chore: Add base classes for chains and agents

README.md

@@ -104,3 +104,16 @@ SimpleMind is inspired by the philosophy of "code for humans" and aims to make w
 ---
 
 SimpleMind: Keep it simple, keep it human.
+
+------------------------
+
+
+## Plugins
+
+
+SimpleMind supports a plugin system to extend its functionality. Currently available plugins:
+
+- **KVPlugin**: Key-Value storage for context management.
+- **BasicMemoryPlugin**: Simple memory storage for conversations.
+
+**Adding a Plugin:**

simplemind/agents/base.py

@@ -0,0 +1,7 @@
+from abc import ABC, abstractmethod
+
+
+class BaseAgent(ABC):
+    @abstractmethod
+    def decide(self, context, *args, **kwargs):
+        pass

simplemind/chains/base.py

@@ -0,0 +1,7 @@
+from abc import ABC, abstractmethod
+
+
+class BaseChain(ABC):
+    @abstractmethod
+    def run(self, input_data):
+        pass

simplemind/chains/reverse_text.py

@@ -0,0 +1,6 @@
+from .base import BaseChain
+
+
+class ReverseTextChain(BaseChain):
+    def run(self, input_data):
+        return input_data[::-1]

simplemind/client.py

@@ -0,0 +1,35 @@
+from typing import Optional
+from simplemind.models import Conversation, AIResponse
+from simplemind.concepts import Context
+from simplemind.integrations.openai import OpenAI
+from simplemind.integrations.anthropic import Anthropic
+
+class Client:
+    def __init__(self, api_key: str, context: Optional[Context] = None):
+        self.api_key = api_key
+        self.context = context or Context()
+        self.providers = self._initialize_providers()
+
+    def _initialize_providers(self):
+        return {
+            "openai": OpenAI(api_key=self.api_key),
+            "anthropic": Anthropic(api_key=self.api_key),
+        }
+
+    def create_conversation(self, provider: str = "openai") -> Conversation:
+        if provider not in self.providers:
+            raise ValueError(f"Provider '{provider}' not supported.")
+        return self.providers[provider].create_conversation(initial_message="Hello!", context=self.context.dict())
+
+    def send_message(self, conversation: Conversation, message: str, provider: str = "openai") -> AIResponse:
+        if provider not in self.providers:
+            raise ValueError(f"Provider '{provider}' not supported.")
+        return self.providers[provider].send_message(conversation.id, message)
+
+    @property
+    def available_models(self):
+        available = {}
+        for name, provider in self.providers.items():
+            available[name] = provider.available_models
+        return available
+

simplemind/concepts.py

@@ -1,6 +1,16 @@
-class Context:
-    def __init__(self):
-        self.plugins = [kv, basic_memory]
+from pydantic import BaseModel
+from typing import Dict, Any
+from simplemind.plugins.base import BasePlugin
 
 
-# TODO: explore pluggy for this.
+class Context(BaseModel):
+    plugins: Dict[str, BasePlugin] = {}
+
+    def add_plugin(self, name: str, plugin: BasePlugin):
+        self.plugins[name] = plugin
+
+    def execute_plugin(self, name: str, *args, **kwargs):
+        if name in self.plugins:
+            return self.plugins[name].execute(self, *args, **kwargs)
+        else:
+            raise ValueError(f"Plugin '{name}' not found in context.")

simplemind/config.py

@@ -0,0 +1,13 @@
+from pydantic import BaseSettings
+
+
+class Settings(BaseSettings):
+    openai_api_key: str = ""
+    anthropic_api_key: str = ""
+    default_model: str = "gpt-4"
+
+    class Config:
+        env_file = ".env"
+
+
+settings = Settings()

simplemind/core.py

@@ -1,21 +0,0 @@
-from fastapi import FastAPI, HTTPException
-from pydantic import BaseModel
-from typing import Any
-
-app = FastAPI(title="SimpleMind AI API", description="AI for humans, replacing LangGraph and LangChain for Python users.")
-
-
-
-@app.post("/generate", response_model=AIResponse)
-def generate_response(request: AIRequest):
-    try:
-        # Placeholder for AI generation logic
-        response = {"message": "This would be the AI response."}
-        metadata = {"tokens_used": 50}
-        return AIResponse(response=response, metadata=metadata)
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-
-@app.get("/health")
-def health_check():
-    return {"status": "healthy"}

simplemind/integrations/anthropic.py

@@ -1,41 +1,69 @@
 import os
+from typing import List, Optional
 
 import instructor
 from anthropic import Anthropic as BaseAnthropic
 
 from .base import BaseClientProvider
+from ..models import AIResponse, Conversation
+from ..logger import logger
 
 
 class Anthropic(BaseClientProvider):
-    def __init__(self, *args, **kwargs):
-        super().__init__(*args, **kwargs)
+    def __init__(self, model: str = DEFAULT_MODEL, api_key: Optional[str] = None):
+        super().__init__(model=model, api_key=api_key)
         self.login()
 
     def login(self):
-        """Initialize Anthropic client, with Instructor enabled."""
-
-        # Default to environment variable if not provided.
-        if self._api_key is None:
+        if not self._api_key:
             self._api_key = os.getenv("ANTHROPIC_API_KEY")
-
+        if not self._api_key:
+            raise ValueError("Anthropic API key not provided.")
         base_client = BaseAnthropic(api_key=self._api_key)
         self.client = instructor.from_anthropic(base_client)
-        # assert self.test_connection()
+        if not self.test_connection():
+            raise ConnectionError("Failed to connect to Anthropic API.")
+        logger.info("Logged in to Anthropic successfully.")
 
     @property
-    def available_models(self):
-        """Returns the available models from the Anthropic client."""
+    def available_models(self) -> List[str]:
+        try:
+            return [
+                "claude-3-opus-20240229",
+                "claude-3-5-sonnet-20240620",
+                "claude-3-haiku-20240307",
+            ]
+        except Exception as e:
+            logger.error(f"Error fetching models: {e}")
+            return []
 
-        # TODO: scrape from website or embed
-        return [
-            "claude-3-opus-20240229",
-            "claude-3-5-sonnet-20240620",
-            "claude-3-haiku-20240307",
-            "claude-3-5-sonnet-20240620",
-            "claude-3-5-sonnet-20240620",
+    def test_connection(self) -> bool:
+        models = self.available_models
+        if models:
+            logger.info(f"Available models: {models}")
+            return True
+        logger.warning("No available models found.")
+        return False
+
+    def generate_response(self, conversation: Conversation) -> AIResponse:
+        messages = [
+            {"role": msg.role, "content": msg.content} for msg in conversation.messages
         ]
+        params = {
+            "messages": messages,
+            "model": self.model,
+        }
+        if conversation.context:
+            params["context"] = conversation.context
 
-    # def test_connection(self):
-    #     """Test the connection to Anthropic. Returns True if successful."""
-
-    #     raise NotImplementedError("Anthropic test_connection not implemented.")
+        try:
+            completion = self.client.completions.create(**params)
+            response_text = completion.completion
+            metadata = {"model": completion.model, "usage": completion.usage}
+            logger.info("Generated response from Anthropic.")
+            return AIResponse(
+                text=response_text, response=completion, metadata=metadata
+            )
+        except Exception as e:
+            logger.error(f"Error generating response: {e}")
+            raise e

simplemind/integrations/base.py

@@ -1,6 +1,10 @@
 # import logging
 
 from pydantic import BaseModel
+from typing import Any, Dict, List, Optional
+from ..models import AIResponse, Conversation, Message
+import uuid
+from abc import ABC, abstractmethod
 
 
 DEFAULT_MODEL = "gpt-4o"
@@ -8,55 +12,46 @@ DEFAULT_MODEL = "gpt-4o"
 
 class BaseClientProvider:
 
-    def __init__(self, *, model=DEFAULT_MODEL, api_key=None):
+    def __init__(self, *, model: str = DEFAULT_MODEL, api_key: Optional[str] = None):
         # self.logger = logging.getLogger(self.__class__.__name__)
         self.client = None
         self.model = model
-
-        # Load API key from environment if not provided
         self._api_key = api_key
+        self.conversations: Dict[str, Conversation] = {}
 
+    @abstractmethod
     def login(self):
         """Initializes the AI provider client."""
-
         msg = "This method must be implemented by the AI provider client."
         raise NotImplementedError(msg)
 
-    def test_connection(self):
+    @abstractmethod
+    def test_connection(self) -> bool:
         """Tests the connection to the AI provider client."""
-
         msg = "This method must be implemented by the AI provider client."
         raise NotImplementedError(msg)
 
-    # def generate_response(self, request):
-    #     """Generates a response from the AI provider client."""
-
-    #     msg = "This method must be implemented by the AI provider client."
-    #     raise NotImplementedError(msg)
-
+    @abstractmethod
     def health_check(self):
         """Checks the health of the AI provider client."""
-
         msg = "This method must be implemented by the AI provider client."
         raise NotImplementedError(msg)
 
-    @property
-    def available_models(self):
+    @abstractmethod
+    def available_models(self) -> List[str]:
         """Returns the available models from the AI provider client."""
-
         msg = "This method must be implemented by the AI provider client."
-
         raise NotImplementedError(msg)
 
-    def message(self, message, **kwargs):
+    @abstractmethod
+    def message(self, message: str, **kwargs) -> AIResponse:
         """Generates a response from the AI provider client."""
-
         msg = "This method must be implemented by the AI provider client."
         raise NotImplementedError(msg)
 
+    # Uncomment and implement additional methods as needed
     # def features(self):
     #     """Returns the features of the AI provider client."""
-
     #     msg = "This method must be implemented by the AI provider client."
     #     raise NotImplementedError(msg)
 
@@ -71,3 +66,51 @@ class BaseClientProvider:
 
     # def start_conversation(self, model, message, **kwargs):
     #     pass
+
+    def create_conversation(
+        self, initial_message: str, context: Optional[Dict[str, Any]] = None
+    ) -> Conversation:
+        conv_id = str(uuid.uuid4())
+        conversation = Conversation(
+            id=conv_id,
+            messages=[Message(role="user", content=initial_message)],
+            context=context or {},
+        )
+        self.conversations[conv_id] = conversation
+        return conversation
+
+    def send_message(
+        self,
+        conversation_id: str,
+        message: str,
+        context_update: Optional[Dict[str, Any]] = None,
+    ) -> AIResponse:
+        if conversation_id not in self.conversations:
+            raise ValueError("Conversation ID does not exist.")
+
+        conversation = self.conversations[conversation_id]
+        conversation.messages.append(Message(role="user", content=message))
+
+        if context_update:
+            conversation.context.update(context_update)
+
+        response = self.generate_response(conversation)
+        conversation.messages.append(Message(role="assistant", content=response.text))
+        return response
+
+    def generate_response(self, conversation: Conversation) -> AIResponse:
+        """Generates a response based on the conversation."""
+        raise NotImplementedError(
+            "This method must be implemented by the AI provider client."
+        )
+
+    def get_conversation(self, conversation_id: str) -> Conversation:
+        if conversation_id not in self.conversations:
+            raise ValueError("Conversation ID does not exist.")
+        return self.conversations[conversation_id]
+
+
+class BasePlugin(ABC):
+    @abstractmethod
+    def execute(self, context, *args, **kwargs):
+        pass

simplemind/integrations/openai.py

@@ -1,68 +1,67 @@
 import os
-
+from typing import Optional, List
 import instructor
 from openai import OpenAI as BaseOpenAI
 
 from .base import BaseClientProvider
-from ..models import AIResponse
-
+from ..models import AIResponse, Conversation
+from ..logger import logger
+from simplemind.config import settings
 
 class OpenAI(BaseClientProvider):
-
-    def __init__(self, *args, **kwargs):
-        super().__init__(*args, **kwargs)
+    def __init__(self, model: str = DEFAULT_MODEL, api_key: Optional[str] = None):
+        super().__init__(model=model, api_key=api_key)
         self.login()
 
     def login(self):
-        """Initialize OpenAI client, with Instructor enabled."""
-
-        # Default to environment variable if not provided.
-        if self._api_key is None:
-            self._api_key = os.getenv("OPENAI_API_KEY")
-
+        if not self._api_key:
+            self._api_key = settings.openai_api_key
+        if not self._api_key:
+            raise ValueError("OpenAI API key not provided.")
         self.client = BaseOpenAI(api_key=self._api_key)
         self.instructor_client = instructor.from_openai(self.client)
-        assert self.test_connection()
+        if not self.test_connection():
+            raise ConnectionError("Failed to connect to OpenAI API.")
+        logger.info("Logged in to OpenAI successfully.")
 
     @property
-    def available_models(self):
-        """Returns the available models from the OpenAI client."""
+    def available_models(self) -> List[str]:
+        try:
+            return [model.id for model in self.client.models.list()]
+        except Exception as e:
+            logger.error(f"Error fetching models: {e}")
+            return []
 
-        def gen():
-            for model in self.client.models.list():
-                yield model.id
+    def test_connection(self) -> bool:
+        try:
+            models = self.available_models
+            if models:
+                logger.info(f"Available models: {models}")
+                return True
+            else:
+                logger.warning("No available models found.")
+                return False
+        except Exception as e:
+            logger.error(f"Error testing connection: {e}")
+            return False
 
-        return [g for g in gen()]
-
-    def test_connection(self):
-        """Test the connection to OpenAI. Returns True if successful."""
-
-        return bool(len(self.available_models))
-
-    def message(self, message, *, response_model=False, **kwargs):
-        """Generates a response from the OpenAI client."""
-        use_instructor = bool(response_model)
-
-        client = self.instructor_client if use_instructor else self.client
-
-        # Parameters for the OpenAI client.
+    def generate_response(self, conversation: Conversation) -> AIResponse:
+        messages = [
+            {"role": msg.role, "content": msg.content} for msg in conversation.messages
+        ]
         params = {
-            "messages": [{"role": "user", "content": message}],
+            "messages": messages,
             "model": self.model,
         }
-        params.update(kwargs)
+        if conversation.context:
+            params["context"] = conversation.context
 
-        if use_instructor:
-            params["response_model"] = response_model
-
-        # Make the request to OpenAI.
-        completion = client.chat.completions.create(**params)
-
-        if use_instructor:
-            return completion.model_dump()
-
-        else:
-            return AIResponse(
-                response=completion,
-                text=completion.choices[0].message.content,
-            )
+        try:
+            completion = self.client.chat.completions.create(**params)
+            response_text = completion.choices[0].message.content
+            metadata = {"model": completion.model, "usage": completion.usage}
+            logger.info("Generated response from OpenAI.")
+            return AIResponse(text=response_text, response=completion, metadata=metadata)
+        except Exception as e:
+            logger.error(f"Error generating response: {e}")
+            raise e

simplemind/logger.py

@@ -0,0 +1,15 @@
+import logging
+
+def setup_logger(name: str) -> logging.Logger:
+    logger = logging.getLogger(name)
+    if not logger.hasHandlers():
+        logger.setLevel(logging.INFO)
+        handler = logging.StreamHandler()
+        formatter = logging.Formatter('[%(asctime)s] %(levelname)s - %(message)s')
+        handler.setFormatter(formatter)
+        logger.addHandler(handler)
+    return logger
+
+# Initialize a global logger
+logger = setup_logger("simplemind")
+

simplemind/models.py

@@ -1,10 +1,11 @@
 from pydantic import BaseModel
-from typing import Any, ClassVar
+from typing import Any, Dict, List, Optional
+import uuid
 
 
 class AIRequest(BaseModel):
     text: str
-    parameters: dict = {}
+    parameters: Dict[str, Any] = {}
 
     def __str__(self):
         return self.text
@@ -13,7 +14,30 @@ class AIRequest(BaseModel):
 class AIResponse(BaseModel):
     text: str
     response: Any
-    metadata: dict = {}
+    metadata: Dict[str, Any] = {}
 
     def __str__(self):
         return self.text
+
+
+class Message(BaseModel):
+    role: str  # "user" or "assistant"
+    content: str
+
+
+class Conversation(BaseModel):
+    id: str
+    messages: List[Message] = []
+    context: Optional[Dict[str, Any]] = {}
+
+
+class ConversationRequest(BaseModel):
+    conversation_id: Optional[str] = None
+    message: str
+    context_update: Optional[Dict[str, Any]] = None
+
+
+class ConversationResponse(BaseModel):
+    conversation_id: str
+    messages: List[Message]
+    metadata: Dict[str, Any] = {}

simplemind/plugins/basic_memory.py

@@ -0,0 +1,10 @@
+from .base import BasePlugin
+
+
+class BasicMemoryPlugin(BasePlugin):
+    def __init__(self):
+        self.memory = []
+
+    def execute(self, context, message):
+        self.memory.append(message)
+        return self.memory

simplemind/plugins/kv.py

@@ -0,0 +1,10 @@
+from .base import BasePlugin
+
+
+class KVPlugin(BasePlugin):
+    def __init__(self):
+        self.store = {}
+
+    def execute(self, context, key, value):
+        self.store[key] = value
+        return self.store

simplemind/vector_store/faiss_store.py

@@ -0,0 +1,19 @@
+import faiss
+import numpy as np
+from typing import List
+
+
+class FAISSStore:
+    def __init__(self, dimension: int):
+        self.dimension = dimension
+        self.index = faiss.IndexFlatL2(dimension)
+        self.ids = []
+
+    def add_embeddings(self, embeddings: np.ndarray, ids: List[str]):
+        self.index.add(embeddings)
+        self.ids.extend(ids)
+
+    def search(self, query_embedding: np.ndarray, top_k: int = 5):
+        distances, indices = self.index.search(query_embedding, top_k)
+        results = [(self.ids[idx], distances[i]) for i, idx in enumerate(indices[0])]
+        return results

t.py

@@ -1,54 +1,34 @@
 from pprint import pprint
 from pydantic import BaseModel
 import simplemind
-
-context = None
-
-openai = simplemind.integrations.OpenAI()
+from simplemind.concepts import Context
+from simplemind.plugins.kv import KVPlugin
+from simplemind.plugins.basic_memory import BasicMemoryPlugin
+from simplemind.chains.reverse_text import ReverseTextChain
+from simplemind.client import Client
 
 
-class YearlyData(BaseModel):
-    year: int
-    events: list[str]
+class MyContext(Context):
+    def __init__(self):
+        super().__init__()
+        self.add_plugin("kv", KVPlugin())
+        self.add_plugin("basic_memory", BasicMemoryPlugin())
 
 
-class ProjectData(BaseModel):
-    name: str
-    description: str
-    url: str
-    github_url: str
+# Initialize context and client
+context = MyContext()
+aiclient = Client(api_key="YOUR_API_KEY", context=context)
 
+# Test connection and available models
+print(aiclient.available_models)
 
-class BioData(BaseModel):
-    bio: str
-    spouse_name: str
-    history: list[YearlyData]
-    fun_facts: list[str]
-    # age: int
-    # occupation: str
-    # bio: str
-    # affiliations: list[str]
+# Example usage
+conversation = aiclient.create_conversation(provider="openai")
+response = aiclient.send_message(
+    conversation, "Who is Kenneth Reitz?", provider="openai"
+)
+print(response)
 
-
-class PersonData(BaseModel):
-    bio: BioData
-    projects: list[ProjectData]
-    yearly_breakdown: list[YearlyData]
-
-
-print(openai.test_connection())
-print(openai.available_models)
-
-print()
-print()
-message = "who is kenneth reitz?"
-
-print(f"> {message}")
-pprint(openai.message(message, response_model=BioData))
-
-# claude = simplemind.integrations.Anthropic()
-
-# # print(claude.test_connection())
-# # print(claude.available_models)
-
-# claude.login()
+reverse_chain = ReverseTextChain()
+result = reverse_chain.run("Hello, World!")
+print(result)  # Output: !dlroW ,olleH

t2.py

@@ -1,32 +1,68 @@
-import instructor
+from pprint import pprint
 from pydantic import BaseModel
-from openai import OpenAI
+import simplemind
+from simplemind.vector_store.faiss_store import FAISSStore
+import numpy as np
+
+context = None
+
+openai = simplemind.integrations.OpenAI()
 
 
-class ProjectInfo(BaseModel):
+class YearlyData(BaseModel):
+    year: int
+    events: list[str]
+
+
+class ProjectData(BaseModel):
     name: str
     description: str
     url: str
     github_url: str
 
 
-# Define your desired output structure
-class UserInfo(BaseModel):
-    name: str
-    age: int
+class BioData(BaseModel):
     bio: str
-    projects: list[ProjectInfo]
+    spouse_name: str
+    history: list[YearlyData]
+    fun_facts: list[str]
+    # age: int
+    # occupation: str
+    # bio: str
+    # affiliations: list[str]
 
 
-# Patch the OpenAI client
-client = instructor.from_openai(OpenAI())
+class PersonData(BaseModel):
+    bio: BioData
+    projects: list[ProjectData]
+    yearly_breakdown: list[YearlyData]
 
-# Extract structured data from natural language
-user_info = client.chat.completions.create(
-    model="gpt-4o",
-    response_model=UserInfo,
-    messages=[{"role": "user", "content": "who is kennethreitz?"}],
-)
 
-print(user_info.model_dump())
-# > 30
+print(openai.test_connection())
+print(openai.available_models)
+
+print()
+print()
+message = "who is kenneth reitz?"
+
+print(f"> {message}")
+pprint(openai.message(message, response_model=BioData))
+
+# claude = simplemind.integrations.Anthropic()
+
+# # print(claude.test_connection())
+# # print(claude.available_models)
+
+# claude.login()
+
+vector_store = FAISSStore(dimension=768)  # Example dimension for embeddings
+
+# Add embeddings
+embeddings = np.random.random((10, 768)).astype('float32')
+ids = [f"doc_{i}" for i in range(10)]
+vector_store.add_embeddings(embeddings, ids)
+
+# Search
+query_embedding = np.random.random((1, 768)).astype('float32')
+results = vector_store.search(query_embedding, top_k=3)
+print(results)

t3.py

@@ -0,0 +1,15 @@
+from simplemind.concepts import Context
+
+# from simplemind.plugins.default_plugin import DefaultPlugin
+
+# Initialize the context
+ctx = Context()
+
+# Add and initialize the DefaultPlugin
+# ctx.add_plugin(DefaultPlugin, "DefaultPlugin")
+
+# Execute the DefaultPlugin with some data
+# ctx.execute_plugin("DefaultPlugin", {"key": "value"})
+
+# Shutdown all plugins
+# ctx.shutdown_plugins()

tests/test_openai.py

@@ -0,0 +1,26 @@
+import unittest
+from unittest.mock import patch, MagicMock
+from simplemind.integrations.openai import OpenAI
+
+
+class TestOpenAIProvider(unittest.TestCase):
+    @patch("simplemind.integrations.openai.BaseOpenAI")
+    def setUp(self, mock_openai):
+        self.mock_openai = mock_openai.return_value
+        self.mock_openai.models.list.return_value = [MagicMock(id="gpt-4")]
+        self.provider = OpenAI(api_key="test_api_key", model="gpt-4")
+
+    def test_available_models(self):
+        models = self.provider.available_models
+        self.assertIn("gpt-4", models)
+
+    def test_test_connection_success(self):
+        self.assertTrue(self.provider.test_connection())
+
+    def test_generate_response_not_implemented(self):
+        with self.assertRaises(NotImplementedError):
+            self.provider.generate_response(None)
+
+
+if __name__ == "__main__":
+    unittest.main()