feat: added parsing dimensions

pipeline/beam_etl/helpers/dimensions.py

@@ -0,0 +1,67 @@
+"""Functions to parse the dimensions string into a dictionary that represents
+those dimenions in cm
+"""
+
+import logging
+from typing import Dict, Optional
+import re
+
+
+UNIT_CONVERSIONS = {"inches": 2.54, "feet": 30.48, "cm": 1}
+
+
+def parse_dimensions_measure(dimensions: str, measure: str) -> Optional[Dict]:
+    """Using a regex, parse a measurement out of a dimensions string
+    I expect to find a value of the form `1.2 inches (W)` specifying the value,
+    unit and measurement.
+    Return a dictionary representing the parsed value and its unit.
+    """
+    expr = rf"(?P<value>\d*[.,]?\d*)\s+(?P<unit>[a-zA-Z]*)\s+\({measure}\)"
+    if match := re.search(expr, dimensions):
+        match_value = match.group("value")
+        try:
+            value = float(match_value)
+        except ValueError:
+            logging.error("could not parse value `%s` as a float for a dimension")
+            return None
+        return {
+            "value": value,
+            "unit": match.group("unit").lower(),
+        }
+    return None
+
+
+def units_to_cm(value: float, unit: str) -> Optional[float]:
+    """Convert a given dimension unit into centimeters.
+    If unrecognized unit, return None"""
+    try:
+        conversion = UNIT_CONVERSIONS[unit]
+    except KeyError:
+        logging.error("unrecognized unit: %s", unit)
+        return None
+    return value * conversion
+
+
+def parse_dimensions(dimensions: Optional[str]) -> Dict[str, Optional[float]]:
+    """Parse a string representing dimensions"""
+    if dimensions is None:
+        return {
+            "height": None,
+            "width": None,
+            "depth": None,
+        }
+    height = parse_dimensions_measure(dimensions, "H")
+    width = parse_dimensions_measure(dimensions, "W")
+    depth = parse_dimensions_measure(dimensions, "D")
+    parsed_dimensions = {
+        "height": height,
+        "width": width,
+        "depth": depth,
+    }
+    result = {}
+    for key, value in parsed_dimensions.items():
+        if value is None:
+            result[key] = value
+        else:
+            result[key] = units_to_cm(**value)
+    return result

pipeline/beam_etl/tests/test_dimensions.py

@@ -0,0 +1,113 @@
+"""Test the `parse_dimensions` function and its helpers"""
+
+from helpers.dimensions import parse_dimensions, parse_dimensions_measure, units_to_cm
+
+
+def test_none():
+    """Test None value"""
+    assert parse_dimensions(None) == {"height": None, "width": None, "depth": None}
+
+
+def test_parse_dimensions():
+    """Test a normal example from the sample file"""
+    dimensions_str = "23 inches (H) x 1 inches (W) x 23 inches (D)"
+    assert parse_dimensions_measure(dimensions_str, "W") == {
+        "unit": "inches",
+        "value": 1.0,
+    }
+    assert parse_dimensions_measure(dimensions_str, "H") == {
+        "unit": "inches",
+        "value": 23.0,
+    }
+    assert parse_dimensions_measure(dimensions_str, "D") == {
+        "unit": "inches",
+        "value": 23.0,
+    }
+    assert parse_dimensions(dimensions_str) == {
+        "depth": 58.42,
+        "height": 58.42,
+        "width": 2.54,
+    }
+
+
+def test_parse_dimensions_comma():
+    """Test a normal example from the sample file, but the separator is a comma"""
+    dimensions_str = "23 inches (H), 1 inches (W), 23 inches (D)"
+    assert parse_dimensions_measure(dimensions_str, "W") == {
+        "unit": "inches",
+        "value": 1.0,
+    }
+    assert parse_dimensions_measure(dimensions_str, "H") == {
+        "unit": "inches",
+        "value": 23.0,
+    }
+    assert parse_dimensions_measure(dimensions_str, "D") == {
+        "unit": "inches",
+        "value": 23.0,
+    }
+    assert parse_dimensions(dimensions_str) == {
+        "depth": 58.42,
+        "height": 58.42,
+        "width": 2.54,
+    }
+
+
+def test_parse_dimensions_feet():
+    """Test a normal example from the sample file, but the units is feet"""
+    dimensions_str = "23 feet (H) x 1 feet (W) x 23 feet (D)"
+    assert parse_dimensions_measure(dimensions_str, "W") == {
+        "unit": "feet",
+        "value": 1.0,
+    }
+    assert parse_dimensions_measure(dimensions_str, "H") == {
+        "unit": "feet",
+        "value": 23.0,
+    }
+    assert parse_dimensions_measure(dimensions_str, "D") == {
+        "unit": "feet",
+        "value": 23.0,
+    }
+    assert parse_dimensions(dimensions_str) == {
+        "depth": 701.04,
+        "height": 701.04,
+        "width": 30.48,
+    }
+
+
+def test_parse_dimensions_missing():
+    """Test a normal example from the sample file but some measurement is missing"""
+    dimensions_str = "23 inches (H) x 23 inches (D)"
+    assert parse_dimensions_measure(dimensions_str, "W") is None
+    assert parse_dimensions_measure(dimensions_str, "H") == {
+        "unit": "inches",
+        "value": 23.0,
+    }
+    assert parse_dimensions_measure(dimensions_str, "D") == {
+        "unit": "inches",
+        "value": 23.0,
+    }
+    assert parse_dimensions(dimensions_str) == {
+        "depth": 58.42,
+        "height": 58.42,
+        "width": None,
+    }
+
+
+def test_units_to_cm_inches():
+    """Test `units_to_cm` where the units is inches"""
+    assert units_to_cm(value=0.5, unit="inches") == 1.27
+
+
+def test_units_to_cm_feet():
+    """Test `units_to_cm` where the units is inches"""
+    assert units_to_cm(value=0.5, unit="feet") == 15.24
+
+
+def test_units_to_cm_cm():
+    """Test `units_to_cm` where the units is already cm"""
+    assert units_to_cm(value=0.5, unit="cm") == 0.5
+
+
+def test_units_to_cm_unrecognized():
+    """Test `units_to_cm` where the units are not recognized"""
+    assert units_to_cm(value=0.5, unit="yard") is None