Skip to content

Commit e67a022

Browse files
committed
break up some filters to make them (overall) faster
Signed-off-by: John Seekins <[email protected]>
1 parent 1f8186d commit e67a022

File tree

1 file changed

+12
-4
lines changed

1 file changed

+12
-4
lines changed

scraper.py

Lines changed: 12 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -165,15 +165,23 @@ def _clean_street(self, street: str, locality: str = "") -> Tuple[str, bool]:
165165
{"match": "4909 FM 2826", "replace": "4909 Farm to Market Road", "locality": "ROBSTOWN"},
166166
{"match": "6920 DIGITAL RD", "replace": "11541 Montana Avenue", "locality": "EL PASO"},
167167
# default matches should come last
168-
{"match": "'s", "replace": "", "locality": ""},
169-
{"match": ".", "replace": "", "locality": ""},
170-
{"match": ",", "replace": "", "locality": ""},
171168
]
172169
cleaned = False
173170
for f in street_filters:
174171
if (f["match"] in street) and ((f["locality"] and f["locality"] == locality) or not f["locality"]):
175172
street = street.replace(f["match"], f["replace"])
176173
cleaned = True
174+
break
175+
# simpler loop for default cleanup
176+
default_matches = [
177+
{"match": "'s", "replace": ""},
178+
{"match": ".", "replace": ""},
179+
{"match": ",", "replace": ""},
180+
]
181+
for f in default_matches:
182+
if f["match"] in street:
183+
street = street.replace(f["match"], f["replace"])
184+
cleaned = True
177185
return street, cleaned
178186

179187
def _repair_zip(self, zip_code: int, locality: str) -> Tuple[str, bool]:
@@ -185,7 +193,7 @@ def _repair_zip(self, zip_code: int, locality: str) -> Tuple[str, bool]:
185193
cleaned = False
186194
if len(zcode) == 4:
187195
zcode = f"0{zcode}"
188-
cleaned = True
196+
return zcode, cleaned
189197
matches = [
190198
{"match": "89512", "replace": "89506", "locality": "Reno"},
191199
{"match": "82901", "replace": "82935", "locality": "Rock Springs"},

0 commit comments

Comments
 (0)