Create vlm.py

vlm/vlm.py

@@ -0,0 +1,55 @@
+import os
+from typing import Any
+from vlm.vlm_interface import VLMIterface
+import google.generativeai as genai
+from dotenv import load_dotenv
+
+# Load environment variables from a .env file
+load_dotenv()
+# Fetch Google API key from environment variables
+_api = os.getenv("GOOGLE_API_KEY")
+NAME = "gemini-1.5-flash"
+
+class VLM(VLMIterface):
+    """
+    This class implements the interface for a Visual Language Model (VLM) using Google's generative AI API.
+    It inherits from the VLMInterface and configures the Google API to interact with the generative model.
+    """
+
+    def __init__(self) -> None:
+        """
+        Initialize the VLM object by configuring the Google generative AI model using the provided API key.
+        """
+        super().__init__()
+        # Configure the generative AI model with the provided API key
+        genai.configure(api_key=_api)
+        # Initialize the model (Gemini 1.5 Flash) from Google's generative AI
+        self.vlm = genai.GenerativeModel(NAME)
+
+    def execution(self) -> Any | None:
+        """
+        Execute the VLM model and return the model object.
+        If there is an error (e.g., invalid API key or connection issue), handle the exception gracefully.
+        """
+        try:
+            return self.vlm
+        except Exception as e:
+            print(f"Something went wrong with the API: {e}")
+            return None
+    
+    def model_name(self) -> str:
+        return NAME
+
+    def __str__(self) -> str:
+        """
+        Return a human-readable string representation of the VLM object.
+        Useful for informal descriptions or when printing the object.
+        """
+        return f"VLM Model: {self.vlm.model_name}"
+
+    def __repr__(self) -> str:
+        """
+        Return an official string representation of the VLM object.
+        This should ideally be a valid Python expression that can recreate the object.
+        """
+        return f"VLM(api_key='****', model_name='{self.vlm.model_name}')"