aimakerspace/vectordatabase.py

from enum import Enum
import numpy as np
import uuid
from collections import defaultdict
from typing import List, Tuple, Callable
from aimakerspace.openai_utils.embedding import EmbeddingModel
import asyncio
from qdrant_client import models, QdrantClient
from qdrant_client.models import PointStruct,VectorParams,Distance

collection_name = "embedding_collection"

def cosine_similarity(vector_a: np.array, vector_b: np.array) -> float:
    """Computes the cosine similarity between two vectors."""
    dot_product = np.dot(vector_a, vector_b)
    norm_a = np.linalg.norm(vector_a)
    norm_b = np.linalg.norm(vector_b)
    return dot_product / (norm_a * norm_b)


def euclidean_distance(vector_a: np.array, vector_b: np.array) -> float:
    """Computes the Euclidean distance between two vectors."""
    return np.sqrt(np.sum((vector_a - vector_b) ** 2))


def manhattan_distance(vector_a: np.array, vector_b: np.array) -> float:
    """Computes the Manhattan distance between two vectors."""
    return np.sum(np.abs(vector_a - vector_b))


def minkowski_distance(vector_a: np.array, vector_b: np.array, p: float) -> float:
    """
    Computes the Minkowski distance between two vectors.

    Parameters:
    vector_a (np.array): First vector.
    vector_b (np.array): Second vector.
    p (float): The order of the norm. For example, p=1 gives Manhattan distance, p=2 gives Euclidean distance.

    Returns:
    float: Minkowski distance between vector_a and vector_b.
    """
    # Ensure the input vectors are NumPy arrays
    vector_a = np.asarray(vector_a)
    vector_b = np.asarray(vector_b)

    # Compute Minkowski distance
    distance = np.sum(np.abs(vector_a - vector_b) ** p) ** (1 / p)
    return distance


class DistanceMeasure(Enum):
    COSINE_SIMILARITY = cosine_similarity
    EUCLIDEAN_DISTANCE = euclidean_distance
    MANHATTAN_DISTANCE = manhattan_distance
    MINKOWSKI_DISTANCE = minkowski_distance


class VectorDatabaseOptions(Enum):
    DICTIONARY = "dictionary"
    QDRANT = "qdrant"


class VectorDatabase:
    def __init__(
        self,
        vector_db_options: VectorDatabaseOptions,
        embedding_model: EmbeddingModel = None,
    ):
        self.vectors = None
        self.vector_db_options = vector_db_options
        self.embedding_model = embedding_model or EmbeddingModel()
        if vector_db_options == VectorDatabaseOptions.DICTIONARY:
            self.vectors = defaultdict(np.array)
        if vector_db_options == VectorDatabaseOptions.QDRANT:
            self.qdrant_client = QdrantClient(":memory:")
            vector_params = VectorParams(
                size=embedding_model.dimensions,  # vector size 
                distance="Cosine"  # distance metric
            )
            self.qdrant_client.recreate_collection(
                collection_name=collection_name,
                vectors_config={"default": vector_params},
            )

    def insert(self, key: str, vector: np.array) -> None:
        idx = str(uuid.uuid4())
        payload = {"text": key}  

        point = PointStruct(
            id=idx,
            vector={"default": vector.tolist()},
            payload=payload
        )
        # Insert the vector into Qdrant with the associated document
        self.qdrant_client.upsert(
            collection_name=collection_name,
            points=[point]
        )
        print(f"Inserted vector with ID {idx}: {vector}")
        

    def search(
        self,
        query_vector: np.array,
        k: int,
        distance_measure: Callable = cosine_similarity,
    ) -> List[Tuple[str, float]]:
        # if isinstance(query_vector, list):
        #     query_vector = np.array(query_vector)
        print(f"Searching in collection: {collection_name} with vector: {query_vector}")
        collection_info = self.qdrant_client.get_collection(collection_name)
        print(f"Collection info: {collection_info}")

        search_results = self.qdrant_client.search(
            collection_name=collection_name,
            query_vector=query_vector,
            limit=k
        )
        return [(result.payload['text'], result.score) for result in search_results]

    def search_by_text(
        self,
        query_text: str,
        k: int,
        distance_measure: Callable = cosine_similarity,
        return_as_text: bool = False,
    ) -> List[Tuple[str, float]]:
        query_vector = self.embedding_model.get_embedding(query_text)
        results = self.search(query_vector, k, distance_measure)
        return [result[0] for result in results] if return_as_text else results
       

    def retrieve_from_key(self, key: str) -> np.array:
        return self.vectors.get(key, None)

    async def abuild_from_list(self, list_of_text: List[str]) -> "VectorDatabase":
        embeddings = await self.embedding_model.async_get_embeddings(list_of_text)
        for text, embedding in zip(list_of_text, embeddings):
            self.insert(text, np.array(embedding))
        return self


if __name__ == "__main__":
    list_of_text = [
        "I like to eat broccoli and bananas.",
        "I ate a banana and spinach smoothie for breakfast.",
        "Chinchillas and kittens are cute.",
        "My sister adopted a kitten yesterday.",
        "Look at this cute hamster munching on a piece of broccoli.",
    ]

    vector_db = VectorDatabase()
    vector_db = asyncio.run(vector_db.abuild_from_list(list_of_text))
    k = 2

    searched_vector = vector_db.search_by_text("I think fruit is awesome!", k=k)
    print(f"Closest {k} vector(s):", searched_vector)

    retrieved_vector = vector_db.retrieve_from_key(
        "I like to eat broccoli and bananas."
    )
    print("Retrieved vector:", retrieved_vector)

    relevant_texts = vector_db.search_by_text(
        "I think fruit is awesome!", k=k, return_as_text=True
    )
    print(f"Closest {k} text(s):", relevant_texts)