Better output for centuries in format_wikibase_time + tests.

matcher/utils.py

@@ -7,6 +7,7 @@ import user_agents
 import re
 import pattern.en
 from datetime import date
+from num2words import num2words
 
 metres_per_mile = 1609.344
 feet_per_metre = 3.28084
@@ -187,4 +188,5 @@ def format_wikibase_time(v):
     if p == 9:
         return t[1:5]
     if p == 7:
-        return t[1:5]
+        century = ((int(t[:5]) - 1) // 100) + 1
+        return num2words(century, to="ordinal_num") + " century"

tests/test_utils.py

@@ -0,0 +1,19 @@
+from matcher import utils
+
+def test_pluralize_label():
+    v = {"language": "en", "value": "building"}
+    assert utils.pluralize_label(v) == "buildings"
+
+    v = {"language": "en", "value": "mine"}
+    assert utils.pluralize_label(v) == "mines"
+
+def test_format_wikibase_time_year():
+    v = {"time": "+1950-00-00T00:00:00Z", "precision": 9}
+    assert utils.format_wikibase_time(v) == "1950"
+
+def test_format_wikibase_time_century():
+    v = {"time": "+0800-00-00T00:00:00Z", "precision": 7}
+    assert utils.format_wikibase_time(v) == "8th century"
+
+    v = {"time": "+1950-00-00T00:00:00Z", "precision": 7}
+    assert utils.format_wikibase_time(v) == "20th century"