#!/usr/bin/env python3 from parse import parse_mstrsen, parse_vacsen, mergesen import pandas as pd import sys import os ######################## ## ARGUMENTS ## if not len(sys.argv)==3: print("Need Vacancy Seniority.pdf and Master Seniority.html files as inputs!") sys.exit("Incorrent Arguments. " + sys.argv[0] + " aborted.") elif not sys.argv[1].lower().endswith('.pdf'): sys.exit("Vacancy Seniority file is not a PDF file." + sys.argv[0] + " aborted.") elif not sys.argv[2].lower().endswith('.html'): sys.exit("Master Seniority file is not an HTML file." + sys.argv[0] + " aborted.") else: if os.path.exists(sys.argv[1]): VacancyFile=sys.argv[1] else: sys.exit("Vacancy Seniority file not found. " + sys.argv[0] + " aborted.") if os.path.exists(sys.argv[2]): MasterSenFile=sys.argv[2] else: sys.exit("Master Seniority file not found. " + sys.argv[0] + " aborted.") ######################## print("Starting " + sys.argv[0]) df_mastersen = parse_mstrsen(MasterSenFile) df_domsen = parse_vacsen(VacancyFile) df_mergesen = mergesen(df_domsen, df_mastersen) print("Parsing Complete!") print("-----------------") print("Files are formatted for additional analysis") print("Try running alalyze.py for processing")