support base_url path for model providers (#608)

* adding support for base_url

* updated docs

* fixed tests for config generator

* making fixes based on PR comments

---------

Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-288.local>

arch/tools/cli/config_generator.py

@@ -8,7 +8,14 @@ from urllib.parse import urlparse
 from copy import deepcopy
 
 
-SUPPORTED_PROVIDERS = [
+SUPPORTED_PROVIDERS_WITH_BASE_URL = [
+    "azure_openai",
+    "ollama",
+    "qwen",
+    "amazon_bedrock",
+]
+
+SUPPORTED_PROVIDERS_WITHOUT_BASE_URL = [
     "arch",
     "deepseek",
     "groq",
@@ -17,15 +24,15 @@ SUPPORTED_PROVIDERS = [
     "gemini",
     "anthropic",
     "together_ai",
-    "azure_openai",
     "xai",
-    "ollama",
     "moonshotai",
     "zhipu",
-    "qwen",
-    "amazon_bedrock",
 ]
 
+SUPPORTED_PROVIDERS = (
+    SUPPORTED_PROVIDERS_WITHOUT_BASE_URL + SUPPORTED_PROVIDERS_WITH_BASE_URL
+)
+
 
 def get_endpoint_and_port(endpoint, protocol):
     endpoint_tokens = endpoint.split(":")
@@ -189,12 +196,9 @@ def validate_and_render_schema():
             provider = model_name_tokens[0]
 
             # Validate azure_openai and ollama provider requires base_url
-            if (
-                provider == "azure_openai"
-                or provider == "ollama"
-                or provider == "qwen"
-                or provider == "amazon_bedrock"
-            ) and model_provider.get("base_url") is None:
+            if (provider in SUPPORTED_PROVIDERS_WITH_BASE_URL) and model_provider.get(
+                "base_url"
+            ) is None:
                 raise Exception(
                     f"Provider '{provider}' requires 'base_url' to be set for model {model_name}"
                 )
@@ -245,11 +249,11 @@ def validate_and_render_schema():
             if model_provider.get("base_url", None):
                 base_url = model_provider["base_url"]
                 urlparse_result = urlparse(base_url)
-                url_path = urlparse_result.path
-                if url_path and url_path != "/":
-                    raise Exception(
-                        f"Please provide base_url without path, got {base_url}. Use base_url like 'http://example.com' instead of 'http://example.com/path'."
-                    )
+                base_url_path_prefix = urlparse_result.path
+                if base_url_path_prefix and base_url_path_prefix != "/":
+                    # we will now support base_url_path_prefix. This means that the user can provide base_url like http://example.com/path and we will extract /path as base_url_path_prefix
+                    model_provider["base_url_path_prefix"] = base_url_path_prefix
+
                 if urlparse_result.scheme == "" or urlparse_result.scheme not in [
                     "http",
                     "https",

arch/tools/test/test_config_generator.py

@@ -243,14 +243,13 @@ listeners:
     timeout: 30s
 
 llm_providers:
-
   - model: custom/gpt-4o
 
 """,
     },
     {
-        "id": "base_url_no_prefix",
-        "expected_error": "Please provide base_url without path",
+        "id": "base_url_with_path_prefix",
+        "expected_error": None,
         "arch_config": """
 version: v0.1.0
 
@@ -264,7 +263,7 @@ listeners:
 llm_providers:
 
   - model: custom/gpt-4o
-    base_url: "http://custom.com/test"
+    base_url: "http://custom.com/api/v2"
     provider_interface: openai
 
 """,
@@ -322,8 +321,7 @@ def test_validate_and_render_schema_tests(monkeypatch, arch_config_test_case):
     monkeypatch.setenv("TEMPLATE_ROOT", "../")
 
     arch_config = arch_config_test_case["arch_config"]
-    expected_error = arch_config_test_case["expected_error"]
-    test_id = arch_config_test_case["id"]
+    expected_error = arch_config_test_case.get("expected_error")
 
     arch_config_schema = ""
     with open("../arch_config_schema.yaml", "r") as file:
@@ -346,9 +344,14 @@ def test_validate_and_render_schema_tests(monkeypatch, arch_config_test_case):
     ]
     with mock.patch("builtins.open", m_open):
         with mock.patch("config_generator.Environment"):
-            with pytest.raises(Exception) as excinfo:
+            if expected_error:
+                # Test expects an error
+                with pytest.raises(Exception) as excinfo:
+                    validate_and_render_schema()
+                assert expected_error in str(excinfo.value)
+            else:
+                # Test expects success - no exception should be raised
                 validate_and_render_schema()
-            assert expected_error in str(excinfo.value)
 
 
 def test_convert_legacy_llm_providers():