16-add-image-endpoint

README.md

@@ -3,7 +3,17 @@
 > [!CAUTION]
 > AI Backend is still in Development. You will find bugs and broken/unfinished features.
 
-## ✨ Installation and Configuration
+## 🌟 Overview
+
+ai-backend is a backend for AI-powered applications. It leverages FastAPI and Ollama to provide a robust API for natural language processing tasks.
+
+## 🚀 Installation and Configuration
+
+###  Prerequisites
+
+- Python 3.12
+- pip
+- git
 
 ### Installation for Development
 
@@ -32,6 +42,8 @@ pip install -r requirements.txt
 fastapi dev src/main.py
 ```
 
+## 📖 Documentation
+
 ### OpenAPI Documentation
 
 The OpenAPI documentation is available at `/docs`. It is automatically generated from the code.
@@ -52,6 +64,19 @@ curl -X POST "http://localhost:8000/message/" -H "Content-Type: application/json
 
 // WIP
 
+## 🧪 Testing
+
+To run the test suite:
+
+1. Ensure that both the AI Backend and Ollama services are running.
+2. Execute the following command:
+
+```bash
+pytest
+```
+
+This will run all tests in the `tests/` directory.
+
 ## 📝 Contributing
 
 // WIP
@@ -62,6 +87,14 @@ curl -X POST "http://localhost:8000/message/" -H "Content-Type: application/json
 - [Ollama](https://ollama.com/)
 - [Pydantic](https://pydantic-docs.helpmanual.io/)
 
-## 😊 License
+## 📄 License
+
+This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
+
+## 🙏 Acknowledgements
+
+- Special thanks to the FastAPI and Ollama communities for their excellent tools and documentation
+
+---
 
-[MIT](https://choosealicense.com/licenses/mit/)
+For more information, please [open an issue](https://github.com/Dino-Kupinic/ai-backend/issues) or contact the maintainers.

pytest.ini

@@ -0,0 +1,2 @@
+[pytest]
+pythonpath = .

src/internal/models.py

@@ -0,0 +1,17 @@
+from enum import Enum
+
+
+class TextModel(Enum):
+    """
+    Enum for text models.
+    """
+
+    LLAMA3 = "llama3"
+
+
+class ImageModel(Enum):
+    """
+    Enum for image models.
+    """
+
+    LLAVA = "llava"

src/internal/query.py

@@ -0,0 +1,106 @@
+from abc import ABC, abstractmethod
+from typing import overload, Any, Union
+
+import ollama
+from fastapi import HTTPException
+
+from src.internal.models import TextModel, ImageModel
+from src.types.alias import LLMResponse
+
+
+class Query(ABC):
+    """
+    Abstract class for querying a model
+    """
+
+    @abstractmethod
+    def query(self, prompt: str, model: Any, **kwargs) -> LLMResponse:
+        """
+        Query the model with the given prompt
+        :param prompt: The prompt to query the model with
+        :param model: The model to query
+        :return: A generator of the model's responses
+        """
+        pass
+
+
+class TextQuery(Query):
+    """
+    Query a text model
+    """
+
+    @overload
+    def query(self, prompt: str, model: TextModel, **kwargs) -> LLMResponse: ...
+
+    @overload
+    def query(self, prompt: str, model: str, **kwargs) -> LLMResponse: ...
+
+    def query(self, prompt: str, model: Union[TextModel, str], **kwargs) -> LLMResponse:
+        """
+        Query the model with the given prompt
+        :param prompt: The prompt to query the model with
+        :param model: The model to query (TextModel instance or string)
+        :return: A generator of the model's responses
+        """
+        try:
+            model_name = model.value if isinstance(model, TextModel) else model
+            messages = [{"role": "user", "content": prompt}]
+            yield from self._text_llm_call(model_name, messages)
+        except Exception as e:
+            print(f"Unexpected error: {str(e)}")
+            raise HTTPException(
+                status_code=500, detail=f"Internal Server Error: {str(e)}"
+            )
+
+    @staticmethod
+    def _text_llm_call(model: str, messages: list) -> LLMResponse:
+        for chunk in ollama.chat(
+            model,
+            messages=messages,
+            stream=True,
+        ):
+            yield chunk["message"]["content"]
+
+
+class ImageQuery(Query):
+    """
+    Query an image model
+    """
+
+    @overload
+    def query(self, prompt: str, model: ImageModel, **kwargs) -> LLMResponse: ...
+
+    @overload
+    def query(self, prompt: str, model: str, **kwargs) -> LLMResponse: ...
+
+    def query(
+        self, prompt: str, model: Union[ImageModel, str], **kwargs
+    ) -> LLMResponse:
+        """
+        Query the model with the given prompt and images
+        :param prompt: The prompt to query the model with
+        :param model: The model to query (ImageModel instance)
+        :return: A generator of the model's responses
+        """
+        try:
+            images = kwargs.get("images", [])
+            if not images:
+                raise ValueError("Images must be provided for an image query.")
+
+            model_name = model.value if isinstance(model, ImageModel) else model
+            messages = [{"role": "user", "content": prompt, "images": images}]
+            yield from self._image_llm_call(model_name, messages)
+        except Exception as e:
+            print(f"Unexpected error: {str(e)}")
+            raise HTTPException(
+                status_code=500, detail=f"Internal Server Error: {str(e)}"
+            )
+
+    @staticmethod
+    def _image_llm_call(model: str, messages: list) -> LLMResponse:
+        for chunk in ollama.chat(
+            model,
+            messages=messages,
+            stream=True,
+        ):
+            yield chunk["message"]["content"]

src/main.py

@@ -1,7 +1,7 @@
 from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
 
-from src.routers import message
+from src.routers import message, image
 
 app = FastAPI(
     title="AI Backend",
@@ -20,3 +20,4 @@
 )
 
 app.include_router(message.router)
+app.include_router(image.router)

src/routers/image.py

@@ -0,0 +1,30 @@
+from fastapi import APIRouter, HTTPException
+from fastapi.responses import StreamingResponse
+
+from src.internal.models import ImageModel
+from src.internal.query import ImageQuery
+from src.types.requests import Image
+
+router = APIRouter(prefix="/image", tags=["image"])
+
+image_query = ImageQuery()
+
+
+@router.post("/")
+async def send_message(image: Image):
+    try:
+        if isinstance(image.model, ImageModel):
+            model = image.model
+        elif isinstance(image.model, str):
+            model = image.model
+        else:
+            raise HTTPException(status_code=400, detail="Invalid model type")
+
+        return StreamingResponse(
+            image_query.query(prompt=image.prompt, model=model, images=image.images),
+            media_type="text/plain",
+        )
+    except HTTPException as http_exc:
+        raise http_exc
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Internal Server Error: {str(e)}")

src/routers/message.py

@@ -1,12 +1,30 @@
-from fastapi import APIRouter
+from fastapi import APIRouter, HTTPException
 from fastapi.responses import StreamingResponse
 
-from src.internal.queries import query
+from src.internal.query import TextQuery
 from src.types.requests import Message
+from src.internal.models import TextModel
 
 router = APIRouter(prefix="/message", tags=["message"])
 
+text_query = TextQuery()
+
 
 @router.post("/")
 async def send_message(message: Message):
-    return StreamingResponse(query(message.text))
+    try:
+        if isinstance(message.model, TextModel):
+            model = message.model
+        elif isinstance(message.model, str):
+            model = message.model
+        else:
+            raise HTTPException(status_code=400, detail="Invalid model type")
+
+        return StreamingResponse(
+            text_query.query(prompt=message.prompt, model=model),
+            media_type="text/plain",
+        )
+    except HTTPException as http_exc:
+        raise http_exc
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Internal Server Error: {str(e)}")

src/types/alias.py

@@ -0,0 +1,3 @@
+from typing import TypeAlias, Generator
+
+LLMResponse: TypeAlias = Generator[str, None, None]

src/types/requests.py

@@ -1,5 +1,20 @@
 from pydantic import BaseModel
 
+from src.internal.models import TextModel, ImageModel
+
 
 class Message(BaseModel):
-    text: str
+    """
+    Request body for sending a message to a model
+    """
+    prompt: str
+    model: TextModel
+
+
+class Image(BaseModel):
+    """
+    Request body for sending an image to a model
+    """
+    prompt: str
+    model: ImageModel
+    images: list[str]

tests/assets/test_images.json

@@ -0,0 +1,44 @@
+[
+  {
+    "prompt": "What can you see in the image?",
+    "model": "llava",
+    "images": ["images/bird.jpg"],
+    "answers": [
+      "bird",
+      "animal",
+      "tree",
+      "yellow",
+      "grey",
+      "wings"
+    ]
+  },
+  {
+    "prompt": "What can you see in the image?",
+    "model": "llava",
+    "images": ["images/cat.jpg"],
+    "answers": [
+      "cat",
+      "animal",
+      "black",
+      "white",
+      "furry",
+      "ears",
+      "green"
+    ]
+  },
+  {
+    "prompt": "What can you see in the image?",
+    "model": "llava",
+    "images": ["images/house.webp"],
+    "answers": [
+      "house",
+      "building",
+      "blue",
+      "white",
+      "windows",
+      "door",
+      "roof",
+      "plants"
+    ]
+  }
+]