gctools/misc/convert_mail_csv.py

import csv
from pathlib import Path

from misc.file_backup import copy_file_with_timestamp


def count_unique_emails(csv_file):
    emails = set()
    with open(csv_file, "r") as file:
        reader = csv.reader(file, delimiter=";")
        for row in reader:
            if row[2] in ["Empfaenger", ""]:
                continue
            emails.add(row[2])
    with open(str(Path(csv_file).parent) + "\\email_empfaenger.csv", "w") as fwh:
        for email in sorted(emails):
            fwh.write(email + "\n")
    return len(emails)


def convert_csv_in_folder(folder, export_csv, all_reports=False):
    row_count = 0

    with open(export_csv, "w") as fwh:
        fwh.write("Versand;Datei;Empfaenger\n")
        for file in Path(folder).glob("*.csv"):
            with file.open("r") as frh:
                reader = csv.reader(frh, delimiter=";")
                for row in reader:
                    if len(row) < 2 or row[1] in ["Empfaenger"]:
                        continue
                    if row[1] == "":
                        if all_reports:
                            fwh.write(file.name + ";" + row[0] + ";\n")
                            row_count += 1
                        continue

                    email_list = row[1].lower().split(",")
                    for email in sorted(email_list):
                        fwh.write(file.name + ";" + row[0] + ";" + email.strip() + "\n")
                        row_count += 1
    return row_count


def convert_combined_to_folder(csv_file, export_folder):
    result = {}
    with open(csv_file, "r") as frh:
        reader = csv.reader(frh, delimiter=";")

        for row in reader:
            if row[0] in ["Versand"]:
                # Kopfzeile ignorieren
                continue
            if row[0] not in result:
                result[row[0]] = {}
            if row[1] not in result[row[0]]:
                result[row[0]][row[1]] = []
            result[row[0]][row[1]].append(row[2])

            # if row[0] != current_file:
            #     if fwh is not None:
            #         fwh.close()
            #     current_file = row[0]
            #     fwh = open(f"{export_folder}\\{current_file}", "w")

            # fwh.write(row[1] + ";" + row[2] + "\n")
            # fwh.close()
    # json.dump(result, open("export2\\test.json", "w"), indent=2)
    for file, reports in result.items():
        print(file)
        with open(f"{export_folder}\\{file}", "w") as fwh:
            fwh.write("Datei;Empfaenger\n")
            for report, mails in reports.items():
                fwh.write(report + ";" + ",".join(sorted(mails)) + "\n")


def mail_csv_backup(base_dir: str) -> None:
    for f in Path(base_dir).glob("*.csv"):
        # print("* " + f.name)
        copy_file_with_timestamp(str(f), str(f.parent) + "\\backup")


def convert_mail_csv(csv_file):
    pass


def convert_folder_to_combined_csv(path: str):
    print("Sicherung der Versand-Intervall-Dateien")
    mail_csv_backup(path)

    export_csv = path + "\\info\\Versand_kombiniert.csv"
    row_count = convert_csv_in_folder(path, export_csv)
    print(f"Anzahl der Datei-Empfaenger-Kombinationen: {row_count}")
    row_count = count_unique_emails(export_csv)
    print(f"Anzahl der eindeutigen E-Mail-Adressen: {row_count}")
    export_csv2 = path + "\\info\\Versand_kombiniert_komplett.csv"
    convert_csv_in_folder(path, export_csv2, True)


def workflow2():
    convert_combined_to_folder("info\\Versand_kombiniert.csv", "export2")


if __name__ == "__main__":
    convert_folder_to_combined_csv()
    workflow2()