pr3-s2/s2_a3.py

import re
import json
import traceback

index = 0


def parse_zeile(zeile):
    global index
    index = index + 1

    elemente = []
    aktuelles_element = ""
    is_escaped = False
    for c in zeile:
        if c == "\"":
            is_escaped = not is_escaped
        elif c == "," and not is_escaped:
            elemente.append(aktuelles_element)
            aktuelles_element = ""
        else:
            aktuelles_element = aktuelles_element + c

    elemente.append(aktuelles_element)

    name = elemente[0]
    adresse = elemente[1]
    geburtsdatum = elemente[2]
    # rufnummer = elemente[3]

    person = {}

    person['Index'] = index

    # Maurizio Nette
    # Ing. Charlotte Sölzer
    # Dipl.-Ing. Friedericke Hecker
    # Univ.Prof. Siglinde Herrmann
    name_matches = re.search(r'(?:([^ ]+) )?([^ ]+) ([^ ]+)(?: ([^ ]+))?', name)
    person['Titel'] = [name_matches.group(1)]
    person['Vorname'] = [name_matches.group(2)]
    if name_matches.group(4) is None:
        person['Zweitname'] = [None]
        person['Nachname'] = [name_matches.group(3)]
    else:
        person['Zweitname'] = [name_matches.group(3)]
        person['Nachname'] = [name_matches.group(4)]

    # Bärbel-Heydrich-Straße 45,76368 Freital
    # Augustin-Weinhold-Weg 14,05392 Sankt Goarshausen
    # Lorenz-Lindner-Allee 46,61054 Nürtingen
    # Eckard-Sölzer-Gasse 13,68137 Uelzen
    adress_matches = re.search(r'([^0-9]+) ([0-9]+),([0-9]{5,10}) (.+)', adresse)
    person['Straße'] = [adress_matches.group(1)]
    person['Hausnummer'] = [adress_matches.group(2)]
    person['PLZ'] = [adress_matches.group(3)]
    person['Wohnort'] = [adress_matches.group(4)]

    birth_matches = re.search(r'([0-9]{2,4})[ .-]([0-9]{2})[ .-]([0-9]{2,4})', geburtsdatum)
    if len(birth_matches.group(1)) == 4:
        person['Geburtsdatum'] = [birth_matches.group(3) + "." + birth_matches.group(2) + "." + birth_matches.group(1)]
    else:
        person['Geburtsdatum'] = [birth_matches.group(1) + "." + birth_matches.group(2) + "." + birth_matches.group(3)]

    return person


personen = []

with open("Personen.txt", 'r') as file:
    next(file)
    for zeile in file:
        zeile = zeile.replace("\n", "")

        try:
            zeile_parsed = parse_zeile(zeile)
            personen.append(zeile_parsed)
        except:
            print("Fehler in Zeile:", zeile)
            print(traceback.format_exc())

with open("PersonenNeu.json", 'w') as file:
    json.dump(personen, file)