Feature: Add "generous" as a token indicating an approximate amount.

Author: strangetom

ingredient_parser/en/_constants.py

@@ -419,6 +419,7 @@
     "nearly",
     "roughly",
     "~",
+    "generous",
 ]
 # Tokens that indicate an amount is singular
 SINGULAR_TOKENS = ["each"]

ingredient_parser/en/postprocess.py

@@ -1551,7 +1551,7 @@ def _fallback_pattern(
     def _is_approximate(
         self, i: int, tokens: list[str], labels: list[str], idx: list[int]
     ) -> bool:
-        """Return True is token at current index is approximate.
+        """Return True if token at current index is approximate.
 
         This is determined by the token label being QTY and the previous token being in
         a list of approximate tokens.
@@ -1611,6 +1611,11 @@ def _is_approximate(
             self.consumed.append(idx[i - 1])
             self.consumed.append(idx[i - 2])
             return True
+        elif labels[i] == "UNIT" and tokens[i - 1].lower() in APPROXIMATE_TOKENS:
+            # For cases like "2 generous cups"
+            # Mark i - 1 element as consumed
+            self.consumed.append(idx[i - 1])
+            return True
 
         return False
 

tests/postprocess/test_is_approximate.py

@@ -82,6 +82,22 @@ def test_is_approximate_nearly(self):
         assert p._is_approximate(1, tokens, labels, idx)
         assert p.consumed == [0]
 
+    def test_is_approximate_generous(self):
+        """
+        Test that QTY at index is indicated as approximate
+        """
+        sentence = "6 generous cups orange juice"
+        tokens = ["6", "generous", "cups", "orange", "juice"]
+        pos_tags = ["CD", "JJ", "NNS", "NN", "NN"]
+        labels = ["QTY", "UNIT", "UNIT", "B_NAME_TOK", "I_NAME_TOK"]
+        idx = [0, 1, 2, 3, 4]
+
+        p = PostProcessor(
+            sentence, tokens, pos_tags, labels, [0] * len(tokens), custom_units={}
+        )
+        assert p._is_approximate(2, tokens, labels, idx)
+        assert p.consumed == [1]
+
     def test_not_approximate(self):
         """
         Test that QTY at index is not indicated as approximate