features/reports/full_report.py

"""전체 리포트 생성 탭."""
from concurrent.futures import ThreadPoolExecutor, as_completed

import pandas as pd
import requests
import streamlit as st
import streamlit.components.v1 as components

from core.api_client import ChainShiftClient


def _build_one(
    api_key: str,
    campaign_id: int,
    feat_key: str,
    start_date: str,
    end_date: str,
    enable_insights: bool,
    homepage_urls: list[str] | None,
) -> tuple[str, dict | None, str | None]:
    """Worker thread — st.* 호출 금지. 독립 HTTP 클라이언트로 feature 빌드."""
    try:
        thread_client = ChainShiftClient(api_key=api_key)
        result = thread_client.build_html_feature(
            campaign_id=campaign_id,
            feature=feat_key,
            start_date=start_date,
            end_date=end_date,
            enable_insights=enable_insights,
            enable_action_items=True,
            homepage_urls=homepage_urls if feat_key == "homepage-citations" else None,
        )
        return feat_key, result, None
    except Exception as e:
        return feat_key, None, str(e)


@st.cache_data(ttl=60)
def _fetch_report_history(
    campaign_id: int, page: int = 1, page_size: int = 20,
    _api_key: str = "", _access_token: str = "",
) -> dict:
    """Cached fetch for HTML report history."""
    client = ChainShiftClient(api_key=_api_key or None, access_token=_access_token or None)
    return client.get_html_report_history(campaign_id, page=page, page_size=page_size)


AVAILABLE_FEATURES = [
    ("overview", "1. 가시성 분석 개요"),
    ("visibility", "2. AI 검색 가시성"),
    ("citations", "3. 인용 출처 분석"),
    ("citation-trends", "4. 인용 출처 시계열"),
    ("content-types", "5. 콘텐츠 유형"),
    ("sentiment", "6. 브랜드 감정"),
    ("homepage-citations", "7. 홈페이지 인용률"),
]


def render(client: ChainShiftClient, base_ctx: dict, start_date: str, end_date: str):
    """전체 리포트 생성 섹션."""
    st.markdown("#### 📄 전체 HTML 리포트 생성")
    st.caption("7개 Feature를 포함한 통합 HTML 리포트를 생성합니다. LLM 인사이트로 컨설턴트 톤의 분석을 추가할 수 있습니다.")

    with st.expander("⚙️ 리포트 옵션", expanded=True):
        st.markdown("**포함할 Feature 선택**")
        selected_features = []
        col1, col2 = st.columns(2)
        for i, (feat_key, feat_label) in enumerate(AVAILABLE_FEATURES):
            with col1 if i < 4 else col2:
                if st.checkbox(feat_label, value=True, key=f"reports:full_feat_{feat_key}"):
                    selected_features.append(feat_key)

        st.markdown("---")

        enable_insights = st.checkbox(
            "🤖 LLM 인사이트 생성",
            value=True,
            help="Gemini API를 사용하여 컨설턴트 톤의 분석 인사이트를 추가합니다 (생성 시간 증가)",
        )

        st.text_area(
            "Homepage URLs (줄바꿈 구분, 선택)",
            help="홈페이지 인용률 분석에 사용할 URL 목록",
            key="reports:homepage_urls_input",
            height=80,
        )

    # Parse homepage URLs from text area
    homepage_urls_raw = st.session_state.get("reports:homepage_urls_input", "")
    homepage_urls = [u.strip() for u in homepage_urls_raw.splitlines() if u.strip()] or None

    # Feature display name lookup
    _feat_display = dict(AVAILABLE_FEATURES)

    if st.button("🚀 전체 리포트 생성", key="reports:generate_html_btn", type="primary", disabled=not selected_features):
        campaign_id = base_ctx["campaign_id"]
        total = len(selected_features)
        progress_bar = st.progress(0, text="리포트 생성 준비 중...")
        status_container = st.container()

        built_features: list[dict] = []
        skipped_features: list[str] = []
        total_build_ms = 0

        # Phase 1: Build features in parallel (I/O-bound HTTP calls)
        with ThreadPoolExecutor(max_workers=total) as executor:
            futures = {
                executor.submit(
                    _build_one, client.api_key, campaign_id, feat_key,
                    start_date, end_date, enable_insights, homepage_urls,
                ): feat_key
                for feat_key in selected_features
            }
            completed = 0
            for future in as_completed(futures):
                feat_key = futures[future]
                feat_label = _feat_display.get(feat_key, feat_key)
                completed += 1
                fk, result, error = future.result()
                if error:
                    skipped_features.append(feat_key)
                    with status_container:
                        st.caption(f"  {feat_label} 실패: {error}")
                elif result and result.get("success"):
                    feat_resp = result["data"]
                    built_features.append(feat_resp["feature_data"])
                    build_ms = feat_resp.get("build_time_ms", 0)
                    total_build_ms += build_ms
                    insight_tag = " +인사이트" if feat_resp.get("insights_generated") else ""
                    with status_container:
                        st.caption(f"  {feat_label} ({build_ms/1000:.1f}s{insight_tag})")
                else:
                    skipped_features.append(feat_key)
                    with status_container:
                        st.caption(f"  {feat_label} 건너뜀")
                progress_bar.progress(
                    completed / (total + 1),
                    text=f"({completed}/{total}) 빌드 완료...",
                )

        # Restore original feature order for rendering
        feat_order = {k: i for i, k in enumerate(selected_features)}
        built_features.sort(key=lambda f: feat_order.get(f.get("feature_id", ""), 99))

        if not built_features:
            progress_bar.empty()
            st.error("모든 Feature 생성에 실패했습니다.")
        else:
            # Phase 2: Render final report
            progress_bar.progress(
                total / (total + 1),
                text="HTML 리포트 조립 중...",
            )
            try:
                render_result = client.render_html_report(
                    campaign_id=campaign_id,
                    features_data=built_features,
                    start_date=start_date,
                    end_date=end_date,
                    enable_insights=enable_insights,
                    enable_action_items=True,
                    output_mode="url",
                )

                if render_result.get("success"):
                    data = render_result.get("data") or {}
                    if not isinstance(data, dict):
                        progress_bar.empty()
                        st.error("리포트 렌더링 실패: 서버 응답이 비정상입니다.")
                    else:
                        render_ms = data.get("generation_time_ms", 0)

                        # Download HTML from Supabase Storage URL directly.
                        # url mode avoids Vercel 4.5MB response body limit.
                        html_content = ""
                        html_url = data.get("html_url") or ""
                        if html_url:
                            try:
                                dl_resp = requests.get(html_url, timeout=30)
                                dl_resp.raise_for_status()
                                dl_resp.encoding = "utf-8"
                                html_content = dl_resp.text
                            except Exception as dl_err:
                                st.warning(f"HTML 다운로드 실패, URL 링크로 대체: {dl_err}")

                        if not html_url and not html_content:
                            progress_bar.empty()
                            st.error("리포트 렌더링 실패: 스토리지 URL이 반환되지 않았습니다.")
                        else:
                            # Persist results in session_state for rerun survival
                            st.session_state["full_report_result"] = {
                                "html_content": html_content,
                                "html_url": html_url,
                                "report_id": data.get("report_id", "")[:8],
                                "features_generated": data.get("features_generated", []),
                                "file_size_kb": round(data.get("file_size_bytes", 0) / 1024, 1),
                                "total_sec": round((total_build_ms + render_ms) / 1000, 1),
                                "insights_generated": data.get("insights_generated", False),
                                "skipped_features": skipped_features,
                                "campaign_id": campaign_id,
                                "start_date": start_date,
                                "end_date": end_date,
                            }
                            progress_bar.empty()
                            st.rerun()
                else:
                    progress_bar.empty()
                    st.error("리포트 렌더링 실패: " + str(render_result.get("error", "Unknown error")))

            except Exception as e:
                progress_bar.empty()
                st.error(f"리포트 렌더링 오류: {e}")
    elif not selected_features:
        st.warning("최소 1개 이상의 Feature를 선택하세요.")

    # ── Results display (persists across reruns via session_state) ──
    report_state = st.session_state.get("full_report_result")
    if report_state:
        html_content = report_state["html_content"]
        report_id = report_state["report_id"]
        features_generated = report_state["features_generated"]
        file_size_kb = report_state["file_size_kb"]
        total_sec = report_state["total_sec"]
        skipped = report_state["skipped_features"]
        r_campaign_id = report_state["campaign_id"]
        r_start = report_state["start_date"]
        r_end = report_state["end_date"]

        if skipped:
            st.warning(f"일부 Feature를 건너뛰고 리포트를 생성했습니다: {', '.join(skipped)}")
        st.success("리포트가 생성되었습니다!")

        st.markdown(f"""
        <div style="background: linear-gradient(135deg, #E5E2FF 0%, #F0FDFA 100%);
                    padding: 20px; border-radius: 12px; margin: 16px 0;">
            <h4 style="margin: 0 0 12px 0; color: #2B239B;">리포트 생성 완료</h4>
            <p style="margin: 8px 0;"><strong>Report ID:</strong> {report_id}...</p>
            <p style="margin: 8px 0;"><strong>Features:</strong> {len(features_generated)}개 ({len(skipped)}개 건너뜀)</p>
            <p style="margin: 8px 0;"><strong>파일 크기:</strong> {file_size_kb} KB</p>
            <p style="margin: 8px 0;"><strong>생성 시간:</strong> {total_sec}초</p>
            <p style="margin: 8px 0;"><strong>LLM 인사이트:</strong> {'포함' if report_state.get('insights_generated') else '미포함'}</p>
        </div>
        """, unsafe_allow_html=True)

        html_url = report_state.get("html_url", "")

        if html_content:
            col_open, col_download, col_clear = st.columns(3)
            with col_open:
                if html_url:
                    st.link_button("새 창에서 보기", html_url, use_container_width=True)
                else:
                    st.button("새 창에서 보기", disabled=True, use_container_width=True, key="reports:open_disabled")
            with col_download:
                file_name = f"AI_가시성_리포트_{r_campaign_id}_{r_start}_{r_end}.html"
                st.download_button(
                    label="HTML 다운로드",
                    data=b'\xef\xbb\xbf' + html_content.lstrip('\ufeff').encode("utf-8"),
                    file_name=file_name,
                    mime="text/html; charset=utf-8",
                    use_container_width=True,
                    key="reports:full_report_download",
                )
            with col_clear:
                if st.button("초기화", key="reports:clear_result", use_container_width=True):
                    del st.session_state["full_report_result"]
                    st.rerun()

            with st.expander("리포트 미리보기", expanded=False):
                components.html(html_content, height=800, scrolling=True)
        elif html_url:
            # Fallback: HTML download failed, show direct link
            col_link, col_clear = st.columns(2)
            with col_link:
                st.link_button("리포트 열기 (외부 링크)", html_url, use_container_width=True)
            with col_clear:
                if st.button("초기화", key="reports:clear_result", use_container_width=True):
                    del st.session_state["full_report_result"]
                    st.rerun()

    st.markdown("---")

    # Report History
    st.markdown("##### 📋 생성 이력")
    try:
        history_result = _fetch_report_history(
            base_ctx["campaign_id"],
            _api_key=base_ctx.get("api_key", ""),
            _access_token=base_ctx.get("access_token", ""),
        )
        if history_result.get("success"):
            history_data = history_result["data"]
            reports = history_data.get("items", [])
            total = history_data.get("total", 0)

            if reports:
                st.markdown(f"총 **{total}**건의 리포트가 생성되었습니다.")

                history_rows = []
                for r in reports:
                    status_emoji = "✅" if r.get("status") == "completed" else "❌"
                    features = r.get("features_included", [])
                    file_size_kb = round(r.get("file_size_bytes", 0) / 1024, 1)
                    history_rows.append({
                        "생성일": r.get("created_at", "")[:16].replace("T", " "),
                        "상태": f"{status_emoji}",
                        "Features": f"{len(features)}/{len(AVAILABLE_FEATURES)}",
                        "기간": f"{r.get('start_date', '?')} ~ {r.get('end_date', '?')}",
                        "크기": f"{file_size_kb} KB",
                        "링크": r.get("html_url") or "-",
                    })

                df = pd.DataFrame(history_rows)
                st.dataframe(
                    df,
                    use_container_width=True,
                    hide_index=True,
                    column_config={
                        "링크": st.column_config.LinkColumn("링크", display_text="열기"),
                    },
                )
            else:
                st.info("아직 생성된 리포트가 없습니다.")
    except Exception as e:
        st.warning(f"이력 로드 실패: {e}")