baodan/api/insurance/ppt/template_asset_service.py

"""PPT 模板资产保存、解析与路径解析。"""
from __future__ import annotations

import os
import re
import uuid
import hashlib


BUILTIN_SCHEME = "builtin://"
STORED_SCHEME = "stored://"
MAX_TEMPLATE_BYTES = 30 * 1024 * 1024
BUILTIN_TEMPLATE_ROOT = os.path.join(os.path.dirname(__file__), "template_assets")


def template_asset_sha256(path: str) -> str:
    """计算模板资产指纹，供任务快照和 Worker 二次校验。"""
    digest = hashlib.sha256()
    with open(path, "rb") as source:
        for chunk in iter(lambda: source.read(1024 * 1024), b""):
            digest.update(chunk)
    return digest.hexdigest()


def resolve_template_asset(asset_id: str | None) -> str | None:
    """将数据库中的模板资产标识解析为受控的本地绝对路径。"""
    value = str(asset_id or "")
    if value.startswith(BUILTIN_SCHEME):
        return _safe_join(BUILTIN_TEMPLATE_ROOT, value[len(BUILTIN_SCHEME):])
    if value.startswith(STORED_SCHEME):
        from insurance.config import get_storage_root
        root = os.path.join(get_storage_root(), "ppt_templates")
        return _safe_join(root, value[len(STORED_SCHEME):])
    return None


def save_uploaded_template(file_storage, template_id: str) -> tuple[str, str]:
    """校验并保存上传的 PPTX，返回资产标识和绝对路径。"""
    if not file_storage or not file_storage.filename:
        raise ValueError("请选择 PPTX 模板文件")
    if not file_storage.filename.lower().endswith(".pptx"):
        raise ValueError("仅支持 .pptx 格式")

    stream = file_storage.stream
    stream.seek(0, os.SEEK_END)
    size = stream.tell()
    stream.seek(0)
    if size <= 0:
        raise ValueError("模板文件为空")
    if size > MAX_TEMPLATE_BYTES:
        raise ValueError("模板文件不能超过 30MB")

    safe_id = re.sub(r"[^a-zA-Z0-9_-]", "_", template_id)[:50] or "template"
    relative_path = f"{safe_id}/{uuid.uuid4().hex}.pptx"
    from insurance.config import get_storage_root
    root = os.path.join(get_storage_root(), "ppt_templates")
    absolute_path = _safe_join(root, relative_path)
    os.makedirs(os.path.dirname(absolute_path), exist_ok=True)
    file_storage.save(absolute_path)
    return f"{STORED_SCHEME}{relative_path.replace(os.sep, '/')}", absolute_path


def parse_template_pptx(path: str) -> dict:
    """提取幻灯片标题、页面类型和基础版式信息。"""
    try:
        from pptx import Presentation
    except ImportError as exc:
        raise ValueError("服务端缺少 PPTX 解析组件") from exc

    try:
        presentation = Presentation(path)
    except Exception as exc:
        raise ValueError(f"PPTX 文件无法解析：{exc}") from exc

    slides = []
    total = len(presentation.slides)
    for index, slide in enumerate(presentation.slides):
        title = _extract_slide_title(slide) or f"第 {index + 1} 页"
        slides.append({
            "pageType": _infer_page_type(slide, title, index, total),
            "title": title[:120],
            "narrativeHint": "由源 PPTX 自动解析",
            "sourceSlide": index + 1,
        })

    if not slides:
        raise ValueError("模板中没有可用幻灯片")
    return {
        "slideCount": total,
        "width": round(presentation.slide_width / 914400, 2),
        "height": round(presentation.slide_height / 914400, 2),
        "slidesConfig": slides,
        "requiredPageTypes": [slide["pageType"] for slide in slides],
    }


def delete_stored_template(asset_id: str | None) -> None:
    """删除管理员上传的旧资产；内置模板永不删除。"""
    if not str(asset_id or "").startswith(STORED_SCHEME):
        return
    path = resolve_template_asset(asset_id)
    if path and os.path.isfile(path):
        try:
            os.remove(path)
        except OSError:
            pass


def _safe_join(root: str, relative_path: str) -> str:
    normalized_root = os.path.abspath(root)
    target = os.path.abspath(os.path.join(normalized_root, relative_path))
    try:
        if os.path.commonpath([normalized_root, target]) != normalized_root:
            raise ValueError("模板资产路径非法")
    except ValueError as exc:
        raise ValueError("模板资产路径非法") from exc
    return target


def _extract_slide_title(slide) -> str:
    title_shape = slide.shapes.title
    if title_shape is not None and title_shape.has_text_frame:
        value = title_shape.text.strip()
        if value:
            return value
    for shape in slide.shapes:
        if getattr(shape, "has_text_frame", False):
            value = shape.text.strip()
            if value:
                return value.splitlines()[0].strip()
    return ""


def _infer_page_type(slide, title: str, index: int, total: int) -> str:
    text = title.lower()
    if index == 0:
        return "cover"
    if index == total - 1 or any(word in text for word in ("感谢", "下一步", "thank")):
        return "closing"
    if "组合" in text and "摘要" in text:
        return "combined_summary"
    if "摘要" in text:
        return "policy_summary"
    if any(word in text for word in ("对应表", "年度对应", "计划书对应")):
        return "alignment_table"
    if any(word in text for word in ("启动路径", "两种路径")):
        return "launch_paths"
    if any(word in text for word in ("承接", "现金流接力")):
        return "cashflow_bridge"
    if any(word in text for word in ("时间轴", "时间表", "阶段")):
        return "timeline"
    if any(word in text for word in ("保司", "公司介绍", "公司实力")):
        return "company"
    if any(getattr(shape, "has_chart", False) for shape in slide.shapes):
        return "comparison_chart" if "对比" in text else "chart"
    if any(getattr(shape, "has_table", False) for shape in slide.shapes):
        return "comparison_table" if "对比" in text else "table"
    if any(word in text for word in ("对比", "比较")):
        return "compare"
    if any(word in text for word in ("组合", "协同", "分工")):
        return "synergy"
    if any(word in text for word in ("结论", "建议", "适合")):
        return "conclusion"
    return "guidance"