Merge pull request #167 from climatepolicyradar/fix/sensitive-queries…

…-word-boundaries

Fix/sensitive queries word boundaries

src/cpr_data_access/utils.py

@@ -1,4 +1,5 @@
 import csv
+import re
 from pathlib import Path
 from typing import Any, Union
 
@@ -21,7 +22,9 @@ def is_sensitive_query(text: str, sensitive_terms: set) -> bool:
 
     """
     sensitive_terms_in_query = [
-        term for term in sensitive_terms if term in text.lower()
+        term
+        for term in sensitive_terms
+        if re.findall(r"\b" + re.escape(term) + r"\b", text.lower())
     ]
 
     if sensitive_terms_in_query:

tests/test_utils.py

@@ -22,7 +22,7 @@
         [False, "word but outnumbered"],
         [False, "word another phrase example but with many other items"],
         [True, "word"],
-        [True, "wordle"],
+        [False, "wordle"],
         [True, "test term"],
         [True, "test term word"],
         [True, "test term and"],