gac 3.6.0__py3-none-any.whl → 3.10.10__py3-none-any.whl

gac/providers/README.md

@@ -0,0 +1,437 @@
+# AI Provider Architecture
+
+This directory contains AI provider implementations for GAC (Git Auto Commit). All providers follow a class-based architecture with shared base classes to reduce code duplication and improve maintainability.
+
+## Architecture Overview
+
+```text
+ProviderProtocol (Protocol - type contract)
+     ↓
+BaseConfiguredProvider (ABC - core logic)
+     ↓
+├── OpenAICompatibleProvider (OpenAI-style APIs)
+├── AnthropicCompatibleProvider (Anthropic-style APIs)
+└── GenericHTTPProvider (Fully custom implementations)
+     ↓
+Concrete Providers (e.g., OpenAIProvider, GeminiProvider)
+```
+
+## Core Components
+
+### BaseConfiguredProvider
+
+Abstract base class implementing the template method pattern. All providers inherit from this class.
+
+**Key Features:**
+
+- Standardized HTTP handling with httpx
+- Common error handling patterns
+- Flexible configuration via ProviderConfig
+- Template methods for customization:
+  - `_get_api_key()` - Load API key from environment
+  - `_build_headers()` - Build request headers
+  - `_build_request_body()` - Build request body
+  - `_get_api_url()` - Get API endpoint URL
+  - `_parse_response()` - Parse API response
+  - `_make_http_request()` - Execute HTTP request
+  - `generate()` - Main entry point
+
+### OpenAICompatibleProvider
+
+Specialized base class for OpenAI-compatible APIs (standard format).
+
+**Request Format:**
+
+```json
+{
+  "model": "gpt-5",
+  "messages": [...],
+  "temperature": 0.7,
+  "max_tokens": 1024
+}
+```
+
+**Response Format:**
+
+```json
+{
+  "choices": [
+    {
+      "message": {
+        "content": "..."
+      }
+    }
+  ]
+}
+```
+
+**Providers Using This Base:**
+
+- OpenAI, DeepSeek, Together, Fireworks, Cerebras, Mistral, Minimax, Moonshot, Groq, OpenRouter
+- Custom OpenAI, Azure OpenAI, LM Studio
+- Kimi Coding, Streamlake, Synthetic, Z.AI
+
+### AnthropicCompatibleProvider
+
+Specialized base class for Anthropic-style APIs.
+
+**Request Format:**
+
+```json
+{
+  "model": "claude-sonnet-4-5",
+  "messages": [...],
+  "system": "You are a helpful assistant",
+  "temperature": 0.7,
+  "max_tokens": 1024
+}
+```
+
+**Response Format:**
+
+```json
+{
+  "content": [
+    {
+      "type": "text",
+      "text": "..."
+    }
+  ]
+}
+```
+
+**Providers Using This Base:**
+
+- Anthropic, Custom Anthropic
+- Claude Code
+
+### GenericHTTPProvider
+
+Base class for providers with completely custom API formats.
+
+**Providers Using This Base:**
+
+- Gemini (Google's unique format)
+- Replicate (async prediction polling)
+
+## Creating a New Provider
+
+### Step 1: Choose the Right Base Class
+
+```python
+from gac.providers.base import OpenAICompatibleProvider, ProviderConfig
+from gac.providers.error_handler import handle_provider_errors
+
+# Most providers fit one of these patterns:
+# 1. OpenAI-compatible format → inherit from OpenAICompatibleProvider
+# 2. Anthropic-compatible format → inherit from AnthropicCompatibleProvider
+# 3. Custom format → inherit from GenericHTTPProvider
+```
+
+### Step 2: Define Provider Configuration
+
+```python
+class MyProvider(OpenAICompatibleProvider):
+    config = ProviderConfig(
+        name="My Provider",
+        api_key_env="MY_PROVIDER_API_KEY",
+        base_url="https://api.myprovider.com/v1/chat/completions",
+    )
+```
+
+### Step 3: Override Template Methods (If Needed)
+
+```python
+class MyProvider(OpenAICompatibleProvider):
+    # Override only what's needed
+
+    def _build_headers(self) -> dict[str, str]:
+        """Custom header handling."""
+        headers = super()._build_headers()
+        # Add provider-specific headers
+        return headers
+
+    def _get_api_url(self, model: str | None = None) -> str:
+        """Custom URL construction."""
+        if model is None:
+            return super()._get_api_url(model)
+        return f"https://custom.endpoint/{model}/chat"
+```
+
+### Step 4: Create Lazy Getter and Decorated Function
+
+```python
+def _get_my_provider() -> MyProvider:
+    """Lazy getter to initialize provider at call time."""
+    return MyProvider(MyProvider.config)
+
+@handle_provider_errors("My Provider")
+def call_my_provider_api(model: str, messages: list[dict], temperature: float, max_tokens: int) -> str:
+    """Call My Provider API."""
+    provider = _get_my_provider()
+    return provider.generate(model=model, messages=messages, temperature=temperature, max_tokens=max_tokens)
+```
+
+### Step 5: Export in `__init__.py`
+
+```python
+# In src/gac/providers/__init__.py
+from .my_provider import call_my_provider_api
+
+# Add to PROVIDER_REGISTRY
+PROVIDER_REGISTRY = {
+    ...
+    "my-provider": call_my_provider_api,
+    ...
+}
+
+# Add to __all__
+__all__ = [
+    ...
+    "call_my_provider_api",
+    ...
+]
+```
+
+## Common Customization Patterns
+
+### Pattern 1: Optional API Key (e.g., Ollama, LM Studio)
+
+```python
+def _get_api_key(self) -> str:
+    """Get optional API key."""
+    api_key = os.getenv(self.config.api_key_env)
+    if not api_key:
+        return ""  # Optional
+    return api_key
+```
+
+### Pattern 2: Custom URL Construction (e.g., Azure OpenAI)
+
+```python
+def _get_api_url(self, model: str | None = None) -> str:
+    """Build custom URL with model in path."""
+    if model is None:
+        return super()._get_api_url(model)
+    return f"{self.endpoint}/openai/deployments/{model}/chat/completions?api-version={self.api_version}"
+```
+
+### Pattern 3: Alternative Environment Variables (e.g., Streamlake)
+
+```python
+def _get_api_key(self) -> str:
+    """Try primary key, then fallback."""
+    api_key = os.getenv(self.config.api_key_env)
+    if api_key:
+        return api_key
+    # Fallback to alternative
+    api_key = os.getenv("ALTERNATIVE_KEY_ENV")
+    if api_key:
+        return api_key
+    raise AIError.authentication_error("No API key found")
+```
+
+### Pattern 4: Model Preprocessing (e.g., Synthetic - adding prefixes)
+
+```python
+def _build_request_body(self, messages: list[dict], temperature: float, max_tokens: int, model: str, **kwargs) -> dict[str, Any]:
+    """Preprocess model name."""
+    data = super()._build_request_body(messages, temperature, max_tokens, model, **kwargs)
+    # Add "hf:" prefix for HuggingFace models
+    data["model"] = f"hf:{model}"
+    return data
+```
+
+### Pattern 5: Custom Response Parsing (e.g., LM Studio with text field fallback)
+
+```python
+def _parse_response(self, response: dict[str, Any]) -> str:
+    """Parse response with fallback."""
+    # Try standard OpenAI format first
+    choices = response.get("choices")
+    if choices:
+        content = choices[0].get("message", {}).get("content")
+        if content:
+            return content
+
+    # Fallback to text field
+    content = choices[0].get("text")
+    if content:
+        return content
+
+    raise AIError.model_error("No content found")
+```
+
+### Pattern 6: System Message Handling (e.g., Claude Code)
+
+```python
+def _build_request_body(self, messages: list[dict], temperature: float, max_tokens: int, model: str, **kwargs) -> dict[str, Any]:
+    """Extract and handle system messages."""
+    anthropic_messages = []
+    system_instructions = ""
+
+    for msg in messages:
+        if msg["role"] == "system":
+            system_instructions = msg["content"]
+        else:
+            anthropic_messages.append(msg)
+
+    # Move system instructions to first user message
+    if system_instructions and anthropic_messages:
+        anthropic_messages[0]["content"] = f"{system_instructions}\n\n{anthropic_messages[0]['content']}"
+
+    return {
+        "messages": anthropic_messages,
+        "system": "REQUIRED_FIXED_MESSAGE",
+        "temperature": temperature,
+        "max_tokens": max_tokens,
+        **kwargs,
+    }
+```
+
+### Pattern 7: Async Operations (e.g., Replicate with polling)
+
+```python
+def generate(self, model: str, messages: list[dict], temperature: float = 0.7, max_tokens: int = 1024, **kwargs) -> str:
+    """Override for async/polling operations."""
+    # Create prediction
+    response = httpx.post(url, json=body, headers=headers, ...)
+    prediction_id = response.json()["id"]
+
+    # Poll for completion
+    while elapsed_time < max_wait_time:
+        status_response = httpx.get(f"{url}/{prediction_id}", headers=headers, ...)
+        status_data = status_response.json()
+
+        if status_data["status"] == "succeeded":
+            return status_data["output"]
+        elif status_data["status"] == "failed":
+            raise AIError.model_error("Prediction failed")
+
+        time.sleep(wait_interval)
+        elapsed_time += wait_interval
+
+    raise AIError.timeout_error("Prediction timed out")
+```
+
+## Error Handling
+
+All providers use the `@handle_provider_errors` decorator to normalize error handling:
+
+```python
+from gac.providers.error_handler import handle_provider_errors
+
+@handle_provider_errors("My Provider")
+def call_my_provider_api(...) -> str:
+    # Errors are automatically caught and converted to AIError types
+    pass
+```
+
+**Error Mapping:**
+
+- HTTP 401 → `AIError.authentication_error()`
+- HTTP 429 → `AIError.rate_limit_error()`
+- Timeout → `AIError.timeout_error()`
+- Connection error → `AIError.connection_error()`
+- Other → `AIError.model_error()`
+
+## Testing Providers
+
+Each provider has comprehensive tests in `tests/providers/test_<provider>.py`.
+
+### Test Structure
+
+```python
+class TestProviderImports:
+    """Test imports."""
+    def test_import_provider(self): ...
+
+class TestProviderMocked(BaseProviderTest):
+    """Standard mocked tests (inherited from BaseProviderTest)."""
+    @property
+    def provider_name(self) -> str: return "my-provider"
+
+    @property
+    def provider_module(self) -> str: return "gac.providers.my_provider"
+
+    @property
+    def api_function(self): return call_my_provider_api
+
+    @property
+    def api_key_env_var(self) -> str: return "MY_PROVIDER_API_KEY"
+
+class TestProviderEdgeCases:
+    """Provider-specific edge cases."""
+    def test_custom_behavior(self): ...
+
+@pytest.mark.integration
+class TestProviderIntegration:
+    """Real API tests (skipped by default)."""
+    def test_real_api_call(self): ...
+```
+
+## SSL Verification
+
+All HTTP requests use GAC's SSL verification settings via `get_ssl_verify()`:
+
+```python
+from gac.utils import get_ssl_verify
+
+response = httpx.post(url, ..., verify=get_ssl_verify())
+```
+
+This respects environment configurations for custom certificates.
+
+## Timeout Configuration
+
+All providers use `ProviderDefaults.HTTP_TIMEOUT` for consistency:
+
+```python
+from gac.constants import ProviderDefaults
+
+config = ProviderConfig(
+    name="My Provider",
+    api_key_env="MY_KEY",
+    base_url="https://api.example.com",
+    timeout=ProviderDefaults.HTTP_TIMEOUT,  # Default: 120 seconds
+)
+```
+
+## Provider-Specific Documentation
+
+See individual provider files for detailed documentation:
+
+- `openai.py` - OpenAI API reference
+- `anthropic.py` - Anthropic API reference
+- `azure_openai.py` - Azure OpenAI configuration
+- `gemini.py` - Google Gemini custom format
+- `replicate.py` - Async prediction handling
+- And others...
+
+## Contributing
+
+When adding a new provider:
+
+1. Follow the architecture and patterns above
+2. Write comprehensive tests (unit, mocked, integration)
+3. Update `__init__.py` exports
+4. Document the provider in its docstring
+5. Run `mypy` for type checking: `uv run -- mypy src/gac`
+6. Run tests: `uv run -- pytest tests/providers/test_<provider>.py -v`
+7. Update this README if adding new patterns
+
+## Best Practices
+
+1. **Lazy Initialization**: Use getter functions to initialize providers at call time, not import time. This allows tests to mock environment variables.
+
+2. **Error Preservation**: Always re-raise `AIError` exceptions without wrapping them. The error decorator handles generic exceptions.
+
+3. **Optional Parameters**: Match superclass signatures exactly, especially for `_get_api_url(model: str | None = None)`.
+
+4. **Response Validation**: Always validate responses for null/empty content before returning.
+
+5. **Configuration Over Code**: Use environment variables and `ProviderConfig` rather than hardcoding values.
+
+6. **Documentation**: Include docstrings with API endpoint references and required environment variables.
+
+7. **Test Coverage**: Aim for 100% test coverage of provider logic.

gac/providers/__init__.py

@@ -1,88 +1,80 @@
-"""AI provider implementations for commit message generation."""
+"""AI provider implementations for commit message generation.
 
-from .anthropic import call_anthropic_api
-from .azure_openai import call_azure_openai_api
-from .cerebras import call_cerebras_api
-from .chutes import call_chutes_api
-from .claude_code import call_claude_code_api
-from .custom_anthropic import call_custom_anthropic_api
-from .custom_openai import call_custom_openai_api
-from .deepseek import call_deepseek_api
-from .fireworks import call_fireworks_api
-from .gemini import call_gemini_api
-from .groq import call_groq_api
-from .kimi_coding import call_kimi_coding_api
-from .lmstudio import call_lmstudio_api
-from .minimax import call_minimax_api
-from .mistral import call_mistral_api
-from .moonshot import call_moonshot_api
-from .ollama import call_ollama_api
-from .openai import call_openai_api
-from .openrouter import call_openrouter_api
-from .replicate import call_replicate_api
-from .streamlake import call_streamlake_api
-from .synthetic import call_synthetic_api
-from .together import call_together_api
-from .zai import call_zai_api, call_zai_coding_api
+This module provides a unified interface to all AI providers. Provider classes
+are registered and wrapper functions are auto-generated with error handling.
 
-# Provider registry - single source of truth for all providers
-PROVIDER_REGISTRY = {
-    "anthropic": call_anthropic_api,
-    "azure-openai": call_azure_openai_api,
-    "cerebras": call_cerebras_api,
-    "claude-code": call_claude_code_api,
-    "chutes": call_chutes_api,
-    "custom-anthropic": call_custom_anthropic_api,
-    "custom-openai": call_custom_openai_api,
-    "deepseek": call_deepseek_api,
-    "fireworks": call_fireworks_api,
-    "gemini": call_gemini_api,
-    "groq": call_groq_api,
-    "kimi-coding": call_kimi_coding_api,
-    "lm-studio": call_lmstudio_api,
-    "minimax": call_minimax_api,
-    "mistral": call_mistral_api,
-    "moonshot": call_moonshot_api,
-    "ollama": call_ollama_api,
-    "openai": call_openai_api,
-    "openrouter": call_openrouter_api,
-    "replicate": call_replicate_api,
-    "streamlake": call_streamlake_api,
-    "synthetic": call_synthetic_api,
-    "together": call_together_api,
-    "zai": call_zai_api,
-    "zai-coding": call_zai_coding_api,
-}
+Usage:
+    from gac.providers import PROVIDER_REGISTRY
+
+    # Get the function for a provider
+    func = PROVIDER_REGISTRY["openai"]
+    result = func(model="gpt-4", messages=[...], temperature=0.7, max_tokens=1000)
+"""
+
+# Import provider classes for registration
+from .anthropic import AnthropicProvider
+from .azure_openai import AzureOpenAIProvider
+from .cerebras import CerebrasProvider
+from .chutes import ChutesProvider
+from .claude_code import ClaudeCodeProvider
+from .custom_anthropic import CustomAnthropicProvider
+from .custom_openai import CustomOpenAIProvider
+from .deepseek import DeepSeekProvider
+from .fireworks import FireworksProvider
+from .gemini import GeminiProvider
+from .groq import GroqProvider
+from .kimi_coding import KimiCodingProvider
+from .lmstudio import LMStudioProvider
+from .minimax import MinimaxProvider
+from .mistral import MistralProvider
+from .moonshot import MoonshotProvider
+from .ollama import OllamaProvider
+from .openai import OpenAIProvider
+from .openrouter import OpenRouterProvider
+from .qwen import QwenProvider
+from .registry import (
+    PROVIDER_REGISTRY,
+    register_provider,
+)
+from .replicate import ReplicateProvider
+from .streamlake import StreamlakeProvider
+from .synthetic import SyntheticProvider
+from .together import TogetherProvider
+from .zai import ZAICodingProvider, ZAIProvider
+
+# Register all providers - this populates PROVIDER_REGISTRY automatically
+register_provider("anthropic", AnthropicProvider)
+register_provider("azure-openai", AzureOpenAIProvider)
+register_provider("cerebras", CerebrasProvider)
+register_provider("chutes", ChutesProvider)
+register_provider("claude-code", ClaudeCodeProvider)
+register_provider("custom-anthropic", CustomAnthropicProvider)
+register_provider("custom-openai", CustomOpenAIProvider)
+register_provider("deepseek", DeepSeekProvider)
+register_provider("fireworks", FireworksProvider)
+register_provider("gemini", GeminiProvider)
+register_provider("groq", GroqProvider)
+register_provider("kimi-coding", KimiCodingProvider)
+register_provider("lm-studio", LMStudioProvider)
+register_provider("minimax", MinimaxProvider)
+register_provider("mistral", MistralProvider)
+register_provider("moonshot", MoonshotProvider)
+register_provider("ollama", OllamaProvider)
+register_provider("openai", OpenAIProvider)
+register_provider("openrouter", OpenRouterProvider)
+register_provider("qwen", QwenProvider)
+register_provider("replicate", ReplicateProvider)
+register_provider("streamlake", StreamlakeProvider)
+register_provider("synthetic", SyntheticProvider)
+register_provider("together", TogetherProvider)
+register_provider("zai", ZAIProvider)
+register_provider("zai-coding", ZAICodingProvider)
 
 # List of supported provider names - derived from registry keys
 SUPPORTED_PROVIDERS = sorted(PROVIDER_REGISTRY.keys())
 
 __all__ = [
-    "call_anthropic_api",
-    "call_azure_openai_api",
-    "call_cerebras_api",
-    "call_chutes_api",
-    "call_claude_code_api",
-    "call_custom_anthropic_api",
-    "call_custom_openai_api",
-    "call_deepseek_api",
-    "call_fireworks_api",
-    "call_gemini_api",
-    "call_groq_api",
-    "call_kimi_coding_api",
-    "call_lmstudio_api",
-    "call_minimax_api",
-    "call_mistral_api",
-    "call_moonshot_api",
-    "call_ollama_api",
-    "call_openai_api",
-    "call_openrouter_api",
-    "call_replicate_api",
-    "call_streamlake_api",
-    "call_synthetic_api",
-    "call_together_api",
-    "call_zai_api",
-    "call_zai_coding_api",
     "PROVIDER_REGISTRY",
     "SUPPORTED_PROVIDERS",
+    "register_provider",
 ]

gac/providers/anthropic.py

@@ -1,51 +1,17 @@
-"""Anthropic AI provider implementation."""
+"""Anthropic AI provider for gac."""
 
-import os
+from gac.providers.base import AnthropicCompatibleProvider, ProviderConfig
 
-import httpx
 
-from gac.errors import AIError
+class AnthropicProvider(AnthropicCompatibleProvider):
+    """Anthropic Claude API provider."""
 
+    config = ProviderConfig(
+        name="Anthropic",
+        api_key_env="ANTHROPIC_API_KEY",
+        base_url="https://api.anthropic.com/v1",
+    )
 
-def call_anthropic_api(model: str, messages: list[dict], temperature: float, max_tokens: int) -> str:
-    """Call Anthropic API directly."""
-    api_key = os.getenv("ANTHROPIC_API_KEY")
-    if not api_key:
-        raise AIError.authentication_error("ANTHROPIC_API_KEY not found in environment variables")
-
-    url = "https://api.anthropic.com/v1/messages"
-    headers = {"x-api-key": api_key, "anthropic-version": "2023-06-01", "content-type": "application/json"}
-
-    # Convert messages to Anthropic format
-    anthropic_messages = []
-    system_message = ""
-
-    for msg in messages:
-        if msg["role"] == "system":
-            system_message = msg["content"]
-        else:
-            anthropic_messages.append({"role": msg["role"], "content": msg["content"]})
-
-    data = {"model": model, "messages": anthropic_messages, "temperature": temperature, "max_tokens": max_tokens}
-
-    if system_message:
-        data["system"] = system_message
-
-    try:
-        response = httpx.post(url, headers=headers, json=data, timeout=120)
-        response.raise_for_status()
-        response_data = response.json()
-        content = response_data["content"][0]["text"]
-        if content is None:
-            raise AIError.model_error("Anthropic API returned null content")
-        if content == "":
-            raise AIError.model_error("Anthropic API returned empty content")
-        return content
-    except httpx.HTTPStatusError as e:
-        if e.response.status_code == 429:
-            raise AIError.rate_limit_error(f"Anthropic API rate limit exceeded: {e.response.text}") from e
-        raise AIError.model_error(f"Anthropic API error: {e.response.status_code} - {e.response.text}") from e
-    except httpx.TimeoutException as e:
-        raise AIError.timeout_error(f"Anthropic API request timed out: {str(e)}") from e
-    except Exception as e:
-        raise AIError.model_error(f"Error calling Anthropic API: {str(e)}") from e
+    def _get_api_url(self, model: str | None = None) -> str:
+        """Get Anthropic API URL with /messages endpoint."""
+        return f"{self.config.base_url}/messages"