feat: scripts de projectos vindos do Hub (podcast, alojadamaria, clip, ocr, etc.)

Movidos do vault Hub para centralizar scripts. Hub mantem symlinks. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-28 20:53:29 +01:00
parent e810bbb114
commit 6035542b67
27 changed files with 4246 additions and 0 deletions
@@ -0,0 +1,141 @@
+#!/usr/bin/env python3
+"""
+Captura screenshots de alojadamaria.com para auditoria visual SEO/UX
+"""
+from playwright.sync_api import sync_playwright
+import json
+import time
+import os
+
+BASE_URL = "https://alojadamaria.com/"
+OUTPUT_DIR = "/media/ealmeida/Dados/Hub/03-Propostas/ALojaDaMaria/screenshots/alojadamaria"
+
+os.makedirs(OUTPUT_DIR, exist_ok=True)
+
+VIEWPORTS = {
+    "desktop": {"width": 1440, "height": 900},
+    "mobile": {"width": 375, "height": 812},
+}
+
+PAGES = {
+    "homepage": BASE_URL,
+    "categoria": BASE_URL + "product-category/novidades/",
+    "contacto": BASE_URL + "contactos/",
+}
+
+def capturar(page, url, nome, viewport):
+    """Captura acima da dobra e página completa"""
+    print(f"  -> A capturar: {nome} ({viewport['width']}x{viewport['height']})")
+    try:
+        page.goto(url, wait_until="networkidle", timeout=30000)
+        time.sleep(2)
+
+        # Fechar pop-ups comuns (cookie consent, newsletter)
+        for selector in [
+            "button[class*='close']",
+            "button[class*='dismiss']",
+            "[class*='cookie'] button",
+            "[id*='cookie'] button",
+            "[class*='popup-close']",
+            ".pum-close",
+            "button[aria-label*='Close']",
+            "button[aria-label*='close']",
+        ]:
+            try:
+                el = page.query_selector(selector)
+                if el and el.is_visible():
+                    el.click()
+                    time.sleep(0.5)
+            except Exception:
+                pass
+
+        # Above the fold (viewport apenas)
+        page.screenshot(
+            path=f"{OUTPUT_DIR}/{nome}_atf.png",
+            full_page=False,
+            clip={"x": 0, "y": 0, "width": viewport["width"], "height": viewport["height"]},
+        )
+
+        # Página completa
+        page.screenshot(
+            path=f"{OUTPUT_DIR}/{nome}_full.png",
+            full_page=True,
+        )
+
+        # Recolher metadados
+        title = page.title()
+        h1_els = page.query_selector_all("h1")
+        h1_texts = [el.inner_text().strip() for el in h1_els if el.is_visible()]
+
+        nav_visible = bool(page.query_selector("nav, [class*='nav'], [class*='menu']"))
+
+        ctas = []
+        for sel in ["a[class*='btn'], a[class*='button'], button[class*='btn'], .add-to-cart, [class*='cta']"]:
+            els = page.query_selector_all(sel)
+            for el in els[:5]:
+                try:
+                    if el.is_visible():
+                        ctas.append(el.inner_text().strip()[:50])
+                except Exception:
+                    pass
+
+        popup_visible = bool(page.query_selector(".pum-overlay, [class*='popup'][style*='display: block'], [class*='modal'][style*='display: block']"))
+
+        # Dimensões do logo
+        logo = page.query_selector("img[class*='logo'], a[class*='logo'] img, header img, .site-logo img")
+        logo_info = None
+        if logo:
+            try:
+                bb = logo.bounding_box()
+                logo_info = bb
+            except Exception:
+                pass
+
+        return {
+            "url": url,
+            "title": title,
+            "h1": h1_texts,
+            "nav_visible": nav_visible,
+            "ctas_sample": ctas[:8],
+            "popup_detected": popup_visible,
+            "logo_bounding_box": logo_info,
+        }
+    except Exception as e:
+        print(f"  ERRO: {e}")
+        return {"error": str(e)}
+
+
+def main():
+    resultados = {}
+
+    with sync_playwright() as p:
+        browser = p.chromium.launch(headless=True)
+
+        for device_name, viewport in VIEWPORTS.items():
+            print(f"\n[{device_name.upper()}] {viewport['width']}x{viewport['height']}")
+            context = browser.new_context(
+                viewport=viewport,
+                user_agent="Mozilla/5.0 (compatible; AuditBot/1.0)",
+                locale="pt-PT",
+            )
+            page = context.new_page()
+
+            for page_name, url in PAGES.items():
+                chave = f"{device_name}_{page_name}"
+                print(f"  Página: {page_name}")
+                dados = capturar(page, url, chave, viewport)
+                resultados[chave] = dados
+
+            context.close()
+
+        browser.close()
+
+    with open(f"{OUTPUT_DIR}/metadados.json", "w", encoding="utf-8") as f:
+        json.dump(resultados, f, ensure_ascii=False, indent=2)
+
+    print("\nCaptura concluída. Ficheiros em:", OUTPUT_DIR)
+    return resultados
+
+
+if __name__ == "__main__":
+    main()