Centralize base URL normalization and validation

Co-authored-by: Shri Sukhani <shrisukhani@users.noreply.github.com>

Author: cursoragent

hyperbrowser/client/base.py

@@ -87,24 +87,8 @@ def _build_url(self, path: str) -> str:
         normalized_query_suffix = (
             f"?{normalized_parts.query}" if normalized_parts.query else ""
         )
-        if not isinstance(self.config.base_url, str):
-            raise HyperbrowserError("base_url must be a string")
-        normalized_base_url = self.config.base_url.strip().rstrip("/")
-        if not normalized_base_url:
-            raise HyperbrowserError("base_url must not be empty")
-
+        normalized_base_url = ClientConfig.normalize_base_url(self.config.base_url)
         parsed_base_url = urlparse(normalized_base_url)
-        if (
-            parsed_base_url.scheme not in {"https", "http"}
-            or not parsed_base_url.netloc
-        ):
-            raise HyperbrowserError(
-                "base_url must start with 'https://' or 'http://' and include a host"
-            )
-        if parsed_base_url.query or parsed_base_url.fragment:
-            raise HyperbrowserError(
-                "base_url must not include query parameters or fragments"
-            )
         base_has_api_suffix = parsed_base_url.path.rstrip("/").endswith("/api")
 
         if normalized_path_only == "/api" or normalized_path_only.startswith("/api/"):

hyperbrowser/config.py

@@ -18,15 +18,24 @@ class ClientConfig:
     def __post_init__(self) -> None:
         if not isinstance(self.api_key, str):
             raise HyperbrowserError("api_key must be a string")
-        if not isinstance(self.base_url, str):
-            raise HyperbrowserError("base_url must be a string")
         self.api_key = self.api_key.strip()
         if not self.api_key:
             raise HyperbrowserError("api_key must not be empty")
-        self.base_url = self.base_url.strip().rstrip("/")
-        if not self.base_url:
+        self.base_url = self.normalize_base_url(self.base_url)
+        self.headers = normalize_headers(
+            self.headers,
+            mapping_error_message="headers must be a mapping of string pairs",
+        )
+
+    @staticmethod
+    def normalize_base_url(base_url: str) -> str:
+        if not isinstance(base_url, str):
+            raise HyperbrowserError("base_url must be a string")
+        normalized_base_url = base_url.strip().rstrip("/")
+        if not normalized_base_url:
             raise HyperbrowserError("base_url must not be empty")
-        parsed_base_url = urlparse(self.base_url)
+
+        parsed_base_url = urlparse(normalized_base_url)
         if (
             parsed_base_url.scheme not in {"https", "http"}
             or not parsed_base_url.netloc
@@ -38,10 +47,7 @@ def __post_init__(self) -> None:
             raise HyperbrowserError(
                 "base_url must not include query parameters or fragments"
             )
-        self.headers = normalize_headers(
-            self.headers,
-            mapping_error_message="headers must be a mapping of string pairs",
-        )
+        return normalized_base_url
 
     @classmethod
     def from_env(cls) -> "ClientConfig":
@@ -67,4 +73,4 @@ def resolve_base_url_from_env(raw_base_url: Optional[str]) -> str:
             return "https://api.hyperbrowser.ai"
         if not raw_base_url.strip():
             raise HyperbrowserError("HYPERBROWSER_BASE_URL must not be empty when set")
-        return raw_base_url
+        return ClientConfig.normalize_base_url(raw_base_url)

tests/test_config.py

@@ -288,3 +288,24 @@ def test_client_config_resolve_base_url_from_env_defaults_and_rejects_blank():
         HyperbrowserError, match="HYPERBROWSER_BASE_URL must not be empty"
     ):
         ClientConfig.resolve_base_url_from_env("   ")
+    with pytest.raises(HyperbrowserError, match="include a host"):
+        ClientConfig.resolve_base_url_from_env("https://")
+
+
+def test_client_config_normalize_base_url_validates_and_normalizes():
+    assert (
+        ClientConfig.normalize_base_url(" https://example.local/custom/api/ ")
+        == "https://example.local/custom/api"
+    )
+
+    with pytest.raises(HyperbrowserError, match="base_url must be a string"):
+        ClientConfig.normalize_base_url(None)  # type: ignore[arg-type]
+
+    with pytest.raises(HyperbrowserError, match="base_url must not be empty"):
+        ClientConfig.normalize_base_url("   ")
+
+    with pytest.raises(HyperbrowserError, match="base_url must start with"):
+        ClientConfig.normalize_base_url("example.local")
+
+    with pytest.raises(HyperbrowserError, match="must not include query parameters"):
+        ClientConfig.normalize_base_url("https://example.local?foo=bar")