onepager/tools/remove-footer-toggle.py

#!/usr/bin/env python3
"""Remove the redundant footer [data-i18n-toggle] block.

After Issue #13 the top-right widget owns toggling. The legacy footer button
+ "Maschinell übersetzt" disclaimer line are now redundant — top-right button
has the same disclaimer in its title tooltip, and ai-disclosure.js still
adds the AI footnote.

Two patterns handled:
  Pattern A: <div style="text-align:center;margin-top:16px..."> ... </div>
  Pattern B (knzlmgmt, schulfrai): <p class="footer" style="margin-top:12px;">...</p>

Idempotent — sites without the patterns are skipped.
"""
import re
import sys
from pathlib import Path

ROOT = Path(__file__).resolve().parent.parent
SITES_DIR = ROOT / "sites"

# Pattern A: a div wrapping the button + br + small disclaimer.
# Allow extra style segments after margin-top:16px (some sites add ;opacity:1 etc.)
PATTERN_A = re.compile(
    r'\s*<div\s+style="text-align:center;margin-top:16px[^"]*">\s*'
    r'<button\s+data-i18n-toggle[\s\S]*?</button>\s*'
    r'<br>\s*<small[^>]*data-de="Maschinell übersetzt"[^>]*>[\s\S]*?</small>\s*'
    r'</div>',
    re.IGNORECASE,
)

# Pattern B: <p class="footer" style="margin-top:12px;">...</p>
PATTERN_B = re.compile(
    r'\s*<p\s+class="footer"\s+style="margin-top:12px;">'
    r'<button\s+data-i18n-toggle[\s\S]*?'
    r'</p>',
    re.IGNORECASE,
)

# Pattern C: bare button (optionally followed by <br><small>...</small>),
# not enclosed in our standard wrappers. Used by sites that placed the
# toggle inline in a footer paragraph (ichbinotto, kainco, keinefreun,
# omakise) or in a custom-classed div (orakil).
PATTERN_C = re.compile(
    r'\s*<button\s+data-i18n-toggle[\s\S]*?</button>'
    r'(?:\s*<br>\s*<small[^>]*data-de="Maschinell übersetzt"[^>]*>[\s\S]*?</small>)?',
    re.IGNORECASE,
)

# Pattern D: <div class="footer-toggle">...</div> (orakil only, but tolerant)
PATTERN_D = re.compile(
    r'\s*<div\s+class="footer-toggle">\s*'
    r'<button\s+data-i18n-toggle[\s\S]*?</button>\s*'
    r'<br>\s*<small[^>]*data-de="Maschinell übersetzt"[^>]*>[\s\S]*?</small>\s*'
    r'</div>',
    re.IGNORECASE,
)


def patch(html):
    new = html
    # Wrappers first (they contain a button); then bare-button fallback.
    new, n_a = PATTERN_A.subn("", new)
    new, n_b = PATTERN_B.subn("", new)
    new, n_d = PATTERN_D.subn("", new)
    new, n_c = PATTERN_C.subn("", new)
    return new, n_a + n_b + n_c + n_d


def main():
    sites = sorted(p for p in SITES_DIR.iterdir() if p.is_dir() and (p / "index.html").exists())
    total_removed = 0
    sites_changed = 0
    sites_with_attr = 0
    sites_residual = []

    for site in sites:
        path = site / "index.html"
        before = path.read_text()
        if "data-i18n-toggle" not in before:
            continue
        sites_with_attr += 1

        # Count old buttons before removal (we expect exactly 1 footer button per site;
        # the new top-right widget injects its own at runtime, not in source).
        before_count = len(re.findall(r"data-i18n-toggle", before))

        after, removed = patch(before)
        after_count = len(re.findall(r"data-i18n-toggle", after))

        if removed > 0:
            path.write_text(after)
            sites_changed += 1
            total_removed += removed
            print(f"  [removed {removed}] {site.name}  (data-i18n-toggle: {before_count} -> {after_count})")
            if after_count > 0:
                sites_residual.append((site.name, after_count))
        else:
            sites_residual.append((site.name, before_count))
            print(f"  [no-match] {site.name}  (still has {before_count} data-i18n-toggle)")

    print(f"\nSites changed: {sites_changed}/{sites_with_attr}")
    print(f"Total wrappers removed: {total_removed}")
    if sites_residual:
        print(f"\nSites with residual data-i18n-toggle attribute (manual review):")
        for name, n in sites_residual:
            print(f"  - {name}: {n}")


if __name__ == "__main__":
    main()