before claude

pricewatch/app/tasks/scrape.py

@@ -157,6 +157,36 @@ def scrape_product(
             )
             success = False
             fetch_error = str(exc)
+        # Si captcha detecte via HTTP, forcer une tentative Playwright.
+        if (
+            fetch_method == FetchMethod.HTTP
+            and use_playwright
+            and snapshot.debug.errors
+            and any("captcha" in error.lower() for error in snapshot.debug.errors)
+        ):
+            logger.info("[FETCH] Captcha detecte, tentative Playwright")
+            pw_result = fetch_playwright(
+                canonical_url,
+                headless=not headful,
+                timeout_ms=timeout_ms,
+                save_screenshot=save_screenshot,
+            )
+            if pw_result.success and pw_result.html:
+                try:
+                    snapshot = store.parse(pw_result.html, canonical_url)
+                    snapshot.debug.method = FetchMethod.PLAYWRIGHT
+                    snapshot.debug.duration_ms = pw_result.duration_ms
+                    snapshot.debug.html_size_bytes = len(pw_result.html.encode("utf-8"))
+                    snapshot.add_note("Captcha detecte via HTTP, fallback Playwright")
+                    success = snapshot.debug.status != DebugStatus.FAILED
+                except Exception as exc:
+                    snapshot.add_note(f"Fallback Playwright echoue: {exc}")
+                    logger.error(f"[PARSE] Exception fallback Playwright: {exc}")
+                    fetch_error = str(exc)
+            else:
+                error = pw_result.error or "Erreur Playwright"
+                snapshot.add_note(f"Fallback Playwright echoue: {error}")
+                fetch_error = error
     else:
         snapshot = ProductSnapshot(
             source=store.store_id,