Skip to content

Commit 23fe9c3

Browse files
committed
more matching
Signed-off-by: John Seekins <[email protected]>
1 parent 9d78fc7 commit 23fe9c3

File tree

1 file changed

+31
-0
lines changed

1 file changed

+31
-0
lines changed

scraper.py

Lines changed: 31 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -95,6 +95,34 @@ def _clean_street(self, street: str, locality: str = "") -> Tuple[str, bool]:
9595
{"match": "351 Elliott Street", "replace": "351 ELLIOTT ST", "locality": "Honolulu"},
9696
{"match": "1 Success Loop Rd", "replace": "1 SUCCESS LOOP DR", "locality": "Berlin"},
9797
{"match": "700 Arch Street", "replace": "700 ARCH ST", "locality": "Philadelphia"},
98+
{"match": "33 NE 4th Street", "replace": "33 NE 4 STREET", "locality": "Miami"},
99+
{"match": "1300 Metropolitan", "replace": "1300 METROPOLITAN AVE", "locality": "Leavenworth"},
100+
{"match": "601 McDonough Blvd SE", "replace": "601 MCDONOUGH BOULEVARD SE", "locality": "Atlanta"},
101+
{"match": "1705 E Hanna Rd", "replace": "1705 EAST HANNA RD", "locality": "Eloy"},
102+
{"match": "2255 East 8th North", "replace": "2255 E 8TH NORTH", "locality": "Mountain Home"},
103+
{"match": "8915 Montana Avenue", "replace": "8915 MONTANA AVE", "locality": "El Paso"},
104+
{"match": "704 E Broadway Street", "replace": "702 E BROADWAY ST", "locality": "Eden"},
105+
{"match": "1300 E Hwy 107", "replace": "1330 HIGHWAY 107", "locality": "La Villa"},
106+
{"match": "216 W. Center Street", "replace": "215 WEST CENTRAL STREET", "locality": "Juneau"},
107+
{"match": "300 El Racho Way ", "replace": "300 EL RANCHO WAY", "locality": "Dilley"},
108+
{"match": "3130 North Oakland Street", "replace": "3130 OAKLAND ST", "locality": "Aurora"},
109+
{"match": "03151 Co. Rd. 24.2", "replace": "3151 ROAD 2425 ROUTE 1", "locality": "Stryker"},
110+
{"match": "20 Hobo Forks Road", "replace": "20 HOBO FORK RD", "locality": "Natchez"},
111+
{"match": "7340 Highway 26 W", "replace": "7340 HIGHWAY 26 WEST", "locality": "Oberlin"},
112+
{"match": "1400 E Fourth Ave", "replace": "1400 E 4TH AVE", "locality": "Anchorage"},
113+
{"match": "3900 N. Powerline Road", "replace": "3900 NORTH POWERLINE ROAD", "locality": "Pompano Beach"},
114+
{"match": "185 E. Michigan Street", "replace": "185 EAST MICHIGAN AVENUE", "locality": "Battle Creek"},
115+
{"match": "601 Central Avenue", "replace": "601 CENTRAL AVE", "locality": "Newport"},
116+
{"match": "501 E Court Avenue", "replace": "501 EAST COURT AVE", "locality": "Jeffersonville"},
117+
{"match": "3200 S. Kings Hwy", "replace": "3700 S KINGS HWY", "locality": "Cushing"},
118+
{"match": "325 Court Street", "replace": "325 COURT STREET", "locality": "Sault Ste. Marie"},
119+
{"match": "301 South Walnut", "replace": "301 SOUTH WALNUT STREET", "locality": "Cottonwood Falls"},
120+
{"match": "830 Pine Hill Road", "replace": "830 PINEHILL ROAD", "locality": "Jena"},
121+
{
122+
"match": "11093 SW Lewis Memorial Dr",
123+
"replace": "11093 SW LEWIS MEMORIAL DRIVE",
124+
"locality": "Bowling Green",
125+
},
98126
# a unique one, 'cause the PHONE NUMBER IS IN THE ADDRESS?!
99127
{"match": "911 PARR BLVD 775 328 3308", "replace": "911 E Parr Blvd", "locality": "RENO"},
100128
# default matches should come last
@@ -155,6 +183,9 @@ def _repair_locality(self, locality: str, administrative_area: str) -> Tuple[str
155183
if locality == "Susupe, Saipan" and administrative_area == "MP":
156184
locality = "SAIPAN"
157185
cleaned = True
186+
if locality == "Cottonwood Falls" and administrative_area == "KS":
187+
locality = "COTTONWOOD FALL"
188+
cleaned = True
158189
return locality, cleaned
159190

160191
def _load_sheet(self) -> dict:

0 commit comments

Comments
 (0)