#!/usr/bin/env python3
"""
fix_article_format.py — Stripped Inline-Styles von BSN-Standard-Klassen.
Entfernt veraltete Box-Formate aus Artikel-HTML.
Usage: python3 fix_article_format.py <artikel.html>
"""
import re, sys

if len(sys.argv) < 2:
    print(f"Usage: {sys.argv[0]} <artikel.html>")
    sys.exit(1)

filepath = sys.argv[1]
with open(filepath) as f:
    html = f.read()

original = html
fixes = 0

# 1. Strip style="" from vorbestell-box elements
html, n = re.subn(
    r'(<div[^>]*class="[^"]*vorbestell-box[^"]*")[^>]*>',
    r'\1>',
    html
)
fixes += n

# 2. Strip style="" from bsn-facts elements
html, n = re.subn(
    r'(<div[^>]*class="[^"]*bsn-facts[^"]*")[^>]*>',
    r'\1>',
    html
)
fixes += n

# 3. Remove <div class="affiliate-box">...</div> blocks entirely
html, n = re.subn(
    r'<div class="affiliate-box"[^>]*>.*?</div>\s*',
    '',
    html,
    flags=re.DOTALL
)
fixes += n

# 4. Strip style="background:..." / style="border:..." from <div>s
#    (but ONLY if they also have a class= — pure style-only divs stay)
html, n = re.subn(
    r'(<div[^>]*class="[^"]*"[^>]*)\s+style="[^"]*(?:background|border)[^"]*"',
    r'\1',
    html
)
fixes += n

# 5. Strip style="" from <p> inside old-format deal boxes
html, n = re.subn(
    r'<p style="[^"]*(?:margin|font-weight|font-size|color:#888)[^"]*"[^>]*>',
    r'<p>',
    html
)
fixes += n

if html != original:
    with open(filepath, 'w') as f:
        f.write(html)
    print(f"✅ {fixes} Fixes angewandt auf {filepath}")
else:
    print(f"ℹ️  Keine Fixes nötig für {filepath}")