"""研发报告 PPT 生成：markdown 报告正文 → .pptx（专业排版，可下载查证）。

触发时机：opp_submit_report 提交人工确认时生成/更新，存项目工作空间 deliverables/。

设计系统（2026-09-11 重做——旧版纯白底+平铺要点，用户反馈"没有排版和风格，很难看"）：
- 统一配色：深海军蓝主色 + 亮蓝强调 + 中性灰阶，全文一致不混色；
- 封面页：深色满版背景 + 大标题 + 方向标签 + 日期 + 数据来源署名；
- 章节页：四节模板（研发场景/输入/输出/成果）各一张深色分隔页，带节序号；
- 内容页：顶部强调条 + 标题 + 正文区，多级要点缩进，粗体引导词单独着色；
- **表格真渲染**：markdown 表格转 pptx Table（表头深蓝白字、隔行浅底），不再拍平成文本行；
- 页脚：页码 + 署名；长内容自动分页（≤7 要点/页）；
- python-pptx 纯后端，无外部模板依赖；import 失败如实降级报错（不静默产出丑文件）。
"""

import logging
import os
import re

logger = logging.getLogger("pipeline.opportunity.ppt")

# ── 设计令牌（配色/字号/版式，集中一处便于统一调风格）──
NAVY = (0x0E, 0x2A, 0x5C)        # 主色：深海军蓝（封面/章节页背景、表头）
NAVY_DARK = (0x08, 0x1C, 0x3E)   # 封面渐变下段
ACCENT = (0x2F, 0x6F, 0xED)      # 强调亮蓝（顶部条、要点符、引导词）
ACCENT_SOFT = (0xE8, 0xF0, 0xFE) # 强调浅底（要点卡片底）
INK = (0x1F, 0x29, 0x37)         # 正文墨色
SLATE = (0x47, 0x55, 0x69)       # 次级正文
MUTED = (0x94, 0xA3, 0xB8)       # 弱化（署名/页码）
LINE = (0xE2, 0xE8, 0xF0)        # 分隔线
ROW_ALT = (0xF6, 0xF8, 0xFC)     # 表格隔行
WHITE = (0xFF, 0xFF, 0xFF)

FONT = "Microsoft YaHei"         # 中文优先（缺失时 PowerPoint 自动回退）

SW, SH = 13.333, 7.5             # 16:9 英寸
MARGIN_X = 0.72
CONTENT_TOP = 1.42
CONTENT_H = 5.18
MAX_BULLETS_PER_PAGE = 7

# 四节模板顺序（与 opp_report_capability.REPORT_SECTIONS 一致）
SECTION_ORDER = ("研发场景", "输入", "输出", "成果")


def _rgb(t):
    from pptx.dml.color import RGBColor
    return RGBColor(t[0], t[1], t[2])


# ══════════════════════ markdown 解析 ══════════════════════

def _split_blocks(md):
    """把 markdown 段切成结构化块：[('h', lvl, text) | ('p', text) | ('ul', lvl, text)
    | ('ol', num, text) | ('table', [rows]) | ('quote', text)]。"""
    blocks = []
    lines = (md or "").split("\n")
    i = 0
    while i < len(lines):
        raw = lines[i].rstrip()
        s = raw.strip()
        if not s:
            i += 1
            continue
        # 表格：当前行是 | 开头且下一行是分隔线
        if s.startswith("|") and i + 1 < len(lines) and re.match(r"^\s*\|?[\s:|-]+\|?\s*$", lines[i + 1]) \
                and "-" in lines[i + 1]:
            rows = [_split_row(s)]
            i += 2
            while i < len(lines) and lines[i].strip().startswith("|"):
                rows.append(_split_row(lines[i].strip()))
                i += 1
            blocks.append(("table", rows))
            continue
        m = re.match(r"^(#{1,4})\s+(.+)$", s)
        if m:
            blocks.append(("h", len(m.group(1)), m.group(2).strip()))
            i += 1
            continue
        m = re.match(r"^>\s*(.+)$", s)
        if m:
            blocks.append(("quote", m.group(1).strip()))
            i += 1
            continue
        m = re.match(r"^[-*+]\s+(.+)$", s)
        if m:
            indent = len(raw) - len(raw.lstrip())
            lvl = 1 if indent < 2 else (2 if indent < 4 else 3)
            blocks.append(("ul", lvl, m.group(1).strip()))
            i += 1
            continue
        m = re.match(r"^(\d+)[.、)]\s*(.+)$", s)
        if m:
            blocks.append(("ol", m.group(1), m.group(2).strip()))
            i += 1
            continue
        if re.match(r"^[-=*_]{3,}$", s):      # 水平线 → 忽略
            i += 1
            continue
        # 普通段落：合并连续行
        para = [s]
        i += 1
        while i < len(lines):
            ns = lines[i].strip()
            if not ns or ns.startswith(("#", "|", ">", "- ", "* ", "+ ")) \
                    or re.match(r"^\d+[.、)]", ns) or re.match(r"^[-=*_]{3,}$", ns):
                break
            para.append(ns)
            i += 1
        blocks.append(("p", " ".join(para)))
    return blocks


def _split_row(line):
    """markdown 表格行 → 单元格列表。"""
    s = line.strip()
    if s.startswith("|"):
        s = s[1:]
    if s.endswith("|"):
        s = s[:-1]
    return [_clean_inline(c.strip()) for c in s.split("|")]


def _clean_inline(s):
    """去除行内 markdown 标记（粗体/斜体/行内码/链接）。"""
    s = s or ""
    s = re.sub(r"\*\*(.+?)\*\*", r"\1", s)
    s = re.sub(r"__(.+?)__", r"\1", s)
    s = re.sub(r"(?<!\*)\*([^*]+)\*(?!\*)", r"\1", s)
    s = re.sub(r"`([^`]+)`", r"\1", s)
    s = re.sub(r"\[([^\]]+)\]\(([^)]+)\)", r"\1", s)   # 链接只留文字（PPT 查证走界面）
    s = re.sub(r"~~(.+?)~~", r"\1", s)
    return s.strip()


def _split_lead(text):
    """拆出粗体引导词：「**结论**：正文」→ ('结论', '正文')；无引导词返回 ('', text)。"""
    m = re.match(r"^\s*\*\*(.+?)\*\*\s*[:：]?\s*(.*)$", text or "")
    if m and m.group(1).strip():
        return m.group(1).strip(), (m.group(2) or "").strip()
    return "", _clean_inline(text)


def _parse_report(content):
    """按四节模板解析（与详情页同口径）：返回 [(节名, markdown)]。

    无四节标题的旧报告 → 整篇归「输出」节（如实呈现，不编造分节）。
    """
    content = content or ""
    pat = re.compile(
        r"^##\s*(?:[一二三四五六七八九十\d]+[、.．]?\s*)?[*]*\s*"
        r"(研发场景|输入|输出|成果)\s*[*]*\s*[:：]?\s*$", re.MULTILINE)
    marks = list(pat.finditer(content))
    if not marks:
        return [("输出", content.strip())]
    out = []
    preamble = content[:marks[0].start()].strip()
    if preamble:
        out.append(("概览", preamble))
    for i, m in enumerate(marks):
        end = marks[i + 1].start() if i + 1 < len(marks) else len(content)
        body = content[m.end():end].strip()
        if body:
            out.append((m.group(1), body))
    return out or [("输出", content.strip())]


# ══════════════════════ 版式基元 ══════════════════════

def _rect(slide, x, y, w, h, color, line_color=None):
    from pptx.enum.shapes import MSO_SHAPE
    from pptx.util import Inches
    sh = slide.shapes.add_shape(MSO_SHAPE.RECTANGLE, Inches(x), Inches(y), Inches(w), Inches(h))
    sh.fill.solid()
    sh.fill.fore_color.rgb = _rgb(color)
    if line_color is None:
        sh.line.fill.background()
    else:
        sh.line.color.rgb = _rgb(line_color)
        sh.line.width = Inches(0.008)
    sh.shadow.inherit = False
    return sh


def _text(slide, x, y, w, h, runs, size=15, color=INK, bold=False,
          align=None, space_after=0, line_spacing=1.15):
    """放一个文本框；runs 可以是 str 或 [(text, {size,color,bold}), ...] 列表。"""
    from pptx.util import Inches, Pt
    box = slide.shapes.add_textbox(Inches(x), Inches(y), Inches(w), Inches(h))
    tf = box.text_frame
    tf.word_wrap = True
    tf.margin_left = tf.margin_right = tf.margin_top = tf.margin_bottom = 0
    p = tf.paragraphs[0]
    if align is not None:
        p.alignment = align
    p.space_after = Pt(space_after)
    p.line_spacing = line_spacing
    if isinstance(runs, str):
        runs = [(runs, {})]
    for txt, style in runs:
        if not txt:
            continue
        r = p.add_run()
        r.text = txt
        r.font.size = Pt(style.get("size", size))
        r.font.bold = style.get("bold", bold)
        r.font.color.rgb = _rgb(style.get("color", color))
        r.font.name = FONT
    return box


def _footer(slide, page_no, total=None):
    """页脚：细分隔线 + 页码 + 署名。"""
    from pptx.util import Inches
    from pptx.enum.text import PP_ALIGN
    _rect(slide, MARGIN_X, SH - 0.52, SW - 2 * MARGIN_X, 0.012, LINE)
    _text(slide, MARGIN_X, SH - 0.44, SW - 2 * MARGIN_X - 1.2, 0.3,
          "商机产线 · 数据来源可查证", size=9, color=MUTED)
    label = "%d" % page_no if not total else "%d / %d" % (page_no, total)
    _text(slide, SW - MARGIN_X - 1.2, SH - 0.44, 1.2, 0.3, label,
          size=9, color=MUTED, align=PP_ALIGN.RIGHT)


def _content_header(slide, title, kicker=""):
    """内容页页眉：强调竖条 + 标题（+ 可选小标签）。"""
    from pptx.util import Inches
    _rect(slide, MARGIN_X, 0.52, 0.085, 0.52, ACCENT)
    _text(slide, MARGIN_X + 0.24, 0.48, SW - 2 * MARGIN_X - 0.3, 0.62,
          _clean_inline(title), size=23, color=NAVY, bold=True)
    if kicker:
        _text(slide, MARGIN_X + 0.24, 1.08, SW - 2 * MARGIN_X - 0.3, 0.28,
              _clean_inline(kicker), size=11, color=MUTED)


# ══════════════════════ 表格渲染（真 Table，非文本） ══════════════════════

def _disp_w(s):
    """显示宽度（CJK=1.0、ASCII≈0.55 字符单位），用于估算换行数。"""
    return sum(1.0 if ord(ch) > 0x2E7F else 0.55 for ch in (s or ""))


def _add_table(slide, rows, top, width=None):
    """渲染 markdown 表格为 pptx Table；返回占用高度（英寸）。超宽自动缩字号。

    行高按内容估算逐行设置（2026-09-11 实测教训：统一行高时 LibreOffice/WPS
    不按换行撑行 → 长文本在行底被截断）。总高超页则截行加提示（不跨页拆表，
    保持单表完整可读）。
    """
    from pptx.util import Inches, Pt
    import math
    if not rows or not rows[0]:
        return 0.0
    width = width or (SW - 2 * MARGIN_X)
    ncol = max(len(r) for r in rows)
    rows = [r + [""] * (ncol - len(r)) for r in rows]
    nrow = len(rows)
    avail = SH - 0.95 - top
    # 列宽：首列略宽（多为名称），其余均分
    first = min(width * 0.30, 4.6)
    rest = (width - first) / max(1, ncol - 1)
    col_w = [first if ci == 0 else rest for ci in range(ncol)]
    fs = 11 if ncol <= 5 else (10 if ncol <= 7 else 9)
    line_h = fs * 1.45 / 72.0          # 每行文本高度（英寸，含行距余量）
    pad = 0.12                          # 单元格上下边距+余量

    def row_height(row_vals):
        lines = 1
        for ci, val in enumerate(row_vals):
            cap = max(2.0, (col_w[ci] - 0.16) * 72.0 / fs)   # 每行可容纳字符单位
            n = 0
            for seg in (val or "")[:180].split("\n"):        # 显式换行强制断行
                n += max(1, math.ceil(_disp_w(seg) / cap))
            lines = max(lines, n)
        return lines * line_h + pad

    heights = [row_height(r) for r in rows]
    total_h = sum(heights)
    if total_h > avail:
        # 从尾行往前砍到放得下（至少留表头+1 数据行+提示行）
        keep = nrow
        while keep > 3 and sum(heights[:keep]) + row_height(
                ["…"] + [""] * (ncol - 1)) > avail:
            keep -= 1
        cut = nrow - keep
        rows = rows[:keep] + [["… 其余 %d 行见平台详情页" % cut] + [""] * (ncol - 1)]
        heights = heights[:keep] + [row_height(rows[-1])]
        nrow = len(rows)
        total_h = sum(heights)
    gf = slide.shapes.add_table(nrow, ncol, Inches(MARGIN_X), Inches(top),
                                Inches(width), Inches(total_h))
    tbl = gf.table
    for ci in range(ncol):
        tbl.columns[ci].width = Inches(col_w[ci])
    for ri, row in enumerate(rows):
        tbl.rows[ri].height = Inches(heights[ri])
        for ci, val in enumerate(row):
            cell = tbl.cell(ri, ci)
            cell.margin_left = cell.margin_right = Inches(0.07)
            cell.margin_top = cell.margin_bottom = Inches(0.03)
            cell.vertical_anchor = 3   # MIDDLE
            cell.fill.solid()
            if ri == 0:
                cell.fill.fore_color.rgb = _rgb(NAVY)
            else:
                cell.fill.fore_color.rgb = _rgb(WHITE if ri % 2 else ROW_ALT)
            tf = cell.text_frame
            tf.word_wrap = True
            p = tf.paragraphs[0]
            r = p.add_run()
            r.text = (val or "")[:180]
            r.font.size = Pt(fs if ri else fs)
            r.font.bold = (ri == 0)
            r.font.name = FONT
            r.font.color.rgb = _rgb(WHITE if ri == 0 else (INK if ci == 0 else SLATE))
    return total_h + 0.18


# ══════════════════════ 页面构建 ══════════════════════

def _cover(prs, title, software, date_str):
    from pptx.util import Inches, Pt
    slide = prs.slides.add_slide(prs.slide_layouts[6])
    _rect(slide, 0, 0, SW, SH, NAVY)
    _rect(slide, 0, SH * 0.62, SW, SH * 0.38, NAVY_DARK)
    _rect(slide, MARGIN_X, 2.28, 1.5, 0.075, ACCENT)          # 强调短线
    _text(slide, MARGIN_X, 1.62, SW - 2 * MARGIN_X, 0.4,
          "RESEARCH REPORT · 研发商机报告", size=13, color=(0x9F, 0xC0, 0xFF))
    _text(slide, MARGIN_X, 2.62, SW - 2 * MARGIN_X - 0.8, 1.9,
          _clean_inline(title) or "研发报告", size=38, color=WHITE, bold=True, line_spacing=1.12)
    if software:
        _rect(slide, MARGIN_X, 4.62, 0.055, 0.42, ACCENT)
        _text(slide, MARGIN_X + 0.22, 4.66, 8.0, 0.4,
              "方向：%s" % _clean_inline(software), size=16, color=(0xC7, 0xD7, 0xF5))
    _text(slide, MARGIN_X, SH - 1.16, SW - 2 * MARGIN_X, 0.34,
          date_str, size=12, color=(0x8F, 0xA6, 0xCC))
    _text(slide, MARGIN_X, SH - 0.82, SW - 2 * MARGIN_X, 0.34,
          "商机产线 · 全部数据来自数据爬取平台（内网采集、已去重、逐条可查证）",
          size=11, color=(0x7B, 0x92, 0xBB))
    return slide


def _section_divider(prs, idx, name, total_sections):
    slide = prs.slides.add_slide(prs.slide_layouts[6])
    _rect(slide, 0, 0, SW, SH, NAVY)
    _rect(slide, 0, 0, 4.35, SH, NAVY_DARK)
    from pptx.util import Inches
    _text(slide, 0.85, 2.55, 2.9, 1.6, "%02d" % idx, size=76, color=ACCENT, bold=True)
    _rect(slide, 5.05, 3.02, 0.9, 0.06, ACCENT)
    _text(slide, 5.05, 3.28, SW - 5.05 - 0.8, 1.1, _clean_inline(name),
          size=32, color=WHITE, bold=True)
    _text(slide, 5.05, 4.42, SW - 5.05 - 0.8, 0.4,
          "SECTION %d OF %d" % (idx, total_sections), size=11, color=(0x8F, 0xA6, 0xCC))
    return slide


def _bullet_slide(prs, title, kicker, items, page_no):
    """要点页：items = [(level, lead, text)]。"""
    from pptx.util import Inches, Pt
    slide = prs.slides.add_slide(prs.slide_layouts[6])
    _rect(slide, 0, 0, SW, SH, WHITE)
    _rect(slide, 0, 0, SW, 0.14, ACCENT)
    _content_header(slide, title, kicker)
    top = CONTENT_TOP + (0.26 if kicker else 0.0)
    y = top
    for lvl, lead, text in items:
        indent = 0.0 if lvl <= 1 else (0.34 if lvl == 2 else 0.68)
        marker = "▪" if lvl <= 1 else "–"
        mcolor = ACCENT if lvl <= 1 else MUTED
        _text(slide, MARGIN_X + indent, y, 0.26, 0.3, marker,
              size=12 if lvl <= 1 else 11, color=mcolor, bold=True)
        runs = []
        if lead:
            runs.append((lead, {"size": 14, "bold": True, "color": NAVY}))
            runs.append(("  ", {}))
        if text:
            runs.append((text, {"size": 14 if lvl <= 1 else 13,
                                "color": INK if lvl <= 1 else SLATE}))
        if not runs:
            runs = [("（空）", {"size": 13, "color": MUTED})]
        _text(slide, MARGIN_X + indent + 0.26, y - 0.02,
              SW - 2 * MARGIN_X - indent - 0.26, 0.72, runs, line_spacing=1.18)
        est = max(1, int(len((lead or "") + (text or "")) / 46) + 1)
        y += 0.30 * est + 0.16
    _footer(slide, page_no)
    return slide


def _table_slide(prs, title, kicker, table_rows, extra_items, page_no):
    from pptx.util import Inches
    slide = prs.slides.add_slide(prs.slide_layouts[6])
    _rect(slide, 0, 0, SW, SH, WHITE)
    _rect(slide, 0, 0, SW, 0.14, ACCENT)
    _content_header(slide, title, kicker)
    top = CONTENT_TOP + (0.26 if kicker else 0.0)
    used = _add_table(slide, table_rows, top)
    if extra_items:
        y = top + used + 0.05
        for lvl, lead, text in extra_items:
            _text(slide, MARGIN_X, y, 0.26, 0.3, "▪", size=12, color=ACCENT, bold=True)
            runs = []
            if lead:
                runs.append((lead, {"size": 13, "bold": True, "color": NAVY}))
                runs.append(("  ", {}))
            if text:
                runs.append((text, {"size": 13, "color": SLATE}))
            _text(slide, MARGIN_X + 0.26, y - 0.02, SW - 2 * MARGIN_X - 0.26, 0.5, runs)
            y += 0.30 * max(1, int(len((lead or "") + (text or "")) / 52) + 1) + 0.10
    _footer(slide, page_no)
    return slide


def _closing(prs, page_no):
    from pptx.util import Inches
    slide = prs.slides.add_slide(prs.slide_layouts[6])
    _rect(slide, 0, 0, SW, SH, NAVY)
    _rect(slide, MARGIN_X, 2.0, 1.2, 0.06, ACCENT)
    _text(slide, MARGIN_X, 2.28, SW - 2 * MARGIN_X, 0.8, "说明与查证", size=30,
          color=WHITE, bold=True)
    notes = [
        "本报告全部数据来自数据爬取平台（内网采集、已去重），未做人工编造。",
        "每条引用均带公告原文 / 采集来源，可在产线平台「📋 数据参考」逐条查证。",
        "报告状态以平台门禁为准：人工确认 → 研发审批；未经审批不得视为立项依据。",
        "功能点与成本估算为模型基于公开数据的推算，落地前需结合团队实际复核。",
    ]
    y = 3.28
    for n in notes:
        _text(slide, MARGIN_X, y, 0.3, 0.3, "▪", size=12, color=ACCENT, bold=True)
        _text(slide, MARGIN_X + 0.3, y - 0.02, SW - 2 * MARGIN_X - 0.3, 0.6,
              n, size=14, color=(0xC7, 0xD7, 0xF5), line_spacing=1.2)
        y += 0.62
    return slide


# ══════════════════════ 主入口 ══════════════════════

def build_report_ppt(title, content, software="", out_path=""):
    """生成 PPT 并保存。返回 (True, 文件路径) 或 (False, 错误)。

    签名保持不变（opp_report_capability._build_report_ppt 调用方零改动）。
    """
    try:
        from pptx import Presentation
        from pptx.util import Inches
    except ImportError:
        return False, "缺少 python-pptx（pip install python-pptx）"
    if not out_path:
        return False, "缺少输出路径"

    import datetime
    date_str = datetime.date.today().strftime("%Y 年 %m 月 %d 日")
    prs = Presentation()
    prs.slide_width = Inches(SW)
    prs.slide_height = Inches(SH)

    _cover(prs, title, software, date_str)

    # 预扫描：算出要渲染的页面，便于一次性算总页数（页脚 x/y）
    pages = []          # [('divider', idx, name) | ('bullets', title, kicker, items)
                        #  | ('table', title, kicker, rows, extra)]
    sections = _parse_report(content)
    sec_idx = 0
    for sec_name, sec_md in sections:
        if sec_name in SECTION_ORDER:
            sec_idx += 1
            pages.append(("divider", sec_idx, sec_name))
        blocks = _split_blocks(sec_md)
        cur_title = sec_name if sec_name not in SECTION_ORDER else sec_name
        buf_items = []          # [(lvl, lead, text)]
        pending_table = None

        def flush(kicker=""):
            if not buf_items and pending_table is None:
                return
            if pending_table is not None:
                pages.append(("table", cur_title, kicker, pending_table, list(buf_items)))
            else:
                # 按页容量切分要点
                chunk = []
                for it in buf_items:
                    chunk.append(it)
                    if len(chunk) >= MAX_BULLETS_PER_PAGE:
                        pages.append(("bullets", cur_title, kicker, chunk))
                        chunk = []
                if chunk:
                    pages.append(("bullets", cur_title, kicker, chunk))
            del buf_items[:]

        for b in blocks:
            kind = b[0]
            if kind == "h":
                flush()
                cur_title = b[2]
            elif kind == "table":
                # 表格独立成页（带上此前累积的要点作为补充说明）
                flush()
                pending_table = b[1]
                pages.append(("table", cur_title, "", pending_table, []))
                pending_table = None
            elif kind == "ul":
                lead, text = _split_lead(b[2])
                buf_items.append((b[1], lead, text))
            elif kind == "ol":
                lead, text = _split_lead(b[2])
                buf_items.append((1, "%s. %s" % (b[1], lead) if lead else "%s." % b[1], text))
            elif kind == "quote":
                buf_items.append((1, "", _clean_inline(b[1])))
            else:   # p
                lead, text = _split_lead(b[1])
                buf_items.append((1, lead, text))
        flush()

    # 渲染（带总页数）
    total_pages = len(pages) + 2      # 封面 + 内容 + 尾页
    page_no = 1
    for pg in pages:
        page_no += 1
        if pg[0] == "divider":
            _section_divider(prs, pg[1], pg[2], sec_idx or 1)
        elif pg[0] == "bullets":
            _bullet_slide(prs, pg[1], pg[2], pg[3], page_no)
        else:
            _table_slide(prs, pg[1], pg[2], pg[3], pg[4], page_no)
    _closing(prs, total_pages)

    os.makedirs(os.path.dirname(out_path), exist_ok=True)
    prs.save(out_path)
    try:
        npages = len(prs.slides._sldIdLst)
    except Exception:
        npages = -1
    logger.info("报告 PPT 已生成: %s (%d 页)", out_path, npages)
    return True, out_path