Merge pull request #80 from bmir-radx/feat/format-number-primitive

Add format_number primitive

Author: matthewhorridge

demo/harmonize_example/README.md

@@ -11,7 +11,7 @@ It includes a small input CSV, a rules JSON file, and a Python script that perfo
 
 ## What the example does
 - Renames `age` to `age_years` (pass-through).
-- Converts `weight_lbs` to `weight_kg` (multiply by 0.453592).
+- Converts `weight_lbs` to `weight_kg` (multiply by 0.453592) and formats to two decimals.
 - Splits `name` (stored as `"Last, First"`) into two new columns:
   - `given_name`
   - `family_name`

demo/harmonize_example/output.csv

@@ -1,4 +1,4 @@
 given_name,family_name,age_years,weight_kg,visit_type_label,source dataset,original_id
-Alice,Smith,10,35.5,baseline,demo,0
+Alice,Smith,10,35.50,baseline,demo,0
 Bob,Jones,5,20.18,follow_up,demo,1
 Carol,Nguyen,8,41.82,screening,demo,2

demo/harmonize_example/rules.json

@@ -16,13 +16,8 @@
           "scaling_factor": 0.453592
         },
         {
-          "operation": "round",
+          "operation": "format_number",
           "precision": 2
-        },
-        {
-          "operation": "cast",
-          "source": "float",
-          "target": "text"
         }
       ]
     }

src/harmonization_framework/harmonization_rule.py

@@ -1,7 +1,7 @@
 from typing import Any, List
 from .element import DataElement
 from .primitives.base import PrimitiveOperation
-from .primitives import PrimitiveVocabulary, Bin, Cast, ConvertDate, ConvertUnits, DoNothing, EnumToEnum, NormalizeText, Offset, Reduce, Round, Scale, Substitute, Threshold, Truncate
+from .primitives import PrimitiveVocabulary, Bin, Cast, ConvertDate, ConvertUnits, DoNothing, EnumToEnum, FormatNumber, NormalizeText, Offset, Reduce, Round, Scale, Substitute, Threshold, Truncate
 
 import json
 
@@ -61,6 +61,8 @@ def from_serialization(cls, serialization):
                     primitive = DoNothing.from_serialization(operation)
                 case PrimitiveVocabulary.ENUM_TO_ENUM.value:
                     primitive = EnumToEnum.from_serialization(operation)
+                case PrimitiveVocabulary.FORMAT_NUMBER.value:
+                    primitive = FormatNumber.from_serialization(operation)
                 case PrimitiveVocabulary.NORMALIZE_TEXT.value:
                     primitive = NormalizeText.from_serialization(operation)
                 case PrimitiveVocabulary.OFFSET.value:

src/harmonization_framework/primitives/__init__.py

@@ -4,6 +4,7 @@
 from .dates import ConvertDate
 from .donothing import DoNothing
 from .enum2enum import EnumToEnum
+from .format_number import FormatNumber
 from .normalize import NormalizeText
 from .offset import Offset
 from .reduce import Reduce

src/harmonization_framework/primitives/format_number.py

@@ -0,0 +1,38 @@
+from .base import PrimitiveOperation, support_iterable
+from typing import Union
+
+class FormatNumber(PrimitiveOperation):
+    """
+    Format numeric values to a fixed number of decimal places.
+
+    Output is a string, intended for stable presentation (e.g., CSV output).
+    """
+    def __init__(self, precision: int):
+        if not isinstance(precision, int):
+            raise TypeError(f"Precision must be an integer, got {type(precision).__name__}")
+        if precision < 0:
+            raise ValueError("Precision must be non-negative")
+        self.precision = precision
+
+    def __str__(self):
+        return f"Format number to {self.precision} decimal places"
+
+    def to_dict(self):
+        """Serialize this operation to a JSON-friendly dict."""
+        return {
+            "operation": "format_number",
+            "precision": self.precision,
+        }
+
+    @support_iterable
+    def transform(self, value: Union[int, float]) -> str:
+        """Format the numeric value to the configured decimal precision."""
+        if not isinstance(value, (int, float)) or isinstance(value, bool):
+            raise TypeError(f"FormatNumber expects a numeric value, got {type(value).__name__}")
+        return f"{value:.{self.precision}f}"
+
+    @classmethod
+    def from_serialization(cls, serialization):
+        """Reconstruct a FormatNumber operation from a serialized dict."""
+        precision = int(serialization["precision"])
+        return FormatNumber(precision)

src/harmonization_framework/primitives/vocabulary.py

@@ -7,6 +7,7 @@ class PrimitiveVocabulary(Enum):
     CONVERT_UNITS = "convert_units"
     DO_NOTHING = "do_nothing"
     ENUM_TO_ENUM = "enum_to_enum"
+    FORMAT_NUMBER = "format_number"
     NORMALIZE_TEXT = "normalize_text"
     OFFSET = "offset"
     REDUCE = "reduce"

tests/test_primitives_serialization.py

@@ -7,6 +7,7 @@
     ConvertUnits,
     DoNothing,
     EnumToEnum,
+    FormatNumber,
     NormalizeText,
     Offset,
     Reduce,
@@ -136,6 +137,29 @@ def test_enum_to_enum_strict_raises_for_missing_value():
         primitive.transform(2)
 
 
+def test_format_number_serialization_and_transform():
+    payload = {"operation": "format_number", "precision": 2}
+
+    roundtrip = FormatNumber.from_serialization(payload)
+    assert roundtrip.to_dict() == payload
+    assert roundtrip.transform(35.5) == "35.50"
+    assert roundtrip.transform(3) == "3.00"
+    assert roundtrip.transform([1.234, 2]) == ["1.23", "2.00"]
+
+
+def test_format_number_rejects_invalid_precision():
+    with pytest.raises(TypeError, match="Precision must be an integer"):
+        FormatNumber("2")  # type: ignore[arg-type]
+    with pytest.raises(ValueError, match="non-negative"):
+        FormatNumber(-1)
+
+
+def test_format_number_rejects_non_numeric():
+    primitive = FormatNumber(2)
+    with pytest.raises(TypeError, match="numeric"):
+        primitive.transform("nope")
+
+
 def test_round_serialization_and_transform():
     primitive = Round(2)
     payload = primitive.to_dict()