group_fqcd_jr/客服agent/_build/_docbuilder.py

# -*- coding: utf-8 -*-
"""客服 Agent 四份交付文档 · 构建器（JSON 配置驱动）

从既有文档抽取「项目 HTML 版式外壳」，再灌入新正文，保证版式与既有体系逐字一致。
用法：python _docbuilder.py <spec.json>
"""
import json
import re
import sys
from pathlib import Path

SRC = Path(r"D:\桌面\金融\开发文档\D3.1-客服Agent需求开发文档与设计方案.html")
OUT_DIR = Path(r"D:\桌面\金融\客服agent")

text = SRC.read_text(encoding="utf-8")

HEAD_END = '<div id="toc-container">'
SHELL_HEAD = text[: text.index(HEAD_END) + len(HEAD_END)]
# ⚠️ 结束锚点必须取「正文第一段注释」的位置：若取 `</div><!-- end doc-container -->`，
# 那在文件里是**最后一行**，会把原文档整段正文一并带进来（曾因此产出 4920 行的错件）。
BODY_START = '<!-- ==================== 0 ==================== -->'
SHELL_MID = text[text.index('  </div>\n</nav>'): text.index(BODY_START)]
SHELL_TAIL = text[text.index('</div><!-- end doc-container -->'):]
assert 'doc-header' in SHELL_MID and len(SHELL_MID) < 4000, "SHELL_MID 抽取异常"
assert SHELL_TAIL.startswith('</div><!-- end doc-container -->')

# 备份一份外壳，便于后续单独构建
Path(r"D:\桌面\金融\_shell_head.html").write_text(SHELL_HEAD, encoding="utf-8")
Path(r"D:\桌面\金融\_shell_mid.html").write_text(SHELL_MID, encoding="utf-8")
Path(r"D:\桌面\金融\_shell_tail.html").write_text(SHELL_TAIL, encoding="utf-8")


def build(spec: dict) -> str:
    head = SHELL_HEAD
    head = head.replace(
        "<title>XX科技·智能财富管家系统 — 智能客服 Agent 需求开发文档与设计方案</title>",
        f"<title>{spec['title']}</title>")
    head = head.replace(
        '<div class="sidebar-subtitle">客服 Agent 专项方案 v2.3 · 双角色 · 身份分离</div>',
        f'<div class="sidebar-subtitle">{spec["subtitle"]}</div>')

    parts = []
    for idx, grp in enumerate(spec["toc"], start=1):
        active = " active" if idx == 1 else ""
        arr = " open" if idx == 1 else ""
        kids = " open" if idx == 1 else ""
        parts.append(f'<div class="toc-group">\n'
                     f'<button class="toc-toggle{active}" data-group="{idx}">'
                     f'<span class="arr{arr}">▶</span><span>{grp["g"]}</span></button>\n'
                     f'<div class="toc-kids{kids}">')
        for href, label in grp["items"]:
            parts.append(f'<a class="toc-item" href="#{href}">{label}</a>')
        parts.append("</div></div>\n")
    toc_html = "\n".join(parts)

    mid = SHELL_MID
    mid = mid.replace(
        '<div id="topbar-title">智能财富管家系统 — 智能客服 Agent 需求开发文档与设计方案</div>',
        f'<div id="topbar-title">{spec["topbar"]}</div>')
    mid = mid.replace('<span id="topbar-badge">v2.3 · 访客 ≠ 角色 · 重构基线</span>',
                      f'<span id="topbar-badge">{spec["badge"]}</span>')
    mid = mid.replace('<div class="doc-eyebrow">专项需求开发文档 · 详细设计方案</div>',
                      f'<div class="doc-eyebrow">{spec["eyebrow"]}</div>')
    mid = mid.replace(
        '<h1 class="doc-title">XX科技·智能财富管家系统<br>智能客服 Agent 需求开发文档与设计方案</h1>',
        f'<h1 class="doc-title">{spec["doctitle"]}</h1>')
    meta_html = "".join(
        f'    <div class="doc-meta-item">\n'
        f'      <span class="doc-meta-label">{k}</span>\n'
        f'      <span class="doc-meta-value">{v}</span>\n'
        f'    </div>\n' for k, v in spec["meta"])
    mid = re.sub(r'  <div class="doc-meta">.*?\n  </div>\n',
                 f'  <div class="doc-meta">\n{meta_html}  </div>\n', mid, flags=re.S)

    tail = SHELL_TAIL
    tail = tail.replace(
        "（<strong>v2.2 双角色 · 重构基线版</strong>）",
        f"（<strong>{spec['closing']}</strong>）")

    # ⚠️ 版式外壳是从旧版式文档抽取的，会把旧品牌字面一并带进来（曾因此漏改 4 份文档的
    # <title>/<h1>/sidebar 品牌）。这里对「外壳」做一次品牌兜底归一，保证任何一次重新生成
    # 都不会回退到旧口径。只作用于 head/mid/tail；body 内可能有意保留历史字面（如 P-10 行），不动。
    BRAND = (
        ("XX科技·智能财富管家系统", "南方基金·智能服务系统"),
        ("XX科技 · 智能金融事业部", "南方基金 · 智能服务事业部"),
        ("XX科技", "南方基金"),
        ("智能财富管家系统", "智能服务系统"),
        ("南方财富管理有限公司", "南方基金管理股份有限公司"),
        ("南方财富", "南方基金"),
        ("400-XXX-XXXX", "400-889-8899"),
        ("400-826-9518", "400-889-8899"),
        ("nanfangwm.com", "nffund.com"),
    )
    head, mid, tail = (
        __import__("functools").reduce(
            lambda s, kv: s.replace(kv[0], kv[1]), BRAND, part)
        for part in (head, mid, tail)
    )

    body = Path(spec["body_file"]).read_text(encoding="utf-8")
    return head + "\n" + toc_html + mid + body + tail


def main() -> None:
    spec = json.loads(Path(sys.argv[1]).read_text(encoding="utf-8"))
    OUT_DIR.mkdir(parents=True, exist_ok=True)
    out = OUT_DIR / spec["out_file"]
    html = build(spec)
    out.write_text(html, encoding="utf-8")
    print(f"OK {out}  ({html.count(chr(10)) + 1} lines)")


if __name__ == "__main__":
    main()