app.py

import json
import logging
import os
import re
import sys

from langchain.embeddings.sentence_transformer import SentenceTransformerEmbeddings
from sentence_transformers import SentenceTransformer
from langchain.text_splitter import CharacterTextSplitter
#from langchain.embeddings import OpenAIEmbeddings
from langchain.vectorstores import Chroma
from langchain.document_loaders import PyPDFLoader
from fastapi.encoders import jsonable_encoder
from dotenv import load_dotenv

load_dotenv()
logging.basicConfig(level=logging.DEBUG)

ABS_PATH = os.path.dirname(os.path.abspath(__file__))
DB_DIR = os.path.join(ABS_PATH, "db")

#embedding_function

def replace_newlines_and_spaces(text):
    # Replace all newline characters with spaces
    text = text.replace("\n", " ")
    # Replace multiple spaces with a single space
    text = re.sub(r'\s+', ' ', text)
    return text


def get_documents():
    return PyPDFLoader("AI-smart-water-management-systems.pdf").load()


def init_chromadb():
    # Delete existing index directory and recreate the directory
    if os.path.exists(DB_DIR):
        import shutil
        shutil.rmtree(DB_DIR, ignore_errors=True)
        os.mkdir(DB_DIR)

    documents = []
    for num, doc in enumerate(get_documents()):
        doc.page_content = replace_newlines_and_spaces(doc.page_content)
        documents.append(doc)

    # Split the documents into chunks
    text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
    texts = text_splitter.split_documents(documents)

    # Select which embeddings we want to use
    #embeddings = OpenAIEmbeddings()
    embeddings = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")

    # Create the vectorestore to use as the index
    vectorstore = Chroma.from_documents(texts, embeddings, persist_directory=DB_DIR)
    #vectorstore.persist()
    
    print(vectorstore)
    #vectorstore = None

    db = vectorstore
    db.get()
    print(len(db.get()["ids"]))

    # Print the list of source files
    for x in range(len(db.get()["ids"])):
        # print(db.get()["metadatas"][x])
        doc = db.get()["metadatas"][x]
        source = doc["source"]
        print(source)

def query_chromadb():
    if not os.path.exists(DB_DIR):
        raise Exception(f"{DB_DIR} does not exist, nothing can be queried")

    # Select which embeddings we want to use
    #embeddings = OpenAIEmbeddings()
    embeddings = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")

    # Load Vector store from local disk
    vectorstore = Chroma(persist_directory=DB_DIR, embedding_function=embeddings)
    #vectorstore.persist()

    result = vectorstore.similarity_search_with_score(query="how to use AI in water conservation?", k=4)
    
    jsonable_result = jsonable_encoder(result)
    print(json.dumps(jsonable_result, indent=2))

def main():
    init_chromadb()
    query_chromadb()

if __name__ == '__main__':
    main()