services/embed_service/embed.py

import os
from services.embed_service.utils import VTTTranscriptLoader, DocumentEmbedder

class EmbeddingService:
    def __init__(self, conf, pinecone, session_key):
        self.conf = conf
        
        if self.conf["embeddings"]["override"]:
            self.sess_key = self.conf["embeddings"]["demo_namespace"]
        else:
            self.sess_key = session_key
        
        self.pinecones = pinecone.run(namespace=self.sess_key)
        self.pc = self.pinecones['connection']
        self.pipeline = self.pinecones['pipeline']
        self.embedder = self.pinecones['embedder']

    def __enter__(self):
        print("Start Embedding Service")
        return self

    def __exit__(self, exc_type, exc_val, exc_tb):
        print("Exiting Embedding Service")
    
    def get_transcripts(self, files):
        # Get filepaths and load them in document format
        filepaths = [file.name for file in files]
        loader = VTTTranscriptLoader(filepaths)
        results = loader.load()
            
        return results
    
    def run(self, files):
        # gets the files, cleans them, and loads them into pinecone
        results = self.get_transcripts(files)

        doc_embedder =  DocumentEmbedder(
            pinecone_pipeline=self.pipeline, 
            files=results,
            embedder=self.embedder
        )
        
        # uploads them into pinecone
        doc_embedder.embed()
        
        return "complete"