fix(site): consolidate homepage search signals

scripts/build-site.py

@@ -18,6 +18,10 @@ def locale_path(locale):
     return f"/{locale}/"
 
 
+def public_path(locale):
+    return "/" if locale == "en-US" else locale_path(locale)
+
+
 def load_catalogs():
     catalogs = {p.stem: json.loads(p.read_text()) for p in sorted((ROOT / "site/locales").glob("*.json"))}
     supported = {p.parent.name for p in (ROOT / "locales").glob("*/opencadstudio.ftl")}
@@ -65,14 +69,14 @@ def build(output):
     script = "const SITE_LOCALES = " + json.dumps(list(catalogs)) + ";\n" + (ROOT / "site/site.js").read_text()
     (output / "site.js").write_text(script, encoding="utf-8")
     js_version = hashlib.sha256(script.encode()).hexdigest()[:12]
-    alternates = '\n'.join(f'<link rel="alternate" hreflang="{lang}" href="{BASE}{locale_path(lang)}" />' for lang in catalogs)
+    alternates = '\n'.join(f'<link rel="alternate" hreflang="{lang}" href="{BASE}{public_path(lang)}" />' for lang in catalogs)
     alternates += f'\n<link rel="alternate" hreflang="x-default" href="{BASE}/" />'
     for locale, messages in catalogs.items():
-        path = locale_path(locale)
-        destination = output / path.lstrip("/")
+        path = public_path(locale)
+        destination = output / locale_path(locale).lstrip("/")
         destination.mkdir(parents=True, exist_ok=True)
         language_links = '\n'.join(
-            f'<a href="{locale_path(lang)}" lang="{lang}" hreflang="{lang}" dir="auto"'
+            f'<a href="{public_path(lang)}" lang="{lang}" hreflang="{lang}" dir="auto"'
             + (' aria-current="page"' if lang == locale else '')
             + f'>{escape(labels["name"])}</a>' for lang, labels in catalogs.items()
         )
@@ -85,6 +89,11 @@ def build(output):
             "downloadUrl": REPO + "/releases/latest", "installUrl": BASE + "/app/",
             "offers": {"@type": "Offer", "price": "0", "priceCurrency": "USD"},
         }
+        website_schema = {
+            "@context": "https://schema.org", "@type": "WebSite",
+            "name": "Open CAD Studio", "alternateName": ["OpenCADStudio", "opencadstudio.com"],
+            "url": BASE + "/",
+        }
         values = {key: escape(value) for key, value in messages.items()}
         values.update(locale=locale, direction="rtl" if locale == "ar-SA" else "ltr",
                       path=path, url=BASE + path, og_locale=locale.replace("-", "_"),
@@ -92,6 +101,8 @@ def build(output):
                       css_version=css_version, js_version=js_version, alternates=alternates,
                       languages=escape(messages["languages"].format(count=len(catalogs))),
                       language_links=language_links,
+                      website_schema=(f'<script type="application/ld+json">{json.dumps(website_schema)}</script>'
+                                      if locale == "en-US" else ""),
                       schema=json.dumps(schema, ensure_ascii=False).replace("<", "\\u003c"))
         rendered = re.sub(r"\{\{(\w+)\}\}", lambda match: values[match[1]], template)
         (destination / "index.html").write_text(rendered, encoding="utf-8")
@@ -101,7 +112,7 @@ def build(output):
         (destination / "site.webmanifest").write_text(json.dumps(localized_manifest, ensure_ascii=False, indent=2) + '\n', encoding="utf-8")
     for filename in ("index.html", "site.webmanifest"):
         shutil.copyfile(output / "en-US" / filename, output / filename)
-    urls = f'<url><loc>{BASE}/</loc></url>\n' + ''.join(f'<url><loc>{BASE}{locale_path(locale)}</loc></url>\n' for locale in catalogs)
+    urls = ''.join(f'<url><loc>{BASE}{public_path(locale)}</loc></url>\n' for locale in catalogs)
     (output / "sitemap.xml").write_text('<?xml version="1.0" encoding="UTF-8"?>\n<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n' + urls + '</urlset>\n')
 
     app = output / "app/index.html"

scripts/test_site.py

@@ -57,7 +57,8 @@ with tempfile.TemporaryDirectory() as directory:
         links = [attrs for tag, attrs in page.tags if tag == 'link']
         assert {link['href'] for link in links if link.get('rel') == 'icon'} == {'/favicon.ico', '/favicon.png', '/favicon.svg'}
         assert {link['hreflang'] for link in links if link.get('rel') == 'alternate'} == supported | {'x-default'}
-        assert next(link['href'] for link in links if link.get('rel') == 'canonical') == f'https://www.opencadstudio.com/{locale}/'
+        expected_path = '/' if locale == 'en-US' else f'/{locale}/'
+        assert next(link['href'] for link in links if link.get('rel') == 'canonical') == f'https://www.opencadstudio.com{expected_path}'
         choices = [attrs for tag, attrs in page.tags if tag == 'a' and 'hreflang' in attrs]
         assert {choice['hreflang'] for choice in choices} == supported
         assert [choice['hreflang'] for choice in choices if choice.get('aria-current') == 'page'] == [locale]
@@ -74,7 +75,7 @@ with tempfile.TemporaryDirectory() as directory:
                     path = output / urlsplit(target).path.lstrip('/')
                     assert path.exists(), (locale, target)
         manifest = json.loads((output / locale / 'site.webmanifest').read_text())
-        assert manifest['lang'] == locale and manifest['start_url'] == f'/{locale}/'
+        assert manifest['lang'] == locale and manifest['start_url'] == expected_path
         for icon in manifest['icons']:
             assert (output / icon['src'].lstrip('/')).exists()
         if locale != 'en-US':
@@ -87,7 +88,13 @@ with tempfile.TemporaryDirectory() as directory:
     assert icon == '/favicon.svg'
     assert (output / icon.lstrip('/')).read_bytes() == (ROOT / 'site/favicon.svg').read_bytes()
     sitemap = ET.parse(output / 'sitemap.xml')
-    assert len(sitemap.getroot()) == len(supported) + 1
+    locations = {url[0].text for url in sitemap.getroot()}
+    assert len(locations) == len(supported)
+    assert 'https://www.opencadstudio.com/' in locations
+    assert 'https://www.opencadstudio.com/en-US/' not in locations
     assert (output / 'index.html').read_bytes() == (output / 'en-US/index.html').read_bytes()
+    homepage_schemas = [json.loads(value) for value in re.findall(r'<script type="application/ld\+json">(.*?)</script>', (output / 'index.html').read_text())]
+    website = next(schema for schema in homepage_schemas if schema['@type'] == 'WebSite')
+    assert website['url'] == 'https://www.opencadstudio.com/' and 'OpenCADStudio' in website['alternateName']
     subprocess.run(['node', 'scripts/test_site.cjs', str(output / 'site.js')], cwd=ROOT, check=True)
 print(f'All {len(supported)} translations, metadata, local links, charts, manifests and icons passed')