PlantAI-rag-new/sdk/app/services/test_docgen_html2docx.py

"""services/test_docgen_html2docx.py —— HTML 同源 DOCX 转换（缺陷 #47a，2026-09-10）。

覆盖：红头居中红字、h1-h3 字号、表格边框+表头灰底、图片缺失降级、空 HTML。
"""
from __future__ import annotations

import sys
from pathlib import Path

sys.path.insert(0, str(Path(__file__).resolve().parents[2]))

import pytest
from docx import Document
from docx.oxml.ns import qn

from app.services.docgen.html2docx import docx_from_html_document

# 模拟 render_html 产物形态：红头 + <main> 内容容器（与 templating.py 一致）
HTML_SAMPLE = """<!DOCTYPE html><html><head><meta charset="utf-8">
<title>测试公文</title></head><body>
<div class="doc-letterhead">某某单位文件</div>
<main>
<h1>一、总则</h1>
<p>这是<b>加粗</b>正文，含<em>斜体</em>与<code>inline_code</code>。</p>
<h2>1.1 适用范围</h2>
<ul><li>列表项一</li><li>列表项二</li></ul>
<h3>1.1.1 细节</h3>
<p>三段内容。</p>
<table><tr><th>字段</th><th>说明</th></tr>
<tr><td>名称</td><td>项目名称</td></tr></table>
<figure><img src="http://127.0.0.1:1/definitely-down.png" alt="样例图"/><figcaption>图1 样例</figcaption></figure>
</main>
</body></html>"""


def _first_run_color(run):
    return run.font.color.rgb


def _cell_shade(cell) -> str:
    """取 w:shd 的 w:fill（无则空串）。"""
    tc_pr = cell._tc.get_or_add_tcPr()
    shd = tc_pr.find(qn("w:shd"))
    return shd.get(qn("w:fill")) if shd is not None else ""


def _run_east_asia(run) -> str:
    rpr = run._element.find(qn("w:rPr"))
    if rpr is None:
        return ""
    rfonts = rpr.find(qn("w:rFonts"))
    if rfonts is None:
        return ""
    return rfonts.get(qn("w:eastAsia")) or ""


def test_letterhead_red_centered(tmp_path):
    out = docx_from_html_document(HTML_SAMPLE, title="测试公文", out_path=tmp_path / "a.docx")
    assert out.exists()
    doc = Document(str(out))
    paras = doc.paragraphs
    assert paras, "应至少有一个红头段落"
    head = paras[0]
    assert "某某单位文件" in head.text
    assert head.alignment is not None and head.alignment.name == "CENTER"
    run = head.runs[0]
    assert str(_first_run_color(run)) == "FF0000"
    # 22pt 黑体
    assert run.font.size is not None and run.font.size.pt == 22.0
    assert _run_east_asia(run) == "黑体"


def test_heading_levels_and_sizes(tmp_path):
    out = docx_from_html_document(HTML_SAMPLE, title="测试公文", out_path=tmp_path / "b.docx")
    doc = Document(str(out))
    styles = [p.style.name for p in doc.paragraphs]
    assert "Heading 1" in styles
    assert "Heading 2" in styles
    assert "Heading 3" in styles
    sizes = {}
    for p in doc.paragraphs:
        if p.style.name.startswith("Heading") and p.runs:
            sizes[p.style.name] = p.runs[0].font.size.pt
    assert sizes.get("Heading 1") == 22.0
    assert sizes.get("Heading 2") == 16.0
    assert sizes.get("Heading 3") == 14.0


def test_table_grid_and_header_shade(tmp_path):
    out = docx_from_html_document(HTML_SAMPLE, title="测试公文", out_path=tmp_path / "c.docx")
    doc = Document(str(out))
    assert len(doc.tables) == 1
    table = doc.tables[0]
    assert table.style.name == "Table Grid"
    assert len(table.rows) == 2 and len(table.columns) == 2
    assert table.cell(0, 0).text == "字段"
    assert _cell_shade(table.cell(0, 0)) == "D9D9D9"
    assert _cell_shade(table.cell(1, 0)) == ""


def test_image_fetch_failure_falls_back_to_alt(tmp_path):
    out = docx_from_html_document(HTML_SAMPLE, title="测试公文", out_path=tmp_path / "d.docx")
    doc = Document(str(out))
    text = "\\n".join(p.text for p in doc.paragraphs)
    assert "图片缺失" in text
    assert "样例图" in text  # alt 文本兜底
    # 表格完整
    assert doc.tables[0].cell(1, 1).text == "项目名称"


def test_empty_html_produces_empty_document(tmp_path):
    out = docx_from_html_document("", title="空文档", out_path=tmp_path / "e.docx")
    assert out.exists()
    doc = Document(str(out))
    assert doc.core_properties.title == "空文档"
    assert len(doc.paragraphs) == 0
    assert len(doc.tables) == 0


def test_bold_italic_code_runs(tmp_path):
    out = docx_from_html_document(HTML_SAMPLE, title="测试公文", out_path=tmp_path / "f.docx")
    doc = Document(str(out))
    bold_found = any(r.font.bold for p in doc.paragraphs for r in p.runs)
    assert bold_found
    code_found = any(r.font.name == "Courier New"
                     for p in doc.paragraphs for r in p.runs)
    assert code_found


# ---------------------------------------------------------------------------
# 缺陷 #47a（2026-09-11）：红头/标题样式从 render_html 产物解析（css 参数）
# ---------------------------------------------------------------------------

HTML_OFFICIAL_STYLED = """<!DOCTYPE html><html><head><meta charset="utf-8">
<title>红头公文</title></head><body>
<div class="doc-cover"><div class="doc-letterhead">某某单位文件</div></div>
<main><div class="doc-body">
<h1 class="doc-title">关于加强安全生产的通知</h1>
<h2>一、总体要求</h2>
<h3>1.1 具体要求</h3>
<p>正文内容。</p>
</div></main>
</body></html>"""

OFFICIAL_CSS = """
.doc-letterhead { color: #c00000; font-size: 30px; font-weight: bold; }
h1.doc-title { font-size: 22px; }
h2 { font-size: 16px; }
h3, h4, h5, h6 { font-size: 15px; }
"""


def test_letterhead_style_from_css_param(tmp_path):
    """红头样式从模板 CSS 解析：#c00000 系颜色、30px→22.5pt 字号。"""
    out = docx_from_html_document(
        HTML_OFFICIAL_STYLED, title="红头公文",
        out_path=tmp_path / "a.docx", css=OFFICIAL_CSS,
    )
    doc = Document(str(out))
    head = doc.paragraphs[0]
    assert "某某单位文件" in head.text
    assert head.alignment.name == "CENTER"
    run = head.runs[0]
    assert str(_first_run_color(run)) == "C00000"
    assert run.font.size is not None and abs(run.font.size.pt - 22.5) < 0.01
    assert run.font.bold is True


def test_heading_sizes_from_css_param(tmp_path):
    """标题字号从模板 CSS 解析：h1 22px→16.5pt、h2 16px→12pt、h3 15px→11.25pt。"""
    out = docx_from_html_document(
        HTML_OFFICIAL_STYLED, title="红头公文",
        out_path=tmp_path / "b.docx", css=OFFICIAL_CSS,
    )
    doc = Document(str(out))
    sizes = {}
    for p in doc.paragraphs:
        if p.style.name.startswith("Heading") and p.runs:
            sizes[p.style.name] = p.runs[0].font.size.pt
    # DOCX 字号按半点（half-point）量化：11.25pt 落 22 半点 = 11.0pt
    assert abs(sizes.get("Heading 1") - 16.5) < 0.01
    assert abs(sizes.get("Heading 2") - 12.0) < 0.01
    assert abs(sizes.get("Heading 3") - 11.25) <= 0.26


def test_no_style_falls_back_to_legacy_constants(tmp_path):
    """无样式 HTML（不传 css）→ 回退 r3 常量（FF0000/22pt 红头，22/16/14pt 标题）。"""
    out = docx_from_html_document(HTML_SAMPLE, title="测试公文",
                                  out_path=tmp_path / "c.docx")
    doc = Document(str(out))
    head = doc.paragraphs[0]
    run = head.runs[0]
    assert str(_first_run_color(run)) == "FF0000"
    assert run.font.size.pt == 22.0
    sizes = {}
    for p in doc.paragraphs:
        if p.style.name.startswith("Heading") and p.runs:
            sizes[p.style.name] = p.runs[0].font.size.pt
    assert sizes.get("Heading 1") == 22.0
    assert sizes.get("Heading 2") == 16.0
    assert sizes.get("Heading 3") == 14.0


def test_report_template_grey_letterhead_not_red(tmp_path):
    """非红头模板（report：灰 #333、26px）不误渲染红头。"""
    report_css = """
    .doc-letterhead { color: #333333; font-size: 26px; font-weight: bold; }
    h1.doc-title { font-size: 22px; }
    """
    out = docx_from_html_document(
        HTML_OFFICIAL_STYLED, title="报告",
        out_path=tmp_path / "d.docx", css=report_css,
    )
    doc = Document(str(out))
    run = doc.paragraphs[0].runs[0]
    assert str(_first_run_color(run)) == "333333"
    assert abs(run.font.size.pt - 19.5) < 0.01  # 26px * 0.75


# r3-47-01 复现样例：真实 render_html 产物形态——
# <main><div class="doc-body">…</div></main> 包裹层 + 嵌套 section
HTML_WRAPPED = """<!DOCTYPE html><html><head><meta charset="utf-8">
<title>水稻病虫害防治方案</title></head><body>
<div class="doc-cover"><div class="doc-letterhead">某某单位文件</div></div>
<main><div class="doc-body">
<h1 class="doc-title">水稻病虫害防治方案</h1>
<section>
<h2>一、病害防治</h2>
<p>第一节正文，共 60 段落之一。</p>
<p>另一个段落。</p>
<h3>1.1 稻瘟病</h3>
<ul><li>防治要点一</li><li>防治要点二</li></ul>
<table class="doc-table"><tr><th>药剂</th><th>用量</th></tr>
<tr><td>三环唑</td><td>20g/亩</td></tr></table>
<figure><img src="http://127.0.0.1:1/down.png" alt="稻瘟病症状"/><figcaption>图1 症状</figcaption></figure>
</section>
<section>
<h2>二、虫害防治</h2>
<p>第二节正文。</p>
</section>
</div></main>
</body></html>"""


def test_wrapped_docbody_structure_preserved(tmp_path):
    """r3-47-01 回归：doc-body 包裹层下钻——标题/段落/表格/列表/图注全保留，
    不得塌缩成单段落。"""
    out = docx_from_html_document(HTML_WRAPPED, title="水稻病虫害防治方案",
                                  out_path=tmp_path / "w.docx")
    doc = Document(str(out))
    styles = [p.style.name for p in doc.paragraphs]
    assert "Heading 1" in styles
    assert styles.count("Heading 2") == 2  # 两个 section 的 h2
    assert "Heading 3" in styles
    assert "List Bullet" in styles
    # 表格完整（2 行 2 列，表头灰底）
    assert len(doc.tables) == 1
    table = doc.tables[0]
    assert len(table.rows) == 2 and table.cell(1, 1).text == "20g/亩"
    assert _cell_shade(table.cell(0, 0)) == "D9D9D9"
    # 段落文本保留（不是全塌成一段）
    texts = "\n".join(p.text for p in doc.paragraphs)
    assert "第一节正文" in texts
    assert "第二节正文" in texts
    # 红头仍正确
    assert doc.paragraphs[0].text == "某某单位文件"


def test_wrapped_structure_counts_match_html(tmp_path):
    """r3-47-01：docx 元素计数与 HTML 结构对应（段落数 >= p 数，标题/表格计数一致）。"""
    from bs4 import BeautifulSoup

    soup = BeautifulSoup(HTML_WRAPPED, "html.parser")
    n_p = len(soup.find_all("p"))
    n_h2 = len(soup.find_all("h2"))
    n_h3 = len(soup.find_all("h3"))
    n_table = len(soup.find_all("table"))
    out = docx_from_html_document(HTML_WRAPPED, title="水稻病虫害防治方案",
                                  out_path=tmp_path / "c.docx")
    doc = Document(str(out))
    styles = [p.style.name for p in doc.paragraphs]
    assert styles.count("Heading 2") == n_h2
    assert styles.count("Heading 3") == n_h3
    assert len(doc.tables) == n_table
    # 段落：p 落 Normal（含图注段），计数不塌缩（>= p 数，图注/图片段会额外加）
    assert styles.count("Normal") >= n_p