fff4fb;
     --panel: #ffffff;
-    --text: #202545;
-    --muted: #616889;
-    --primary: #6d28ff;
-    --secondary: #ff48c4;
-    --accent: #ffe94d;
-    --accent2: #45d8ff;
-    --border: #dde4ff;
+    --text: #2f2a4d;
+    --muted: #736d92;
+    --primary: #7c3aed;
+    --secondary: #ec4899;
+    --third: #f59e0b;
+    --mint: #6ee7b7;
+    --sky: #67e8f9;
+    --border: #eadcff;
 }
 
 html, body, .gradio-container {
-    background: linear-gradient(180deg, #f6f7ff 0%, #fffbe0 100%) !important;
+    background: linear-gradient(180deg, #fffdf3 0%, #fff7fb 55%, #fff2d7 100%) !important;
     color: var(--text) !important;
     font-family: "Segoe UI", Arial, sans-serif !important;
 }
 
 #main_shell {
-    max-width: 1200px;
+    max-width: 1180px;
     margin: 18px auto;
-    padding: 0 10px 18px 10px;
+    padding: 0 12px 18px 12px;
 }
 
 #topbar {
-    background: linear-gradient(90deg, #6d28ff 0%, #ff48c4 60%, #ffe94d 100%);
-    border-radius: 28px;
-    padding: 16px 18px;
-    box-shadow: 0 12px 28px rgba(80, 64, 170, 0.22);
-    border: 2px solid rgba(255,255,255,0.65);
-    margin-bottom: 16px;
+    background: linear-gradient(90deg, #7c3aed 0%, #ec4899 55%, #f4d03f 100%);
+    border-radius: 30px;
+    padding: 18px 20px;
+    box-shadow: 0 12px 28px rgba(167, 85, 190, 0.22);
+    border: 2px solid rgba(255,255,255,0.7);
+    margin-bottom: 18px;
 }
 
 #brand_row {
     display: flex;
     align-items: center;
-    gap: 14px;
+    gap: 16px;
 }
 
 #brand_logo {
-    width: 74px;
-    height: 74px;
-    border-radius: 18px;
-    object-fit: cover;
-    background: white;
-    padding: 4px;
-    box-shadow: 0 6px 18px rgba(0,0,0,0.15);
+    width: 78px;
+    height: 78px;
+    border-radius: 20px;
+    object-fit: contain;
+    background: rgba(255,255,255,0.96);
+    padding: 6px;
+    box-shadow: 0 6px 18px rgba(0,0,0,0.12);
 }
 
 #brand_fallback {
-    width: 74px;
-    height: 74px;
-    border-radius: 18px;
+    width: 78px;
+    height: 78px;
+    border-radius: 20px;
     display: flex;
     align-items: center;
     justify-content: center;
-    background: white;
-    color: #6d28ff;
-    font-size: 24px;
+    background: rgba(255,255,255,0.96);
+    color: #7c3aed;
+    font-size: 28px;
     font-weight: 800;
-    box-shadow: 0 6px 18px rgba(0,0,0,0.15);
+    box-shadow: 0 6px 18px rgba(0,0,0,0.12);
 }
 
 #brand_title {
@@ -427,79 +609,89 @@ html, body, .gradio-container {
     font-size: 15px;
     color: #fffdfd;
     font-weight: 600;
-    margin-top: 2px;
+    margin-top: 3px;
 }
 
 #settings_card, #chat_card {
-    background: rgba(255,255,255,0.95) !important;
-    border: 2px solid var(--border) !important;
-    border-radius: 22px !important;
-    box-shadow: 0 10px 26px rgba(71, 89, 160, 0.10);
+    background: rgba(255,255,255,0.92) !important;
+    border: 2px solid #eadcff !important;
+    border-radius: 24px !important;
+    box-shadow: 0 10px 24px rgba(120, 100, 180, 0.08);
 }
 
 #chatbot {
-    background: linear-gradient(180deg, #fffdf6 0%, #f8fbff 100%) !important;
-    border-radius: 18px !important;
+    background: linear-gradient(180deg, #fffefb 0%, #fff7ff 100%) !important;
+    border-radius: 20px !important;
 }
 
 #chatbot .message.user {
-    background: linear-gradient(90deg, #6d28ff, #9247ff) !important;
+    background: linear-gradient(90deg, #8b5cf6, #ec4899) !important;
     color: white !important;
     border-radius: 18px !important;
 }
 
 #chatbot .message.bot {
-    background: #fff7bf !important;
-    color: #202545 !important;
-    border: 1px solid #eedb66 !important;
+    background: linear-gradient(180deg, #fff8d9 0%, #fffdf3 100%) !important;
+    color: #2f2a4d !important;
+    border: 1px solid #f3df8c !important;
     border-radius: 18px !important;
 }
 
 textarea, input, .wrap textarea {
-    border-radius: 16px !important;
-    border: 2px solid #d8def8 !important;
-    background: white !important;
+    border-radius: 18px !important;
+    border: 2px solid #e3d8ff !important;
+    background: #ffffff !important;
     color: var(--text) !important;
 }
 
 button {
-    border-radius: 16px !important;
+    border-radius: 18px !important;
     border: none !important;
     font-weight: 700 !important;
+    min-height: 46px !important;
 }
 
 #send_btn {
-    background: linear-gradient(90deg, #6d28ff, #ff48c4) !important;
+    background: linear-gradient(90deg, #7c3aed, #ec4899) !important;
     color: white !important;
 }
 
 #clear_btn {
-    background: linear-gradient(90deg, #ffe94d, #ffd930) !important;
-    color: #3a3000 !important;
+    background: linear-gradient(90deg, #fde047, #facc15) !important;
+    color: #4b3b00 !important;
 }
 
 #upload_btn {
-    background: linear-gradient(90deg, #45d8ff, #6ef0c0) !important;
-    color: #07304d !important;
+    background: linear-gradient(90deg, #67e8f9, #6ee7b7) !important;
+    color: #17324d !important;
 }
 
 #reload_btn {
-    background: linear-gradient(90deg, #ffffff, #f2f5ff) !important;
-    color: #334 !important;
-    border: 2px solid #dce3ff !important;
+    background: linear-gradient(90deg, #f9f5ff, #fff2fb) !important;
+    color: #473a69 !important;
+    border: 2px solid #e6d7ff !important;
 }
 
 .small_hint {
     color: var(--muted);
     font-size: 13px;
-    margin-top: -4px;
+}
+
+.gr-file, .gr-file * {
+    background: #fffaf0 !important;
+    color: #2f2a4d !important;
+    border-color: #f3df8c !important;
+}
+
+footer {
+    display: none !important;
 }
 """
 
 
 def build_header_html() -> str:
-    if LOGO_PATH:
-        logo_html = f'<img id="brand_logo" src="/file={LOGO_PATH}" alt="BrainChat logo">'
+    if LOGO_URI:
+        logo_html = f'<img id="brand_logo" src="{LOGO_URI}" alt="BrainChat logo">'
     else:
         logo_html = '<div id="brand_fallback">BC</div>'
 
@@ -516,22 +708,18 @@ def build_header_html() -> str:
     """
 
 
-def respond(message, history, tutor_mode, answer_language, quiz_questions, show_sources):
+def respond(message, history, learning_mode, tutor_mode, answer_language, quiz_questions, show_sources):
     history = history or []
 
     answer_text, sources = get_answer_and_sources(
         message=message,
+        learning_mode=learning_mode,
         tutor_mode=tutor_mode,
         answer_language=answer_language,
         quiz_questions=quiz_questions
     )
 
-    final_text = format_answer(
-        answer_text=answer_text,
-        sources=sources,
-        show_sources=show_sources
-    )
-
+    final_text = format_answer(answer_text, sources, show_sources)
     history.append((message, final_text))
     return history, ""
 
@@ -544,7 +732,7 @@ def reload_materials():
     global KB
     KB = LocalKnowledgeBase()
     KB.load_from_directories()
-    return "Course materials reloaded."
+    return f"Materials reloaded. Loaded {len(KB.docs)} document(s)."
 
 
 def upload_files(files):
@@ -563,8 +751,8 @@ def upload_files(files):
             continue
 
     if added == 0:
-        return "No readable text was found in the uploaded file(s)."
-    return f"{added} file(s) added to the course material."
+        return "No readable text found in uploaded files."
+    return f"{added} file(s) added successfully."
 
 
 with gr.Blocks(css=CUSTOM_CSS, title=APP_TITLE) as demo:
@@ -572,33 +760,51 @@ with gr.Blocks(css=CUSTOM_CSS, title=APP_TITLE) as demo:
         gr.HTML(build_header_html())
 
         with gr.Accordion("Settings", open=False, elem_id="settings_card"):
+            learning_mode = gr.Dropdown(
+                [
+                    "Normal Answer",
+                    "Short Explanation",
+                    "Detailed Explanation",
+                    "Flash Cards",
+                    "Case Study",
+                    "Quiz",
+                    "Revision Questions",
+                ],
+                value="Detailed Explanation",
+                label="Learning Mode"
+            )
+
             tutor_mode = gr.Dropdown(
                 ["Brief", "Detailed"],
                 value="Detailed",
                 label="Tutor Mode"
             )
+
             answer_language = gr.Dropdown(
                 ["Auto", "English", "Spanish"],
                 value="Auto",
                 label="Answer Language"
             )
+
             quiz_questions = gr.Dropdown(
                 ["Auto", "5", "10"],
                 value="Auto",
-                label="Quiz Questions"
+                label="Quiz / Revision Count"
             )
+
             show_sources = gr.Checkbox(
                 value=True,
                 label="Show Sources"
             )
+
             gr.Markdown(
-                "Sources are shown only when useful text is found.",
+                "Choose a learning mode such as flash cards, case study, quiz, short explanation, or detailed explanation.",
                 elem_classes=["small_hint"]
             )
 
         with gr.Column(elem_id="chat_card"):
             chatbot = gr.Chatbot(
-                height=520,
+                height=470,
                 elem_id="chatbot",
                 show_label=False
             )
@@ -607,7 +813,7 @@ with gr.Blocks(css=CUSTOM_CSS, title=APP_TITLE) as demo:
                 file_input = gr.File(
                     file_count="multiple",
                     file_types=[".txt", ".md", ".pdf", ".docx"],
-                    label="",
+                    label="Add material files (optional)",
                     scale=2
                 )
                 msg = gr.Textbox(
@@ -626,13 +832,13 @@ with gr.Blocks(css=CUSTOM_CSS, title=APP_TITLE) as demo:
 
     send_btn.click(
         respond,
-        inputs=[msg, chatbot, tutor_mode, answer_language, quiz_questions, show_sources],
+        inputs=[msg, chatbot, learning_mode, tutor_mode, answer_language, quiz_questions, show_sources],
         outputs=[chatbot, msg]
     )
 
     msg.submit(
         respond,
-        inputs=[msg, chatbot, tutor_mode, answer_language, quiz_questions, show_sources],
+        inputs=[msg, chatbot, learning_mode, tutor_mode, answer_language, quiz_questions, show_sources],
         outputs=[chatbot, msg]
     )