Check that documentation urls are valid (#31188)

* Check that documentation urls are valid

* Validate documentation url in pieces

homeassistant/components/solarlog/manifest.json

@@ -2,7 +2,7 @@
   "domain": "solarlog",
   "name": "Solar-Log",
   "config_flow": true,
-  "documentation": "https://www.home-assistant.io/integration/solarlog",
+  "documentation": "https://www.home-assistant.io/integrations/solarlog",
   "dependencies": [],
   "codeowners": ["@Ernst79"],
   "requirements": ["sunwatcher==0.2.1"]

script/hassfest/manifest.py

@@ -1,11 +1,17 @@
 """Manifest validation."""
 from typing import Dict
+from urllib.parse import urlparse
 
 import voluptuous as vol
 from voluptuous.humanize import humanize_error
 
 from .model import Integration
 
+DOCUMENTATION_URL_SCHEMA = "https"
+DOCUMENTATION_URL_HOST = "www.home-assistant.io"
+DOCUMENTATION_URL_PATH_PREFIX = "/integrations/"
+DOCUMENTATION_URL_EXCEPTIONS = ["https://www.home-assistant.io/hassio"]
+
 SUPPORTED_QUALITY_SCALES = [
     "gold",
     "internal",
@@ -13,6 +19,25 @@ SUPPORTED_QUALITY_SCALES = [
     "silver",
 ]
 
+
+def documentation_url(value: str) -> str:
+    """Validate that a documentation url has the correct path and domain."""
+    if value in DOCUMENTATION_URL_EXCEPTIONS:
+        return value
+
+    parsed_url = urlparse(value)
+    if not parsed_url.scheme == DOCUMENTATION_URL_SCHEMA:
+        raise vol.Invalid("Documentation url is not prefixed with https")
+    if not parsed_url.netloc == DOCUMENTATION_URL_HOST:
+        raise vol.Invalid("Documentation url not hosted at www.home-assistant.io")
+    if not parsed_url.path.startswith(DOCUMENTATION_URL_PATH_PREFIX):
+        raise vol.Invalid(
+            "Documentation url does not begin with www.home-assistant.io/integrations"
+        )
+
+    return value
+
+
 MANIFEST_SCHEMA = vol.Schema(
     {
         vol.Required("domain"): str,
@@ -23,9 +48,9 @@ MANIFEST_SCHEMA = vol.Schema(
             vol.All([vol.All(vol.Schema({}, extra=vol.ALLOW_EXTRA), vol.Length(min=1))])
         ),
         vol.Optional("homekit"): vol.Schema({vol.Optional("models"): [str]}),
-        vol.Required(
-            "documentation"
-        ): vol.Url(),  # pylint: disable=no-value-for-parameter
+        vol.Required("documentation"): vol.All(
+            vol.Url(), documentation_url  # pylint: disable=no-value-for-parameter
+        ),
         vol.Optional("quality_scale"): vol.In(SUPPORTED_QUALITY_SCALES),
         vol.Required("requirements"): [str],
         vol.Required("dependencies"): [str],