"""HTML 美化：注入 MathJax / 样式 / 导航 / 统计"""
import re
import os
import shutil
import urllib.request
# --- 常量 ---
NAV = ('')
STYLE = (''
'')
MATHJAX_TAG = (''
'')
ASSETS_DIR = os.path.expanduser('~/.hermes/mcp/assets')
MATHJAX_FILE = os.path.join(ASSETS_DIR, 'tex-svg.js')
MATHJAX_MIRROR = 'https://registry.npmmirror.com/mathjax/3.2.2/files/es5/tex-svg.js'
def _fix_math(m):
inner = (m.group(1).replace(' ', ' ').replace('&', '&')
.replace('<', '<').replace('>', '>').replace(' ', ' '))
return r'\(' + re.sub(r'[ \t]{2,}', ' ', inner).strip() + r'\)'
def clean_html(content: str, mathjax: bool = True) -> str:
"""原始 HTML → 加 MathJax + 导航 + 样式"""
c = re.sub(r'^<\?xml[^>]*\?>\s*', '', content)
c = re.sub(r']*>', '', c, count=1)
c = re.sub(r'\\\((.*?)\\\)', _fix_math, c, flags=re.S)
inject = STYLE + (MATHJAX_TAG if mathjax else '')
c = c.replace('', inject + '', 1) if '' in c else inject + c
m = re.search(r']*>', c, re.I)
c = (c[:m.end()] + NAV + c[m.end():]) if m else (NAV + c)
return c
def stats(html: str) -> dict:
"""HTML 统计：字符 / 公式 / 图片 / 未包裹 LaTeX"""
stripped = re.sub(r'\\\(.*?\\\)', '', html, flags=re.S)
return {
'chars': len(html),
'formulas': len(re.findall(r'\\\(', html)),
'images': html.count(' str:
"""下载/复用 MathJax 离线文件 → 返回路径"""
os.makedirs(ASSETS_DIR, exist_ok=True)
if not os.path.exists(MATHJAX_FILE) or os.path.getsize(MATHJAX_FILE) < 500_000:
urllib.request.urlretrieve(MATHJAX_MIRROR, MATHJAX_FILE)
return MATHJAX_FILE
def copy_mathjax_to(out_dir: str) -> str:
"""把 MathJax 复制到 out_dir，返回目标路径"""
src = ensure_mathjax()
dst = os.path.join(out_dir, 'tex-svg.js')
shutil.copy(src, dst)
return dst