Bugfix: Allow currency to be suffixed with any number of asterisks, as seen in https://www.budgetbytes.com/chicken-alfredo/

Author: strangetom

ingredient_parser/en/_regex.py

@@ -58,11 +58,11 @@
 )
 
 # Regex pattern to match quantities split by "and" e.g. 1 and 1/2.
-# Capture the whole match, and the quantites before and after the "and".
+# Capture the whole match, and the quantities before and after the "and".
 FRACTION_SPLIT_AND_PATTERN = re.compile(r"((\d+)\sand\s(\d/\d+))")
 
 # Regex pattern to match ranges where the unit appears after both quantities e.g.
-# 100 g - 200 g. This assumes the quantites and units have already been seperated
+# 100 g - 200 g. This assumes the quantities and units have already been separated
 # by a single space and that all number are decimals.
 # This regex matches:
 #   <quantity> <unit> - <quantity> <unit>
@@ -117,7 +117,9 @@
 FRACTION_TOKEN_PATTERN = re.compile(r"^\d*\#\d+\$\d+(?:\-\d*\#\d+\$\d+)?$")
 
 # Regex pattern to match currency within parentheses e.g. ($1.99)
-# Allows optional whitespace after opening parenthesis, before currency symbol, and
+# Allows optional white space after opening parenthesis, before currency symbol, and
 # before closing parenthesis.
+# Also allows the current to be suffixed with any number of asterisk characters because
+# that has been seen on budgetbytes.com.
 currency_pattern = "|".join(re.escape(c) for c in ["$", "£", "€", "¥", "₹"])
-CURRENCY_PATTERN = re.compile(rf"\(\s*(?:{currency_pattern})\s*[0-9.,]+\s*\)")
+CURRENCY_PATTERN = re.compile(rf"\(\s*(?:{currency_pattern})\s*[0-9.,]+\**\s*\)")

tests/preprocess/test_remove_price_annotations.py

@@ -84,6 +84,10 @@ def test_price_annotation_with_mixed_whitespace(self, p):
         input_sentence = "1 cup flour ( \t $ 0.20  )"
         assert p._remove_price_annotations(input_sentence) == "1 cup flour "
 
+    def test_price_annotation_with_asterisk_suffix(self, p):
+        input_sentence = "1 cup flour ($0.20**)"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour "
+
     def test_non_price_parenthetical_remains(self, p):
         input_sentence = "1 cup flour (organic)"
         assert p._remove_price_annotations(input_sentence) == "1 cup flour (organic)"