#!/usr/bin/env python3
"""Verify no double hyphens in visible text content."""
import re

with open('/root/projects/itpp-infrastructure/projects/hotnow-savannah-v2.html') as f:
    html = f.read()

# Remove all HTML tags, comments, and style blocks
clean = html
# Remove style blocks
clean = re.sub(r'<style>.*?</style>', '', clean, flags=re.DOTALL)
# Remove HTML comments
clean = re.sub(r'<!--.*?-->', '', clean, flags=re.DOTALL)
# Remove HTML tags
clean = re.sub(r'<[^>]+>', '', clean)
# Decode HTML entities that contain --
clean = clean.replace('&mdash;', '')

# Find --
matches = list(re.finditer(r'--', clean))
if matches:
    print(f'Found {len(matches)} double hyphens in visible text:')
    for m in matches[:20]:
        ctx = clean[max(0,m.start()-30):m.end()+30]
        print(f'  offset {m.start()}: "...{ctx.strip()}..."')
else:
    print('ZERO double hyphens in visible text content. Clean!')