Add Azure OpenAI support (#49)

Add azure-openai as a new provider option, using the AzureOpenAI client
from the openai package. Users can configure via --provider azure-openai
with --azure-deployment and --api-version options.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: bdqnghi

codewiki/cli/commands/config.py

@@ -85,14 +85,24 @@ def config_group():
 )
 @click.option(
     "--provider",
-    type=click.Choice(['openai-compatible', 'anthropic', 'bedrock'], case_sensitive=False),
+    type=click.Choice(['openai-compatible', 'anthropic', 'bedrock', 'azure-openai'], case_sensitive=False),
     help="LLM provider type (default: openai-compatible)"
 )
 @click.option(
     "--aws-region",
     type=str,
     help="AWS region for Bedrock provider (default: us-east-1)"
 )
+@click.option(
+    "--api-version",
+    type=str,
+    help="Azure OpenAI API version (default: 2024-12-01-preview)"
+)
+@click.option(
+    "--azure-deployment",
+    type=str,
+    help="Azure OpenAI deployment name"
+)
 def config_set(
     api_key: Optional[str],
     base_url: Optional[str],
@@ -104,7 +114,9 @@ def config_set(
     max_token_per_leaf_module: Optional[int],
     max_depth: Optional[int],
     provider: Optional[str] = None,
-    aws_region: Optional[str] = None
+    aws_region: Optional[str] = None,
+    api_version: Optional[str] = None,
+    azure_deployment: Optional[str] = None
 ):
     """
     Set configuration values for CodeWiki.
@@ -139,7 +151,7 @@ def config_set(
     """
     try:
         # Check if at least one option is provided
-        if not any([api_key, base_url, main_model, cluster_model, fallback_model, max_tokens, max_token_per_module, max_token_per_leaf_module, max_depth, provider, aws_region]):
+        if not any([api_key, base_url, main_model, cluster_model, fallback_model, max_tokens, max_token_per_module, max_token_per_leaf_module, max_depth, provider, aws_region, api_version, azure_deployment]):
             click.echo("No options provided. Use --help for usage information.")
             sys.exit(EXIT_CONFIG_ERROR)
 
@@ -187,6 +199,12 @@ def config_set(
         if aws_region is not None:
             validated_data['aws_region'] = aws_region
 
+        if api_version is not None:
+            validated_data['api_version'] = api_version
+
+        if azure_deployment is not None:
+            validated_data['azure_deployment'] = azure_deployment
+
         # Create config manager and save
         manager = ConfigManager()
         manager.load()  # Load existing config if present
@@ -202,7 +220,9 @@ def config_set(
             max_token_per_leaf_module=validated_data.get('max_token_per_leaf_module'),
             max_depth=validated_data.get('max_depth'),
             provider=validated_data.get('provider'),
-            aws_region=validated_data.get('aws_region')
+            aws_region=validated_data.get('aws_region'),
+            api_version=validated_data.get('api_version'),
+            azure_deployment=validated_data.get('azure_deployment')
         )
 
         # Display success messages
@@ -256,7 +276,13 @@ def config_set(
 
         if aws_region:
             click.secho(f"✓ AWS Region: {aws_region}", fg="green")
-        
+
+        if api_version:
+            click.secho(f"✓ API Version: {api_version}", fg="green")
+
+        if azure_deployment:
+            click.secho(f"✓ Azure Deployment: {azure_deployment}", fg="green")
+
         click.echo("\n" + click.style("Configuration updated successfully.", fg="green", bold=True))
 
     except ConfigurationError as e:
@@ -342,6 +368,12 @@ def config_show(output_json: bool):
                 click.echo(f"  Main Model:       {config.main_model or 'Not set'}")
                 click.echo(f"  Cluster Model:    {config.cluster_model or 'Not set'}")
                 click.echo(f"  Fallback Model:   {config.fallback_model or 'Not set'}")
+                click.echo(f"  Provider:         {config.provider}")
+                if config.provider == "bedrock":
+                    click.echo(f"  AWS Region:       {config.aws_region}")
+                elif config.provider == "azure-openai":
+                    click.echo(f"  API Version:      {config.api_version}")
+                    click.echo(f"  Azure Deployment: {config.azure_deployment or 'Not set'}")
             else:
                 click.secho("  Not configured", fg="yellow")
 
@@ -523,7 +555,17 @@ def config_validate(quick: bool, verbose: bool):
 
             try:
                 base_url_lower = (config.base_url or "").lower()
-                if "api.anthropic.com" in base_url_lower:
+                provider = getattr(config, 'provider', 'openai-compatible')
+                if provider == "azure-openai" or ".openai.azure.com" in base_url_lower:
+                    # Use Azure OpenAI SDK
+                    from openai import AzureOpenAI
+                    client = AzureOpenAI(
+                        api_key=api_key,
+                        api_version=config.api_version,
+                        azure_endpoint=config.base_url,
+                    )
+                    client.models.list()
+                elif "api.anthropic.com" in base_url_lower:
                     # Use Anthropic SDK for native Anthropic endpoints
                     import anthropic
                     client = anthropic.Anthropic(api_key=api_key)

codewiki/cli/config_manager.py

@@ -133,7 +133,9 @@ def save(
         max_token_per_leaf_module: Optional[int] = None,
         max_depth: Optional[int] = None,
         provider: Optional[str] = None,
-        aws_region: Optional[str] = None
+        aws_region: Optional[str] = None,
+        api_version: Optional[str] = None,
+        azure_deployment: Optional[str] = None
     ):
         """
         Save configuration to file and keyring.
@@ -149,8 +151,10 @@ def save(
             max_token_per_module: Maximum tokens per module for clustering
             max_token_per_leaf_module: Maximum tokens per leaf module
             max_depth: Maximum depth for hierarchical decomposition
-            provider: LLM provider type (openai-compatible, anthropic, bedrock)
+            provider: LLM provider type (openai-compatible, anthropic, bedrock, azure-openai)
             aws_region: AWS region for Bedrock provider
+            api_version: Azure OpenAI API version
+            azure_deployment: Azure OpenAI deployment name
         """
         # Ensure config directory exists
         try:
@@ -196,6 +200,10 @@ def save(
             self._config.provider = provider
         if aws_region is not None:
             self._config.aws_region = aws_region
+        if api_version is not None:
+            self._config.api_version = api_version
+        if azure_deployment is not None:
+            self._config.azure_deployment = azure_deployment
 
         # Validate configuration (only if base fields are set)
         if self._config.base_url and self._config.main_model and self._config.cluster_model:

codewiki/cli/models/config.py

@@ -113,8 +113,10 @@ class Configuration:
         cluster_model: Model for module clustering
         fallback_model: Fallback model for documentation generation
         default_output: Default output directory
-        provider: LLM provider type (openai-compatible, anthropic, bedrock)
+        provider: LLM provider type (openai-compatible, anthropic, bedrock, azure-openai)
         aws_region: AWS region for Bedrock provider
+        api_version: Azure OpenAI API version
+        azure_deployment: Azure OpenAI deployment name
         max_tokens: Maximum tokens for LLM response (default: 32768)
         max_token_per_module: Maximum tokens per module for clustering (default: 36369)
         max_token_per_leaf_module: Maximum tokens per leaf module (default: 16000)
@@ -128,6 +130,8 @@ class Configuration:
     default_output: str = "docs"
     provider: str = "openai-compatible"
     aws_region: str = "us-east-1"
+    api_version: str = "2024-12-01-preview"
+    azure_deployment: str = ""
     max_tokens: int = 32768
     max_token_per_module: int = 36369
     max_token_per_leaf_module: int = 16000
@@ -155,6 +159,8 @@ def to_dict(self) -> dict:
             'default_output': self.default_output,
             'provider': self.provider,
             'aws_region': self.aws_region,
+            'api_version': self.api_version,
+            'azure_deployment': self.azure_deployment,
             'max_tokens': self.max_tokens,
             'max_token_per_module': self.max_token_per_module,
             'max_token_per_leaf_module': self.max_token_per_leaf_module,
@@ -187,6 +193,8 @@ def from_dict(cls, data: dict) -> 'Configuration':
             default_output=data.get('default_output', 'docs'),
             provider=data.get('provider', 'openai-compatible'),
             aws_region=data.get('aws_region', 'us-east-1'),
+            api_version=data.get('api_version', '2024-12-01-preview'),
+            azure_deployment=data.get('azure_deployment', ''),
             max_tokens=data.get('max_tokens', 32768),
             max_token_per_module=data.get('max_token_per_module', 36369),
             max_token_per_leaf_module=data.get('max_token_per_leaf_module', 16000),
@@ -243,6 +251,8 @@ def to_backend_config(self, repo_path: str, output_dir: str, api_key: str, runti
             fallback_model=self.fallback_model,
             provider=self.provider,
             aws_region=self.aws_region,
+            api_version=self.api_version,
+            azure_deployment=self.azure_deployment,
             max_tokens=self.max_tokens,
             max_token_per_module=self.max_token_per_module,
             max_token_per_leaf_module=self.max_token_per_leaf_module,

codewiki/src/be/llm_services.py

@@ -4,7 +4,7 @@
 Includes a compatibility layer for OpenAI-compatible API proxies that may
 return slightly non-standard responses (e.g. choices[].index = None).
 
-Supports multiple providers: openai-compatible, anthropic, bedrock.
+Supports multiple providers: openai-compatible, anthropic, bedrock, azure-openai.
 """
 import logging
 from openai.types import chat
@@ -174,6 +174,9 @@ def call_llm(
     if provider in ("bedrock", "anthropic"):
         return _call_llm_via_litellm(prompt, config, model, temperature)
 
+    if provider == "azure-openai":
+        return _call_llm_via_azure(prompt, config, model, temperature)
+
     # Default: OpenAI-compatible
     client = create_openai_client(config)
 
@@ -225,3 +228,35 @@ def _call_llm_via_litellm(
         api_key=config.llm_api_key if config.provider != "bedrock" else None,
     )
     return response.choices[0].message.content
+
+
+def _call_llm_via_azure(
+    prompt: str,
+    config: Config,
+    model: str,
+    temperature: float = 0.0
+) -> str:
+    """
+    Call LLM via Azure OpenAI.
+
+    Uses the AzureOpenAI client from the openai package with
+    azure_endpoint, api_version, and deployment name.
+    """
+    from openai import AzureOpenAI
+
+    client = AzureOpenAI(
+        api_key=config.llm_api_key,
+        api_version=config.api_version,
+        azure_endpoint=config.llm_base_url,
+    )
+
+    deployment = config.azure_deployment or model
+    logger.debug("Calling Azure OpenAI deployment %s (api_version=%s)", deployment, config.api_version)
+
+    response = client.chat.completions.create(
+        model=deployment,
+        messages=[{"role": "user", "content": prompt}],
+        temperature=temperature,
+        max_tokens=config.max_tokens,
+    )
+    return response.choices[0].message.content

codewiki/src/config.py

@@ -58,8 +58,10 @@ class Config:
     cluster_model: str
     fallback_model: str = FALLBACK_MODEL_1
     # Provider configuration
-    provider: str = "openai-compatible"  # openai-compatible, anthropic, bedrock
+    provider: str = "openai-compatible"  # openai-compatible, anthropic, bedrock, azure-openai
     aws_region: str = "us-east-1"
+    api_version: str = "2024-12-01-preview"  # Azure OpenAI API version
+    azure_deployment: str = ""  # Azure OpenAI deployment name
     # Max token settings
     max_tokens: int = DEFAULT_MAX_TOKENS
     max_token_per_module: int = DEFAULT_MAX_TOKEN_PER_MODULE
@@ -160,6 +162,8 @@ def from_cli(
         fallback_model: str = FALLBACK_MODEL_1,
         provider: str = "openai-compatible",
         aws_region: str = "us-east-1",
+        api_version: str = "2024-12-01-preview",
+        azure_deployment: str = "",
         max_tokens: int = DEFAULT_MAX_TOKENS,
         max_token_per_module: int = DEFAULT_MAX_TOKEN_PER_MODULE,
         max_token_per_leaf_module: int = DEFAULT_MAX_TOKEN_PER_LEAF_MODULE,
@@ -177,8 +181,10 @@ def from_cli(
             main_model: Primary model
             cluster_model: Clustering model
             fallback_model: Fallback model
-            provider: LLM provider type (openai-compatible, anthropic, bedrock)
+            provider: LLM provider type (openai-compatible, anthropic, bedrock, azure-openai)
             aws_region: AWS region for Bedrock provider
+            api_version: Azure OpenAI API version
+            azure_deployment: Azure OpenAI deployment name
             max_tokens: Maximum tokens for LLM response
             max_token_per_module: Maximum tokens per module for clustering
             max_token_per_leaf_module: Maximum tokens per leaf module
@@ -204,6 +210,8 @@ def from_cli(
             fallback_model=fallback_model,
             provider=provider,
             aws_region=aws_region,
+            api_version=api_version,
+            azure_deployment=azure_deployment,
             max_tokens=max_tokens,
             max_token_per_module=max_token_per_module,
             max_token_per_leaf_module=max_token_per_leaf_module,