Upload 2 files

app.py

@@ -0,0 +1,399 @@
+"""
+CarmelGPT — Hugging Face Spaces deployment
+Carmel Convent Senior Secondary School, Ujjain
+
+HOW TO UPDATE — HF Space → Files → app.py → pencil icon:
+  • Add facts to CUSTOM_QA (Section 2)
+  • Add facts to SCHOOL_FACTS string (Section 3) so Qwen knows them too
+"""
+
+import re
+import difflib
+import gradio as gr
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SECTION 1 — BUILT-IN Q&A  (short keys → reliable matching)
+# Rule: keys have 1-3 important words only. ANY key word found = strong match.
+# ─────────────────────────────────────────────────────────────────────────────
+
+BUILTIN_QA = {
+    "principal":
+        "The principal of Carmel Convent School, Ujjain is **Dr. Sr. Kripa Maria, PhD**. "
+        "She took charge on June 10, 2023.",
+
+    "school name":
+        "The full name is **Carmel Convent Senior Secondary School, Ujjain**.",
+
+    "address location":
+        "**Address:** Agar Road, Chak Kamed, Ujjain, Madhya Pradesh - 456001.  \n"
+        "📞 8871850918 | 📧 carmelkamedujn@yahoo.com | 🌐 www.carmelujjain.com",
+
+    "phone contact mobile":
+        "The phone number is **8871850918**.",
+
+    "email mail":
+        "The email address is **carmelkamedujn@yahoo.com**.",
+
+    "website online":
+        "The official website is **www.carmelujjain.com**.",
+
+    "cbse board affiliated":
+        "Carmel Convent School is affiliated to **CBSE** (Central Board of Secondary Education), "
+        "New Delhi, since February 2007. Affiliation number: **1030281**.",
+
+    "affiliation number":
+        "The CBSE affiliation number is **1030281**.",
+
+    "established founded year":
+        "Carmel Convent School, Ujjain was **established in 1997** by the Sisters of the "
+        "Congregation of Mother of Carmel (C.M.C.). CBSE affiliation: February 2007.",
+
+    "cmc congregation sisters":
+        "**CMC** = Congregation of Mother of Carmel — an indigenous religious congregation. "
+        "CMC entered Madhya Pradesh in 1968. The Mount Carmel Province was established in 1997.",
+
+    "managed management society":
+        "Managed by the **Mount Carmel Education Society**, run by the CMC sisters.",
+
+    "vision mission goal":
+        "The vision is the **all-round development** of every learner — academically, socially, "
+        "culturally, physically, and psychologically.",
+
+    "classes nursery grades":
+        "Classes from **Nursery (age 3+) to Class 12**. Nursery/KG I entry: 60 seats.",
+
+    "streams science commerce":
+        "Classes 11–12 offer four streams:  \n"
+        "• Science PCM (Physics, Chemistry, Mathematics)  \n"
+        "• Science PCB (Physics, Chemistry, Biology)  \n"
+        "• Commerce  \n"
+        "• Arts",
+
+    "medium language english":
+        "The medium of instruction is **English**.",
+
+    "classrooms rooms":
+        "**61 air-conditioned smart classrooms** with digital boards.",
+
+    "teachers staff faculty":
+        "**64 staff members**: 36 PRTs, 15 TGTs, 11 PGTs, 2 non-teaching.",
+
+    "campus area size":
+        "The campus spans **11,412 square metres**.",
+
+    "facilities infrastructure":
+        "Facilities: library, **6 computer labs**, science lab, auditorium, activity room, "
+        "smart classrooms, Wi-Fi, medical centre, transport, NCC, Scouts & Guides, "
+        "indoor and outdoor sports, AC classrooms.",
+
+    "computer labs laboratory":
+        "Yes — **6 computer labs**.",
+
+    "library books":
+        "Yes — Carmel Convent School has a **library**.",
+
+    "wifi internet":
+        "Yes — **Wi-Fi** connectivity across the campus.",
+
+    "transport bus":
+        "Yes — **transport services** are available for students.",
+
+    "coeducational boys girls":
+        "Yes — **co-educational**, welcoming both boys and girls.",
+
+    "admission apply enroll":
+        "**Admission process:**  \n"
+        "1. Register online at carmelujjain.com/registration2025-26  \n"
+        "2. Print and submit with: birth certificate, 5 passport photos, "
+        "Aadhar cards of both parents and student.  \n"
+        "Minimum age for Nursery: **3 years**.",
+
+    "documents required admission":
+        "Documents for admission: birth certificate, 5 passport photos, "
+        "filled form, Aadhar cards of parents and student.",
+
+    "minimum age nursery":
+        "Minimum age for Nursery: **3 years**.",
+
+    "activities ncc sports extracurricular":
+        "Activities: NCC, Scouts & Guides, sports, yoga, debates, drama, music, "
+        "science exhibitions, cooking classes, cultural programs.",
+
+    "fees cost charges":
+        "For fee details, contact: 📞 **8871850918** or visit **www.carmelujjain.com**.",
+}
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SECTION 2 — YOUR CUSTOM Q&A  (add new pairs here anytime)
+# ─────────────────────────────────────────────────────────────────────────────
+
+CUSTOM_QA = {
+    # Format:  "keyword1 keyword2":  "Answer text",
+    # EXAMPLES (remove # to activate):
+    # "commerce rini faculty":    "The Commerce faculty includes **Sr. Rini Maria**.",
+    # "robotics teacher":         "Our Robotics teachers include Ritu Ma'am and others.",
+    # "founder chavara cmc":      "CMC was founded by Blessed Chavara Kuriakose Elias.",
+}
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SECTION 3 — FACTS FOR QWEN  (Qwen reads this as its knowledge base)
+# Update this whenever you add new custom facts above.
+# ─────────────────────────────────────────────────────────────────────────────
+
+SCHOOL_FACTS = """
+CARMEL CONVENT SENIOR SECONDARY SCHOOL, UJJAIN — VERIFIED FACTS
+
+- Full name: Carmel Convent Senior Secondary School, Ujjain
+- Principal: Dr. Sr. Kripa Maria, PhD (joined June 10, 2023)
+- Address: Agar Road, Chak Kamed, Ujjain, Madhya Pradesh - 456001
+- Phone: 8871850918  |  Email: carmelkamedujn@yahoo.com  |  Website: www.carmelujjain.com
+- Board: CBSE, affiliated since February 2007, affiliation number 1030281
+- Established: 1997 by Sisters of the Congregation of Mother of Carmel (CMC)
+- Managed by: Mount Carmel Education Society
+- Type: Co-educational, English medium, CBSE Day School
+- Classes: Nursery (age 3+) to Class 12
+- Class 11-12 streams: Science PCM, Science PCB, Commerce, Arts
+- Campus: 11,412 sq metres, 61 air-conditioned smart classrooms
+- Staff: 64 total (36 PRTs, 15 TGTs, 11 PGTs, 2 non-teaching)
+- Facilities: 6 computer labs, 1 science lab, library, auditorium, Wi-Fi, transport
+- Activities: NCC, Scouts & Guides, sports, yoga, debates, drama, music, cultural programs
+- Admission: Register at carmelujjain.com; minimum age 3 years for Nursery
+- Admission documents: birth certificate, 5 photos, filled form, Aadhar cards of parents+student
+"""
+
+CUSTOM_FACTS = ""  # Add custom facts here as bullet points if needed
+
+ALL_FACTS = SCHOOL_FACTS + CUSTOM_FACTS
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SECTION 4 — TYPO CORRECTION  (do not edit)
+# ─────────────────────────────────────────────────────────────────────────────
+
+CORRECTIONS = {
+    "principle":    "principal",   "prinicpal":   "principal",
+    "princpal":     "principal",   "principel":   "principal",
+    "admision":     "admission",   "admmission":  "admission",
+    "facilites":    "facilities",  "facilty":     "facility",
+    "afiliation":   "affiliation", "affilitation":"affiliation",
+    "establised":   "established", "transprt":    "transport",
+    "tranport":     "transport",   "scince":      "science",
+    "commerse":     "commerce",    "streems":     "streams",
+    "clases":       "classes",     "ujain":       "ujjain",
+    "karmel":       "carmel",      "carmell":     "carmel",
+    "whos":         "who",         "whats":       "what",
+    "hows":         "how",         "wheres":      "where",
+}
+
+KNOWN_WORDS = [
+    "principal", "admission", "facilities", "affiliation", "established",
+    "transport", "science", "commerce", "streams", "classes", "ujjain",
+    "carmel", "cbse", "board", "campus", "library", "laboratory",
+    "teacher", "faculty", "staff", "nursery", "website", "contact",
+    "phone", "email", "fees", "activities", "ncc", "medium", "english",
+    "classrooms", "auditorium", "wifi", "internet", "management",
+]
+
+def correct_typos(text: str):
+    words = text.lower().split()
+    corrected, corrections_made = [], []
+    for word in words:
+        clean = re.sub(r"[^\w]", "", word)
+        if clean in CORRECTIONS:
+            fix = CORRECTIONS[clean]
+            corrected.append(fix)
+            if fix != clean:
+                corrections_made.append((clean, fix))
+        elif len(clean) > 4:
+            matches = difflib.get_close_matches(clean, KNOWN_WORDS, n=1, cutoff=0.82)
+            if matches and matches[0] != clean:
+                corrected.append(matches[0])
+                corrections_made.append((clean, matches[0]))
+            else:
+                corrected.append(clean)
+        else:
+            corrected.append(clean)
+    return " ".join(corrected), corrections_made
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SECTION 5 — MATCHING ENGINE  (do not edit)
+# ─────────────────────────────────────────────────────────────────────────────
+
+ALL_QA = {**BUILTIN_QA, **CUSTOM_QA}
+
+# Words that mean the question IS about this school
+SCHOOL_SIGNALS = {
+    "school", "carmel", "principal", "admission", "teacher", "faculty",
+    "class", "stream", "science", "commerce", "cbse", "board", "campus",
+    "library", "transport", "ncc", "ujjain", "cmc", "fees", "address",
+    "staff", "nursery", "facilities", "established", "founded", "vision",
+    "website", "email", "phone", "contact", "classrooms", "head",
+    "headmistress", "director", "kripa", "medium", "wifi", "auditorium",
+    "affiliation", "congregation", "sisters", "activities", "uniform",
+}
+
+def _score(stored_keywords: str, user_q: str) -> float:
+    """
+    Short-key scoring:
+    - Extract key words (len > 3) from stored key
+    - ANY long word match (len > 6) = strong signal → score 0.9
+    - Fraction of key words found for shorter words
+    """
+    keywords = [w for w in stored_keywords.split() if len(w) > 3]
+    if not keywords:
+        return 0.0
+    matched = [w for w in keywords if w in user_q]
+    if not matched:
+        return 0.0
+    # Long specific word matched → very strong signal
+    if any(len(w) > 6 for w in matched):
+        return 0.9
+    # Short word matched → score by fraction
+    return len(matched) / len(keywords)
+
+def direct_lookup(q: str):
+    best_score, best_answer = 0.0, None
+    for stored_keywords, answer in ALL_QA.items():
+        s = _score(stored_keywords, q)
+        if s > best_score:
+            best_score, best_answer = s, answer
+    return best_answer, best_score
+
+def is_school_question(q: str) -> bool:
+    return bool(set(q.split()) & SCHOOL_SIGNALS)
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SECTION 6 — QWEN FALLBACK  (handles unusual phrasings)
+# ─────────────────────────────────────────────────────────────────────────────
+
+try:
+    from huggingface_hub import InferenceClient
+    _client = InferenceClient("Qwen/Qwen2.5-1.5B-Instruct")
+    QWEN_AVAILABLE = True
+except Exception:
+    QWEN_AVAILABLE = False
+
+QWEN_SYSTEM = f"""You are CarmelGPT — a chatbot ONLY for Carmel Convent Senior Secondary School, Ujjain, India.
+
+RULES (never break):
+1. Answer ONLY using the SCHOOL FACTS below. Never invent names, numbers, or details.
+2. If not in SCHOOL FACTS, say exactly: "I don't have that information. Contact: 8871850918"
+3. If asked about anything unrelated to this school, say exactly: "I can only answer questions about Carmel Convent Senior Secondary School, Ujjain."
+4. Keep answers short and factual.
+
+SCHOOL FACTS:
+{ALL_FACTS}"""
+
+def ask_qwen(question: str):
+    if not QWEN_AVAILABLE:
+        return None
+    try:
+        r = _client.chat_completion(
+            messages=[
+                {"role": "system", "content": QWEN_SYSTEM},
+                {"role": "user",   "content": question},
+            ],
+            max_tokens=250,
+            temperature=0.1,
+        )
+        return r.choices[0].message.content.strip()
+    except Exception:
+        return None
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SECTION 7 — CHAT FUNCTION  (do not edit)
+# ─────────────────────────────────────────────────────────────────────────────
+
+OUT_OF_SCOPE = (
+    "Sorry, I can only answer questions about "
+    "**Carmel Convent Senior Secondary School, Ujjain**.  \n"
+    "Please ask about admissions, facilities, faculty, streams, or other school topics."
+)
+
+NO_INFO = (
+    "I don't have that specific information right now.  \n"
+    "Please contact the school:  \n"
+    "📞 **8871850918** | 📧 carmelkamedujn@yahoo.com | 🌐 www.carmelujjain.com"
+)
+
+def chat(message: str, history) -> str:
+    if not message.strip():
+        return ""
+
+    # Step 1 — Typo correction
+    corrected_q, corrections = correct_typos(message)
+    note = ""
+    if corrections:
+        fixes = ", ".join(f'"{a}" → "{b}"' for a, b in corrections)
+        note = f"*(Correcting spelling: {fixes})*  \n\n"
+
+    # Step 2 — Direct lookup (fast, 100% accurate)
+    answer, score = direct_lookup(corrected_q)
+    if score >= 0.3:
+        return note + answer
+
+    # Step 3 — Scope check
+    if not is_school_question(corrected_q):
+        return OUT_OF_SCOPE
+
+    # Step 4 — Qwen with guardrails (handles unusual phrasings)
+    qwen_ans = ask_qwen(message)
+    if qwen_ans:
+        return note + qwen_ans
+
+    # Step 5 — Fallback
+    return note + NO_INFO
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SECTION 8 — UI
+# ─────────────────────────────────────────────────────────────────────────────
+
+EXAMPLES = [
+    "Who is the principal?",
+    "Whos is the principle?",
+    "Do you have a principal?",
+    "What is the name of your principal?",
+    "Don't you have a principal?",
+    "What streams are available in Class 11?",
+    "How do I get admission?",
+    "What is the CBSE affiliation number?",
+    "Does the school have NCC?",
+    "Who is the President of India?",
+]
+
+with gr.Blocks(title="CarmelGPT") as demo:
+    gr.Markdown("# CarmelGPT")
+    gr.Markdown(
+        "**Carmel Convent Senior Secondary School, Ujjain**  \n"
+        "Ask me anything about the school."
+    )
+    with gr.Row():
+        with gr.Column(scale=3):
+            chatbot = gr.Chatbot(height=460, label="CarmelGPT")
+            msg = gr.Textbox(
+                placeholder="Ask about admissions, principal, facilities...",
+                label="Your question", lines=1)
+            with gr.Row():
+                send_btn  = gr.Button("Ask", variant="primary")
+                clear_btn = gr.Button("Clear")
+        with gr.Column(scale=1):
+            gr.Markdown("### Try these")
+            for q in EXAMPLES:
+                gr.Button(q, size="sm").click(lambda x=q: x, outputs=msg)
+            gr.Markdown("---")
+            gr.Markdown(
+                "**Contact**  \n📞 8871850918  \n"
+                "📧 carmelkamedujn@yahoo.com  \n"
+                "🌐 [carmelujjain.com](http://www.carmelujjain.com)")
+
+    def submit(m, h):
+        if not m.strip():
+            return h, ""
+        reply = chat(m, h)
+        return h + [{"role": "user", "content": m},
+                    {"role": "assistant", "content": reply}], ""
+
+    send_btn.click(submit, [msg, chatbot], [chatbot, msg])
+    msg.submit(submit,    [msg, chatbot], [chatbot, msg])
+    clear_btn.click(lambda: ([], ""), outputs=[chatbot, msg])
+
+demo.launch()

requirements.txt

@@ -0,0 +1,2 @@
+gradio
+huggingface_hub