Process ingredients with prices properly (#33)

This commit looks to add support for preprocessing out prices from
ingredients.

Co-authored-by: tom <tpstrange@gmail.com>
Co-authored-by: Tom Strange <strangetom@users.noreply.github.com>

Author: jmt-gh

ingredient_parser/en/preprocess.py

@@ -169,6 +169,7 @@ def _normalise(self, sentence: str) -> str:
         # List of functions to apply to sentence
         # Note that the order matters
         funcs = [
+            self._remove_price_annotations,
             self._replace_en_em_dash,
             self._replace_html_fractions,
             self._replace_unicode_fractions,
@@ -187,6 +188,25 @@ def _normalise(self, sentence: str) -> str:
             logger.debug(f"{func.__name__}: {sentence}")
 
         return sentence.strip()
+    
+    def _remove_price_annotations(self, sentence: str) -> str:
+        """Remove price annotations like ($0.20), (£1.50), etc. from the sentence.
+
+        Parameters
+        ----------
+        sentence : str
+            Ingredient sentence
+
+        Returns
+        -------
+        str
+            Ingredient sentence with price annotations removed
+        """
+        currencies = ["$", "£", "€", "¥", "₹"]
+        currency_pattern = "|".join(re.escape(c) for c in currencies)
+        # Allow optional whitespace after opening parenthesis and before currency, and after currency
+        pattern = rf"\(\s*(?:{currency_pattern})\s*[0-9.,]+\s*\)"
+        return re.sub(pattern, "", sentence)
 
     def _replace_en_em_dash(self, sentence: str) -> str:
         """Replace en-dashes and em-dashes with hyphens.

tests/preprocess/test_preprocess.py

@@ -22,6 +22,28 @@ def test__repr__(self):
         p = PreProcessor("1/2 cup chicken broth")
         assert repr(p) == 'PreProcessor("1/2 cup chicken broth")'
 
+    def test_debug_output(self, capsys):
+        """
+        Test printed debug output
+        """
+        _ = PreProcessor("1/2 cup chicken broth", show_debug_output=True)
+        captured = capsys.readouterr()
+        assert (
+            captured.out
+            == """_remove_price_annotations: 1/2 cup chicken broth
+_replace_en_em_dash: 1/2 cup chicken broth
+_replace_html_fractions: 1/2 cup chicken broth
+_replace_unicode_fractions: 1/2 cup chicken broth
+combine_quantities_split_by_and: 1/2 cup chicken broth
+_identify_fractions: #1$2 cup chicken broth
+_split_quantity_and_units: #1$2 cup chicken broth
+_remove_unit_trailing_period: #1$2 cup chicken broth
+replace_string_range: #1$2 cup chicken broth
+_replace_dupe_units_ranges: #1$2 cup chicken broth
+_merge_quantity_x: #1$2 cup chicken broth
+_collapse_ranges: #1$2 cup chicken broth
+"""
+        )
 
 def normalise_test_cases() -> list[tuple[str, ...]]:
     """

tests/preprocess/test_remove_price_annotations.py

@@ -0,0 +1,101 @@
+import pytest
+
+from ingredient_parser.en import PreProcessor
+
+@pytest.fixture
+def p():
+    """Define an empty PreProcessor object to use for testing the PreProcessor class methods."""
+    return PreProcessor("")
+
+class TestPreProcessor_remove_price_annotations:
+    def test_remove_dollar_price(self, p):
+        input_sentence = "1 cup flour ($0.20)"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour "
+
+    def test_remove_pound_price(self, p):
+        input_sentence = "2 eggs (£1.50)"
+        assert p._remove_price_annotations(input_sentence) == "2 eggs "
+
+    def test_remove_euro_price(self, p):
+        input_sentence = "3 tomatoes (€2.00)"
+        assert p._remove_price_annotations(input_sentence) == "3 tomatoes "
+
+    def test_remove_yen_price(self, p):
+        input_sentence = "1 onion (¥100)"
+        assert p._remove_price_annotations(input_sentence) == "1 onion "
+
+    def test_remove_rupee_price(self, p):
+        input_sentence = "1 potato (₹10.50)"
+        assert p._remove_price_annotations(input_sentence) == "1 potato "
+
+    def test_multiple_prices(self, p):
+        input_sentence = "1 apple ($0.50) and 1 orange (£0.30)"
+        assert p._remove_price_annotations(input_sentence) == "1 apple  and 1 orange "
+
+    def test_no_price_annotation(self, p):
+        input_sentence = "1 cup sugar"
+        assert p._remove_price_annotations(input_sentence) == "1 cup sugar"
+
+    def test_malformed_price_annotation(self, p):
+        input_sentence = "1 cup flour ($0.20"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour ($0.20"
+
+    def test_price_with_comma(self, p):
+        input_sentence = "1 steak (€1,200.00)"
+        assert p._remove_price_annotations(input_sentence) == "1 steak "
+
+    def test_price_with_multiple_decimals(self, p):
+        input_sentence = "1 cheese ($1.99) and 1 bread ($2.49)"
+        assert p._remove_price_annotations(input_sentence) == "1 cheese  and 1 bread "
+
+    def test_price_annotation_at_start(self, p):
+        input_sentence = "($0.20) 1 cup flour"
+        assert p._remove_price_annotations(input_sentence) == " 1 cup flour"
+
+    def test_price_annotation_in_middle(self, p):
+        input_sentence = "1 cup ($0.20) flour"
+        assert p._remove_price_annotations(input_sentence) == "1 cup  flour"
+
+    def test_price_annotation_at_end(self, p):
+        input_sentence = "1 cup flour ($0.20)"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour "
+
+    def test_price_annotation_with_leading_space(self, p):
+        input_sentence = "1 cup flour ( $0.20)"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour "
+
+    def test_price_annotation_with_inner_spaces(self, p):
+        input_sentence = "1 cup flour ( $ 0.20 )"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour "
+
+    def test_price_annotation_with_multiple_spaces(self, p):
+        input_sentence = "1 cup flour (  $  0.20  )"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour "
+
+    def test_price_annotation_with_tab_spaces(self, p):
+        input_sentence = "1 cup flour (\t$0.20\t)"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour "
+
+    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_non_price_parenthetical_remains(self, p):
+        input_sentence = "1 cup flour (organic)"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour (organic)"
+
+    def test_multiple_non_price_parentheticals(self, p):
+        input_sentence = "2 eggs (free-range) (large)"
+        assert p._remove_price_annotations(input_sentence) == "2 eggs (free-range) (large)"
+
+    def test_mixed_price_and_non_price_parentheticals(self, p):
+        input_sentence = "1 cup flour ($0.20) (organic)"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour  (organic)"
+
+    def test_non_price_parenthetical_with_spaces(self, p):
+        input_sentence = "1 cup flour ( see note )"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour ( see note )"
+
+    def test_non_price_parenthetical_with_numbers(self, p):
+        input_sentence = "1 cup flour (2nd batch)"
+        assert p._remove_price_annotations(input_sentence) == "1 cup flour (2nd batch)"