avoid interpreting integral numbers as coordinates

There are addresses of type '123 W 345 N, Reigh', where 'W 345 N' is
the actual name of the street.
This commit is contained in:
Sarah Hoffmann
2023-09-05 11:26:47 +02:00
parent e1303fb592
commit 18b2a4c204
2 changed files with 7 additions and 3 deletions

View File

@@ -136,10 +136,10 @@ def _deg(axis:str) -> str:
return f"(?P<{axis}_deg>\\d+\\.\\d+)°?" return f"(?P<{axis}_deg>\\d+\\.\\d+)°?"
def _deg_min(axis: str) -> str: def _deg_min(axis: str) -> str:
return f"(?P<{axis}_deg>\\d+)[°\\s]+(?P<{axis}_min>[\\d.]+)?[']*" return f"(?P<{axis}_deg>\\d+)[°\\s]+(?P<{axis}_min>[\\d.]+)[']*"
def _deg_min_sec(axis: str) -> str: def _deg_min_sec(axis: str) -> str:
return f"(?P<{axis}_deg>\\d+)[°\\s]+(?P<{axis}_min>\\d+)['\\s]+(?P<{axis}_sec>[\\d.]+)?[\"″]*" return f"(?P<{axis}_deg>\\d+)[°\\s]+(?P<{axis}_min>\\d+)['\\s]+(?P<{axis}_sec>[\\d.]+)[\"″]*"
COORD_REGEX = [re.compile(r'(?:(?P<pre>.*?)\s+)??' + r + r'(?:\s+(?P<post>.*))?') for r in ( COORD_REGEX = [re.compile(r'(?:(?P<pre>.*?)\s+)??' + r + r'(?:\s+(?P<post>.*))?') for r in (
r"(?P<ns>[NS])\s*" + _deg('lat') + r"[\s,]+" + r"(?P<ew>[EW])\s*" + _deg('lon'), r"(?P<ns>[NS])\s*" + _deg('lat') + r"[\s,]+" + r"(?P<ew>[EW])\s*" + _deg('lon'),

View File

@@ -11,7 +11,11 @@ import pytest
import nominatim.api.v1.helpers as helper import nominatim.api.v1.helpers as helper
@pytest.mark.parametrize('inp', ['', 'abc', '12 23', 'abc -78.90, 12.456 def']) @pytest.mark.parametrize('inp', ['',
'abc',
'12 23',
'abc -78.90, 12.456 def',
'40 N 60 W'])
def test_extract_coords_no_coords(inp): def test_extract_coords_no_coords(inp):
query, x, y = helper.extract_coords_from_query(inp) query, x, y = helper.extract_coords_from_query(inp)