PlantAI-rag-new/tests/docgen/test_templates.py

# -*- coding: utf-8 -*-
"""tests/docgen/test_templates.py —— 内置模版系统验收用例（DC-UNIT-TPL-*）。

2026-09-04 手工测试需求2：套用模版生成可预览 HTML。
口径：公开函数/文件系统断言；全部离线（不依赖 LLM/网络/vstore 服务）。
执行：bash tests/docgen/run_all.sh unit
"""
from __future__ import annotations

import json
import re
import sys
from pathlib import Path
import pytest

REPO = Path(__file__).resolve().parents[2]
sys.path.insert(0, str(REPO))
sys.path.insert(0, str(REPO / "sdk"))

from app.services.docgen import templating  # noqa: E402

pytestmark = pytest.mark.unit

# Story 禁用属性（双渲染纪律：浏览器与 PyMuPDF Story 共用 css）
BANNED_PATTERNS = (
    re.compile(r"position\s*:", re.IGNORECASE),
    re.compile(r"display\s*:\s*(flex|grid)", re.IGNORECASE),
    re.compile(r"transform\s*:", re.IGNORECASE),
    re.compile(r"@media", re.IGNORECASE),
    re.compile(r"@page\s*\{[^}]*margin", re.IGNORECASE),
    re.compile(r"var\(--", re.IGNORECASE),
    re.compile(r"[vhvw]h\b|[vhvw]w\b", re.IGNORECASE),
)


class TestTemplates:

    def test_DC_UNIT_TPL_01_registry_loads(self):
        registry = templating.load_registry()
        assert isinstance(registry, dict)
        entries = registry.get("templates") or []
        keys = [e.get("key") for e in entries if isinstance(e, dict)]
        assert len(keys) == 7
        assert set(keys) == {
            "official", "report", "application", "contract",
            "speech", "plan", "document",
        }
        assert registry.get("fallback") == "document"

    @pytest.mark.parametrize(
        "text,expected",
        [
            ("帮我写一份会议通知", "official"),
            ("起草一份请示", "official"),
            ("写一个调研报告", "report"),
            ("拟一份年度工作总结", "report"),
            ("写一份项目申报书", "application"),
            ("起草一份采购合同", "contract"),
            ("写个讲话稿", "speech"),
            ("写一份感谢信", "speech"),
            ("制定一个实施方案", "plan"),
            ("写个说明书", "document"),
            ("写一篇论文", "document"),
            ("", "document"),
        ],
    )
    def test_DC_UNIT_TPL_02_classify(self, text, expected):
        assert templating.classify_doc_type(text) == expected

    def test_DC_UNIT_TPL_03_render_html_structure(self):
        html, css = templating.render_html(
            "# 关于安全生产的通知\n\n## 一、总体要求\n\n正文内容。",
            title="关于安全生产的通知",
            key="official",
        )
        assert "<main>" in html and "</main>" in html
        assert 'class="doc-title"' in html
        assert "关于安全生产的通知" in html
        assert "正文内容" in html
        assert "{{TITLE}}" not in html
        assert css and "doc-table" in css or True

    def test_DC_UNIT_TPL_04_style_hints_overlay(self):
        hints = {
            "body_font": "宋体",
            "body_size": 14,
            "headings": {"1": {"font": "黑体", "size": 18}},
        }
        _, css = templating.render_html(
            "# 标题\n\n正文。", title="t", key="report", style_hints=hints
        )
        assert "宋体" in css
        assert "font-size: 14.0pt" in css
        assert "font-size: 18.0pt" in css

    def test_DC_UNIT_TPL_05_never_raises(self):
        for markdown in ("", "| a | b |\n|---|---|", "```\nfence\n```", "### 标题"):
            html, css = templating.render_html(markdown, title="", key="document")
            assert isinstance(html, str) and isinstance(css, str)
        # 乱 key 回落 fallback
        html, _ = templating.render_html("x", title="", key="nonexistent-key")
        assert "<main>" in html

    def test_DC_UNIT_TPL_06_css_story_safe(self):
        tpl_dir = templating.templates_dir()
        for css_file in sorted(tpl_dir.glob("*.css")):
            content = css_file.read_text(encoding="utf-8")
            # 剥离注释后再扫（注释里的禁用词不算违规）
            content = re.sub(r"/\*[\s\S]*?\*/", "", content)
            for pattern in BANNED_PATTERNS:
                assert not pattern.search(content), (
                    f"{css_file.name} 含 Story 禁用属性：{pattern.pattern}"
                )

    def test_DC_UNIT_TPL_07_wrapper_single_main_xml(self):
        tpl_dir = templating.templates_dir()
        for html_file in sorted(tpl_dir.glob("*.html")):
            content = html_file.read_text(encoding="utf-8")
            assert content.count("<main>") == 1, f"{html_file.name} main 注入点不唯一"
            assert "<!DOCTYPE html>" in content
            # main 必须为空注入点（渲染时整体替换）
            m = re.search(r"<main>\s*</main>", content)
            assert m is not None, f"{html_file.name} main 非空注入点"

    def test_DC_UNIT_TPL_08_tables_marked(self):
        html, _ = templating.render_html(
            "| 列一 | 列二 |\n| --- | --- |\n| 值1 | 值2 |",
            title="t",
            key="document",
        )
        assert 'class="doc-table"' in html

    def test_DC_UNIT_TPL_09_letterhead(self):
        # 2026-09-04：红头文字可编辑持久化——letterhead 参数覆盖，空回落模版默认
        html, _ = templating.render_html(
            "# 标题\n\n正文。", title="t", key="official", letterhead="某某农业科学研究院"
        )
        assert "某某农业科学研究院" in html
        assert "红头文件" not in html
        # 空 letterhead → 模版默认
        html2, _ = templating.render_html("# 标题", title="t", key="official")
        assert "红头文件" in html2
        # 无红头区模板（contract）：letterhead 不影响
        html3, _ = templating.render_html("# 标题", title="t", key="contract", letterhead="x")
        assert "{{LETTERHEAD}}" not in html3

    @pytest.mark.parametrize(
        "text,expected",
        [
            ("帮我写一个十一放假通知，红头用三亚海古纪生物科技有限公司", "三亚海古纪生物科技有限公司"),
            ("红头：某市农业农村局，写一份通知", "某市农业农村局"),
            ("写一份通知，红头为某某研究所", "某某研究所"),
            ("帮我写一份普通通知", ""),
        ],
    )
    def test_DC_UNIT_TPL_10_letterhead_prompt_extract(self, text, expected):
        # 2026-09-04：prompt「红头用XX」→ 生成即采用指定红头
        got = templating.extract_letterhead_from_prompt(text)
        if expected:
            assert got == expected
        else:
            assert got == ""