refactor: simplify base_url resolution with clear priority order

vlmrun/cli/cli.py

@@ -41,14 +41,28 @@ def version_callback(value: bool) -> None:
         raise typer.Exit()
 
 
-def check_credentials(
-    ctx: typer.Context, api_key: Optional[str], base_url: str
-) -> None:
+def resolve_base_url(base_url: Optional[str]) -> str:
+    """Resolve base_url with priority: arg > env > config > default."""
+    if base_url:
+        return base_url
+    if env_url := os.getenv("VLMRUN_BASE_URL"):
+        return env_url
+    if config_url := get_config().base_url:
+        return config_url
+    return DEFAULT_BASE_URL
+
+
+def resolve_api_key(api_key: Optional[str]) -> Optional[str]:
+    """Resolve api_key with priority: arg > env > config."""
+    if api_key:
+        return api_key
+    if env_key := os.getenv("VLMRUN_API_KEY"):
+        return env_key
+    return get_config().api_key
+
+
+def check_credentials(ctx: typer.Context, api_key: str, base_url: str) -> None:
     """Check if API key is present and show helpful message if missing."""
-    config = get_config()
-    api_key = api_key or config.api_key
-    base_url = base_url or config.base_url
-
     if not api_key:
         console.print("\n[red bold]Error:[/] API key not found! 🔑\n")
         console.print(
@@ -69,7 +83,7 @@ def check_credentials(
         )
         raise typer.Exit(1)
 
-    is_default_url = base_url == DEFAULT_BASE_URL and not os.getenv("VLMRUN_BASE_URL")
+    is_default_url = base_url == DEFAULT_BASE_URL
     if is_default_url and not ctx.meta.get("has_shown_base_url_notice"):
         console.print(
             f"[yellow]Note:[/] Using default API endpoint: [blue]{base_url}[/]\n"
@@ -89,14 +103,12 @@ def main(
     api_key: Optional[str] = typer.Option(
         None,
         "--api-key",
-        envvar="VLMRUN_API_KEY",
-        help="VLM Run API key. Can also be set via VLMRUN_API_KEY environment variable.",
+        help="VLM Run API key. Can also be set via VLMRUN_API_KEY env var or config.",
     ),
     base_url: Optional[str] = typer.Option(
         None,
         "--base-url",
-        envvar="VLMRUN_BASE_URL",
-        help="VLM Run API base URL. Can also be set via VLMRUN_BASE_URL environment variable.",
+        help="VLM Run API base URL. Can also be set via VLMRUN_BASE_URL env var or config.",
     ),
     version: Optional[bool] = typer.Option(
         None,
@@ -119,8 +131,11 @@ def main(
 
     # Skip credential check for config commands (needed to set API key)
     if ctx.invoked_subcommand is not None and ctx.invoked_subcommand != "config":
-        check_credentials(ctx, api_key, base_url)
-        ctx.obj = VLMRun(api_key=api_key, base_url=base_url)
+        # Resolve with priority: arg > env > config > default
+        resolved_api_key = resolve_api_key(api_key)
+        resolved_base_url = resolve_base_url(base_url)
+        check_credentials(ctx, resolved_api_key, resolved_base_url)
+        ctx.obj = VLMRun(api_key=resolved_api_key, base_url=resolved_base_url)
 
 
 # Add subcommands

vlmrun/client/base_requestor.py

@@ -1,7 +1,6 @@
 """VLM Run API requestor implementation."""
 
 from typing import Any, Dict, Tuple, TYPE_CHECKING, Union, Optional
-from urllib.parse import urljoin
 
 if TYPE_CHECKING:
     from vlmrun.types.abstract import VLMRunProtocol
@@ -49,12 +48,13 @@ def __init__(
 
         Args:
             client: VLMRun API instance
-            base_url: Base URL for API
+            base_url: Base URL for API (without trailing slash)
             timeout: Request timeout in seconds
             max_retries: Maximum number of retry attempts
         """
         self._client = client
-        self._base_url = base_url or client.base_url
+        # Normalize: strip trailing slashes for consistent URL building
+        self._base_url = (base_url or client.base_url).rstrip("/")
         self._timeout = timeout
         self._max_retries = (
             max_retries
@@ -131,8 +131,8 @@ def _request_with_retry():
             if "X-Client-Id" not in _headers:
                 _headers["X-Client-Id"] = f"python-sdk-{__version__}"
 
-            # Build full URL
-            full_url = urljoin(self._base_url.rstrip("/") + "/", url.lstrip("/"))
+            # Build full URL (base_url is pre-normalized without trailing slash)
+            full_url = f"{self._base_url}/{url.lstrip('/')}"
 
             try:
                 response = self._session.request(

vlmrun/client/client.py

@@ -32,15 +32,42 @@
 )
 
 
+def _resolve_base_url(base_url: Optional[str]) -> str:
+    """Resolve base_url with priority: arg > env > default.
+
+    Also normalizes the URL by removing trailing slashes.
+    """
+    url = base_url or os.getenv("VLMRUN_BASE_URL") or DEFAULT_BASE_URL
+    return url.rstrip("/")
+
+
+def _resolve_api_key(api_key: Optional[str]) -> str:
+    """Resolve api_key with priority: arg > env.
+
+    Raises ConfigurationError if no API key is found.
+    """
+    resolved = api_key or os.getenv("VLMRUN_API_KEY")
+    if not resolved:
+        raise ConfigurationError(
+            message="Missing API key",
+            error_type="missing_api_key",
+            suggestion="Please provide your VLM Run API key:\n\n"
+            "1. Set it in your code:\n"
+            "   client = VLMRun(api_key='your-api-key')\n\n"
+            "2. Or set the environment variable:\n"
+            "   export VLMRUN_API_KEY='your-api-key'\n\n"
+            "Get your API key at https://app.vlm.run/dashboard",
+        )
+    return resolved
+
+
 @dataclass
 class VLMRun:
     """VLM Run API client.
 
     Attributes:
-        api_key: API key for authentication. Can be provided through constructor
-            or VLMRUN_API_KEY environment variable.
-        base_url: Base URL for API. Defaults to None, which falls back to
-            VLMRUN_BASE_URL environment variable or https://api.vlm.run/v1.
+        api_key: API key for authentication. Priority: arg > VLMRUN_API_KEY env var.
+        base_url: Base URL for API. Priority: arg > VLMRUN_BASE_URL env var > default.
         timeout: Request timeout in seconds. Defaults to 120.0.
         max_retries: Maximum number of retry attempts for failed requests. Defaults to 5.
         files: Files resource for managing files
@@ -54,30 +81,9 @@ class VLMRun:
     max_retries: int = 5
 
     def __post_init__(self):
-        """Initialize the client after dataclass initialization.
-
-        This method handles environment variable fallbacks:
-        - api_key: Falls back to VLMRUN_API_KEY environment variable
-        - base_url: Can be overridden by constructor or VLMRUN_BASE_URL environment variable
-        """
-        # Handle API key first
-        if not self.api_key:  # Handle both None and empty string
-            self.api_key = os.getenv("VLMRUN_API_KEY", None)
-            if not self.api_key:  # Still None or empty after env check
-                raise ConfigurationError(
-                    message="Missing API key",
-                    error_type="missing_api_key",
-                    suggestion="Please provide your VLM Run API key:\n\n"
-                    "1. Set it in your code:\n"
-                    "   client = VLMRun(api_key='your-api-key')\n\n"
-                    "2. Or set the environment variable:\n"
-                    "   export VLMRUN_API_KEY='your-api-key'\n\n"
-                    "Get your API key at https://app.vlm.run/dashboard",
-                )
-
-        # Handle base URL
-        if self.base_url is None:
-            self.base_url = os.getenv("VLMRUN_BASE_URL", DEFAULT_BASE_URL)
+        """Initialize the client after dataclass initialization."""
+        self.api_key = _resolve_api_key(self.api_key)
+        self.base_url = _resolve_base_url(self.base_url)
 
         # Initialize requestor for API key validation
         requestor = APIRequestor(

vlmrun/version.py

@@ -1 +1 @@
-__version__ = "0.5.3"
+__version__ = "0.5.4"