[#71] Output UTF-8 encoded JSON

Author: Bjwebb

flattentool/__init__.py

@@ -4,6 +4,7 @@
 from flattentool.output import FORMATS_SUFFIX
 from flattentool.input import FORMATS as INPUT_FORMATS
 import json
+import codecs
 from decimal import Decimal
 from collections import OrderedDict
 
@@ -122,6 +123,6 @@ def unflatten(input_name, base_json=None, input_format=None, output_name='releas
     else:
         base = OrderedDict()
     base[main_sheet_name] = list(spreadsheet_input.unflatten())
-    with open(output_name, 'w') as fp:
-        json.dump(base, fp, indent=4, default=decimal_default)
+    with codecs.open(output_name, 'w', encoding='utf-8') as fp:
+        json.dump(base, fp, indent=4, default=decimal_default, ensure_ascii=False)
 

flattentool/tests/test_init.py

@@ -174,6 +174,9 @@ def test_unflatten_csv_utf8(tmpdir):
         main_sheet_name='main')
     reloaded_json = json.load(tmpdir.join('release.json'))
     assert reloaded_json == {'main': [{'ocid': '1', 'id': 'éαГ😼𝒞人'}]}
+    # The JSON we output should be UTF-8, rather than escaped ASCII
+    # https://github.com/OpenDataServices/flatten-tool/issues/71
+    assert 'éαГ😼𝒞人' in tmpdir.join('release.json').read_text(encoding='utf-8')
 
 
 def test_unflatten_csv_latin1(tmpdir):