Clean up phone

CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## 6.20.13 - Dec 27, 2024
+
+* Sanitize phone number for US people scrape.
+
 ## 6.20.12 - Nov 22, 2024
 
 * Use transformers to trim incoming strings at import that are too long for DB columns:

openstates/cli/convert_us.py

@@ -1,3 +1,4 @@
+import re
 import typing
 import uuid
 from collections import defaultdict
@@ -28,7 +29,22 @@ def make_org_id(id_: str) -> str:
     return "ocd-organization/" + str(uuid.uuid5(US_UUID_NAMESPACE, id_))
 
 
+def sanitize_phone(phone: str) -> str:
+    """Remove trail text, toll-free phone number or N/A"""
+    if phone.lower() in ["n/a", "same as above"]:
+        return ""
+    # Some phone might appear like (123) 456 7890
+    pattern = r"\((\d{3})\)\s*(\d{3})-(\d{4})"
+    match = re.search(pattern, phone)
+    if match:
+        # Format the first matched number as XXX-XXX-XXXX
+        formatted_number = f"{match.group(1)}-{match.group(2)}-{match.group(3)}"
+        return formatted_number
+    return phone
+
+
 def _fix_bad_dashes(phone: str) -> str:
+    phone = sanitize_phone(phone)
     return phone.replace("–", "-")
 
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "openstates"
-version = "6.20.12"
+version = "6.20.13"
 description = "core infrastructure for the openstates project"
 authors = ["James Turk <[email protected]>"]
 license = "MIT"