Skip to content

Azure OpenAI support #33

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
borisdev wants to merge 1 commit into from
Closed

Azure OpenAI support #33

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 13 additions & 2 deletions .env.example
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@ OPENAI_API_KEY="sk-..."
GEMINI_API_KEY="your-google-ai-studio-key"

# Optional: Provider Preference and Model Mapping
# Controls which provider (google or openai) is preferred for mapping haiku/sonnet.
# Controls which provider (google, openai, or azure) is preferred for mapping haiku/sonnet.
# Defaults to openai if not set.
PREFERRED_PROVIDER="openai"

Expand All @@ -18,4 +18,15 @@ PREFERRED_PROVIDER="openai"
# Example Google mapping:
# PREFERRED_PROVIDER="google"
# BIG_MODEL="gemini-2.5-pro-preview-03-25"
# SMALL_MODEL="gemini-2.0-flash"
# SMALL_MODEL="gemini-2.0-flash"

# If PREFERRED_PROVIDER=azure, these should match your Azure deployment names.
#
# Example Azure mapping:
# PREFERRED_PROVIDER="azure"
# BIG_MODEL="your-deployment-name"
# SMALL_MODEL="your-deployment-name"
# Azure-specific settings:
# AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com"
# AZURE_OPENAI_API_KEY="your-azure-openai-api-key"
# AZURE_API_VERSION="your-api-version"
12 changes: 12 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -141,6 +141,18 @@ BIG_MODEL="gpt-4o" # Example specific model
SMALL_MODEL="gpt-4o-mini" # Example specific model
```


**Example 4: Use Azure OpenAI**
```dotenv
PREFERRED_PROVIDER="azure"
BIG_MODEL="your-gpt4-deployment"
SMALL_MODEL="your-gpt4-mini-deployment"

# Azure OpenAI Configuration
AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com"
AZURE_OPENAI_API_KEY="your-azure-openai-api-key"
AZURE_API_VERSION="2024-02-15-preview"
```
## How It Works 🧩

This proxy works by:
Expand Down
59 changes: 49 additions & 10 deletions server.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,6 +82,18 @@ def format(self, record):
OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
GEMINI_API_KEY = os.environ.get("GEMINI_API_KEY")

# Azure OpenAI configuration
AZURE_OPENAI_ENDPOINT = os.environ.get("AZURE_OPENAI_ENDPOINT")
AZURE_OPENAI_API_KEY = os.environ.get("AZURE_OPENAI_API_KEY")
AZURE_API_VERSION = os.environ.get("AZURE_API_VERSION")
AZURE_DEPLOYMENT_NAME = os.environ.get("AZURE_DEPLOYMENT_NAME")

# List of Azure models (deployment names)
AZURE_MODELS = [
# These are typically deployment names, not model names
# Users configure their own deployment names in Azure OpenAI Studio
]

# Get preferred provider (default to openai)
PREFERRED_PROVIDER = os.environ.get("PREFERRED_PROVIDER", "openai").lower()

Expand Down Expand Up @@ -112,6 +124,12 @@ def format(self, record):
"gemini-2.0-flash"
]

# List of Azure models (deployment names)
AZURE_MODELS = [
# These are typically deployment names, not model names
# Users configure their own deployment names in Azure OpenAI Studio
]

# Helper function to clean schema for Gemini
def clean_gemini_schema(schema: Any) -> Any:
"""Recursively removes unsupported fields from a JSON schema for Gemini."""
Expand Down Expand Up @@ -202,12 +220,17 @@ def validate_model_field(cls, v, info): # Renamed to avoid conflict
clean_v = clean_v[7:]
elif clean_v.startswith('gemini/'):
clean_v = clean_v[7:]
elif clean_v.startswith('azure/'):
clean_v = clean_v[6:]

# --- Mapping Logic --- START ---
mapped = False
# Map Haiku to SMALL_MODEL based on provider preference
if 'haiku' in clean_v.lower():
if PREFERRED_PROVIDER == "google" and SMALL_MODEL in GEMINI_MODELS:
if PREFERRED_PROVIDER == "azure":
new_model = f"azure/{SMALL_MODEL}"
mapped = True
elif PREFERRED_PROVIDER == "google" and SMALL_MODEL in GEMINI_MODELS:
new_model = f"gemini/{SMALL_MODEL}"
mapped = True
else:
Expand All @@ -216,7 +239,10 @@ def validate_model_field(cls, v, info): # Renamed to avoid conflict

# Map Sonnet to BIG_MODEL based on provider preference
elif 'sonnet' in clean_v.lower():
if PREFERRED_PROVIDER == "google" and BIG_MODEL in GEMINI_MODELS:
if PREFERRED_PROVIDER == "azure":
new_model = f"azure/{BIG_MODEL}"
mapped = True
elif PREFERRED_PROVIDER == "google" and BIG_MODEL in GEMINI_MODELS:
new_model = f"gemini/{BIG_MODEL}"
mapped = True
else:
Expand All @@ -237,7 +263,7 @@ def validate_model_field(cls, v, info): # Renamed to avoid conflict
logger.debug(f"📌 MODEL MAPPING: '{original_model}' ➡️ '{new_model}'")
else:
# If no mapping occurred and no prefix exists, log warning or decide default
if not v.startswith(('openai/', 'gemini/', 'anthropic/')):
if not v.startswith(('openai/', 'gemini/', 'anthropic/', 'azure/')):
logger.warning(f"⚠️ No prefix or mapping rule for model: '{original_model}'. Using as is.")
new_model = v # Ensure we return the original if no rule applied

Expand Down Expand Up @@ -275,12 +301,17 @@ def validate_model_token_count(cls, v, info): # Renamed to avoid conflict
clean_v = clean_v[7:]
elif clean_v.startswith('gemini/'):
clean_v = clean_v[7:]
elif clean_v.startswith('azure/'):
clean_v = clean_v[6:]

# --- Mapping Logic --- START ---
mapped = False
# Map Haiku to SMALL_MODEL based on provider preference
if 'haiku' in clean_v.lower():
if PREFERRED_PROVIDER == "google" and SMALL_MODEL in GEMINI_MODELS:
if PREFERRED_PROVIDER == "azure":
new_model = f"azure/{SMALL_MODEL}"
mapped = True
elif PREFERRED_PROVIDER == "google" and SMALL_MODEL in GEMINI_MODELS:
new_model = f"gemini/{SMALL_MODEL}"
mapped = True
else:
Expand All @@ -289,7 +320,10 @@ def validate_model_token_count(cls, v, info): # Renamed to avoid conflict

# Map Sonnet to BIG_MODEL based on provider preference
elif 'sonnet' in clean_v.lower():
if PREFERRED_PROVIDER == "google" and BIG_MODEL in GEMINI_MODELS:
if PREFERRED_PROVIDER == "azure":
new_model = f"azure/{BIG_MODEL}"
mapped = True
elif PREFERRED_PROVIDER == "google" and BIG_MODEL in GEMINI_MODELS:
new_model = f"gemini/{BIG_MODEL}"
mapped = True
else:
Expand All @@ -309,7 +343,7 @@ def validate_model_token_count(cls, v, info): # Renamed to avoid conflict
if mapped:
logger.debug(f"📌 TOKEN COUNT MAPPING: '{original_model}' ➡️ '{new_model}'")
else:
if not v.startswith(('openai/', 'gemini/', 'anthropic/')):
if not v.startswith(('openai/', 'gemini/', 'anthropic/', 'azure/')):
logger.warning(f"⚠️ No prefix or mapping rule for token count model: '{original_model}'. Using as is.")
new_model = v # Ensure we return the original if no rule applied

Expand Down Expand Up @@ -533,9 +567,9 @@ def convert_anthropic_to_litellm(anthropic_request: MessagesRequest) -> Dict[str

# Cap max_tokens for OpenAI models to their limit of 16384
max_tokens = anthropic_request.max_tokens
if anthropic_request.model.startswith("openai/") or anthropic_request.model.startswith("gemini/"):
if anthropic_request.model.startswith("openai/") or anthropic_request.model.startswith("gemini/") or anthropic_request.model.startswith("azure/"):
max_tokens = min(max_tokens, 16384)
logger.debug(f"Capping max_tokens to 16384 for OpenAI/Gemini model (original value: {anthropic_request.max_tokens})")
logger.debug(f"Capping max_tokens to 16384 for OpenAI/Gemini/Azure model (original value: {anthropic_request.max_tokens})")

# Create LiteLLM request dict
litellm_request = {
Expand Down Expand Up @@ -1110,13 +1144,18 @@ async def create_message(
elif request.model.startswith("gemini/"):
litellm_request["api_key"] = GEMINI_API_KEY
logger.debug(f"Using Gemini API key for model: {request.model}")
elif request.model.startswith("azure/"):
litellm_request["api_key"] = AZURE_OPENAI_API_KEY
litellm_request["api_base"] = AZURE_OPENAI_ENDPOINT
litellm_request["api_version"] = AZURE_API_VERSION
logger.debug(f"Using Azure OpenAI API key for model: {request.model}")
else:
litellm_request["api_key"] = ANTHROPIC_API_KEY
logger.debug(f"Using Anthropic API key for model: {request.model}")

# For OpenAI models - modify request format to work with limitations
if "openai" in litellm_request["model"] and "messages" in litellm_request:
logger.debug(f"Processing OpenAI model request: {litellm_request['model']}")
if ("openai" in litellm_request["model"] or "azure" in litellm_request["model"]) and "messages" in litellm_request:
logger.debug(f"Processing OpenAI/Azure model request: {litellm_request['model']}")

# For OpenAI models, we need to convert content blocks to simple strings
# and handle other requirements
Expand Down
104 changes: 96 additions & 8 deletions tests.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -361,6 +361,83 @@ def test_request(test_name, request_data, check_tools=False):
traceback.print_exc()
return False

def test_azure_model_mapping():
"""Test Azure model mapping specifically."""
print(f"\n{'='*20} RUNNING AZURE MODEL MAPPING TEST {'='*20}")

# Test different Azure configurations
azure_tests = [
{
"name": "direct_azure",
"model": "azure/my-deployment",
"expected_prefix": "azure/",
"env_vars": {}
}
]

success_count = 0

for test_case in azure_tests:
print(f"\n--- Testing {test_case['name']} ---")

# Create test request
test_data = {
"model": test_case["model"],
"max_tokens": 100,
"messages": [{"role": "user", "content": "Test message"}]
}

try:
# We'll just test that the proxy accepts the request and processes it
# without necessarily getting a real response (since we may not have Azure configured)
print(f"Testing model: {test_case['model']}")

# Use dummy headers for Azure-only testing if no API key is set
test_headers = proxy_headers
if not ANTHROPIC_API_KEY:
test_headers = {
"x-api-key": "dummy-key-for-azure-testing",
"anthropic-version": ANTHROPIC_VERSION,
"content-type": "application/json",
}

proxy_response = httpx.post(PROXY_API_URL, headers=test_headers, json=test_data, timeout=10)

print(f"Status code: {proxy_response.status_code}")

# Accept both success and certain types of failures (like missing Azure config)
# The important thing is that the model mapping logic works
if proxy_response.status_code == 200:
print("✅ Request succeeded - Azure mapping worked")
success_count += 1
elif proxy_response.status_code in [400, 401, 403, 422, 500]:
response_text = proxy_response.text
print(f"❌ Azure test failed (HTTP {proxy_response.status_code})")
print(f"Response: {response_text}")

# Check if it's a config issue and provide helpful message
if "azure" in response_text.lower() or "deployment" in response_text.lower():
print("💡 This looks like an Azure configuration issue.")
elif proxy_response.status_code == 500:
print("💡 HTTP 500 - This could be missing Azure config or incorrect deployment name.")

print("💡 To fix Azure configuration, ensure you have these in your .env:")
print(" AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com/")
print(" AZURE_OPENAI_API_KEY=your-azure-api-key")
print(" AZURE_API_VERSION=your-api-version")
print(" BIG_MODEL=your-deployment-name")
print(" SMALL_MODEL=your-deployment-name")
# Don't increment success_count - this is a failure
else:
print(f"❌ Unexpected status code: {proxy_response.status_code}")
print(f"Response: {proxy_response.text}")

except Exception as e:
print(f"❌ Exception during test: {e}")

print(f"\nAzure mapping tests: {success_count}/{len(azure_tests)} passed")
return success_count == len(azure_tests)

# ================= STREAMING TESTS =================

class StreamStats:
Expand Down Expand Up @@ -636,8 +713,18 @@ async def run_tests(args):
# Track test results
results = {}

# First run non-streaming tests
# Run Azure-specific tests first (unless we're doing streaming-only)
if not args.streaming_only:
print("\n\n=========== RUNNING AZURE INTEGRATION TESTS ===========\n")
azure_result = test_azure_model_mapping()
results["azure_model_mapping"] = azure_result

# If Azure-only is specified, skip other tests
if args.azure_only:
print("\n\n=========== AZURE-ONLY MODE - SKIPPING OTHER TESTS ===========\n")

# First run non-streaming tests
elif not args.streaming_only:
print("\n\n=========== RUNNING NON-STREAMING TESTS ===========\n")
for test_name, test_data in TEST_SCENARIOS.items():
# Skip streaming tests
Expand All @@ -658,7 +745,7 @@ async def run_tests(args):
results[test_name] = result

# Now run streaming tests
if not args.no_streaming:
if not args.no_streaming and not args.azure_only:
print("\n\n=========== RUNNING STREAMING TESTS ===========\n")
for test_name, test_data in TEST_SCENARIOS.items():
# Only select streaming tests, or force streaming
Expand Down Expand Up @@ -695,19 +782,20 @@ async def run_tests(args):
return False

async def main():
# Check that API key is set
if not ANTHROPIC_API_KEY:
print("Error: ANTHROPIC_API_KEY not set in .env file")
return

# Parse command-line arguments
# Parse command-line arguments first
parser = argparse.ArgumentParser(description="Test the Claude-on-OpenAI proxy")
parser.add_argument("--no-streaming", action="store_true", help="Skip streaming tests")
parser.add_argument("--streaming-only", action="store_true", help="Only run streaming tests")
parser.add_argument("--simple", action="store_true", help="Only run simple tests (no tools)")
parser.add_argument("--tools-only", action="store_true", help="Only run tool tests")
parser.add_argument("--azure-only", action="store_true", help="Only run Azure integration tests")
args = parser.parse_args()

# Check that API key is set (unless we're only testing Azure)
if not ANTHROPIC_API_KEY and not args.azure_only:
print("Error: ANTHROPIC_API_KEY not set in .env file")
return

# Run tests
success = await run_tests(args)
sys.exit(0 if success else 1)
Expand Down