baodan/api/insurance/ppt/quality_checker.py

"""PPT 生成质量检查服务。

自动检查 PPTX 文件质量，并生成人工确认清单。
"""
import logging

logger = logging.getLogger(__name__)


class QualityChecker:
    """PPT 质量检查器。"""

    def check(self, pptx_path: str, slides_data: dict | None,
              extractions: list, expected_slide_count: int) -> dict:
        """执行质量检查，返回 qualityReport。

        Args:
            pptx_path: 生成的 PPTX 文件路径
            slides_data: parse_slides 输出的结构化数据（可为 None）
            extractions: 提取数据列表
            expected_slide_count: 预期幻灯片数

        Returns:
            {"auto": [...], "manual": [...], "summary": {...}}
        """
        auto = []
        manual = []

        # 1. 文件可打开
        auto.append(self._check_file_openable(pptx_path))

        # 2. 页数一致
        actual_count = 0
        if slides_data and slides_data.get("slides"):
            actual_count = len(slides_data["slides"])
        auto.append(self._check_slide_count(actual_count, expected_slide_count))

        # 3. 无空白页
        if slides_data and slides_data.get("slides"):
            auto.append(self._check_blank_pages(slides_data["slides"]))

        # 4. 必需页面存在
        if slides_data and slides_data.get("slides"):
            auto.append(self._check_required_pages(slides_data["slides"]))

        # 5. 关键数据出现
        auto.append(self._check_key_data(slides_data, extractions))

        # 6. 文本溢出估算
        if slides_data and slides_data.get("slides"):
            overflow = self._check_text_overflow(slides_data["slides"])
            auto.append(overflow)

        # 人工确认项
        manual = [
            {"key": "confirm_customer_info", "confirmed": False,
             "message": "请确认客户姓名和产品名称与原计划书一致"},
            {"key": "confirm_premium_data", "confirmed": False,
             "message": "请确认保费金额、缴费年期和回本年份正确"},
            {"key": "confirm_benefit_table", "confirmed": False,
             "message": "请确认利益演示表数据与原计划书一致"},
            {"key": "confirm_compliance", "confirmed": False,
             "message": "请确认营销表述和合规内容符合监管要求"},
        ]

        pass_count = sum(1 for a in auto if a["status"] == "pass")
        fail_count = sum(1 for a in auto if a["status"] == "fail")
        warn_count = sum(1 for a in auto if a["status"] == "warn")

        return {
            "auto": auto,
            "manual": manual,
            "summary": {
                "pass": pass_count,
                "fail": fail_count,
                "warn": warn_count,
                "manualPending": len(manual),
            },
        }

    @staticmethod
    def _check_file_openable(pptx_path: str) -> dict:
        """检查 PPTX 可否正常打开。"""
        try:
            from pptx import Presentation
            prs = Presentation(pptx_path)
            count = len(prs.slides)
            if count == 0:
                return {"key": "file_openable", "status": "fail",
                        "message": "PPTX 文件可以打开但不包含幻灯片", "page": None}
            return {"key": "file_openable", "status": "pass",
                    "message": f"PPTX 文件可正常打开（{count} 页）", "page": None}
        except Exception as e:
            return {"key": "file_openable", "status": "fail",
                    "message": f"PPTX 文件无法打开: {e}", "page": None}

    @staticmethod
    def _check_slide_count(actual: int, expected: int) -> dict:
        """检查页数是否一致。"""
        if expected <= 0:
            return {"key": "slide_count_match", "status": "warn",
                    "message": "预期页数未知，无法比较", "page": None}
        if actual == expected:
            return {"key": "slide_count_match", "status": "pass",
                    "message": f"页数一致（{actual} 页）", "page": None}
        if abs(actual - expected) <= 1:
            return {"key": "slide_count_match", "status": "warn",
                    "message": f"页数略有偏差: 实际 {actual} 页，预期 {expected} 页",
                    "page": None}
        return {"key": "slide_count_match", "status": "fail",
                "message": f"页数不一致: 实际 {actual} 页，预期 {expected} 页",
                "page": None}

    @staticmethod
    def _check_blank_pages(slides: list) -> dict:
        """检查是否有空白或近似空白页。"""
        blank_pages = []
        for i, slide in enumerate(slides):
            shapes = slide.get("shapes", [])
            has_content = False
            for shape in shapes:
                stype = shape.get("type", "")
                # 文本框有内容
                if stype == "textbox":
                    for para in shape.get("paragraphs", []):
                        if para.get("text", "").strip():
                            has_content = True
                            break
                # 图片、表格也算内容
                elif stype in ("image", "table"):
                    has_content = True
                # 有填充色的矩形也算内容（封面背景等）
                elif stype == "rect" and shape.get("fill"):
                    has_content = True
                # 分组形状中的子元素
                elif stype == "group" and shape.get("children"):
                    has_content = True
                if has_content:
                    break
            if not has_content:
                blank_pages.append(i + 1)

        if not blank_pages:
            return {"key": "no_blank_pages", "status": "pass",
                    "message": "所有页面均包含内容（文本/图片/表格/形状）", "page": None}
        if len(blank_pages) <= 2:
            return {"key": "no_blank_pages", "status": "warn",
                    "message": f"第 {', '.join(str(p) for p in blank_pages)} 页可能为空白页",
                    "page": blank_pages[0]}
        return {"key": "no_blank_pages", "status": "fail",
                "message": f"{len(blank_pages)} 页为空白: 第 {', '.join(str(p) for p in blank_pages)} 页",
                "page": blank_pages[0]}

    @staticmethod
    def _check_required_pages(slides: list) -> dict:
        """检查必需页面类型是否存在（封面、数据页、总结页）。"""
        if len(slides) < 2:
            return {"key": "required_pages_present", "status": "warn",
                    "message": "幻灯片数量较少，无法判断必需页面", "page": None}

        # 简单启发式：第 1 页应含封面关键词，最后 1 页应含总结关键词
        first_texts = []
        last_texts = []
        for shape in slides[0].get("shapes", []):
            for para in shape.get("paragraphs", []):
                first_texts.append(para.get("text", ""))
        for shape in slides[-1].get("shapes", []):
            for para in shape.get("paragraphs", []):
                last_texts.append(para.get("text", ""))

        first_text = " ".join(first_texts).lower()
        last_text = " ".join(last_texts).lower()

        cover_keywords = ["方案", "计划", "客户", "保险", "财富"]
        summary_keywords = ["总结", "总结", "summary", "感谢", "联系"]

        has_cover = any(kw in first_text for kw in cover_keywords)
        has_summary = any(kw in last_text for kw in summary_keywords)

        if has_cover and has_summary:
            return {"key": "required_pages_present", "status": "pass",
                    "message": "封面页和总结页均存在", "page": None}
        missing = []
        if not has_cover:
            missing.append("封面")
        if not has_summary:
            missing.append("总结")
        return {"key": "required_pages_present", "status": "warn",
                "message": f"未检测到 {'、'.join(missing)} 页", "page": 1 if not has_cover else len(slides)}

    @staticmethod
    def _check_key_data(slides_data: dict | None, extractions: list) -> dict:
        """检查关键数据是否出现在幻灯片中。"""
        if not slides_data or not slides_data.get("slides"):
            return {"key": "key_data_present", "status": "warn",
                    "message": "无法获取幻灯片内容进行数据核对", "page": None}

        # 收集所有幻灯片文本
        all_text_parts = []
        for slide in slides_data["slides"]:
            for shape in slide.get("shapes", []):
                for para in shape.get("paragraphs", []):
                    text = para.get("text", "").strip()
                    if text:
                        all_text_parts.append(text)
                # 表格
                for row in shape.get("rows", []):
                    for cell in row:
                        text = cell.get("text", "").strip()
                        if text:
                            all_text_parts.append(text)
        all_text = " ".join(all_text_parts)

        # 从提取数据中获取关键字段
        missing = []
        for ext in extractions:
            if ext.get("status") not in ("success", "partial") or not ext.get("data"):
                continue
            data = ext["data"]
            # 客户姓名
            name = (data.get("insured", {}) or {}).get("name", "")
            if name and name not in all_text:
                missing.append(f"客户姓名({name})")
            # 产品名称
            product = data.get("product_name", "")
            if product and product not in all_text:
                missing.append(f"产品名称({product})")
            # 年缴保费
            policy = data.get("policy", {}) or {}
            premium = policy.get("annual_premium") or policy.get("annualPremium", 0)
            if premium and str(premium) not in all_text:
                missing.append(f"年缴保费({premium})")

        if not missing:
            return {"key": "key_data_present", "status": "pass",
                    "message": "关键数据（客户名、产品名、保费）均已出现", "page": None}
        return {"key": "key_data_present", "status": "warn",
                "message": f"未找到: {', '.join(missing)}", "page": None}

    @staticmethod
    def _check_text_overflow(slides: list) -> dict:
        """估算文本是否可能溢出文本框。"""
        overflow_pages = []
        for i, slide in enumerate(slides):
            for shape in slide.get("shapes", []):
                if shape.get("type") != "textbox":
                    continue
                box_w = shape.get("w", 0)
                box_h = shape.get("h", 0)
                if box_w <= 0 or box_h <= 0:
                    continue
                total_text_len = 0
                max_font = 14
                for para in shape.get("paragraphs", []):
                    text = para.get("text", "")
                    total_text_len += len(text)
                    max_font = max(max_font, para.get("fontSize", 14))
                # 粗略估算：每个中文字符占 fontSize 宽度，每行能放 box_w / fontSize 个字符
                if max_font <= 0:
                    continue
                chars_per_line = max(1, box_w / max_font)
                est_lines = total_text_len / chars_per_line
                line_height = max_font * 1.4
                est_height = est_lines * line_height
                if est_height > box_h * 1.3:  # 超出 30% 视为溢出
                    overflow_pages.append(i + 1)
                    break

        if not overflow_pages:
            return {"key": "text_overflow", "status": "pass",
                    "message": "未检测到明显文本溢出", "page": None}
        return {"key": "text_overflow", "status": "warn",
                "message": f"第 {', '.join(str(p) for p in overflow_pages)} 页可能存在文本溢出",
                "page": overflow_pages[0]}