Improve cromulent.extract.normalize_dimension to handle values with only partial units (#87).

Author: kasei

cromulent/extract.py

@@ -22,6 +22,13 @@
 
 #mark - Dimensions
 
+NEXT_FINER_DIMENSION_UNIT = {
+	'inches': None,
+	'feet': 'inches',
+	'cm': None,
+	'fr_feet': 'fr_inches',
+	'fr_inches': 'ligne'
+}
 NUMBER_PATTERN = r'((?:\d+\s+\d+/\d+)|(?:\d+/\d+)|(?:\d+(?:[.,]\d+)?))'
 UNIT_PATTERN = r'''('|"|d(?:[.]?|uymen)|pouc[e.]s?|in(?:ch(?:es)?|[.]?)|'''\
 				r'''pieds?|v[.]?|voeten|f(?:eet|oot|t[.]?)|cm)'''
@@ -122,6 +129,8 @@ def parse_simple_dimensions(value, which=None):
 	value = value.strip()
 	dims = []
 # 	warnings.warn('DIMENSION: %s' % (value,))
+
+	last_unit = None
 	for match in re.finditer(DIMENSION_RE, value):
 		# warnings.warn('--> match %s' % (match,))
 		matched_value = _canonical_value(match.group(2))
@@ -130,10 +139,13 @@ def parse_simple_dimensions(value, which=None):
 			return None
 		unit_value = match.group(3)
 		matched_unit = _canonical_unit(unit_value)
+		if matched_unit is None:
+			matched_unit = NEXT_FINER_DIMENSION_UNIT.get(last_unit)
 		if unit_value and not matched_unit:
 			warnings.warn('*** not a recognized unit: %s' % (unit_value,))
 		which = _canonical_which(which)
 		dim = Dimension(value=matched_value, unit=matched_unit, which=which)
+		last_unit = matched_unit
 		dims.append(dim)
 	if not dims:
 		return None

tests/test_dimensions.py

@@ -3,7 +3,7 @@
 import unittest
 import pprint
 from datetime import datetime
-from cromulent.extract import Dimension
+from cromulent.extract import Dimension, normalized_dimension_object
 import cromulent.extract
 
 class TestDateCleaners(unittest.TestCase):
@@ -45,6 +45,10 @@ def test_parse_simple_dimensions(self):
 			'8 1/4 pouces': [Dimension('8.25', 'inches', None)],
 			'8 1/8 pouces': [Dimension('8.125', 'inches', None)],
 			'1': [Dimension('1', None, None)],
+			
+			# values without a unit that follow values with a unit stay in the same system but using the next-finer unit
+			'2 pieds 3': [Dimension('2', 'feet', None), Dimension('3', 'inches', None)],
+			"1' 3": [Dimension('1', 'feet', None), Dimension('3', 'inches', None)],
 		}
 
 		for value, expected in tests.items():
@@ -88,5 +92,21 @@ def test_dimension_cleaner(self):
 			else:
 				self.assertIsNone(dims)
 
+	def test_normalize_dimension(self):
+		tests = {
+			'1 ft, 2 in': ('1 feet, 2 inches', Dimension(value='14.0', unit='inches', which=None)),
+			'8 1/2 pouces': ('8.5 inches', Dimension(value='8.5', unit='inches', which=None)),
+			'1 pied 7 pouces': ('1 feet, 7 inches', Dimension(value='19.0', unit='inches', which=None)),
+			'2 pied 1/2 pouces': ('2 feet, 0.5 inches', Dimension(value='24.5', unit='inches', which=None)),
+			"4' 8": ('4 feet, 8 inches', Dimension(value='56.0', unit='inches', which=None)),
+			"1 pied 2": ('1 feet, 2 inches', Dimension(value='14.0', unit='inches', which=None)),
+		}
+		for value, expected in tests.items():
+			elabel, edim = expected
+			dims = cromulent.extract.parse_simple_dimensions(value)
+			dim, label = normalized_dimension_object(dims)
+			self.assertEqual(label, elabel)
+			self.assertEqual(dim, edim)
+
 if __name__ == '__main__':
 	unittest.main()