From a6a1b4d03f138a9ffceb3c52cc1d520828ba801a Mon Sep 17 00:00:00 2001
From: Eric Duminil <eric.duminil@gmail.com>
Date: Thu, 4 Jul 2024 09:32:04 +0200
Subject: [PATCH] Adding cache, and precision
---
get_coordinates_by_zipcode.py | 28 +++++++++++++++++-----------
requirements.txt | 1 +
test_zipcode.py | 1 +
3 files changed, 19 insertions(+), 11 deletions(-)
diff --git a/get_coordinates_by_zipcode.py b/get_coordinates_by_zipcode.py
index 8b37ce7..12cff76 100644
--- a/get_coordinates_by_zipcode.py
+++ b/get_coordinates_by_zipcode.py
@@ -22,7 +22,6 @@ Also accepts multiple Zipcodes, or Zipcode prefix.
"""
# TODO: Write tests
-# TODO: Use caching for shapes
# TODO: Rename functions
import argparse
@@ -35,37 +34,49 @@ from shapely.ops import unary_union
INPUT_FOLDER = Path('plz')
PLZ_FILENAME = 'plz-5stellig.geojson'
PLZ_SHAPE_FILE = INPUT_FOLDER / PLZ_FILENAME
+PRECISION = 10 # [m]
+ONE_DEGREE = 40e6 / 360 # [m]
+
+PLZ_SHAPES = None
def _download_plz_shapes_if_needed():
if not PLZ_SHAPE_FILE.exists():
from tqdm import tqdm
import requests
+ print("Downloading %s..." % PLZ_FILENAME)
URL = "https://downloads.suche-postleitzahl.org/v2/public/" + PLZ_FILENAME
response = requests.get(URL, stream=True)
INPUT_FOLDER.mkdir(exist_ok=True)
with open(PLZ_SHAPE_FILE, "wb") as handle:
for data in tqdm(response.iter_content(chunk_size=1024), unit='kB'):
handle.write(data)
+ print(' Done')
def _get_plz_shapes():
- print("Parsing %s..." % PLZ_FILENAME)
+ global PLZ_SHAPES
+ if PLZ_SHAPES:
+ return PLZ_SHAPES
+
_download_plz_shapes_if_needed()
try:
+ print("Parsing %s..." % PLZ_FILENAME)
with open(PLZ_SHAPE_FILE) as f:
print(' Done')
- return json.load(f)
+ PLZ_SHAPES = json.load(f)
+ return PLZ_SHAPES
except json.decoder.JSONDecodeError:
PLZ_SHAPE_FILE.unlink()
raise AttributeError(f"{PLZ_FILENAME} seems to be damaged. Removing it. Please try again!")
-def _get_coordinates(data, plz_patterns):
+def get_coordinates_by_zipcode(plz_patterns, precision=PRECISION):
+ plz_shapes = _get_plz_shapes()
geometries = []
for plz_pattern in plz_patterns:
found = False
- for plz_geojson in data['features']:
+ for plz_geojson in plz_shapes['features']:
if re.match(plz_pattern, plz_geojson['properties']['plz']):
found = True
@@ -85,7 +96,7 @@ def _get_coordinates(data, plz_patterns):
raise AttributeError(f"Sorry, no information could be found for PLZ={plz_pattern}")
merged = unary_union(geometries)
- wkt_polygon = merged.simplify(0.0001).wkt
+ wkt_polygon = merged.simplify(precision / ONE_DEGREE).wkt
print(wkt_polygon)
try:
import pyperclip
@@ -99,11 +110,6 @@ def _get_coordinates(data, plz_patterns):
return wkt_polygon
-def get_coordinates_by_zipcode(plzs):
- plz_shapes = _get_plz_shapes()
- return _get_coordinates(plz_shapes, plzs)
-
-
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Get WKT geometry for desired PLZs')
parser.add_argument('plzs', metavar='PLZ', type=str, nargs='+',
diff --git a/requirements.txt b/requirements.txt
index 49ddecb..0083c46 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,3 +1,4 @@
pyproj
requests
pyperclip
+tqdm
diff --git a/test_zipcode.py b/test_zipcode.py
index e3342a7..f70d40b 100644
--- a/test_zipcode.py
+++ b/test_zipcode.py
@@ -7,6 +7,7 @@ class TestClass(unittest.TestCase):
def test_method(self):
self.assertEqual('foo'.upper(), 'FOO')
get_coordinates_by_zipcode.get_coordinates_by_zipcode(['70567'])
+ get_coordinates_by_zipcode.get_coordinates_by_zipcode(['70567'], 1_000)
if __name__ == '__main__':
--
GitLab