abstractcore 2.6.0__tar.gz → 2.6.2__tar.gz

{abstractcore-2.6.0 → abstractcore-2.6.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: abstractcore
-Version: 2.6.0
+Version: 2.6.2
 Summary: Unified interface to all LLM providers with essential infrastructure for tool calling, streaming, and model management
 Author-email: Laurent-Philippe Albou <contact@abstractcore.ai>
 Maintainer-email: Laurent-Philippe Albou <contact@abstractcore.ai>
@@ -795,6 +795,66 @@ summarizer document.pdf --provider anthropic --model claude-3-5-sonnet
 
 **Complete guide**: [Centralized Configuration](docs/centralized-config.md)
 
+### Environment Variables
+
+AbstractCore supports environment variables for provider base URLs, enabling remote servers, Docker deployments, and non-standard ports:
+
+```bash
+# Ollama on remote server
+export OLLAMA_BASE_URL="http://192.168.1.100:11434"
+# Alternative: OLLAMA_HOST is also supported
+export OLLAMA_HOST="http://192.168.1.100:11434"
+
+# LMStudio on non-standard port
+export LMSTUDIO_BASE_URL="http://localhost:1235/v1"
+
+# OpenAI-compatible proxy
+export OPENAI_BASE_URL="https://api.portkey.ai/v1"
+
+# Anthropic proxy
+export ANTHROPIC_BASE_URL="https://api.portkey.ai/v1"
+```
+
+**Priority**: Programmatic `base_url` parameter > Runtime configuration > Environment variable > Default value
+
+**Provider discovery**: `get_all_providers_with_models()` automatically respects these environment variables when checking provider availability.
+
+### Programmatic Configuration
+
+Configure provider settings at runtime without environment variables:
+
+```python
+from abstractcore.config import configure_provider, get_provider_config, clear_provider_config
+from abstractcore import create_llm
+
+# Set provider base URL programmatically
+configure_provider('ollama', base_url='http://192.168.1.100:11434')
+
+# All future create_llm() calls automatically use the configured URL
+llm = create_llm('ollama', model='llama3:8b')  # Uses http://192.168.1.100:11434
+
+# Query current configuration
+config = get_provider_config('ollama')
+print(config)  # {'base_url': 'http://192.168.1.100:11434'}
+
+# Clear configuration (revert to env var / default)
+configure_provider('ollama', base_url=None)
+# Or clear all providers
+clear_provider_config()
+```
+
+**Use Cases**:
+- **Web UI Settings**: Configure providers through settings pages
+- **Docker Startup**: Read from custom env vars and configure programmatically
+- **Testing**: Set mock server URLs for integration tests
+- **Multi-tenant**: Configure different base URLs per tenant
+
+**Priority System**:
+1. Constructor parameter (highest): `create_llm("ollama", base_url="...")`
+2. Runtime configuration: `configure_provider('ollama', base_url="...")`
+3. Environment variable: `OLLAMA_BASE_URL`
+4. Default value (lowest): `http://localhost:11434`
+
 ## Documentation
 
 **📚 Complete Documentation:** [docs/](docs/) - Full documentation index and navigation guide

{abstractcore-2.6.0 → abstractcore-2.6.2}/README.md

@@ -683,6 +683,66 @@ summarizer document.pdf --provider anthropic --model claude-3-5-sonnet
 
 **Complete guide**: [Centralized Configuration](docs/centralized-config.md)
 
+### Environment Variables
+
+AbstractCore supports environment variables for provider base URLs, enabling remote servers, Docker deployments, and non-standard ports:
+
+```bash
+# Ollama on remote server
+export OLLAMA_BASE_URL="http://192.168.1.100:11434"
+# Alternative: OLLAMA_HOST is also supported
+export OLLAMA_HOST="http://192.168.1.100:11434"
+
+# LMStudio on non-standard port
+export LMSTUDIO_BASE_URL="http://localhost:1235/v1"
+
+# OpenAI-compatible proxy
+export OPENAI_BASE_URL="https://api.portkey.ai/v1"
+
+# Anthropic proxy
+export ANTHROPIC_BASE_URL="https://api.portkey.ai/v1"
+```
+
+**Priority**: Programmatic `base_url` parameter > Runtime configuration > Environment variable > Default value
+
+**Provider discovery**: `get_all_providers_with_models()` automatically respects these environment variables when checking provider availability.
+
+### Programmatic Configuration
+
+Configure provider settings at runtime without environment variables:
+
+```python
+from abstractcore.config import configure_provider, get_provider_config, clear_provider_config
+from abstractcore import create_llm
+
+# Set provider base URL programmatically
+configure_provider('ollama', base_url='http://192.168.1.100:11434')
+
+# All future create_llm() calls automatically use the configured URL
+llm = create_llm('ollama', model='llama3:8b')  # Uses http://192.168.1.100:11434
+
+# Query current configuration
+config = get_provider_config('ollama')
+print(config)  # {'base_url': 'http://192.168.1.100:11434'}
+
+# Clear configuration (revert to env var / default)
+configure_provider('ollama', base_url=None)
+# Or clear all providers
+clear_provider_config()
+```
+
+**Use Cases**:
+- **Web UI Settings**: Configure providers through settings pages
+- **Docker Startup**: Read from custom env vars and configure programmatically
+- **Testing**: Set mock server URLs for integration tests
+- **Multi-tenant**: Configure different base URLs per tenant
+
+**Priority System**:
+1. Constructor parameter (highest): `create_llm("ollama", base_url="...")`
+2. Runtime configuration: `configure_provider('ollama', base_url="...")`
+3. Environment variable: `OLLAMA_BASE_URL`
+4. Default value (lowest): `http://localhost:11434`
+
 ## Documentation
 
 **📚 Complete Documentation:** [docs/](docs/) - Full documentation index and navigation guide

abstractcore-2.6.2/abstractcore/config/__init__.py

@@ -0,0 +1,33 @@
+"""
+AbstractCore Configuration Module
+
+Provides configuration management and command-line interface for AbstractCore.
+"""
+
+from .vision_config import handle_vision_commands, add_vision_arguments
+from .manager import get_config_manager
+
+
+def configure_provider(provider: str, **kwargs) -> None:
+    """Configure runtime settings for a provider."""
+    get_config_manager().configure_provider(provider, **kwargs)
+
+
+def get_provider_config(provider: str) -> dict:
+    """Get runtime configuration for a provider."""
+    return get_config_manager().get_provider_config(provider)
+
+
+def clear_provider_config(provider: str = None) -> None:
+    """Clear runtime provider configuration."""
+    get_config_manager().clear_provider_config(provider)
+
+
+__all__ = [
+    'handle_vision_commands',
+    'add_vision_arguments',
+    'get_config_manager',
+    'configure_provider',
+    'get_provider_config',
+    'clear_provider_config'
+]

{abstractcore-2.6.0 → abstractcore-2.6.2}/abstractcore/config/manager.py

@@ -136,6 +136,7 @@ class ConfigurationManager:
         self.config_dir = Path.home() / ".abstractcore" / "config"
         self.config_file = self.config_dir / "abstractcore.json"
         self.config = self._load_config()
+        self._provider_config: Dict[str, Dict[str, Any]] = {}  # Runtime config (not persisted)
 
     def _load_config(self) -> AbstractCoreConfig:
         """Load configuration from file or create default."""
@@ -437,6 +438,52 @@ class ConfigurationManager:
         """Check if local_files_only should be forced for transformers."""
         return self.config.offline.force_local_files_only
 
+    def configure_provider(self, provider: str, **kwargs) -> None:
+        """
+        Configure runtime settings for a provider.
+
+        Args:
+            provider: Provider name ('ollama', 'lmstudio', 'openai', 'anthropic')
+            **kwargs: Configuration options (base_url, timeout, etc.)
+
+        Example:
+            configure_provider('ollama', base_url='http://192.168.1.100:11434')
+        """
+        provider = provider.lower()
+        if provider not in self._provider_config:
+            self._provider_config[provider] = {}
+
+        for key, value in kwargs.items():
+            if value is None:
+                # Remove config (revert to env var / default)
+                self._provider_config[provider].pop(key, None)
+            else:
+                self._provider_config[provider][key] = value
+
+    def get_provider_config(self, provider: str) -> Dict[str, Any]:
+        """
+        Get runtime configuration for a provider.
+
+        Args:
+            provider: Provider name
+
+        Returns:
+            Dict with configured settings, or empty dict if no config
+        """
+        return self._provider_config.get(provider.lower(), {}).copy()
+
+    def clear_provider_config(self, provider: Optional[str] = None) -> None:
+        """
+        Clear runtime provider configuration.
+
+        Args:
+            provider: Provider name, or None to clear all
+        """
+        if provider is None:
+            self._provider_config.clear()
+        else:
+            self._provider_config.pop(provider.lower(), None)
+
 
 # Global instance
 _config_manager = None

{abstractcore-2.6.0 → abstractcore-2.6.2}/abstractcore/providers/lmstudio_provider.py

@@ -2,6 +2,7 @@
 LM Studio provider implementation (OpenAI-compatible API).
 """
 
+import os
 import httpx
 import json
 import time
@@ -23,14 +24,19 @@ from ..events import EventType
 class LMStudioProvider(BaseProvider):
     """LM Studio provider using OpenAI-compatible API"""
 
-    def __init__(self, model: str = "local-model", base_url: str = "http://localhost:1234/v1", **kwargs):
+    def __init__(self, model: str = "local-model", base_url: Optional[str] = None, **kwargs):
         super().__init__(model, **kwargs)
         self.provider = "lmstudio"
 
         # Initialize tool handler
         self.tool_handler = UniversalToolHandler(model)
 
-        self.base_url = base_url.rstrip('/')
+        # Base URL priority: parameter > LMSTUDIO_BASE_URL > default
+        self.base_url = (
+            base_url or
+            os.getenv("LMSTUDIO_BASE_URL") or
+            "http://localhost:1234/v1"
+        ).rstrip('/')
 
         # Get timeout value - None means unlimited timeout
         timeout_value = getattr(self, '_timeout', None)

{abstractcore-2.6.0 → abstractcore-2.6.2}/abstractcore/providers/ollama_provider.py

@@ -3,6 +3,7 @@ Ollama provider implementation.
 """
 
 import json
+import os
 import httpx
 import time
 from typing import List, Dict, Any, Optional, Union, Iterator, AsyncIterator, Type
@@ -23,11 +24,17 @@ from ..events import EventType
 class OllamaProvider(BaseProvider):
     """Ollama provider for local models with full integration"""
 
-    def __init__(self, model: str = "qwen3:4b-instruct-2507-q4_K_M", base_url: str = "http://localhost:11434", **kwargs):
+    def __init__(self, model: str = "qwen3:4b-instruct-2507-q4_K_M", base_url: Optional[str] = None, **kwargs):
         super().__init__(model, **kwargs)
         self.provider = "ollama"
 
-        self.base_url = base_url.rstrip('/')
+        # Base URL priority: parameter > OLLAMA_BASE_URL > OLLAMA_HOST > default
+        self.base_url = (
+            base_url or
+            os.getenv("OLLAMA_BASE_URL") or
+            os.getenv("OLLAMA_HOST") or
+            "http://localhost:11434"
+        ).rstrip('/')
         self.client = httpx.Client(timeout=self._timeout)
         self._async_client = None  # Lazy-loaded async client
 

{abstractcore-2.6.0 → abstractcore-2.6.2}/abstractcore/providers/registry.py

@@ -354,6 +354,8 @@ class ProviderRegistry:
 
         This is used by the factory to create provider instances.
         """
+        from ..config import get_provider_config
+
         provider_info = self.get_provider_info(provider_name)
         if not provider_info:
             available_providers = ", ".join(self.list_provider_names())
@@ -362,8 +364,14 @@ class ProviderRegistry:
         provider_class = self.get_provider_class(provider_name)
         model = model or provider_info.default_model
 
+        # Get runtime config for this provider
+        runtime_config = get_provider_config(provider_name)
+
+        # Merge: runtime_config < kwargs (user kwargs take precedence)
+        merged_kwargs = {**runtime_config, **kwargs}
+
         try:
-            return provider_class(model=model, **kwargs)
+            return provider_class(model=model, **merged_kwargs)
         except ImportError as e:
             # Re-raise import errors with helpful message
             if provider_info.installation_extras:

{abstractcore-2.6.0 → abstractcore-2.6.2}/abstractcore/utils/version.py

@@ -11,4 +11,4 @@ including when the package is installed from PyPI where pyproject.toml is not av
 
 # Package version - update this when releasing new versions
 # This must be manually synchronized with the version in pyproject.toml
-__version__ = "2.6.0"
+__version__ = "2.6.2"

{abstractcore-2.6.0 → abstractcore-2.6.2}/abstractcore.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: abstractcore
-Version: 2.6.0
+Version: 2.6.2
 Summary: Unified interface to all LLM providers with essential infrastructure for tool calling, streaming, and model management
 Author-email: Laurent-Philippe Albou <contact@abstractcore.ai>
 Maintainer-email: Laurent-Philippe Albou <contact@abstractcore.ai>
@@ -795,6 +795,66 @@ summarizer document.pdf --provider anthropic --model claude-3-5-sonnet
 
 **Complete guide**: [Centralized Configuration](docs/centralized-config.md)
 
+### Environment Variables
+
+AbstractCore supports environment variables for provider base URLs, enabling remote servers, Docker deployments, and non-standard ports:
+
+```bash
+# Ollama on remote server
+export OLLAMA_BASE_URL="http://192.168.1.100:11434"
+# Alternative: OLLAMA_HOST is also supported
+export OLLAMA_HOST="http://192.168.1.100:11434"
+
+# LMStudio on non-standard port
+export LMSTUDIO_BASE_URL="http://localhost:1235/v1"
+
+# OpenAI-compatible proxy
+export OPENAI_BASE_URL="https://api.portkey.ai/v1"
+
+# Anthropic proxy
+export ANTHROPIC_BASE_URL="https://api.portkey.ai/v1"
+```
+
+**Priority**: Programmatic `base_url` parameter > Runtime configuration > Environment variable > Default value
+
+**Provider discovery**: `get_all_providers_with_models()` automatically respects these environment variables when checking provider availability.
+
+### Programmatic Configuration
+
+Configure provider settings at runtime without environment variables:
+
+```python
+from abstractcore.config import configure_provider, get_provider_config, clear_provider_config
+from abstractcore import create_llm
+
+# Set provider base URL programmatically
+configure_provider('ollama', base_url='http://192.168.1.100:11434')
+
+# All future create_llm() calls automatically use the configured URL
+llm = create_llm('ollama', model='llama3:8b')  # Uses http://192.168.1.100:11434
+
+# Query current configuration
+config = get_provider_config('ollama')
+print(config)  # {'base_url': 'http://192.168.1.100:11434'}
+
+# Clear configuration (revert to env var / default)
+configure_provider('ollama', base_url=None)
+# Or clear all providers
+clear_provider_config()
+```
+
+**Use Cases**:
+- **Web UI Settings**: Configure providers through settings pages
+- **Docker Startup**: Read from custom env vars and configure programmatically
+- **Testing**: Set mock server URLs for integration tests
+- **Multi-tenant**: Configure different base URLs per tenant
+
+**Priority System**:
+1. Constructor parameter (highest): `create_llm("ollama", base_url="...")`
+2. Runtime configuration: `configure_provider('ollama', base_url="...")`
+3. Environment variable: `OLLAMA_BASE_URL`
+4. Default value (lowest): `http://localhost:11434`
+
 ## Documentation
 
 **📚 Complete Documentation:** [docs/](docs/) - Full documentation index and navigation guide

abstractcore-2.6.0/abstractcore/config/__init__.py

@@ -1,10 +0,0 @@
-"""
-AbstractCore Configuration Module
-
-Provides configuration management and command-line interface for AbstractCore.
-"""
-
-from .vision_config import handle_vision_commands, add_vision_arguments
-from .manager import get_config_manager
-
-__all__ = ['handle_vision_commands', 'add_vision_arguments', 'get_config_manager']