1143 lines
54 KiB
Python
Executable File
1143 lines
54 KiB
Python
Executable File
#!/usr/bin/env python3
|
|
|
|
__version__ = '1.0.0'
|
|
__website__ = 'https://www.fb.gv.at/Funk/amateurfunkdienst.html'
|
|
|
|
import argparse
|
|
import os
|
|
import mariadb
|
|
import sys
|
|
import time
|
|
import pypdf
|
|
from pypdf import PdfReader
|
|
import re # regular expression
|
|
import datetime
|
|
import configparser
|
|
|
|
from selenium import webdriver
|
|
from selenium.webdriver.common.by import By
|
|
from selenium.webdriver.chrome.service import Service as ChromiumService
|
|
|
|
def call_description():
|
|
print(f'Download and Parse the Austrian Callbook Version {__version__}')
|
|
|
|
def call_configparser():
|
|
config = configparser.ConfigParser()
|
|
try:
|
|
configfile = get_active_path('config.ini')
|
|
ret = config.read(configfile)
|
|
if not ret:
|
|
print(f'The file {configfile} was not found. We create a new config file.')
|
|
config['Common']={
|
|
'description': 'Download and Parse the Austrian Callbook',
|
|
'author': 'Thomas Kuschel, OE3TKT (OExTKT)',
|
|
'version': __version__,
|
|
'created': datetime.datetime.now()
|
|
}
|
|
config['MariaDB'] = {
|
|
'user': 'om',
|
|
'password': 'oe3tkt',
|
|
'host': '127.0.0.1',
|
|
'port': 3306,
|
|
'database': 'callbook'}
|
|
with open(get_active_path('config.ini'), 'w') as configfile:
|
|
config.write(configfile, False)
|
|
print(f'Configuration file "config.ini" created.')
|
|
|
|
except FileNotFoundError:
|
|
sys.exit(8)
|
|
except Exception as e:
|
|
print('Error: {}'.format(e), file=sys.stderr)
|
|
sys.exit(9)
|
|
# print(config.sections()) ## ['Common', 'MariaDB']
|
|
return config
|
|
|
|
def call_parser():
|
|
parser = argparse.ArgumentParser(
|
|
description='Download and Parse the Austrian Callbook',
|
|
epilog=f'''
|
|
Written by Thomas Kuschel,
|
|
Version {__version__}
|
|
'''
|
|
)
|
|
parser.add_argument('-i', '--interactive', action='store_true', default=False)
|
|
# parser.add_argument('-s', '--server' default=__website__, required=False)
|
|
parser.add_argument('-V', '--version', action='version', version='{} {}'.format(os.path.split(__file__)[1],__version__))
|
|
parser.add_argument('-v', '--verbose', action='append_const', const = 1)
|
|
# Rufzeichenliste_AT_Stand_010624.pdf
|
|
parser.add_argument('-p', '--path', default='', help= 'skip the download if the specified path to a PDF file exists')
|
|
# parser.add_argument('-t', '--type', default='' , help='specify the output, supported types are [ CSV | JSON ]') # not implemented yet
|
|
parser.add_argument('-o', '--output', default='', help='specify the file where the data are written to, default stdout')
|
|
parser.add_argument('-m', '--mariadb', help='SQL interface to MariaDB (MySql) format "<IP-Address>:<Port> <User> <Passwd>" or defined in .config')
|
|
parser.add_argument('url', metavar='URL', nargs='?', default=__website__)
|
|
|
|
opt = parser.parse_args()
|
|
opt.verbose = 0 if opt.verbose is None else sum(opt.verbose)
|
|
if opt.path != '' and os.path.dirname(opt.path) == '':
|
|
opt.path = get_active_path(opt.path)
|
|
return opt
|
|
|
|
def call_website(url,verbose,path='',interactive=False,output='',mariadb=''):
|
|
if path:
|
|
if os.path.exists(path):
|
|
return path
|
|
else:
|
|
print(f'The given path "{path}" does not exist.')
|
|
sys.exit(3)
|
|
if(interactive):
|
|
print('Interactive')
|
|
driver=webdriver.Chrome()
|
|
else:
|
|
print('Headless Script')
|
|
options = webdriver.ChromeOptions()
|
|
options.add_argument('--headless')
|
|
options.add_argument('--no-sandbox')
|
|
options.add_argument('--disable-dev-shm-usage')
|
|
driver = webdriver.Chrome(options=options)
|
|
|
|
driver.get(url)
|
|
print(driver.title)
|
|
# elements = driver.find_elements(By.XPATH,'//a[contains(@href,"Rufzeichen")]')
|
|
elements = driver.find_elements(By.PARTIAL_LINK_TEXT,"Rufzeichen")
|
|
|
|
if elements:
|
|
element = elements[0]
|
|
href = element.get_attribute('href')
|
|
filename = element.click() # take the first one
|
|
else:
|
|
print('Sorry, no Link containing "Rufzeichen" found.')
|
|
driver.close()
|
|
sys.exit(2)
|
|
|
|
print(element.text)
|
|
# print(href)
|
|
if(interactive):
|
|
time.sleep(300)
|
|
else:
|
|
time.sleep(40)
|
|
|
|
driver.close()
|
|
return os.path.basename(href)
|
|
|
|
def remove_first_quote_if_odd(text, verbose = 0):
|
|
double_quote_cnt = text.count('"')
|
|
# single_quote_cnt = text.count("'")
|
|
|
|
if (double_quote_cnt % 2 != 0): # or (single_quote_cnt % 2 != 0):
|
|
# Find and remove the first quote
|
|
for i, char in enumerate(text):
|
|
if char in ['"']: # ['"', "'"]:
|
|
if (verbose > 1):
|
|
print(text)
|
|
text = text[:i] + text[i+1:]
|
|
if (verbose > 1):
|
|
print(text)
|
|
break
|
|
return text
|
|
|
|
def is_clubstation(call):
|
|
assert(len(call) > 3)
|
|
if call[3].upper() == 'X' or call.upper() == 'OE5SIX': # special case with OE5SIX (Clubstation)
|
|
return True
|
|
|
|
return False
|
|
|
|
def replace_substring_with_line(path, search_substring, verbose=0):
|
|
|
|
try:
|
|
if not replace_substring_with_line.lines:
|
|
with open(path, 'r') as file:
|
|
replace_substring_with_line.lines = file.readlines()
|
|
|
|
for line in replace_substring_with_line.lines:
|
|
if search_substring[0:46].lower() in line.lower():
|
|
return line.strip()
|
|
|
|
except FileNotFoundError:
|
|
print(f'The file {path} was not found.')
|
|
except Exception as e:
|
|
print(f'An error occurred: {e}')
|
|
|
|
return search_substring
|
|
|
|
replace_substring_with_line.lines = None
|
|
|
|
def gender_substring(path, search_substring, verbose=0):
|
|
try:
|
|
if not gender_substring.lines:
|
|
with open(path, 'r') as file:
|
|
gender_substring.lines = file.readlines()
|
|
|
|
for line in gender_substring.lines:
|
|
if line[2:].strip() == search_substring: # search from position 2 and remove all spaces or \n chars
|
|
return line[0] # return the char of gender i.e. 'f' or 'm'
|
|
except FileNotFoundError:
|
|
print(f'The file {path} was not found.')
|
|
except Exception as e:
|
|
print(f'An error occurred: {e}')
|
|
|
|
return 'x' # not found, unknown gender
|
|
|
|
gender_substring.lines = None
|
|
|
|
def get_active_path(file):
|
|
return os.path.join(os.path.dirname(os.path.abspath(__file__)), file)
|
|
|
|
def get_gender(firstnames, surname, call, verbose=0):
|
|
|
|
# load the .gender file:
|
|
genderfile = '.gender'
|
|
gender = 'x'
|
|
gpath = get_active_path(genderfile)
|
|
if os.path.exists(gpath):
|
|
firstname = firstnames.split(' ', 1)[0]
|
|
gender = gender_substring(gpath, firstname, verbose=0) # only check 1st/firstname of name, important when there are more than 1 firstnames
|
|
if gender == 'x':
|
|
if verbose > 0:
|
|
get_gender.cnt += 1
|
|
print(f'({get_gender.cnt}){call} "{firstname}" [{firstnames} {surname}] not found in file {genderfile} - gender "x" is set.')
|
|
|
|
return gender
|
|
|
|
get_gender.cnt = 0
|
|
|
|
def call_split_name(fullname, call, verbose):
|
|
|
|
assert(len(fullname) > 1)
|
|
|
|
name = fullname.split(' ', 1)
|
|
surname = name[0]
|
|
# several special cases like surname "de Lijezer", "van Dijk", "el Shamaa", "da Silva", etc.
|
|
match surname.lower():
|
|
case 'de' | 'el' | 'da':
|
|
name = fullname[3:].split(' ',1)
|
|
surname = surname.lower() + ' ' + name[0]
|
|
if verbose > 1:
|
|
print(f'## {fullname} --> {surname} ##')
|
|
|
|
case 'van' | 'von' :
|
|
name = fullname[4:].split(' ',1)
|
|
surname = surname.lower() + ' ' + name[0]
|
|
if surname.lower() in ['van der', 'von der', 'van den']: # e.g. "van der Meulen", "Walther von der Vogelweide", "Annie van den Berg"
|
|
name = fullname[8:].split(' ',1)
|
|
surname = surname.lower() + ' ' + name[0]
|
|
if verbose > 1:
|
|
print(f'## {fullname} --> {surname} ##')
|
|
case 'della' : # Ancient Italian noble family "della Rowere"
|
|
name = fullname[6:].split(' ',1)
|
|
surname = surname.lower() + ' ' + name[0]
|
|
if verbose > 1:
|
|
print(f'## {fullname} --> {surname} ##')
|
|
case 'senarclens' : # Senarclens de Grancy
|
|
if fullname.lower().startswith('senarclens de grancy'):
|
|
name = fullname[21:].split(' ',1)
|
|
surname = 'Senarclens de Grancy'
|
|
if verbose > 1:
|
|
print(f'## {fullname} --> {surname} ##')
|
|
|
|
if len(name) > 1:
|
|
firstname = name[1].lstrip() # FIX when there are more than 1 space b/w surname and firstname
|
|
else:
|
|
firstname = '<unknown>'
|
|
|
|
# In Austria the call suffix starting with Y is an YL (young lady)
|
|
# if call[3].upper() == 'Y':
|
|
if False:
|
|
gender = 'f'
|
|
else:
|
|
gender = get_gender(firstname, surname, call, verbose)
|
|
|
|
return firstname, surname, gender
|
|
|
|
def fix_call(call,verbose=1):
|
|
fixfile = '.typo_call'
|
|
path = get_active_path(fixfile)
|
|
if os.path.exists(path):
|
|
if not fix_call.lines:
|
|
with open(path,'r') as file:
|
|
fix_call.lines = file.readlines()
|
|
if verbose > 0:
|
|
print(f'File "{fixfile}":')
|
|
for line in fix_call.lines:
|
|
print(f'>> {line.rstrip()}')
|
|
print('>> ** EOF **')
|
|
for line in fix_call.lines[1:]: # starting with line 1
|
|
words = line.rstrip().split()
|
|
if len(words) != 2:
|
|
continue
|
|
if words[0] == call:
|
|
if verbose > 0:
|
|
print(f'Corr typo call: {words[0]} => {words[1]}')
|
|
return words[1]
|
|
|
|
return call
|
|
fix_call.lines = None
|
|
|
|
def fix_typo(call, fullname, verbose=1):
|
|
fixtypofile = '.typo_callbook'
|
|
path = get_active_path(fixtypofile)
|
|
if os.path.exists(path):
|
|
try:
|
|
if not fix_typo.lines:
|
|
with open(path, 'r') as file:
|
|
fix_typo.lines = file.readlines()
|
|
if verbose > 0:
|
|
print(f'File "{fixtypofile}":')
|
|
for line in fix_typo.lines:
|
|
print(f'>> {line.rstrip()}')
|
|
print('>> ** EOF **')
|
|
|
|
for line in fix_typo.lines[4:]: # starting with line 4
|
|
if len(fix_typo.spaces) == 0: # not initialized
|
|
if line[0] == '*':
|
|
words = line.split()
|
|
assert len(words) == 4 # i.e. '*, call, nachname, vorname'
|
|
fix_typo.spaces = [line.index(words[1]), line.index(words[2]), line.index(words[3])]
|
|
else:
|
|
if call in line[2:8]:
|
|
if verbose > 1:
|
|
print(f'Call: {call} found')
|
|
match line[0]:
|
|
case '#':
|
|
if verbose > 1:
|
|
print(line.rstrip())
|
|
case 'F' | 'N':
|
|
if verbose > 0:
|
|
print(line.rstrip())
|
|
print(fullname)
|
|
firstname1, surname1, gender1 = call_split_name(fullname, call, 0)
|
|
fullname2 = line[fix_typo.spaces[1]:fix_typo.spaces[2]-1].rstrip() + ' ' + line[fix_typo.spaces[2]:-1]
|
|
if verbose > 0:
|
|
print(fullname2)
|
|
firstname2, surname2, gender2 = call_split_name(fullname2, call, 0)
|
|
# Hardening: at a minimum, either the firstnames or the surenames must fit
|
|
fix_cnt = 0
|
|
if (firstname1 != firstname2):
|
|
fix_cnt += 1
|
|
if (surname1 != surname2):
|
|
fix_cnt += 1
|
|
# when the surname is splitted and wrong written:
|
|
if line[0] == 'N' and firstname2 in fullname:
|
|
if fix_cnt > 1: fix_cnt = 1
|
|
if fix_cnt == 0:
|
|
print(f'It is fixed! You can remove the line with the item {call} from the file {fixtypofile}!')
|
|
elif fix_cnt > 1:
|
|
print(f'Something went wrong, there are several bugs. Check line with call {call} in file {fixtypofile}!')
|
|
else:
|
|
fullname = fullname2
|
|
case 'X': # exchange the surname with firstname
|
|
if verbose > 0:
|
|
print(line.rstrip())
|
|
print(fullname)
|
|
firstname1, surname1, gender1 = call_split_name(fullname, call, 0)
|
|
fullname2 = line[fix_typo.spaces[1]:fix_typo.spaces[2]-1].rstrip() + ' ' + line[fix_typo.spaces[2]:-1]
|
|
if verbose > 0:
|
|
print(fullname2)
|
|
firstname2, surname2, gender2 = call_split_name(fullname2, call, 0)
|
|
fix_cnt = 0
|
|
if (firstname1 == firstname2) and (surname1 == surname2):
|
|
print(f'It is fixed! You can remove the line with the item {call} from the file {fixtypofile}!')
|
|
elif (firstname1 != surname2) or (surname1 != firstname2):
|
|
print(f'Something went wrong, there are several bugs. Check line with call {call} in file {fixtypofile}!')
|
|
else:
|
|
fullname = fullname2
|
|
|
|
except FileNotFoundError:
|
|
print(f'The file {path} was not found.')
|
|
except Exception as e:
|
|
print(f'An error occurred: {e}')
|
|
|
|
return fullname
|
|
fix_typo.lines = None
|
|
fix_typo.spaces = []
|
|
|
|
def isinteger(s):
|
|
for char in s:
|
|
if not char.isdigit():
|
|
return False
|
|
return True
|
|
|
|
def call_change_city(location, address, verbose):
|
|
post_file = '.post_code'
|
|
path = get_active_path(post_file)
|
|
if not call_change_city.lines:
|
|
if os.path.exists(path):
|
|
try:
|
|
with open(path, 'r') as file:
|
|
call_change_city.lines = file.readlines()
|
|
if verbose > 0:
|
|
print(f'File "{post_file}":')
|
|
for line in call_change_city.lines:
|
|
print(f'>> {line.rstrip()}')
|
|
print('>> ** EOF **')
|
|
# print(f'{path}')
|
|
except FileNotFoundError:
|
|
print(f'The file {path} was not found.')
|
|
except Exception as e:
|
|
print(f'An error occurred: {e}')
|
|
else:
|
|
return 0,'',''
|
|
|
|
postal_code = 0
|
|
city = location
|
|
addr,city = corr_street(address,city)
|
|
street = addr
|
|
|
|
for line in call_change_city.lines:
|
|
line = line.rstrip() # remove the \n
|
|
locpart = line.split(' ',1)
|
|
if len(locpart) < 2:
|
|
continue
|
|
exchange = locpart[1].split('#')
|
|
citypart = exchange[0].split('|',1)
|
|
if (citypart[0] != location): # A hit? Else...
|
|
continue
|
|
if len(citypart) > 1:
|
|
# addr = address[0:len(citypart[1])]
|
|
if (citypart[1] != addr[0:len(citypart[1])]):
|
|
continue
|
|
postal_code = locpart[0]
|
|
if len(exchange) > 1:
|
|
city = exchange[1]
|
|
if len(exchange) > 2 and len(citypart) > 1:
|
|
street = addr.replace(citypart[1],exchange[2])
|
|
# print(f'PLZ: {postal_code}, City: {city}')
|
|
break
|
|
if postal_code != 0:
|
|
if verbose > 0:
|
|
if address == addr:
|
|
print(f'location/address: {location}|{addr} => {postal_code} {city}|{street}')
|
|
else:
|
|
print(f'location/address: {location}|({address}){addr} => {postal_code} {city}|{street}')
|
|
else:
|
|
call_change_city.cnt += 1
|
|
if address == addr:
|
|
print(f'[{call_change_city.cnt}] loc/addr: {location}|{addr} => {postal_code} {city}|{street}')
|
|
else:
|
|
print(f'[{call_change_city.cnt}] loc/addr: {location}|({address}){addr} => {postal_code} {city}|{street}')
|
|
|
|
return postal_code, city, street
|
|
|
|
call_change_city.lines = None
|
|
call_change_city.cnt = 0
|
|
|
|
def corr_street(street,city):
|
|
street = street.replace('strasse','straße')
|
|
street = street.replace('Strasse','Straße')
|
|
street = street.replace('str ','straße ')
|
|
street = street.replace('Str ','Straße ') # Brünner Str 221/4/5
|
|
street = street.replace('str.','straße')
|
|
street = street.replace('Stg. ','Stiege ')
|
|
street = street.replace('g. ','gasse ')
|
|
street = street.replace('-G ','-Gasse ')
|
|
street = street.replace('Stiege ','Stg. ')
|
|
street = street.replace('Dr. ','Dr.-')
|
|
# Special, stupid fixes
|
|
street = street.replace('Str.1','Straße 1') # Peter-Berner-Str.10/1/10
|
|
street = street.replace('Str.5','Straße 5') # Wagramer Str.59/14/2
|
|
street = street.replace('Str.6','Straße 6')
|
|
street = street.replace('Str.8','Straße 8')
|
|
street = street.replace('Str.','Straße') # Brünner Str. 106/43
|
|
street = street.replace('Str7','Straße 7') # Hetzendorfer Str79/2/14
|
|
street = street.replace('Str8','Straße 8') # Theodor-Kramer-Str8/274
|
|
street = street.replace('Markhofg.13-15/','Markhofgasse 13-15/5/14') # Mautner-Markhofg.13-15/ (OE1JJB)
|
|
street = street.replace('g.2','gasse 2') # Spittelbreiteng.23/7/2/
|
|
street = street.replace('g.3','gasse 3') # Rathausg.32, Wulkaprodersdorf
|
|
street = street.replace('g.4','gasse 4') # Siebenbrunneng.42/1/18
|
|
street = street.replace('Absbergg ','Absberggasse ')
|
|
street = street.replace('Alfred Huthg ','Alfred-Huth-Gasse ')
|
|
street = street.replace('Altmannsdfstraße ','Altmannsdorfer Straße ') # Altmannsdfstr 170 23/7
|
|
street = street.replace('Amthofensiedlung1','Amthofensiedlung 1')
|
|
street = street.replace('Anningerg ','Anningergasse ')
|
|
street = street.replace('Antonig ','Antonigasse ') # 7063 Oggau am Neusiedler See
|
|
street = street.replace('Arbeiterg ','Arbeitergasse ')
|
|
street = street.replace('Arnoldg ','Arnoldgasse ')
|
|
street = street.replace('Bellinig ','Bellinigasse ')
|
|
street = street.replace('Birkeng ','Birkengasse ')
|
|
street = street.replace('Bucheng ','Buchengasse ')
|
|
street = street.replace('Canisiusg ','Canisiusgasse ')
|
|
street = street.replace('Czartoryskig ','Czartoryskigasse ')
|
|
street = street.replace('Denglerg ','Denglergasse ')
|
|
street = street.replace('Endemanng ','Endemanngasse ')
|
|
street = street.replace('Ernst Ludwigg ','Ernst-Ludwig-Gasse ')
|
|
street = street.replace('Feldg ','Feldgasse ')
|
|
street = street.replace('Flachg ','Flachgasse ')
|
|
street = street.replace('Gatterburgg ','Gatterburggasse ')
|
|
street = street.replace('Georg Sigl G ','Georg-Sigl-Gasse ')
|
|
street = street.replace('Goldegg ','Goldeggasse ')
|
|
street = street.replace('Gregoryg ','Gregorygasse ')
|
|
street = street.replace('Grundlg ','Grundlgasse ')
|
|
street = street.replace('Gusenleithnerg ','Gusenleithnergasse ')
|
|
street = street.replace('Sendemast Hegerber','Sendemast Hegerberg')
|
|
street = street.replace('Hermanng ','Hermanngasse ') # 7, Neubau, Wien
|
|
street = street.replace('Hutteng ','Huttengasse ')
|
|
street = street.replace('Isbaryg ','Isbarygasse ')
|
|
street = street.replace('Johannesg ','Johannesgasse ')
|
|
street = street.replace('Karl Stummerg ','Karl-Stummer-Gasse ') # Wieselburg
|
|
street = street.replace('Kaunitzg ','Kaunitzgasse ')
|
|
street = street.replace('Keferg ','Kefergasse ')
|
|
street = street.replace('Keinerg ','Keinergasse ')
|
|
street = street.replace('Khekg ','Khekgasse ')
|
|
street = street.replace('Lacknerg ','Lacknergasse ')
|
|
street = street.replace('Langsulzg ','Langsulzgasse ')
|
|
street = street.replace('Lavantg ','Lavantgasse')
|
|
street = street.replace('Leiteng ','Leitengasse ')
|
|
street = street.replace('Leopoldig ','Leopoldigasse ')
|
|
street = street.replace('Lichtensterng ','Lichtensterngasse ')
|
|
street = street.replace('Linseng ','Linsengasse ')
|
|
street = street.replace('Markowskyg ','Markowskygasse ')
|
|
street = street.replace('Mölkerg ','Mölkergasse ')
|
|
street = street.replace('Mommseng ','Mommsengasse ')
|
|
street = street.replace('Neilreichg ','Neilreichgasse ')
|
|
street = street.replace(' Ors Sendeanlage',' ORS-Sendeanlage')
|
|
street = street.replace('Orf Gebäude','ORF-Gebäude')
|
|
street = street.replace('Othellog ','Othellogasse ')
|
|
street = street.replace('Pehamg ','Pehamgasse ')
|
|
street = street.replace('Pfeilg ','Pfeilgasse ')
|
|
street = street.replace('Pfenninggeldg ','Pfenninggeldgasse ')
|
|
street = street.replace('Pohlg ','Pohlgasse ')
|
|
street = street.replace('Reinprechtsdf ','Reinprechtsdorfer ')
|
|
street = street.replace('Ringelseeg ','Ringelseegasse ')
|
|
street = street.replace('Robert-Hamerlingg ','Robert-Hamerling-Gasse ')
|
|
street = street.replace('Romanog ','Romanogasse ')
|
|
street = street.replace('Rombergg ','Romberggasse ')
|
|
street = street.replace('Roseng ','Rosengasse ')
|
|
street = street.replace('Rötzerg ','Rötzergasse ')
|
|
street = street.replace('Ruzickag ','Ruzickagasse ')
|
|
street = street.replace('Schützpl ','Schützplatz ')
|
|
street = street.replace('Schiffmühlenst ','Schiffmühlenstraße ')
|
|
street = street.replace('Schirrmanng ','Schirrmanngasse ')
|
|
street = street.replace('Schlussnig ','Schlußnig ')
|
|
street = street.replace('Seutterg ','Seuttergasse ')
|
|
street = street.replace('Siedlerg ','Siedlergasse ')
|
|
street = street.replace('Sigmund-Haffnerg ','Sigmund-Haffner-Gasse ')
|
|
street = street.replace('Sillerg ','Sillergasse ')
|
|
street = street.replace('Speckbacherg ','Speckbachergasse ')
|
|
street = street.replace('Spengerg ','Spengergasse ')
|
|
street = street.replace('Starhembergg ','Starhemberggasse ')
|
|
street = street.replace('Stolbergg ','Stolberggasse ')
|
|
street = street.replace('Straßmeyerg ','Straßmeyergasse ')
|
|
street = street.replace('Zanaschkag ','Zanaschkagasse ')
|
|
street = street.replace('Zentag ','Zentagasse ')
|
|
street = street.replace('aße1','aße 1')
|
|
street = street.replace('aße2','aße 2') #Ziegelhofstrasse29/16
|
|
street = street.replace('aße3','aße 3')
|
|
street = street.replace('aße5','aße 5') #Josefstädterstraße56/2
|
|
street = street.replace('aße6','aße 6')
|
|
street = street.replace('aße8','aße 8')
|
|
street = street.replace('aße9','aße 9')
|
|
street = street.replace('dorf3','dorf 3') # Füllersdorf 37
|
|
street = street.replace('An Der Lan Straße','An-der-Lan-Straße')
|
|
street = street.replace('An-Der-','An-der-') # An-der-Lan-Straße
|
|
street = street.replace('An Der Furt','An-der-Furt')
|
|
street = street.replace('An Der Leiten','An der Leiten') # 6200 Jenbach
|
|
street = street.replace('Auf Der ','Auf der ')
|
|
#street = street.replace('Auf Der Kohlwiese','Auf der Kohlwiese') # Walding
|
|
street = street.replace(' (Klg Am Am', ' (Klg Am Ameisbach)')
|
|
# Typos
|
|
street = street.replace('Achtungvierzigerplatz ','Achtundvierzigerplatz ')
|
|
street = street.replace('Aflinger Straße','Aflingerstraße')
|
|
street = street.replace('Anton-Bruckner- Straße','Anton-Bruckner-Straße')
|
|
street = street.replace('Arsenal Objet ','Arsenal Objekt ')
|
|
street = street.replace('Basler G ','Basler Gasse ')
|
|
street = street.replace('Bierhaeuselberggasse ','Bierhäuselberggasse ')
|
|
street = street.replace('Breitenfrt ','Breitenfurter ')
|
|
street = street.replace('Brigitt.-L148-154/12/17','Brigittenauer Lände 148-154/12/17')
|
|
street = street.replace('Brigitt L ','Brigittenauer Lände ')
|
|
street = street.replace('Dr.Sumpfstraße ','Dr.-Stumpf-Straße ')
|
|
street = street.replace('Fesstgasse ','Feßtgasse ')
|
|
street = street.replace('Fuchsenfeld Gasse ','Fuchsenfeldgasse')
|
|
street = street.replace('Gußriegelstraße','Gussriegelstraße') # seit 1999-12-17 Gemeinderatsbeschluss Wien
|
|
street = street.replace('Hardegggasse','Hardeggasse')
|
|
street = street.replace('Hoehenrainweg ','Höhenrainweg ') # Ehrwald
|
|
street = street.replace('Hoeggen ','Höggen ')
|
|
street = street.replace('Hoergas ','Hörgas ')
|
|
street = street.replace('H.-Collin-Straße','Heinrich-Collin-Straße')
|
|
street = street.replace('Hptst ','Hauptstraße ')
|
|
street = street.replace('Hernalser Hptstraße ','Hernalser Hauptstraße ') # Hernalser Hptstr.230/14
|
|
street = street.replace('Hütteldfstr1','Hütteldorfer Straße 1') # Hütteldfstr180-186/5/29
|
|
street = street.replace('In Der ','In der ')
|
|
street = street.replace('Iz Nö-Süd Straße','IZ NÖ-Süd Straße')
|
|
street = street.replace('Iz Nö-Süd, Straße','IZ NÖ-Süd Straße')
|
|
street = street.replace(', Obj. ',', Objekt ')
|
|
street = street.replace(' Obj. ',', Objekt ')
|
|
street = street.replace('Jaegerstraße ','Jägerstraße ')
|
|
street = street.replace('Kaiserebersdorferst ','Kaiserebersdorfer Straße ')
|
|
street = street.replace('Kaiser Ebersd.Straße','Kaiserebersdorfer Straße')
|
|
street = street.replace('Koepfleweg','Köpfleweg') # Riezlern
|
|
street = street.replace('Kuernbergblick','Kürnbergblick')
|
|
street = street.replace('Loewenthalgasse ','Löwenthalgasse ')
|
|
street = street.replace('Loeblichgasse','Löblichgasse')
|
|
if (city=='Graz'):
|
|
street = street.replace('Maderspergerstraße','Maderspergergasse')
|
|
street = street.replace('Neulerchfd.Straße','Neulerchenfelder Straße')
|
|
street = street.replace('Neustift Am Walde','Neustift am Walde')
|
|
street = street.replace('Nussdorfer','Nußdorfer')
|
|
street = street.replace('Oberweissenbach','Oberweißenbach')
|
|
street = street.replace('Oehlermuehle ','Öhlermühle ')
|
|
street = street.replace('Pernersdorfergasse','Pernerstorfergasse') # Wien 10
|
|
street = street.replace('Puechlgasse','Püchlgasse')
|
|
street = street.replace('Rossauer Lände','Roßauer Lände')
|
|
street = street.replace('Rossbachstraße','Roßbachstraße')
|
|
street = street.replace('Russbergstraße','Rußbergstraße')
|
|
street = street.replace('Salzlaende ','Salzlände ')
|
|
street = street.replace('Schieszstandstraße','Schießstandstraße') # 6200 Jenbach
|
|
street = street.replace('Schliessmanngasse ','Schließmanngasse ')
|
|
street = street.replace('Schlossmuehlgasse ','Schloßmühlgasse ') # Wiener Neudorf
|
|
street = street.replace('Schluesselamtsgasse ','Schlüsselamtsgasse ') # Krems
|
|
if (city == 'Innsbruck'):
|
|
street = street.replace('Schmidgasse','Schmiedgasse')
|
|
street = street.replace('Schuchardstraße','Schuchardtstraße')
|
|
street = street.replace('Schuetzenstraße','Schützenstraße')
|
|
street = street.replace('Simmering/Blockkraftw','Simmering Blockkraftwerk')
|
|
street = street.replace('Spesingerstraße ','Speisingerstraße') # typo in Wien
|
|
street = street.replace('St. Hubertusgasse ','St.-Hubertus-Gasse ')
|
|
street = street.replace('St.Michael-G.1','St.-Michael-Gasse 1')
|
|
street = street.replace('St. Veit-Gasse','St.-Veit-Gasse') # 1130 Wien
|
|
street = street.replace('Stoegergasse','Stögergasse') # 2000 Stockerau
|
|
street = street.replace('Sturbergasse','Strubergasse') # Sturbergasse 60, Salzburg
|
|
street = street.replace('Toescherweg','Töscherweg') # 8144 Tobelbad
|
|
street = street.replace('Universitaetsstraße','Universitätsstraße') # 6020 Innsbruck
|
|
street = street.replace('Voegelebichl','Vögelebichl') # 6020 Innsbruck
|
|
street = street.replace('Waelderstraße','Wälderstraße') # 6900 Bregenz
|
|
street = street.replace('Weissenbachgraben ','Weißenbachgraben ') # Waidhofen an der Ybbs
|
|
if (street == 'Rum,Serlesstraße 22'):
|
|
city = 'Rum'
|
|
street = 'Serlesstraße 22'
|
|
# Spelling VIPs
|
|
street = street.replace('10. Oktober Straße','10.-Oktober-Straße')
|
|
street = street.replace('12.Februarstraße','12.-Februar-Straße')
|
|
street = street.replace('12. Novemberstraße','12.-November-Straße')
|
|
street = street.replace('Ada Christengasse','Ada-Christen-Gasse')
|
|
street = street.replace('Adalbert Stifter Straße','Adalbert-Stifter-Straße')
|
|
street = street.replace('Adolf Netsch Straße','Adolf-Netsch-Straße')
|
|
street = street.replace('Alexander Groß-Gasse','Alexander-Groß-Gasse')
|
|
street = street.replace('Alfons Petzoldgasse','Alfons-Petzold-Gasse')
|
|
street = street.replace('Alfred Jägerweg','Alfred-Jäger-Weg')
|
|
street = street.replace('Alois Kellner Straße','Alois-Keller-Straße')
|
|
street = street.replace('Am Langen Felde','Am langen Felde')
|
|
street = street.replace('Am Muellerbach','Am Müllerbach')
|
|
street = street.replace('Andreas Grein Gasse','Andreas-Grein-Gasse')
|
|
street = street.replace('Andreas Hoferstraße','Andreas-Hofer-Straße') # Landeck
|
|
street = street.replace('Andreas Urteil Weg','Andreas-Urteil-Weg')
|
|
street = street.replace('Anton Baumgartner Straße','Anton-Baumgartner-Straße')
|
|
street = street.replace('Anton Baumgartnerstraße','Anton-Baumgartner-Straße')
|
|
street = street.replace('Anton-Baumgartnerstraße','Anton-Baumgartner-Straße')
|
|
street = street.replace('A Baumgartnerstraße','Anton-Baumgartner-Straße')
|
|
street = street.replace('A. Baumgartnerstraße','Anton-Baumgartner-Straße')
|
|
street = street.replace('Anton Benya Straße','Anton-Benya-Straße')
|
|
street = street.replace('Anton Böck Gasse','Anton-Böck-Gasse')
|
|
street = street.replace('Anton Brucknergasse','Anton-Bruckner-Gasse')
|
|
street = street.replace('Anton Bruckner-Straße','Anton-Bruckner-Straße') # 6300 Wörgl
|
|
street = street.replace('Anton-Ehrenfriedstraße','Anton-Ehrenfried-Straße') # 2020 Hollabrunn
|
|
street = street.replace('Anton Haglg','Anton-Hagl-Gasse') # Gablitz
|
|
street = street.replace('Anton-Hobbinger-Gasse','Anton-Hobbiger-Gasse') # Heidenreichstein
|
|
street = street.replace('Anton Keller Weg','Anton-Keller-Weg')
|
|
street = street.replace('Anton Kriegergasse','Anton-Krieger-Gasse')
|
|
street = street.replace('Anton Mühlbacherstraße','Anton-Mühlbacher-Straße')
|
|
street = street.replace('Anton Regner-Straße','Anton-Regner-Straße')
|
|
street = street.replace('Anton Ullrich Gasse','Anton-Ullrich-Gasse') # Heidenreichstein
|
|
street = street.replace('A-Weiser Straße','Anton-Weiser-Straße')
|
|
street = street.replace('A.W. Pragergasse','Anton-Wenzel-Prager-Gasse')
|
|
street = street.replace('Baron Karl Gasse','Baron-Karl-Gasse')
|
|
street = street.replace('Bertha V Suttnerstraße','Bertha-von-Suttner-Straße')
|
|
street = street.replace('Bgm. Dr.-Haberl Gasse','Bgm.-Dr.-Haberl-Gasse') # Wiener Neustadt
|
|
street = street.replace('Bischof Faberplatz','Bischof-Faber-Platz')
|
|
street = street.replace('Brunner-Lehensteingasse','Brunner-Lehenstein-Gasse')
|
|
street = street.replace('Camillo - Schulzstraße','Camillo-Schulz-Straße') # 4600 Wels
|
|
street = street.replace('C Holzmeisterstraße','Clemens-Holzmeister-Straße')
|
|
street = street.replace('Carl Maria Von Weberg ','Carl-Maria-von-Weber-Gasse ')
|
|
street = street.replace('C.V.Hoetzendorfstraße','Conrad-von-Hötzendorf-Straße')
|
|
street = street.replace('Ch. Gudenus Straße','Ch.-Gudenus-Straße') # Hausleiten, eig. Christoph von Gudenus 1688-1800
|
|
street = street.replace('Christoph Gudenus Straße','Ch.-Gudenus-Straße') # auch Christoph-Gudenus-Straße
|
|
street = street.replace('Dr A Lemisch-Straße','Dr.-Arthur-Lemisch-Straße')
|
|
street = street.replace('Dechant Pfeifer Straße','Dechant-Pfeifer-Straße')
|
|
street = street.replace('Dechant Pfeiffer Straße','Dechant-Pfeifer-Straße')
|
|
street = street.replace('Dr.-Ambros Giner Weg','Dr.-Ambros-Giner-Weg') # 6065 Thaur
|
|
street = street.replace('Dr.-Eugen Zehme Straße','Dr.-Eugen-Zehme-Straße') # 5111 Bürmoos
|
|
street = street.replace('Dr Fuchs-Gasse','Doktor-Fuchs-Gasse') # 2000 Stockerau
|
|
street = street.replace('Dr.-Habermayergasse','Doktor-Habermayer-Gasse') # 2700 Wiener Neustadt
|
|
street = street.replace('Dr.Ignaz Weber G.','Dr.-Ignaz-Weber-Gasse ') # 2353 Guntramsdorf
|
|
street = street.replace('Dr Karl Rennerstraße','Doktor-Karl-Renner-Straße')
|
|
street = street.replace('Dr.-Karl Schrattenbauerweg','Dr.-Karl-Schrattenbauer-Weg')
|
|
street = street.replace('Dr Kucherweg','Dr.-Kucher-Weg') # 9061 Klagenfurt
|
|
street = street.replace('Dr.-Pascher Straße','Dr.-Pascher-Straße') # 5280 Braunau am Inn
|
|
street = street.replace('Dr.-Robert Griedl Weg','Dr.-Robert-Griedl-Weg') # 8301 Laßnitzhöhe
|
|
street = street.replace('Dr.R.Schuh-Straße','Dr.-Rudolf-Schuh-Straße')
|
|
street = street.replace('Dr.Stumpf Straße','Dr.-Stumpf-Straße')
|
|
street = street.replace('Dr.-Theodor-Körner Straße','Dr.-Theodor-Körner-Straße') # 9523 Landskron
|
|
street = street.replace('Dr.-Viktor Adler Straße','Dr.-Viktor-Adler-Straße') # 2000 Stockerau
|
|
street = street.replace('Eduard-Kittenbergergasse','Eduard-Kittenberger-Gasse')
|
|
street = street.replace('Eduard Pötzl Gasse','Eduard-Pötzl-Gasse')
|
|
street = street.replace('Erich Friedstraße','Erich-Fried-Straße')
|
|
street = street.replace('Erne-Seder Gasse','Erne-Seder-Gasse')
|
|
street = street.replace('Ernst Ludwig Gasse','Ernst-Ludwig-Gasse')
|
|
street = street.replace('Ernst Mach Straße','Ernst-Mach-Straße')
|
|
street = street.replace('Erzherzog Eugen Straße','Erzherzog-Eugen-Straße')
|
|
street = street.replace('Erzherzog Eugenstraße','Erzherzog-Eugen-Straße')
|
|
street = street.replace('Erzherzog Karl Straße','Erzherzog-Karl-Straße')
|
|
street = street.replace('Feldmarschall Conrad Platz','Feldmarschall-Conrad-Platz') # Klagenfurt
|
|
street = street.replace('F-Asenbauer-Gasse','Franz-Asenbauer-Gasse')
|
|
street = street.replace('Franz Lehar-Gasse','Franz-Lehar-Gasse')
|
|
street = street.replace('Fritz Preglgasse','Fritz-Pregl-Gasse')
|
|
street = street.replace('Franz Asenbauergasse','Franz-Asenbauer-Gasse')
|
|
street = street.replace('Franz Asenbauer-Gasse','Franz-Asenbauer-Gasse')
|
|
street = street.replace('Franz Blasser-Straße','Franz-Plasser-Straße') # 5280 Braunau am Inn
|
|
street = street.replace('Franz Fischerstraße','Franz-Fischer-Straße') # 6020 Innsbruck
|
|
street = street.replace('Franz-Jaenkl-Straße','Franz-Jänkl-Straße')
|
|
street = street.replace('Frau Hitt Straße','Frau-Hitt-Straße')
|
|
street = street.replace('F.J. Weizeneggerweg','Franz-Josef-Weizenegger-Weg')
|
|
street = street.replace('Fred Zinnemannplatz','Franz-Zinnemann-Platz')
|
|
street = street.replace('Fridtjof-Nansenstraße','Fridtjof-Nansen-Straße') # 9800 Spittal an der Drau
|
|
street = street.replace('Friedrich Attlmayr Straße','Friedrich-Attlmayr-Straße') # 6600 Reutte
|
|
street = street.replace('Friedrich Hegel-Gasse','Friedrich-Hegel-Gasse') # 2232 Deutsch-Wagram
|
|
street = street.replace('Friedr.Kaiser-Gasse','Friedrich-Kaiser-Gasse')
|
|
street = street.replace('Friedrich Schiller Straße','Friedrich-Schiller-Straße') # 2340 Mödling
|
|
street = street.replace('Friedrich Schmidt Platz','Friedrich-Schmidt-Platz')
|
|
street = street.replace('Friedrich Teller Gasse','Friedrich-Teller-Gasse')
|
|
street = street.replace('Georg Bilgeri Straße','Georg-Bilgeri-Straße')
|
|
street = street.replace('Georg Lora Straße','Georg-Lora-Straße')
|
|
street = street.replace('G.-Grinninger-Straße','Georg-Grinninger-Straße')
|
|
street = street.replace('Gerhart Ellertstraße','Gerhart-Ellert-Straße') # 9400 Wolfsberg
|
|
street = street.replace('Hans Untermüller Straße','Hans-Untermüller-Straße') # 6020 Innsbruck
|
|
street = street.replace('H.V.Hofmannst.Weg','Hugo-von-Hofmannsthal-Weg')
|
|
street = street.replace('Heiligenstaedter Straße','Heiligenstädter Straße')
|
|
street = street.replace('Heinrich Leflergasse','Heinrich-Lefler-Gasse')
|
|
street = street.replace('Herzmanovsky Orlandogasse','Herzmanovsky-Orlando-Gasse')
|
|
street = street.replace('Ing Julius Raab-Straße','Ing.-Julius-Raab-Straße')
|
|
street = street.replace('Ingenuin-Fischlerstraße','Ingenuin-Fischler-Straße')
|
|
street = street.replace('Jakob-Gschielgasse','Jakob-Gschiel-Gasse') # 8052 Graz
|
|
street = street.replace('Jakob Wendlergasse','Jakob-Wendler-Gasse') # 8350 Fehring
|
|
street = street.replace('Jakob Sereiniggstraße','Jakob-Sereinigg-Straße')
|
|
street = street.replace('J.-F.-Perkonig-Straße','Josef-Friedrich-Perkonig-Straße') # 9170 Ferlach
|
|
street = street.replace('J.Marbergerstraße','Josef-Marberger-Straße') # 6424 Silz
|
|
street = street.replace('J. Fux Straße','Johann-Fux-Straße')
|
|
street = street.replace('J.Haagnstraße','Julius-Haagn-Straße')
|
|
street = street.replace('Johann Böhmstraße','Johann-Böhm-Straße')
|
|
street = street.replace('Joh.Kopfmüllerstraße','Johann-Kopfmüller-Straße') # 5580 Tamsweg
|
|
street = street.replace('Johann Neumayer Gasse','Johann-Neumayer-Gasse')
|
|
street = street.replace('J. Schmidtstraße','Johann-Schmidt-Straße') # 3512 Mautern an der Donau
|
|
street = street.replace('Joh.Schmidtstraße','Johann-Schmidt-Straße') # 3512 Mautern an der Donau
|
|
street = street.replace('Johann Weber Straße','Johann-Weber-Straße')
|
|
street = street.replace('J.W.Kleinstraße','Johann-Wilhelm-Klein-Straße')
|
|
street = street.replace('Johannes Gutenberg Straße','Johannes-Gutenberg-Straße') # Wiener Neustadt
|
|
street = street.replace('Josef-Prechtlweg','Josef-Prechtl-Weg') # Braunau am Inn
|
|
street = street.replace('Josef Schwarz Straße','Josef-Schwarz-Straße')
|
|
street = street.replace('Josef Sirowy-Straße','Josef-Sirowy-Straße')
|
|
street = street.replace('Kaiserschuetzenstraße','Kaiserschützenstraße')
|
|
street = street.replace('Kaiser Max Straße','Kaiser-Max-Straße') # 6060 Hall in Tirol
|
|
street = street.replace('K.Billerhartstraße','Karl-Billerhart-Straße')
|
|
street = street.replace('Karl Schwarz Straße','Karl-Schwarz-Straße')
|
|
street = street.replace('Karl Steiger-Straße','Karl-Steiger-Straße')
|
|
street = street.replace('Konr.Hopferwieserg.','Konrad-Hopferwieser-Gasse ')
|
|
street = street.replace('Kuno-Brandauerstraße','Kuno-Brandauer-Straße')
|
|
street = street.replace('Laaer Berg Straße','Laaer-Berg-Straße')
|
|
street = street.replace('Laaerbergstraße','Laaer-Berg-Straße')
|
|
street = street.replace('Laaerberg Straße','Laaer-Berg-Straße')
|
|
if (city == 'Sollenau'):
|
|
street = street.replace('Gartst.M.The.Laercheng','Lärchengasse 5') # qrz.com
|
|
city = 'Gartenstadt Maria Theresia'
|
|
street = street.replace('Leonard Bernstein Straße','Leonard-Bernstein-Straße')
|
|
street = street.replace('L. Kunschakgasse','Leopold-Kunschak-Gasse') # 2225 Zistersdorf
|
|
street = street.replace('Ludo Hartmanngasse','Ludo-Hartmann-Gasse')
|
|
street = street.replace('Ludwig Laab Straße','Ludwig-Laab-Straße') # 2000 Stockerau
|
|
street = street.replace('Margaretengtl','Margaretengürtel')
|
|
street = street.replace('Mautner-Markhofgasse','Mautner-Markhof-Gasse')
|
|
street = street.replace('Maurer Lange-Gasse','Maurer Lange Gasse')
|
|
street = street.replace('Mayr-Melnhofstraße','Mayr-Melnhof-Straße') # 8700 Leoben
|
|
if (city == 'Bregenz'):
|
|
street = street.replace('Michael Gaismeierstraße','Michael-Gaismayr-Straße') # typo # Bregenz, in Innsbruck Gaismair
|
|
street = street.replace('Michl Felder Straße','Michl-Felder-Straße')
|
|
street = street.replace('Michl Felderstraße','Michl-Felder-Straße')
|
|
street = street.replace('M Unterlercher-Straße','Michael-Unterlercher-Straße')
|
|
street = street.replace('M Wachberger Straße','Michael-Wachberger-Straße')
|
|
if (city == 'Kapfenberg'):
|
|
street = street.replace('Mühlbacherstraße','Anton-Mühlbacher-Straße')
|
|
street = street.replace('Nico Dostal Straße','Nico-Dostal-Straße')
|
|
if (city == 'Braunau am Inn'):
|
|
street = street.replace('Pascherstraße','Dr.-Pascher-Straße')
|
|
street = street.replace('Nikolaus Pfeifaufstraße','Nikolaus-Pfeifauf-Straße') # 6200 Jenbach
|
|
street = street.replace('Per Albin Hanssonstraße','Per-Albin-Hansson-Straße')
|
|
street = street.replace('Peter Altenberg-Gasse','Peter-Altenberg-Gasse')
|
|
street = street.replace('Peter Anich Straße','Peter-Anich-Straße') # 6300 Wörgl
|
|
street = street.replace('Peter Henleinweg','Peter-Henlein-Weg')
|
|
street = street.replace('Peter Jordanstraße','Peter-Jordan-Straße')
|
|
street = street.replace('Peter Pfenningerstraße','Peter-Pfenninger-Straße')
|
|
street = street.replace('Pfarrer Matz Gasse','Pfarrer-Matz-Gasse')
|
|
street = street.replace('Prinz Eugen Straße','Prinz-Eugen-Straße')
|
|
street = street.replace('Prinz Eugenstraße','Prinz-Eugen-Straße')
|
|
street = street.replace('Raaberbahngasse','Raaber-Bahn-Gasse')
|
|
street = street.replace('Richard Straussgasse','Richard-Strauss-Gasse')
|
|
street = street.replace('Richard Wagnergasse','Richard-Wagner-Gasse')
|
|
street = street.replace('Robert Ahlfeld Straße','Robert-Ahlfeld-Straße') # 2000 Stockerau
|
|
street = street.replace('Robert-Häuserstraße','Robert-Häuser-Straße') # 2544 Leobersdorf/Siebenhaus
|
|
street = street.replace('Robert Kochstraße','Robert-Koch-Straße') # 4600 Wels
|
|
street = street.replace('Rudolf Waisenhorn-Gasse','Rudolf-Waisenhorn-Gasse')
|
|
street = street.replace('Schuettelstraße','Schüttelstraße')
|
|
street = street.replace('Sebastian Bauer Straße','Sebastian-Bauer-Straße')
|
|
street = street.replace('St. Bartholomäusplatz','St.-Bartholomäus-Platz')
|
|
street = street.replace('St.Gerold Weg','Sankt-Gerold-Weg') # 6830 Brederis
|
|
street = street.replace('Theodor Koernerstraße','Theodor-Körner-Straße') # 8010 Graz
|
|
street = street.replace('Theodor Storm Straße','Theodor-Storm-Straße') # 8041,8042 Graz
|
|
street = street.replace('Th. Pampichler-Straße','Theresia-Pampichler-Straße') # 2000 Stockerau
|
|
street = street.replace('Thomas Edisonstraße','Thomas-Edison-Straße')
|
|
street = street.replace('Thomas Morusgasse','Thomas-Morus-Gasse')
|
|
street = street.replace('Tuerkenstraße','Türkenstraße')
|
|
street = street.replace('Veichtlbauerweg','Johann-Veichtlbauer-Weg') # 4910 Ried im Innkreis
|
|
street = street.replace('Viktor Dankl Straße','Viktor-Dankl-Straße') # 6020 Innsbruck
|
|
street = street.replace('Weissgerberlände','Weißgerberlände')
|
|
street = street.replace('Wilhelm Kressplatz','Wilhelm-Kreß-Platz')
|
|
|
|
if (city == 'Deutsch-Wagram'):
|
|
street = street.replace('Wilhelm Kressgasse','Wilhelm-Kress-Gasse') # Deutsch-Wagram
|
|
else:
|
|
street = street.replace('Wilhelm Kressgasse','Wilhelm-Kreß-Gasse') # Wien 22, Guntramsdorf, Pressbaum
|
|
street = street.replace('Wladimir Kolda Gasse','Wladimir Kolda-Gasse') #3400 Klosterneuburg
|
|
# Streets with location names
|
|
street = street.replace('Alser-Straße','Alser Straße')
|
|
street = street.replace('Altenbergerstraße','Altenberger Straße')
|
|
street = street.replace('Altmannsdorferstraße','Altmannsdorfer Straße')
|
|
street = street.replace('Atzgersdorferstraße','Atzgersdorfer Straße')
|
|
street = street.replace('Breitenfurterstraße','Breitenfurter Straße')
|
|
street = street.replace('Breitenleerstraße','Breitenleer Straße')
|
|
street = street.replace('Breitenseerstraße','Breitenseer Straße')
|
|
street = street.replace('Dresdnerstraße','Dresdner Straße')
|
|
street = street.replace('Feldkirchnerstraße','Feldkirchner Straße')
|
|
street = street.replace('Freistaedterstraße','Freistädter Straße')
|
|
street = street.replace('Freistädterstraße','Freistädter Straße')
|
|
street = street.replace('Gaalerstraße','Gaaler Straße') # 8720 Knittelfeld
|
|
street = street.replace('Gerasdorferstraße','Gerasdorfer Straße')
|
|
street = street.replace('Groedigerstraße ','Grödiger Straße ')
|
|
street = street.replace('Grödigerstraße','Grödiger Straße')
|
|
street = street.replace('Gumpendorferstraße','Gumpendorfer Straße')
|
|
street = street.replace('Harmannsdorferstraße','Harmannsdorfer Straße')
|
|
street = street.replace('Heiligenstädterstraße','Heiligenstädter Straße')
|
|
street = street.replace('Hermannstädterweg','Hermannstädter Weg') # 9020 Klagenfurt
|
|
street = street.replace('Hetzendorferstraße','Hetzendorfer Straße')
|
|
street = street.replace('Hörtendorferstraße','Hörtendorfer Straße')
|
|
street = street.replace('Italienerstraße','Italiener Straße')
|
|
street = street.replace('Jedleseerstraße','Jedleseer Straße')
|
|
street = street.replace('Josefstädterstraße','Josefstädter Straße')
|
|
street = street.replace('Kahlenbergerstraße','Kahlenberger Straße')
|
|
street = street.replace('Kagranerplatz','Kagraner Platz')
|
|
street = street.replace('Kaiserebersdorferstraße','Kaiserebersdorfer Straße')
|
|
street = street.replace('Kirchbergerstraße','Kirchberger Straße')
|
|
street = street.replace('Königstetterstraße','Königstetter Straße')
|
|
street = street.replace('Korneuburgerstraße','Korneuburger Straße') # 2103 Langenzersdorf
|
|
street = street.replace('Krottendorferstraße','Krottendorfer Straße') # 2103 Langenzersdorf
|
|
street = street.replace('Langenzersdorferstraße','Langenzersdorfer Straße')
|
|
street = street.replace('Leobendorferstraße','Leobendorfer Straße') # 2100 Unterrohrbach
|
|
street = street.replace('Lerchenfelderstraße','Lerchenfelder Straße')
|
|
street = street.replace('Liesingerstraße','Liesinger Straße') # 3012 Wolfsgraben
|
|
street = street.replace('Linzerstraße ','Linzer Straße ')
|
|
street = street.replace('Mattighofenerstraße','Mattighofner Straße') # typo ! Braunau am Inn
|
|
street = street.replace('Millstätterstraße','Millstätter Straße') #9523 Landskron (Villach)
|
|
street = street.replace('Münchnerstraße','Münchner Straße') # 6100 Seefeld in Tirol
|
|
street = street.replace('Neudorferstraße','Neudorfer Straße')
|
|
street = street.replace('Nußdorferstraße','Nußdorfer Straße')
|
|
street = street.replace('Ottakringerstraße','Ottakringer Straße')
|
|
street = street.replace('Penzingerstraße','Penzinger Straße')
|
|
street = street.replace('Pischeldorferstraße','Pischeldorfer Straße') # 9020 Klagenfurt
|
|
street = street.replace('Pontlatzerstraße','Pontlatzer Straße')
|
|
street = street.replace('Schillingsdorferstraße','Schillingsdorfer Straße')
|
|
street = street.replace('Schönbrunnerstraße','Schönbrunner Straße')
|
|
street = street.replace('Schweizerstraße','Schweizer Straße') # 2020 Hollabrunn
|
|
street = street.replace('Sechshauserstraße','Sechshauser Straße')
|
|
street = street.replace('Speisingerstraße','Speisinger Straße')
|
|
street = street.replace('Stammersdorferstraße','Stammersdorfer Straße')
|
|
street = street.replace('Stockerauerstraße','Stockerauer Straße') # 2100
|
|
street = street.replace('Strassganger Straße','Straßganger Straße')
|
|
street = street.replace('St. Veiterstraße','St. Veiter Straße')
|
|
street = street.replace('St.-Veiter-Straße','St. Veiter Straße')
|
|
street = street.replace('Traundorferstraße','Traundorfer Straße')
|
|
street = street.replace('Vösendorferstraße','Vösendorfer Straße')
|
|
street = street.replace('Währingerstraße','Währinger Straße')
|
|
street = street.replace('Waidhofnerstraße','Waidhofner Straße')
|
|
street = street.replace('Weimarerstraße','Weimarer Straße')
|
|
street = street.replace('Wienerstraße','Wiener Straße')
|
|
street = street.replace('Wr Neustädterstraße','Wiener Neustädter Straße')
|
|
return street,city
|
|
def corr_city(city,postal_code):
|
|
city = city.replace('Aigen Im Mühlkreis','Aigen-Schlägl') # Gemeinde Aigen im Mühlkreis und Schlägl 2015 zusammengelegt
|
|
city = city.replace('Aigen/Muehlkr.','Aigen-Schlägl')
|
|
#city = city.replace('Alberndorf In Der Riedmark','Alberndorf in der Riedmark')
|
|
city = city.replace(' In Der ',' in der ')
|
|
#city = city.replace('Allerheiligen Bei Wildon','Allerheiligen bei Wildon')
|
|
city = city.replace(' Bei ',' bei ')
|
|
#city = city.replace('Altenberg Bei Linz','Altenberg bei Linz')
|
|
#city = city.replace('Altenmarkt An Der Triesting','Altenmarkt an der Triesting')
|
|
city = city.replace(' An Der ',' an der ')
|
|
#city = city.replace('Altenmarkt Im Pongau','Altenmarkt im Pongau')
|
|
city = city.replace(' Im ',' im ')
|
|
# city = city.replace('Altenmarkt','Altenmarkt an der Triesting') # does not work, use:
|
|
if (city == 'Altenmarkt'): city = 'Altenmarkt an der Triesting'
|
|
city = city.replace('Altmuenster','Altmünster')
|
|
#city = city.replace('Annaberg Im Lammertal','Annaberg im Lammertal')
|
|
#city = city.replace('Arzl Im Pitztal','Arzl im Pitztal')
|
|
#city = city.replace('Aschach An Der Donau','Aschach an der Donau')
|
|
city = city.replace('Aschau/,Zillertal','Aschau')
|
|
#city = city.replace('Bad Goisern Am Hallstättersee','Bad Goisern am Hallstättersee')
|
|
city = city.replace(' Am ',' am ')
|
|
if (city == 'Bad Goisern'): city = 'Bad Goisern am Hallstättersee'
|
|
#city = city.replace('Bad Sankt Leonhard Im Lavanttal','Bad Sankt Leonhard im Lavanttal')
|
|
#city = city.replace('Breitenfurt Bei Wien','Breitenfurt bei Wien')
|
|
if (city == 'Breitenfurt'): city = 'Breitenfurt bei Wien'
|
|
#city = city.replace('Bruck An Der Leitha','Bruck an der Leitha')
|
|
city = city.replace(' An Der ',' an der ')
|
|
#city = city.replace('Bruck An Der Mur','Bruck an der Mur')
|
|
#city = city.replace('Bruck Am Ziller','Bruck am Ziller')
|
|
#city = city.replace('Brunn Am Gebirge','Brunn am Gebirge')
|
|
#city = city.replace('Brunn An Der Wild','Brunn an der Wild')
|
|
#city = city.replace('Bubendorf Im Burgenland','Bubendorf im Burgenland')
|
|
city = city.replace(' Im ',' im ')
|
|
if (city == 'Dellach'): city = 'Dellach im Drautal'
|
|
if (city =='Fliess'): city = 'Fließ'
|
|
#city = city.replace('Hadersdorf Am Kamp','Hadersdorf am Kamp')
|
|
#city = city.replace('Hall In Tirol','Hall in Tirol')
|
|
city = city.replace(' In ',' in ')
|
|
#city = city.replace('Laa An Der Thaya','Laa an der Thaya')
|
|
if (city == 'Rohrbach'): city = 'Rohrbach-Berg'
|
|
#city = city.replace('Sankt Jakob Im Rosental','Sankt Jakob im Rosental')
|
|
#city = city.replace('Sankt Marein Bei Graz-Markt','Sankt Marein bei Graz')
|
|
#city = city.replace('Sankt Veit An Der Glan','Sankt Veit an der Glan')
|
|
#city = city.replace('Schwarzau Am Steinfeld','Schwarzau am Steinfeld')
|
|
#city = city.replace('Steindorf Am Ossiacher See','Steindorf am Ossiacher See')
|
|
#city = city.replace('Steinhaus Am Semmering','Steinhaus am Semmering')
|
|
if (city == 'Uttendorf' and postal_code == '5261'):
|
|
city = 'Helpfau-Uttendorf'
|
|
# TYPO
|
|
city = city.replace('Breittenfurt','Breitenfurt')
|
|
city = city.replace('Bsichofshofen','Bischofshofen')
|
|
city = city.replace('Maria-Anzbach','Maria Anzbach') # Namensänderung der Marktgemeinde 2022 Landesgesetzblatt Nr. 44/2022
|
|
city = city.replace('Sankt Andrä/Wördern','Sankt Andrä-Wördern')
|
|
city = city.replace('St.Andrä-Wördern','Sankt Andrä-Wördern')
|
|
city = city.replace('Sankt Andrä Wördern','Sankt Andrä-Wördern')
|
|
city = city.replace('Waidhofen/Ybbs','Waidhofen an der Ybbs')
|
|
#city = city.replace('Wr.Neustadt','Wiener Neustadt')
|
|
city = city.replace('Wr Neustadt','Wiener Neustadt')
|
|
city = city.replace('Wörther See','Wörthersee')
|
|
# Postleitzahl Änderungen
|
|
if (postal_code == '8071' and city == 'Gössendorf'): postal_code = '8077'
|
|
if (postal_code == '6064'): postal_code = '6063' # Postamt Rum Dorf aufgelassen, nur mehr 6063
|
|
if (postal_code == '6020' and city == 'Rum'): postal_code = '6063' # Serlesstraße in Rum
|
|
return city, postal_code
|
|
def call_postal_code(location, address, verbose):
|
|
street = address
|
|
if location == '' or location[0] == '*':
|
|
return '', '', street
|
|
p = location.split(' ',1)
|
|
if isinteger(p[0]) and int(p[0]) > 1000 and int(p[0]) < 10000:
|
|
postal_code = p[0]
|
|
city, postal_code = corr_city(p[1].strip(),postal_code)
|
|
street,city = corr_street(address,city)
|
|
else:
|
|
postal_code, city, street = call_change_city(location, address, verbose)
|
|
|
|
# street = corr_street(street)
|
|
# add city and street to the database
|
|
return postal_code, city, street
|
|
|
|
def call_data_record(line, mod_date, verbose, cur):
|
|
|
|
# we have to split the record with a cost-intensive regular expression
|
|
# record = re.split('OE[0-9][A-Z]{1,3}[ \t]{3,20}',line) # this does not work 100%
|
|
# record = re.findall(r'(OE[0-9][A-Z]{1,3})[ \t]{2,12}([A-ZÄÖÜ].+[ ]?.*[ ]?.*[ ]?.*)[ \t]{3,30}(.{3,30})[ \t]{3,30}([1,3,4])', line)
|
|
# record = re.search(r'(OE[0-9][A-Z]{1,3})[ ]{2,12}([. ]+)[ ]{3,50}([. ]+)[ ]{3,50}([1-4])', line)
|
|
# record = re.search(r'^(OE[0-9][A-Z]{1,3})[ \t]{2,20}([\w ]{1,12})[ ]{3,50}(.*)([1-4]{1})$', line)
|
|
|
|
# Never split Addresses containing 2 or 3 spaces, also several records contain no address or no location
|
|
records = re.split(r'[ ]{4,65}', line)
|
|
# [records for record in records]
|
|
|
|
if verbose > 2 :
|
|
print(f'Record length: {len(records)}')
|
|
|
|
for m in records:
|
|
print(m)
|
|
|
|
# HARDENING:
|
|
assert(len(records) == 5)
|
|
# OE Call:
|
|
call = records[0]
|
|
call = fix_call(call,verbose)
|
|
match = re.search(r'^(OE[0-9][A-Z]{1,3})', call)
|
|
assert(match.string == call)
|
|
fullname = records[1]
|
|
location = records[2]
|
|
address = records[3]
|
|
permit_class = records[4]
|
|
fullname = fix_typo(call, fullname, verbose)
|
|
firstname = ''
|
|
surname = ''
|
|
# If there is a clubstation
|
|
if is_clubstation(call):
|
|
# Name starting with only one quotation marks e.g. " -- remove that one:
|
|
fullname = remove_first_quote_if_odd(fullname, verbose) # only found @ clubstations
|
|
clubstationfile = '.callbook_club'
|
|
if verbose > 1:
|
|
print(f'Call: {call}, Name: {fullname}')
|
|
path = os.path.join(os.path.dirname(os.path.abspath(__file__)), clubstationfile)
|
|
if os.path.exists(path):
|
|
fullname = replace_substring_with_line(path, fullname, verbose)
|
|
gender = '*'
|
|
firstname = fullname.strip()
|
|
elif fullname[0] == '*':
|
|
gender = '*'
|
|
else: # Try to split the YL or OMs Name, guess the gender
|
|
firstname, surname, gender = call_split_name(fullname, call, verbose)
|
|
if verbose > 1:
|
|
if gender == '*':
|
|
print(f'Call: {call}, Name: {fullname}, Gender: {gender}')
|
|
else:
|
|
print(f'Call: {call}, First Name: {firstname}, Surname: {surname}, Gender: {gender}')
|
|
print(f'Location: {location}, Address: {address}, Permit: {permit_class}')
|
|
|
|
created = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
|
|
call_data_record.cnt += 1 # increment the User_id
|
|
user_id = call_data_record.cnt
|
|
|
|
statement = f'SELECT `call`, `firstname`, `surname`, `gender` FROM `callbook_user` WHERE `call` = "{call}" AND `active` = 1'
|
|
cur.execute(statement)
|
|
result = cur.fetchall()
|
|
if not result:
|
|
postal_code,city,street = call_postal_code(location, address, verbose)
|
|
statement = "INSERT INTO `callbook_address`(`location`,`address`,`postal_code`,`city`,`street`) VALUES (%s,%s,%s,%s,%s)"
|
|
data = (location,address,postal_code,city,street)
|
|
try: cur.execute(statement, data)
|
|
except mariadb.Error as e:
|
|
print(f'\n[WARN] MySQLError during execute statement\n\tArgs: {e.args}')
|
|
except Exception as e:
|
|
print('Error: {}'.format(e), file=sys.stderr)
|
|
|
|
statement = "INSERT INTO `callbook_user`(`user_id`,`call`,`firstname`,`surname`,`gender`,`created`,`modified`,`address_id`) VALUES (%s, %s, %s, %s, %s, %s, %s, LAST_INSERT_ID())"
|
|
data = (user_id,call,firstname,surname,gender,created,created)
|
|
try:
|
|
cur.execute(statement, data)
|
|
except mariadb.Error as e:
|
|
print(f'\n[WARN] MySQLError during execute statement\n\tArgs: {e.args}')
|
|
except Exception as e:
|
|
print('Error: {}'.format(e), file=sys.stderr)
|
|
else:
|
|
for r in result:
|
|
print(f'Call: {call} {firstname} {surname} {gender} <- ')
|
|
|
|
call_data_record.cnt = 0
|
|
|
|
def call_analyse_pdf(file, verbose, cur):
|
|
|
|
# Define a regular expression to match tables
|
|
|
|
reader = PdfReader(file)
|
|
meta = reader.metadata
|
|
if verbose:
|
|
print(verbose)
|
|
print(' Pages:', len(reader.pages))
|
|
# All of the following could be None!
|
|
print(f' Author: {meta.author}')
|
|
print(f' Creator: {meta.creator}')
|
|
print(f'Producer: {meta.producer}')
|
|
print(f' Subject: {meta.subject}')
|
|
print(f' Title: {meta.title}')
|
|
print(f' Created: {meta.creation_date}')
|
|
print(f'Modified: {meta.modification_date}')
|
|
|
|
for page in reader.pages:
|
|
page_text = page.extract_text(extraction_mode="layout", layout_mode_space_vertically=False)
|
|
lines = page_text.strip().splitlines()
|
|
for line in lines[3:-2]:
|
|
line = line.strip()
|
|
|
|
if verbose >= 2:
|
|
print(line)
|
|
call_data_record(line, meta.modification_date,verbose, cur)
|
|
|
|
def exec_sql_file(cursor, sql_file):
|
|
statement = ''
|
|
try:
|
|
for line in open(sql_file):
|
|
if line.strip().startswith('--'): # ignore sql comment lines
|
|
continue
|
|
if line.strip().endswith(';'): # keep appending lines that don't end in ';'
|
|
statement += line
|
|
try:
|
|
cursor.execute(statement)
|
|
except mariadb.Error as e: # (OperationalError, ProgrammingError) as e:
|
|
print(f'\n[WARN] MySQLError during execute statement\n\tArgs: {e.args}')
|
|
statement = ''
|
|
else:
|
|
statement += line
|
|
|
|
except FileNotFoundError:
|
|
print(f'The file {path} was not found.')
|
|
except Exception as e:
|
|
print('Error: {}'.format(e), file=sys.stderr)
|
|
|
|
def call_create_database(configdb):
|
|
database = configdb.get('database','callbook')
|
|
print(f'Try to create the database "{database}" (if you have enough privileges)')
|
|
conn = mariadb.connect(
|
|
host = configdb.get('host','127.0.0.1'),
|
|
user = configdb.get('user','om'),
|
|
password = configdb.get('password','oe3tkt')
|
|
)
|
|
# create a cursor object
|
|
cur = conn.cursor()
|
|
# creating database
|
|
try:
|
|
cur.execute('CREATE DATABASE ' + database)
|
|
except mariadb.Error as e:
|
|
print(f'\n[ERROR] MySQLError during execute statement\n\tArgs: {e.args}')
|
|
sys.exit(5)
|
|
cur.execute('SHOW DATABASES')
|
|
databaseList = cur.fetchall()
|
|
#for database in databaseList:
|
|
# print(database)
|
|
if not databaseList:
|
|
conn.close()
|
|
sys.exit(5)
|
|
cur.execute('USE ' + database)
|
|
|
|
return conn
|
|
|
|
if __name__ == '__main__':
|
|
args = call_parser()
|
|
config = call_configparser()
|
|
try:
|
|
filename = call_website(**vars(args))
|
|
if args.verbose > 1:
|
|
print(f'Filename: {filename}')
|
|
|
|
except Exception as e:
|
|
print('Error: {}'.format(e), file=sys.stderr)
|
|
sys.exit(1)
|
|
|
|
try:
|
|
configdb = config['MariaDB'] # fetch from config file, then do with fallbacks:
|
|
conn = mariadb.connect(
|
|
user = configdb.get('user', 'om'),
|
|
password = configdb.get('password','oe3tkt'),
|
|
host = configdb.get('host','127.0.0.1'),
|
|
port = configdb.getint('port',3306),
|
|
database = configdb.get('database','callbook')
|
|
)
|
|
except mariadb.Error as e:
|
|
print(f'Error connectiong to MariaDB platform: {e}')
|
|
conn = call_create_database(configdb)
|
|
|
|
print(f'Connected to MariaDB API Version {mariadb.mariadbapi_version}.')
|
|
|
|
print(datetime.datetime.now(datetime.UTC))
|
|
# Get Cursor
|
|
cur = conn.cursor()
|
|
cur.execute('SELECT VERSION()')
|
|
print(cur.fetchone()[0])
|
|
sql_file = '.sql_init'
|
|
path = os.path.join(os.path.dirname(os.path.abspath(__file__)), sql_file)
|
|
exec_sql_file(cur, path)
|
|
|
|
call_analyse_pdf(filename,args.verbose,cur)
|
|
conn.commit()
|
|
cur.close()
|
|
conn.close()
|
|
sys.exit(0)
|