app.py

"""
PaderBot — Streamlit UI v4.

"""

import json

import streamlit as st
import streamlit.components.v1 as components
from paderbot import PaderBot

st.set_page_config(
    page_title="PaderBot — Q&A for International Students",
    page_icon="🎓",
    layout="wide",
)


@st.cache_resource(show_spinner="Loading PaderBot (first request takes ~10s)...")
def get_bot():
    return PaderBot()


EXAMPLES_EN = [
    "What is BAföG and can international students get it?",
    "How do I apply to the Computer Science Master's program?",
    "Where can students live in Paderborn?",
    "Do I need to know German to study at Paderborn?",
]

EXAMPLES_DE = [
    "Wie bewerbe ich mich für einen Masterstudiengang?",
    "Welche Englisch-sprachigen Master-Programme gibt es?",
    "Wo kann ich in Paderborn wohnen?",
    "Was kostet ein Semester an der Universität Paderborn?",
]


# Session state
if "history" not in st.session_state:
    st.session_state.history = []
if "pending_question" not in st.session_state:
    st.session_state.pending_question = None
if "edit_text" not in st.session_state:
    st.session_state.edit_text = None
if "input_key_id" not in st.session_state:
    st.session_state.input_key_id = 0


def submit_question(q: str):
    st.session_state.pending_question = q.strip()


def edit_last_question():
    if not st.session_state.history:
        return
    last = st.session_state.history.pop()
    st.session_state.edit_text = last["question"]
    st.session_state.input_key_id += 1


def copy_to_clipboard_button(text: str, button_label: str = "📋 Copy answer"):
    """Theme-adaptive copy button. Uses Streamlit's CSS variables so colors
    work on both light and dark themes."""
    safe_text = json.dumps(text)
    html = f"""
    <button onclick='navigator.clipboard.writeText({safe_text}).then(
        () => {{
            const original = this.innerText;
            this.innerText = "✓ Copied";
            setTimeout(() => this.innerText = original, 1500);
        }}
    )'
    style="
        background: var(--secondary-background-color, #f0f2f6);
        color: var(--text-color, #262730);
        border: 1px solid var(--text-color, rgba(0,0,0,0.2));
        padding: 6px 14px;
        border-radius: 6px;
        cursor: pointer;
        font-size: 13px;
        font-weight: 500;
        font-family: inherit;
    "
    onmouseover="this.style.opacity='0.85'"
    onmouseout="this.style.opacity='1'">
      {button_label}
    </button>
    """
    components.html(html, height=42)


# ============================================================
# Top bar — title + help (right)
# ============================================================
header_col1, header_col2 = st.columns([6, 1], vertical_alignment="center")
with header_col1:
    st.title("Ask PaderBot")
with header_col2:
    with st.popover("❓ Help", use_container_width=True):
        st.markdown("**How to use PaderBot**")
        st.markdown(
            "- 👈 Click any example question in the sidebar, or type your own.\n"
            "- 🌐 Answers in English or German — bot replies in the language of your question.\n"
            "- 📚 Each answer is grounded in scraped sources, with clickable citations.\n\n"
            "**Working with past answers**\n"
            "- ✏️ Click **Edit** on the latest question to refine and re-ask.\n"
            "- 📋 Click **Copy answer** to copy text for use elsewhere.\n"
            "- 🖱️ Select text in any answer with your mouse to copy part of it.\n\n"
            "**Note on follow-ups**\n"
            "Each question is answered independently — follow-ups should reference "
            "their topic explicitly. Instead of *\"How long does it take?\"*, ask "
            "*\"How long does the Computer Science Master's program take?\"*"
        )

st.caption(
    "Antworten auf Englisch oder Deutsch. Bot antwortet in der Sprache der Frage. "
    "·  Answers in English or German. Bot replies in the language of your question."
)


# Sidebar
with st.sidebar:
    st.title("🎓 PaderBot")
    st.caption("A Q&A assistant for prospective international students at Paderborn University.")

    st.divider()
    st.subheader("What I can help with")
    st.markdown(
        "- **Application & admissions** to Master's programs\n"
        "- **Visa, language requirements, fees**\n"
        "- **Housing** via Studierendenwerk Paderborn\n"
        "- **Financing** (BAföG, scholarships)\n"
        "- **Master's program details**\n"
    )

    st.divider()
    st.subheader("💡 Example questions")
    st.caption("Click any to ask immediately")
    st.caption("English")
    for ex in EXAMPLES_EN:
        st.button(ex, key=f"en_{ex}", on_click=submit_question, args=(ex,), use_container_width=True)

    st.caption("Deutsch")
    for ex in EXAMPLES_DE:
        st.button(ex, key=f"de_{ex}", on_click=submit_question, args=(ex,), use_container_width=True)

    if st.session_state.history:
        st.divider()
        if st.button("🗑️ Clear session", use_container_width=True):
            st.session_state.history = []
            st.session_state.pending_question = None
            st.rerun()

    st.divider()
    st.caption(
        "**About.** Built as a learning project. "
        "PaderBot is grounded in ~94 scraped pages from Paderborn University and Studierendenwerk Paderborn. "
        "Each question is answered independently to ensure citation accuracy. "
        "Always verify critical decisions against official sources."
    )
    st.caption("Source: [GitHub](#)")


# Input

default_value = st.session_state.edit_text or ""
st.session_state.edit_text = None

input_key = f"input_text_{st.session_state.input_key_id}"

# Use a form so the input only submits on explicit Enter or button click,
# NOT on focus loss (which was causing half-typed questions to fire).
with st.form(key=f"ask_form_{st.session_state.input_key_id}", clear_on_submit=True):
    user_input = st.text_input(
        "Your question / Deine Frage",
        value=default_value,
        placeholder="e.g., How do I apply to a Master's program? / Wie bewerbe ich mich?",
        key=input_key,
    )
    submitted = st.form_submit_button("Ask / Fragen", type="primary", use_container_width=False)

    if submitted and user_input.strip():
        submit_question(user_input.strip())
        st.session_state.input_key_id += 1
        st.rerun()


# Process pending question
pending = st.session_state.pending_question
if pending:
    bot = get_bot()
    with st.spinner(f'Answering: "{pending[:60]}..."'):
        result = bot.query(pending)
    st.session_state.history.append({"question": pending, "result": result})
    st.session_state.pending_question = None


# Render single Q&A
def render_qa(question: str, result: dict, is_latest: bool):
    header_cols = st.columns([10, 1])
    with header_cols[0]:
        st.markdown(f"### 🙋 {question}")
    with header_cols[1]:
        if is_latest:
            st.button(
                "✏️ Edit",
                key=f"edit_{len(st.session_state.history)}",
                on_click=edit_last_question,
                help="Edit and re-ask this question",
            )

    if result["refused"]:
        # Refusal: show the message centered, NO sources panel, NO copy button
        st.info(result["answer"])
        st.caption(
            "PaderBot only answers questions grounded in its scraped sources. "
            "Try rephrasing, or ask about applications, housing, programs, or financing."
        )
    else:
        # Real answer: two-column layout with answer + sources
        col_answer, col_sources = st.columns([3, 2])

        with col_answer:
            st.markdown(result["answer"])

            if result["retrieval_query"] != question:
                with st.expander("🔍 Search query used internally"):
                    st.code(result["retrieval_query"], language=None)

        with col_sources:
            st.markdown("**Sources**")
            for s in result["sources"]:
                lang_badge = "🇬🇧" if s["language"] == "en" else "🇩🇪"
                title = s["title"][:75]
                st.markdown(f"**[{s['rank']}]** {lang_badge} [{title}]({s['url']})")

            with st.expander("📄 View retrieved context"):
                for c in result["contexts"]:
                    preview = c["text"][:500]
                    more = "..." if len(c["text"]) > 500 else ""
                    st.markdown(f"**[{c['rank']}]** *{c['title'][:80]}*")
                    st.text(preview + more)
                    st.divider()

    st.divider()
    

# Render history
if st.session_state.history:
    for idx, entry in enumerate(reversed(st.session_state.history)):
        render_qa(entry["question"], entry["result"], is_latest=(idx == 0))
else:
    # Minimal empty state — just a pointer. Tips are in the ❓ Help popover.
    st.info("👈 Click an example question in the sidebar, or type your own above.")