process_pdf.py

from utils import move_files
from pdfparser_hq import process_pdf_hq
from pdfparser_performance import process_pdf_performance


def process_pdf(input, path_pdf_input, data_dumpster, pdf_processing_mode):
    """_summary_

    Args:
        input (_type_): Company name
        path_pdf_input (_type_): path where uploaded pdfs are located
        data_dumpster (_type_): path where company data is stored
        pdf_processing_mode (_type_): processing mode, quality or performance
    """
    dest = f"./{data_dumpster}/{input.replace(' ','-')}/pdf"
    src = path_pdf_input
    pdf_parser_mapper = {
    "quality" : process_pdf_hq,
    "performance" : process_pdf_performance
    }
    move_files(src, dest, "pdf")
    pdf_processor = pdf_parser_mapper[pdf_processing_mode]
    pdf_processor(dest)
    move_files(f"{dest}/extracted_pdf", f"./{data_dumpster}/{input.replace(' ','-')}", filetype="txt")