core/journal_reading/upload.py

import tempfile
import os
from llama_parse import LlamaParse
from llama_index.core.node_parser import SimpleNodeParser


class JournalUploader:
    def __init__(self):
        pass

    def parser_journal(self):
        if local_file_name is None:
            local_file_name = "downloaded_pdf_file.pdf"  # Default file name

        try:
            # Create a temporary directory to store the file
            temp_dir = tempfile.mkdtemp()
            file_path = os.path.join(temp_dir, local_file_name)
            
            with open(file_path, 'wb') as temp_file:
                self.s3_client.download_fileobj(self.bucket_name, object_name, temp_file)
                
            documents = LlamaParse(result_type="markdown").load_data(file_path)

            return documents
        
        except Exception as e:
            # Handle specific exceptions or fallback to generic one
            print(f"Error reading PDF file: {e}")
            raise RuntimeError(f"Failed to process the uploaded file: {e}")