import os, re, gradio as gr
from build_index import ensure_chunks, kick_off_background_build, read_status
from retrieval import Retriever
from evidence import extract_evidence

# ---- Fast boot: prepare chunks; build FAISS in background
ensure_chunks()
kick_off_background_build()

# ---- Load retriever (BM25 works immediately; FAISS used when ready)
retriever = Retriever(embed_model_name=os.getenv("EMBED_MODEL", "intfloat/multilingual-e5-small"))
TOP_K = int(os.getenv("TOP_K", "4"))  # internal default (no UI control)

# -------- Helpers --------
def _status_text():
    st = read_status()
    phase = st.get("phase","unknown")
    if phase == "ready":
        return "الفهرس جاهز (FAISS)"
    if phase == "embedding":
        return f"بناء الفهرس في الخلفية… {st.get('done',0)}/{st.get('total',0)}"
    if phase == "chunks_ready":
        return f"تم تجهيز المقاطع ({st.get('total','?')}). جارٍ بدء التضمين…"
    if phase == "waiting_data":
        return f"{st.get('msg','البيانات غير موجودة')}"
    if phase == "error":
        return f"{st.get('msg','خطأ في الفهرس')}"
    return "حالة غير معروفة"

def _clean(s):
    return (s or "").strip()

def _sentences(text):
    parts = re.split(r"(?<=[\.\!\؟\!])\s+", text or "")
    return [p.strip() for p in parts if p.strip()]

def _wrap_quran(s: str) -> str:
    """Ensure Qur'an lines are shown with ornate brackets."""
    s = s.strip()
    if "﴿" in s and "﴾" in s:
        return s
    # if it's already quoted without brackets, wrap it
    return f"﴿{s}﴾"

# -------- Evidence & verdict selection --------
NEG_PATTERNS = [
    r"\bلا\s+يجوز\b", r"\bلا\s+يحل\b", r"\bلا\s+يصح\b",
    r"\bحرام\b", r"\bمحرم\b", r"\bيحرم\b", r"\bمنع\b", r"\bممنوع\b", r"\bباطل\b"
]
POS_PATTERNS = [
    r"\bيجوز\b", r"\bحلال\b", r"\bمباح\b", r"\bلا\s+بأس\b", r"\bلا\s+حرج\b",
    r"\bسنة\b", r"\bمستحب\b", r"\bواجب\b"
]
EXCEPTION_HINTS = [r"\bإلا\b", r"\bللدعوة\b", r"\bضرورة\b", r"\bحاجة\b", r"\bأمن\s+الفتنة\b", r"\bمع\s+الضوابط\b"]

def _count_matches(text, patterns):
    return sum(1 for p in patterns if re.search(p, text))

def _infer_verdict_general(hits):
    """General purpose verdict detector from retrieved text."""
    text = " ".join(h["chunk"] for h in hits)
    text = re.sub(r"\s+", " ", text)

    neg = _count_matches(text, NEG_PATTERNS)
    pos = _count_matches(text, POS_PATTERNS)
    has_exception = _count_matches(text, EXCEPTION_HINTS) > 0

    if neg == 0 and pos == 0:
        return None  # unclear

    if neg > pos:
        if has_exception:
            return "الأصل المنع، ويُستثنى للضرورة/الدعوة مع أمن الفتنة"
        return "حرام"
    if pos > neg:
        if has_exception:
            return "الأصل الجواز مع الضوابط"
        return "يجوز"

    # tie-break: prefer المنع if any explicit negation is present
    if neg > 0:
        return "حرام"
    if pos > 0:
        return "يجوز"
    return None

def _pick_snippets(hits, limit=3):
    """Prefer explicit Qur'an/Hadith/Ijma lines, then strong verdict sentences."""
    snippets = []
    # 1) Explicit evidence
    for h in hits:
        ev = extract_evidence(h["chunk"])
        # Qur'an first, wrapped in ornate brackets
        for q in ev["quran"]:
            snippets.append((_wrap_quran(q), h))
            if len(snippets) >= limit: return snippets
        # Hadith
        for hd in ev["hadith"]:
            snippets.append((hd, h))
            if len(snippets) >= limit: return snippets
        # Ijma'
        for ij in ev["ijma"]:
            snippets.append((ij, h))
            if len(snippets) >= limit: return snippets

    # 2) Strong rule-like sentences (negative or positive)
    neg_or_pos = re.compile("(" + "|".join(p.strip(r"\b") for p in [*NEG_PATTERNS, *POS_PATTERNS]) + ")", re.I)
    for h in hits:
        for sent in _sentences(h["chunk"]):
            if neg_or_pos.search(sent):
                snippets.append((sent, h))
                if len(snippets) >= limit: return snippets

    # 3) Fallback: first informative sentence from top hits
    for h in hits:
        for sent in _sentences(h["chunk"]):
            if len(sent) > 25:
                snippets.append((sent, h))
                if len(snippets) >= limit: return snippets
    return snippets

def _format_main_answer(hits):
    # Try general verdict; if still None but we do have hits, choose a cautious closest form.
    verdict = _infer_verdict_general(hits)
    if verdict is None:
        if hits:
            verdict = "الأقرب: المنع"  # closest conservative reading when signals exist but weak
        else:
            verdict = "غير واضح"

    snippets = _pick_snippets(hits, limit=3)

    lines = [f"### الحكم: **{verdict}**"]
    if snippets:
        lines.append("\n**توضيح مختصر (من النصوص المسترجعة):**")
        for (txt, _h) in snippets:
            lines.append(f"> {txt}")
    return "\n".join(lines)

def _format_refs(hits, limit=4):
    out = []
    for i, h in enumerate(hits[:limit], 1):
        title = _clean(h.get("title")) or "مرجع"
        link = _clean(h.get("link"))
        cat  = _clean(h.get("category"))
        if link:
            out.append(f"{i}. **{title}** — {cat} — [الرابط]({link})")
        else:
            out.append(f"{i}. **{title}** — {cat}")
    return "\n".join(out) if out else "—"

def _format_context(hits, limit_chars=400):
    blocks = []
    for i, h in enumerate(hits, 1):
        title = _clean(h.get("title")) or f"المصدر {i}"
        link  = _clean(h.get("link"))
        chunk = _clean(h.get("chunk"))
        snippet = (chunk[:limit_chars].rstrip() + "…") if len(chunk) > limit_chars else chunk
        hdr = f"**{i}. {title}**" + (f" — [الرابط]({link})" if link else "")
        blocks.append(hdr + "\n\n> " + snippet.replace("\n", "\n> "))
    return "\n\n---\n\n".join(blocks) if blocks else "—"

# -------- Gradio callbacks --------
def answer_fn(question: str):
    if not question or not question.strip():
        return "يرجى كتابة السؤال.", "—", "—"
    # If FAISS finished while app runs, re-init for better retrieval
    global retriever
    if getattr(retriever, "faiss", None) is None and os.path.exists("index/faiss.index"):
        try:
            retriever = Retriever(embed_model_name=os.getenv("EMBED_MODEL", "intfloat/multilingual-e5-small"))
        except Exception:
            pass

    hits = retriever.search(question, k=TOP_K)

    if not hits:
        # Only here print غير واضح per your requirement
        return "### الحكم: **غير واضح**\n\n> لم نعثر على نصوص ذات صلة في فتاوى ابن باز.", "—", "—"

    main_md = _format_main_answer(hits)
    refs_md = _format_refs(hits)
    ctx_md  = _format_context(hits)
    return main_md, refs_md, ctx_md

def status_fn():
    return _status_text()

# ---- RTL + centered layout
css = """
.gradio-container { direction: rtl; text-align: right; }
textarea, input, .wrap, .prose, .markdown-body { direction: rtl; text-align: right; }
.centered { max-width: 880px; margin-left: auto !important; margin-right: auto !important; }
.centered .gr-textbox, .centered .gr-textbox textarea { text-align: right; }
.centered .gr-button { display: block; margin: 0.5rem auto; }
.footer { font-size: 12px; color: #6b7280; text-align: center; margin-top: 1rem; }
"""

with gr.Blocks(css=css, theme=gr.themes.Soft()) as demo:
    with gr.Column(elem_classes=["centered"]):
        # Title & description
        gr.Markdown("## بيان - مساعد الفتاوى للشيخ ابن باز")
        gr.Markdown("يظهر النتائج بناء على الفتاوى في موقع الشيخ ابن باز - لأغراض تعليمية فقط -")

        # Input
        q = gr.Textbox(label="سؤالك", placeholder="اكتب سؤالك هنا…", lines=3)
        btn = gr.Button("أجب")

        # Main answer (verdict-first)
        answer_md = gr.Markdown(label="الإجابة")

        # Optional panels (collapsed)
        with gr.Accordion("المراجع (اضغط للعرض)", open=False):
            refs_md = gr.Markdown()
        with gr.Accordion("النصوص المستعملة (اختياري)", open=False):
            ctx_md = gr.Markdown()

        # Footer: FAISS status at the very bottom (collapsed, manual refresh)
        with gr.Accordion("حالة الفهرسة (FAISS) — اضغط للتحديث", open=False):
            status_md = gr.Markdown(value=_status_text())
            refresh_btn = gr.Button("تحديث الحالة")
            refresh_btn.click(status_fn, inputs=None, outputs=status_md)

        gr.Markdown(
            "المحتوى المستعمل من موقع سماحة الشيخ عبدالعزيز بن باز "
            "[binbaz.org.sa](https://binbaz.org.sa). "
            "جميع الحقوق محفوظة لأصحابها. هذا المنتج تعليمي غير رسمي.",
            elem_classes=["footer"]
        )

    btn.click(answer_fn, inputs=[q], outputs=[answer_md, refs_md, ctx_md])

if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)