#!/usr/bin/env python3 """Audit 404 GSC — teste les URLs candidates et propose une action.""" from __future__ import annotations import re import urllib.error import urllib.parse import urllib.request from dataclasses import dataclass BASE = "https://www.arnaud-merigeau.fr" # URLs réelles repérées (logs Googlebot + .htaccess cassé + crawl interne) CANDIDATES: list[tuple[str, str]] = [ # Portfolio / feeds ("/realisations/feed/", "feed archive portfolio"), ("/realisations/olow/feed/", "feed projet"), ("/realisations/marketplace-produits-locaux/feed/", "feed projet"), ("/realisations/new-york-french-connection/feed/", "feed projet supprimé"), ("/realisations/boutique-prestashop-soin-spa/", "projet retiré"), ("/realisations/theme-prestashop-wide/", "redirect htaccess cassé"), ("/realisations-sitemap.xml", "sitemap obsolète"), # Articles / slugs ("/10-themes-prestashop-1-8-gratuits-et-responsive", "slug article"), ("/10-themes-prestashop-1-7-gratuits-et-responsive/", "article existant"), ("/ajouter-recaptcha-prestashop-tuto-gratuitfeed/", "URL typo"), ("/ajouter-recaptcha-prestashop-tuto-gratuit/", "slug probable"), ("/migration-prestashop-1-7-vers-9-sans-perdre-seo/", "brouillon pilier"), ("/tarif-freelance-prestashop-2026/", "brouillon pilier"), # Catégories / pagination ("/blog/category/referencement/", "structure /blog/category/"), ("/category/referencement/", "structure WP"), ("/blog/category/vitesse-de-chargement/", "structure /blog/category/"), ("/articles/page/11/", "redirect legacy"), ("/articles/page/12/", "redirect legacy"), ("/articles/page/13/", "redirect legacy"), ("/blog/page/11/", "cible redirect articles"), ("/category/prestashop/page/17/", "pagination"), ("/tag/prestashop-2/page/15/", "pagination"), ("/boutique/page/3/", "pagination boutique"), # URLs datées ("/2009/12/03/multiwizz-ou-le-sms-gratuit/?cid=245", "redirect ?cid cassé"), ("/2009/12/03/multiwizz-ou-le-sms-gratuit/", "URL datée"), ("/multiwizz-ou-le-sms-gratuit/", "cible attendue"), ("/2009/07/multiwizz-ou-le-sms-gratuit/", "URL datée variante"), ("/2010/10/22/piles-plus/?cid=161", "redirect ?cid cassé"), ("/2010/10/22/piles-plus/", "URL datée"), ("/2011/07/14/20-themes-prestashop-1-3-1-4-gratuits/?cid=284", "redirect ?cid cassé"), ("/20-themes-prestashop-1-3-1-4-gratuits/", "cible attendue"), ("/2011/07/13/tuto-integration-du-bouton-google-1-dans-prestashop/", "URL datée"), ("/2012/02/02/rassurance-client-sur-votre-boutique-prestashop-5-points-importants", "sans slash final"), # Query legacy ("/?post_type=product&p=2196", "ID produit legacy"), ("/?post_type=portfolio&p=4124", "ID portfolio legacy"), ("/?p=15803", "brouillon"), ("/?author=2", "archive auteur par ID"), ("/author/amerigeau/", "archive auteur"), # Divers ("/boutique/theme-prestashop-1-4-wide/", "cible redirect theme"), ("/confidentialite/", "redirect mentions"), ("/a-propos", "page about"), ("/a-propos/", "page about"), ("/product-category-sitemap.xml", "sitemap WC"), ("/sitemap.xml", "sitemap racine"), ("/sitemap_index.xml", "sitemap Yoast"), ] @dataclass class Result: path: str note: str status: int | str final_url: str action: str def fetch(path: str) -> tuple[int | str, str, str]: url = BASE + path if path.startswith("/") else path req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0 GSC-404-Audit/1.0"}) try: with urllib.request.urlopen(req, timeout=20) as resp: return resp.status, resp.geturl(), resp.headers.get("Location", "") except urllib.error.HTTPError as e: return e.code, url, e.headers.get("Location", "") except Exception as exc: return str(exc), url, "" def suggest(path: str, status: int | str, final: str, note: str) -> str: if status in (301, 302, 308): if final.rstrip("/") != (BASE + path.split("?")[0]).rstrip("/"): return f"Redirection OK → {final}" return f"Redirige mais vérifier cible: {final}" if status == 200: return "OK — aucune action" if status == 404: if "feed" in path: return "301 → /realisations/ ou activer feed + noindex" if "soin-spa" in path or "theme-prestashop-wide" in path: return "301 → /realisations/" if "1-8-gratuits" in path: return "301 → /10-themes-prestashop-1-7-gratuits-et-responsive/" if "recaptcha" in path and "feed" in path: return "301 → /ajouter-recaptcha-prestashop-tuto-gratuit/" if "migration-prestashop" in path or "tarif-freelance" in path: return "Publier le brouillon OU retirer les liens internes en attendant" if "/blog/category/" in path: slug = path.split("/")[-2] if path.endswith("/") else path.split("/")[-1] return f"301 → /category/{slug}/" if "realisations-sitemap" in path: return "301 → /portfolio-sitemap.xml" if "articles/page/" in path: n = re.search(r"/page/(\d+)/", path) return f"RewriteRule articles → blog (page {n.group(1) if n else '?'})" if re.match(r"/\d{4}/", path): slug = path.rstrip("/").split("/")[-1].split("?")[0] return f"301 → /{slug}/ (RewriteRule date legacy)" if "?post_type=" in path or "?p=" in path: return "301 → /boutique/ ou /blog/ (IDs obsolètes)" if "boutique/theme-prestashop" in path: return "301 → /boutique/ (produit retiré)" if path in ("/a-propos", "/a-propos/"): return "301 → /savoir-faire/ ou /contact/" if "product-category-sitemap" in path: return "Ignorer (scan bot) ou 410" return "Créer redirection 301 manuelle" if status == 403: return "Bloquer / ignorer (bot)" return "À analyser" def main() -> None: results: list[Result] = [] for path, note in CANDIDATES: status, final, _ = fetch(path) results.append(Result(path, note, status, final, suggest(path, status, final, note))) need_action = [r for r in results if "OK — aucune" not in r.action and "Ignorer" not in r.action] print(f"Testées: {len(results)} | Actions recommandées: {len(need_action)}\n") print("STATUS\tACTION\tURL") for r in sorted(need_action, key=lambda x: str(x.status)): print(f"{r.status}\t{r.action}\t{r.path}") print("\n--- Pages OK (200 ou redirect sain) ---") for r in results: if r not in need_action: print(f"{r.status}\t{r.path} → {r.final_url if r.final_url != BASE + r.path else '200'}") if __name__ == "__main__": main()