html/api/wem-thumb-worker.py

#!/usr/bin/env python3
"""
wem-thumb-worker.py — Generate thumbnail + metadata for a page
Called by wem-screen-thumb.php in background or by cron refresh.
Usage: python3 wem-thumb-worker.py <path>
Example: python3 wem-thumb-worker.py ethica-chatbot.html

Output:
  /var/www/html/api/screenshots/wem/<md5>.png (320x200)
  Updates /var/www/html/api/wem-page-meta.json with {title, http, mtime, ts}

Doctrine 84.
"""
import sys, os, hashlib, json, time, subprocess
from pathlib import Path

if len(sys.argv) < 2:
    print("Usage: wem-thumb-worker.py <path>")
    sys.exit(1)

raw = sys.argv[1]

# Accept either: bare path (ethica-chatbot.html) OR full URL
# Resolve URL to the right upstream (S204 vs S95)
import re as _re
m_wevads = _re.match(r'^https?://wevads\.weval-consulting\.com/(.+)$', raw)
m_main = _re.match(r'^https?://(?:www\.)?weval-consulting\.com/(.+)$', raw)
m_ethica = _re.match(r'^https?://ethica\.wevup\.app/(.+)$', raw)

if m_wevads:
    # Route via S95 directly (WireGuard 10.1.0.3:5890) — faster than going through CF
    path = m_wevads.group(1)
    url_for_meta = raw
    url_for_fetch = f"http://10.1.0.3:5890/{path}"
    key = f"wevads/{path}"  # meta key prefix to disambiguate from S204
elif m_main:
    path = m_main.group(1)
    url_for_meta = raw
    url_for_fetch = f"http://127.0.0.1:5890/{path}"
    key = path
elif m_ethica:
    path = m_ethica.group(1)
    url_for_meta = raw
    url_for_fetch = f"http://127.0.0.1:5890/{path}"  # ethica.wevup.app proxies locally
    key = f"ethica/{path}"
else:
    # bare path → S204 local
    path = raw.lstrip('/')
    url_for_meta = f"https://weval-consulting.com/{path}"
    url_for_fetch = f"http://127.0.0.1:5890/{path}"
    key = path

h = hashlib.md5(key.encode()).hexdigest()
thumb_out = f"/var/www/html/api/screenshots/wem/{h}.png"
meta_file = "/var/www/html/api/wem-page-meta.json"
lockfile = f"/tmp/wem-thumb-{h}.lock"

# Load existing meta
try:
    with open(meta_file) as f:
        meta = json.load(f)
except Exception:
    meta = {}

url = url_for_fetch

# For vhost routing on S95 without Host header, fallback to public URL via CF
# This is slower but Host header breaks Playwright's connection reuse
if 'wevads' in key:
    url = url_for_meta  # use https://wevads.weval-consulting.com/... (public via CF)
elif 'ethica/' in key:
    url = url_for_meta  # use https://ethica.wevup.app/... (public via CF)

host_header = None  # not needed with public URLs

try:
    from playwright.sync_api import sync_playwright

    with sync_playwright() as p:
        browser = p.chromium.launch(headless=True, args=['--no-sandbox', '--disable-dev-shm-usage'])
        context = browser.new_context(viewport={'width': 1024, 'height': 640}, ignore_https_errors=True)
        page = context.new_page()

        if host_header:
            page.set_extra_http_headers({'Host': host_header})

        http_status = 0
        try:
            resp = page.goto(url, wait_until='domcontentloaded', timeout=12000)
            http_status = resp.status if resp else 0
        except Exception as e:
            http_status = -1
            print(f"nav error: {e}")

        # Give a beat for any fast JS render
        try:
            page.wait_for_timeout(1500)
        except Exception:
            pass

        # Get real title
        try:
            title = page.title() or path
        except Exception:
            title = path
        title = title.strip()[:120]

        # Screenshot full viewport (1024x640) then we resize
        png_tmp = f"/tmp/wem-raw-{h}.png"
        try:
            page.screenshot(path=png_tmp, full_page=False, type='png')
        except Exception as e:
            print(f"screenshot error: {e}")
            browser.close()
            sys.exit(2)

        browser.close()

    # Resize to 320x200 using Python Pillow (fast) or ImageMagick fallback
    try:
        from PIL import Image
        img = Image.open(png_tmp)
        # Crop top portion (aspect 16:10 from 1024x640 is already 16:10, just resize)
        img.thumbnail((640, 400), Image.LANCZOS)
        # Pad/crop to 320x200
        img = img.resize((320, 200), Image.LANCZOS)
        img.save(thumb_out, 'PNG', optimize=True)
    except Exception as e:
        # Fallback: ImageMagick
        subprocess.run(['convert', png_tmp, '-resize', '320x200!', thumb_out], check=False)

    # Cleanup tmp
    try: os.unlink(png_tmp)
    except Exception: pass

    # Chown for web
    try:
        subprocess.run(['chown', 'www-data:www-data', thumb_out], check=False)
    except Exception:
        pass

    # Get file mtime (only for S204-local files)
    try:
        file_path = f"/var/www/html/{path}"
        mtime = time.strftime('%d%b', time.localtime(os.path.getmtime(file_path))) if os.path.exists(file_path) else ''
    except Exception:
        mtime = ''

    # Update meta — key is the URL-derived key used by the HTML for lookup
    meta[key] = {
        'title': title,
        'http': http_status,
        'mtime': mtime,
        'thumb': f"/api/screenshots/wem/{h}.png",
        'url': url_for_meta,
        'ts': int(time.time())
    }

    # Atomic write
    tmp_meta = meta_file + '.tmp'
    with open(tmp_meta, 'w') as f:
        json.dump(meta, f, indent=2, ensure_ascii=False)
    os.rename(tmp_meta, meta_file)
    try:
        subprocess.run(['chown', 'www-data:www-data', meta_file], check=False)
    except Exception:
        pass

    print(f"OK {path} http={http_status} title={title[:40]}")

except Exception as e:
    print(f"FATAL {path}: {e}")
    sys.exit(3)
finally:
    try: os.unlink(lockfile)
    except Exception: pass