Bugfix: Support collapsing ranges with duplicate units where the unit…

…s are not identical but are synonyms.

ingredient_parser/en/_constants.py

@@ -427,3 +427,16 @@
 ]
 # Tokens that indicate an amount is singular
 SINGULAR_TOKENS = ["each"]
+
+# List of sets, where each set contains the synonyms that represent the same unit.
+UNIT_SYNONYMS = [
+    {"cup", "c"},
+    {"gram", "g", "gm"},
+    {"kilogram", "kg"},
+    {"litre", "liter", "l"},
+    {"ounce", "oz"},
+    {"pound", "lb"},
+    {"quart", "qt"},
+    {"tablespoon", "tbsp", "tbs", "tb"},
+    {"teaspoon", "tsp"},
+]

ingredient_parser/en/_utils.py

@@ -10,7 +10,7 @@
 
 from .._common import download_nltk_resources, is_float, is_range
 from ..dataclasses import IngredientAmount
-from ._constants import UNITS
+from ._constants import FLATTENED_UNITS_LIST, UNIT_SYNONYMS, UNITS
 from ._regex import FRACTION_SPLIT_AND_PATTERN, STRING_RANGE_PATTERN
 
 UREG = pint.UnitRegistry()
@@ -236,6 +236,50 @@ def convert_to_pint_unit(unit: str, imperial_units: bool = False) -> str | pint.
     return unit
 
 
+def is_unit_synonym(unit1: str, unit2: str) -> bool:
+    """Check if given units are synonyms.
+
+    Parameters
+    ----------
+    unit1 : str
+        First unit to check.
+    unit2 : str
+        Second unit to check.
+
+    Returns
+    -------
+    bool
+        True if units are synonyms.
+
+    Examples
+    --------
+    >>> is_unit_synonym("oz", "ounce")
+    True
+
+    >>> is_unit_synonym("cups", "c")
+    True
+
+    >>> is_unit_synonym("kg", "g")
+    False
+    """
+    # If not in units list, then cannot be unit synonyms
+    if unit1 not in FLATTENED_UNITS_LIST or unit2 not in FLATTENED_UNITS_LIST:
+        return False
+
+    # Make singular if plural
+    if unit1 in UNITS.keys():
+        unit1 = UNITS[unit1]
+
+    if unit2 in UNITS.keys():
+        unit2 = UNITS[unit2]
+
+    for synonyms in UNIT_SYNONYMS:
+        if unit1 in synonyms and unit2 in synonyms:
+            return True
+
+    return False
+
+
 def combine_quantities_split_by_and(text: str) -> str:
     """Combine fractional quantities split by 'and' into single value.
 

ingredient_parser/en/preprocess.py

@@ -30,6 +30,7 @@
 )
 from ._utils import (
     combine_quantities_split_by_and,
+    is_unit_synonym,
     replace_string_range,
     stem,
     tokenize,
@@ -464,7 +465,7 @@ def _replace_dupe_units_ranges(self, sentence: str) -> str:
 
         for full_match, quantity1, unit1, quantity2, unit2 in matches:
             # We are only interested if the both captured units are the same
-            if unit1 != unit2:
+            if not is_unit_synonym(unit1, unit2):
                 continue
 
             # If capture unit not in units list, abort
@@ -809,6 +810,23 @@ def _is_inside_parentheses(self, index: int) -> bool:
 
         return False
 
+    def _is_example(self, index: int) -> bool:
+        """Return True is the token is part of an example in the sentence.
+
+        Examples are indicated using phrases like "such as", "for example"
+
+        Parameters
+        ----------
+        index : int
+            Index of token to check
+
+        Returns
+        -------
+        bool
+            True if index is part of an example, else False
+        """
+        return False
+
     def _is_ambiguous_unit(self, token: str) -> bool:
         """Return True if token is in AMBIGUOUS_UNITS list.
 

tests/test_utils.py

@@ -2,6 +2,7 @@
     UREG,
     combine_quantities_split_by_and,
     convert_to_pint_unit,
+    is_unit_synonym,
     pluralise_units,
     replace_string_range,
 )
@@ -200,3 +201,17 @@ def test_second_starts_with_zero(self):
         """
         input_sentence = "Type 1 or 00 flour"
         assert replace_string_range(input_sentence) == "Type 1 or 00 flour"
+
+
+class Test_is_unit_synonym:
+    def test_singular(self):
+        assert is_unit_synonym("oz", "ounce")
+
+    def test_plural_singular(self):
+        assert is_unit_synonym("cups", "c")
+
+    def test_plural(self):
+        assert is_unit_synonym("lbs", "pounds")
+
+    def test_not_synonym(self):
+        assert not is_unit_synonym("kg", "gram")