Fixed #22350 -- Consistently serialize bytes and text in migrations.

Thanks to @treyhunner and Loïc for their suggestions and review.

13 files changed, 76 insertions, 12 deletions

diff --git a/django/db/migrations/state.py b/django/db/migrations/state.py
index ab2bd8e6d8..d146bfd8bc 100644
--- a/django/db/migrations/state.py
+++ b/django/db/migrations/state.py
@@ -227,7 +227,7 @@ class ModelState(object):
         body['__module__'] = "__fake__"
         # Then, make a Model object
         return type(
-            self.name,
+            str(self.name),
             bases,
             body,
         )
diff --git a/django/db/migrations/writer.py b/django/db/migrations/writer.py
index d9f9c734a2..7696aaf339 100644
--- a/django/db/migrations/writer.py
+++ b/django/db/migrations/writer.py
@@ -53,8 +53,10 @@ class OperationWriter(object):
                     self.feed('%s={' % arg_name)
                     self.indent()
                     for key, value in arg_value.items():
+                        key_string, key_imports = MigrationWriter.serialize(key)
                         arg_string, arg_imports = MigrationWriter.serialize(value)
-                        self.feed('%s: %s,' % (repr(key), arg_string))
+                        self.feed('%s: %s,' % (key_string, arg_string))
+                        imports.update(key_imports)
                         imports.update(arg_imports)
                     self.unindent()
                     self.feed('},')
@@ -122,7 +124,7 @@ class MigrationWriter(object):
                 dependencies.append("        migrations.swappable_dependency(settings.%s)," % dependency[1])
                 imports.add("from django.conf import settings")
             else:
-                dependencies.append("        %s," % repr(dependency))
+                dependencies.append("        %s," % self.serialize(dependency)[0])
         items["dependencies"] = "\n".join(dependencies) + "\n" if dependencies else ""
 
         # Format imports nicely
@@ -131,7 +133,7 @@ class MigrationWriter(object):
 
         # If there's a replaces, make a string for it
         if self.migration.replaces:
-            items['replaces_str'] = "\n    replaces = %s\n" % repr(self.migration.replaces)
+            items['replaces_str'] = "\n    replaces = %s\n" % self.serialize(self.migration.replaces)[0]
 
         return (MIGRATION_TEMPLATE % items).encode("utf8")
 
@@ -185,6 +187,12 @@ class MigrationWriter(object):
         More advanced than repr() as it can encode things
         like datetime.datetime.now.
         """
+        # FIXME: Ideally Promise would be reconstructible, but for now we
+        # use force_text on them and defer to the normal string serialization
+        # process.
+        if isinstance(value, Promise):
+            value = force_text(value)
+
         # Sequences
         if isinstance(value, (list, set, tuple)):
             imports = set()
@@ -229,11 +237,20 @@ class MigrationWriter(object):
         elif isinstance(value, SettingsReference):
             return "settings.%s" % value.setting_name, set(["from django.conf import settings"])
         # Simple types
-        elif isinstance(value, six.integer_types + (float, six.binary_type, six.text_type, bool, type(None))):
+        elif isinstance(value, six.integer_types + (float, bool, type(None))):
             return repr(value), set()
-        # Promise
-        elif isinstance(value, Promise):
-            return repr(force_text(value)), set()
+        elif isinstance(value, six.binary_type):
+            value_repr = repr(value)
+            if six.PY2:
+                # Prepend the `b` prefix since we're importing unicode_literals
+                value_repr = 'b' + value_repr
+            return value_repr, set()
+        elif isinstance(value, six.text_type):
+            value_repr = repr(value)
+            if six.PY2:
+                # Strip the `u` prefix since we're importing unicode_literals
+                value_repr = value_repr[1:]
+            return value_repr, set()
         # Decimal
         elif isinstance(value, decimal.Decimal):
             return repr(value), set(["from decimal import Decimal"])
@@ -286,6 +303,8 @@ class MigrationWriter(object):
 
 MIGRATION_TEMPLATE = """\
 # encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import models, migrations
 %(imports)s
 
diff --git a/tests/migrations/test_migrations/0001_initial.py b/tests/migrations/test_migrations/0001_initial.py
index 344bebdfe3..3544610eb4 100644
--- a/tests/migrations/test_migrations/0001_initial.py
+++ b/tests/migrations/test_migrations/0001_initial.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_migrations/0002_second.py b/tests/migrations/test_migrations/0002_second.py
index 736e844825..5da381677c 100644
--- a/tests/migrations/test_migrations/0002_second.py
+++ b/tests/migrations/test_migrations/0002_second.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_migrations_2/0001_initial.py b/tests/migrations/test_migrations_2/0001_initial.py
index 6172f011e8..cb15a04956 100644
--- a/tests/migrations/test_migrations_2/0001_initial.py
+++ b/tests/migrations/test_migrations_2/0001_initial.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_migrations_conflict/0001_initial.py b/tests/migrations/test_migrations_conflict/0001_initial.py
index 344bebdfe3..3544610eb4 100644
--- a/tests/migrations/test_migrations_conflict/0001_initial.py
+++ b/tests/migrations/test_migrations_conflict/0001_initial.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_migrations_conflict/0002_conflicting_second.py b/tests/migrations/test_migrations_conflict/0002_conflicting_second.py
index 15ea1f063a..cf83d7b036 100644
--- a/tests/migrations/test_migrations_conflict/0002_conflicting_second.py
+++ b/tests/migrations/test_migrations_conflict/0002_conflicting_second.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_migrations_conflict/0002_second.py b/tests/migrations/test_migrations_conflict/0002_second.py
index ace9a83347..0595de83f0 100644
--- a/tests/migrations/test_migrations_conflict/0002_second.py
+++ b/tests/migrations/test_migrations_conflict/0002_second.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_migrations_squashed/0001_initial.py b/tests/migrations/test_migrations_squashed/0001_initial.py
index 344bebdfe3..3544610eb4 100644
--- a/tests/migrations/test_migrations_squashed/0001_initial.py
+++ b/tests/migrations/test_migrations_squashed/0001_initial.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_migrations_squashed/0001_squashed_0002.py b/tests/migrations/test_migrations_squashed/0001_squashed_0002.py
index 742be641aa..b357e3f6f8 100644
--- a/tests/migrations/test_migrations_squashed/0001_squashed_0002.py
+++ b/tests/migrations/test_migrations_squashed/0001_squashed_0002.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_migrations_squashed/0002_second.py b/tests/migrations/test_migrations_squashed/0002_second.py
index ace9a83347..0595de83f0 100644
--- a/tests/migrations/test_migrations_squashed/0002_second.py
+++ b/tests/migrations/test_migrations_squashed/0002_second.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_migrations_unmigdep/0001_initial.py b/tests/migrations/test_migrations_unmigdep/0001_initial.py
index f4c11b4657..ff19085563 100644
--- a/tests/migrations/test_migrations_unmigdep/0001_initial.py
+++ b/tests/migrations/test_migrations_unmigdep/0001_initial.py
@@ -1,3 +1,6 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
 from django.db import migrations, models
 
 
diff --git a/tests/migrations/test_writer.py b/tests/migrations/test_writer.py
index c3889df0e7..4d552532d4 100644
--- a/tests/migrations/test_writer.py
+++ b/tests/migrations/test_writer.py
@@ -1,9 +1,9 @@
 # encoding: utf8
-
 from __future__ import unicode_literals
 
 import datetime
 import os
+import tokenize
 
 from django.core.validators import RegexValidator, EmailValidator
 from django.db import models, migrations
@@ -59,7 +59,11 @@ class WriterTests(TestCase):
         self.assertSerializedEqual(1)
         self.assertSerializedEqual(None)
         self.assertSerializedEqual(b"foobar")
+        string, imports = MigrationWriter.serialize(b"foobar")
+        self.assertEqual(string, "b'foobar'")
         self.assertSerializedEqual("föobár")
+        string, imports = MigrationWriter.serialize("foobar")
+        self.assertEqual(string, "'foobar'")
         self.assertSerializedEqual({1: 2})
         self.assertSerializedEqual(["a", 2, True, None])
         self.assertSerializedEqual(set([2, 3, "eighty"]))
@@ -92,15 +96,15 @@ class WriterTests(TestCase):
         # Classes
         validator = RegexValidator(message="hello")
         string, imports = MigrationWriter.serialize(validator)
-        self.assertEqual(string, "django.core.validators.RegexValidator(message=%s)" % repr("hello"))
+        self.assertEqual(string, "django.core.validators.RegexValidator(message='hello')")
         self.serialize_round_trip(validator)
         validator = EmailValidator(message="hello")  # Test with a subclass.
         string, imports = MigrationWriter.serialize(validator)
-        self.assertEqual(string, "django.core.validators.EmailValidator(message=%s)" % repr("hello"))
+        self.assertEqual(string, "django.core.validators.EmailValidator(message='hello')")
         self.serialize_round_trip(validator)
         validator = deconstructible(path="custom.EmailValidator")(EmailValidator)(message="hello")
         string, imports = MigrationWriter.serialize(validator)
-        self.assertEqual(string, "custom.EmailValidator(message=%s)" % repr("hello"))
+        self.assertEqual(string, "custom.EmailValidator(message='hello')")
         # Django fields
         self.assertSerializedFieldEqual(models.CharField(max_length=255))
         self.assertSerializedFieldEqual(models.TextField(null=True, blank=True))
@@ -153,6 +157,17 @@ class WriterTests(TestCase):
         # Just make sure it runs for now, and that things look alright.
         result = self.safe_exec(output)
         self.assertIn("Migration", result)
+        # In order to preserve compatibility with Python 3.2 unicode literals
+        # prefix shouldn't be added to strings.
+        tokens = tokenize.generate_tokens(six.StringIO(str(output)).readline)
+        for token_type, token_source, (srow, scol), _, line in tokens:
+            if token_type == tokenize.STRING:
+                self.assertFalse(
+                    token_source.startswith('u'),
+                    "Unicode literal prefix found at %d:%d: %r" % (
+                        srow, scol, line.strip()
+                    )
+                )
 
     def test_migration_path(self):
         test_apps = [