Properly serialize Fluent placeables generated by Pretranslation (moz…

…illa#3061)

After Fluent pretranslations are created, we need to parse and serialize them again in order to make sure they are stored in the canonical form (e.g. with spaces within curly braces).

Fix test:
- The following code only applies to non-FTL strings: "Empty strings and strings containing whitespace only do not need translation".

pontoon/pretranslation/pretranslate.py

@@ -58,6 +58,10 @@ def get_pretranslations(entity, locale, preserve_placeables=False):
 
         pretranslation = serializer.serialize_entry(entry)
 
+        # Parse and serialize pretranslation again in order to assure cannonical style
+        parsed_pretranslation = parser.parse_entry(pretranslation)
+        pretranslation = serializer.serialize_entry(parsed_pretranslation)
+
         authors = [services[service] for service in pretranslate.services]
         author = max(set(authors), key=authors.count) if authors else services["tm"]
 
@@ -82,7 +86,7 @@ def get_pretranslations(entity, locale, preserve_placeables=False):
 
 
 def get_pretranslated_data(source, locale, preserve_placeables):
-    # Empty strings do not need translation
+    # Empty strings and strings containing whitespace only do not need translation
     if re.search("^\\s*$", source):
         return source, "tm"
 

pontoon/pretranslation/tests/test_pretranslate.py

@@ -47,6 +47,22 @@ def test_get_pretranslations_empty_string(entity_a, locale_b, tm_user):
     assert response == [("", None, tm_user)]
 
 
+@pytest.mark.django_db
+def test_get_pretranslations_whitespace(entity_a, locale_b, tm_user):
+    # Entity.string is an empty string
+    entity_a.string = " "
+    response = get_pretranslations(entity_a, locale_b)
+    assert response == [(" ", None, tm_user)]
+
+    entity_a.string = "\t"
+    response = get_pretranslations(entity_a, locale_b)
+    assert response == [("\t", None, tm_user)]
+
+    entity_a.string = "\n"
+    response = get_pretranslations(entity_a, locale_b)
+    assert response == [("\n", None, tm_user)]
+
+
 @pytest.mark.django_db
 def test_get_pretranslations_tm_match(entity_a, entity_b, locale_b, tm_user):
     # 100% TM match exists
@@ -145,32 +161,6 @@ def test_get_pretranslations_fluent_empty(
     assert response == [(pretranslated_string, None, gt_user)]
 
 
-@pytest.mark.django_db
-def test_get_pretranslations_fluent_whitespace(
-    fluent_resource, google_translate_locale, tm_user
-):
-    # Various types of whitespace should be preserved
-    input_string = dedent(
-        """
-        whitespace =
-            { $count ->
-                [0] { "" }
-                [1] { " " }
-                *[other] { "\t" } { "\n" }
-            }
-    """
-    )
-    fluent_entity = EntityFactory(resource=fluent_resource, string=input_string)
-
-    output_string = input_string
-
-    # Re-serialize to match whitespace
-    pretranslated_string = serializer.serialize_entry(parser.parse_entry(output_string))
-
-    response = get_pretranslations(fluent_entity, google_translate_locale)
-    assert response == [(pretranslated_string, None, tm_user)]
-
-
 @patch("pontoon.pretranslation.pretranslate.get_google_translate_data")
 @pytest.mark.django_db
 def test_get_pretranslations_fluent_accesskeys_no_attribute_source(