More flexibility for units (#43)

Allows both the raised and plain notation to be used for powers in units.

Also add ha (hectare) as a known unit.

tests/test_helpers.py

@@ -6,6 +6,7 @@
 import mock
 
 from wikidataStuff.helpers import (
+    get_unit_q,
     is_number,
     is_int,
     is_pos_int,
@@ -241,3 +242,20 @@ def test_fill_cache_redirects_and_warns(self):
         self.warning.assert_called_once_with(
             'fill_cache is deprecated. Use fill_cache_wdqs instead.')
         self.assertEqual(result, expected)
+
+
+class TestGetUnitQ(unittest.TestCase):
+
+    """Test get_unit_q()."""
+
+    def test_get_unit_q_empty(self):
+        self.assertEqual(get_unit_q(''), None)
+
+    def test_get_unit_q_match(self):
+        self.assertEqual(get_unit_q('km2'), 'Q712226')
+
+    def test_get_unit_q_no_match(self):
+        self.assertEqual(get_unit_q('foo'), None)
+
+    def test_get_unit_q_raised(self):
+        self.assertEqual(get_unit_q('km²'), 'Q712226')

wikidataStuff/helpers.py

@@ -501,6 +501,9 @@ def get_unit_q(unit):
     """
     Given a unit abbreviation return the appropriate qid.
 
+    Powers can be given as either a plain number or a raised number, e.g.
+    km2 or km².
+
     @param unit: the unit abbreviation to look up
     @type unit: str
     @return: Q-value of matching wikidata entry or None if not mapped
@@ -511,8 +514,13 @@ def get_unit_q(unit):
         'km': 'Q828224',
         'cm': 'Q174728',
         'mm': 'Q174789',
-        'km²': 'Q712226'
+        'km2': 'Q712226',
+        'ha': 'Q35852'
     }
+
+    # standardise input
+    unit = unit.replace('²', '2').replace('³', '3')
+
     return units.get(unit)