Files
elwig-misc/data/plz.py
2022-11-29 22:46:52 +01:00

70 lines
1.9 KiB
Python
Executable File

#!/bin/env python3
from typing import List, Tuple
import argparse
import requests
import re
import xlrd
import tempfile
import os
URL = 'https://www.post.at/g/c/postlexikon'
BUTTON = re.compile(r'title="PLZ Bestimmungsort" href="(.*?)"')
Row = Tuple[int, str, int, str, int, str]
def get_excel_url() -> str:
r = requests.get(URL, headers={'User-Agent': 'Mozilla/5.0'})
if r.status_code != 200:
raise RuntimeError(f'Unexpected response: {r.status_code} {r.reason}')
matches = BUTTON.findall(r.text)
if len(matches) == 0:
raise RuntimeError('Unable to find url of file')
return matches[0]
def download_excel() -> List[Row]:
f_name = None
try:
f = tempfile.NamedTemporaryFile(delete=False)
with requests.get(get_excel_url(), stream=True) as r:
if r.status_code != 200:
raise RuntimeError(f'Unexpected response: {r.status_code} {r.reason}')
for chunk in r.iter_content(chunk_size=8192):
f.write(chunk)
f_name = f.name
f.close()
rows = []
wb = xlrd.open_workbook(f_name)
sheet = wb.sheet_by_index(0)
for r in range(1, sheet.nrows):
row = sheet.row_values(r)
rows.append((int(row[0]), row[1], int(row[2]), row[3], int(row[4]), row[5]))
return rows
finally:
if f_name is not None:
os.remove(f_name)
def write_sql(data: List[Row]) -> None:
with open('plz.sql', 'wb') as f:
f.write(b"\nINSERT INTO AT_plz VALUES\n")
for plz, dest, okz, _, _, _ in data:
f.write(f"({plz:4}, {okz:5}, '{dest}'),\n".encode('utf8'))
f.seek(-2, 1)
f.write(b';\n')
if __name__ == '__main__':
parser = argparse.ArgumentParser()
args = parser.parse_args()
print('Downloading PLZ data from www.post.at')
write_sql(download_excel())
print('Successfully created plz.sql!')