Merge pull request #96 from MaineDSA/fix_price_extraction

bmos · web-flow · commit 80b0f3af8df7 · 2026-02-11T11:45:05.000-05:00
handle new price format
diff --git a/src/scraper.py b/src/scraper.py
@@ -34,7 +34,7 @@ class ZillowCardParser:
     _PATTERN_BED_NUM = re.compile(r"\d+")
 
     PRICE_CLEANUP_PATTERNS: ClassVar[list[tuple[str, str, re.RegexFlag]]] = [
-        (r"\+?\s*\d+\s*bds?(?:\s|$)", "", re.IGNORECASE),
+        (r"\s+\d+\s*bds?(?:\s|$)", "", re.IGNORECASE),
         (r"\+?\s*bd(?:\s|$)", "", re.IGNORECASE),
         (r"\s+", " ", re.NOFLAG),
     ]

Original file line number	Diff line number	Diff line change
`@@ -34,7 +34,7 @@ class ZillowCardParser:`
`34`	`34`	`_PATTERN_BED_NUM = re.compile(r"\d+")`
`35`	`35`
`36`	`36`	`PRICE_CLEANUP_PATTERNS: ClassVar[list[tuple[str, str, re.RegexFlag]]] = [`
`37`		`- (r"\+?\s\d+\sbds?(?:\s\|$)", "", re.IGNORECASE),`
	`37`	`+ (r"\s+\d+\s*bds?(?:\s\|$)", "", re.IGNORECASE),`
`38`	`38`	`(r"\+?\s*bd(?:\s\|$)", "", re.IGNORECASE),`
`39`	`39`	`(r"\s+", " ", re.NOFLAG),`
`40`	`40`	`]`