Disable "check unicode returns" under Python 3

Disabled the "unicode returns" check that runs on dialect startup when
running under Python 3, which for many years has occurred in order to test
the current DBAPI's behavior for whether or not it returns Python Unicode
or Py2K strings for the VARCHAR and NVARCHAR datatypes.  The check still
occurs by default under Python 2, however the mechanism to test the
behavior will be removed in SQLAlchemy 2.0 when Python 2 support is also
removed.

This logic was very effective when it was needed, however now that Python 3
is standard, all DBAPIs are expected to return Python 3 strings for
character datatypes.  In the unlikely case that a third party DBAPI does
not support this, the conversion logic within :class:`.String` is still
available and the third party dialect may specify this in its upfront
dialect flags by setting the dialect level flag ``returns_unicode_strings``
to one of :attr:`.String.RETURNS_CONDITIONAL` or
:attr:`.String.RETURNS_BYTES`, both of which will enable Unicode conversion
even under Python 3.

As part of this change, disabling testing of the doctest tutorials under
Python 2.

Fixes: #5315
Change-Id: I1260e894611409d3b7fe1a92bd90c52043bbcf19

9 files changed, 185 insertions, 21 deletions

diff --git a/doc/build/changelog/unreleased_14/5315.rst b/doc/build/changelog/unreleased_14/5315.rst
new file mode 100644
index 000000000..ff0073c80
--- /dev/null
+++ b/doc/build/changelog/unreleased_14/5315.rst
@@ -0,0 +1,21 @@
+.. change::
+    :tags: change, performance, engine, py3k
+    :tickets: 5315
+
+    Disabled the "unicode returns" check that runs on dialect startup when
+    running under Python 3, which for many years has occurred in order to test
+    the current DBAPI's behavior for whether or not it returns Python Unicode
+    or Py2K strings for the VARCHAR and NVARCHAR datatypes.  The check still
+    occurs by default under Python 2, however the mechanism to test the
+    behavior will be removed in SQLAlchemy 2.0 when Python 2 support is also
+    removed.
+
+    This logic was very effective when it was needed, however now that Python 3
+    is standard, all DBAPIs are expected to return Python 3 strings for
+    character datatypes.  In the unlikely case that a third party DBAPI does
+    not support this, the conversion logic within :class:`.String` is still
+    available and the third party dialect may specify this in its upfront
+    dialect flags by setting the dialect level flag ``returns_unicode_strings``
+    to one of :attr:`.String.RETURNS_CONDITIONAL` or
+    :attr:`.String.RETURNS_BYTES`, both of which will enable Unicode conversion
+    even under Python 3.
diff --git a/doc/build/core/tutorial.rst b/doc/build/core/tutorial.rst
index 1f475abe3..a504f8578 100644
--- a/doc/build/core/tutorial.rst
+++ b/doc/build/core/tutorial.rst
@@ -150,7 +150,7 @@ each table first before creating, so it's safe to call multiple times:
 .. sourcecode:: pycon+sql
 
     {sql}>>> metadata.create_all(engine)
-    SE...
+    PRAGMA...
     CREATE TABLE users (
         id INTEGER NOT NULL,
         name VARCHAR,
diff --git a/doc/build/orm/tutorial.rst b/doc/build/orm/tutorial.rst
index e77964a7b..c08daa7fd 100644
--- a/doc/build/orm/tutorial.rst
+++ b/doc/build/orm/tutorial.rst
@@ -209,7 +209,6 @@ the actual ``CREATE TABLE`` statement:
 .. sourcecode:: python+sql
 
     >>> Base.metadata.create_all(engine)
-    SELECT ...
     PRAGMA main.table_info("users")
     ()
     PRAGMA temp.table_info("users")
diff --git a/lib/sqlalchemy/engine/default.py b/lib/sqlalchemy/engine/default.py
index 20f731116..d9b4cdda6 100644
--- a/lib/sqlalchemy/engine/default.py
+++ b/lib/sqlalchemy/engine/default.py
@@ -94,12 +94,12 @@ class DefaultDialect(interfaces.Dialect):
     if util.py3k:
         supports_unicode_statements = True
         supports_unicode_binds = True
-        returns_unicode_strings = True
+        returns_unicode_strings = sqltypes.String.RETURNS_UNICODE
         description_encoding = None
     else:
         supports_unicode_statements = False
         supports_unicode_binds = False
-        returns_unicode_strings = False
+        returns_unicode_strings = sqltypes.String.RETURNS_UNKNOWN
         description_encoding = "use_encoding"
 
     name = "default"
@@ -325,7 +325,14 @@ class DefaultDialect(interfaces.Dialect):
         except NotImplementedError:
             self.default_isolation_level = None
 
-        self.returns_unicode_strings = self._check_unicode_returns(connection)
+        if self.returns_unicode_strings is sqltypes.String.RETURNS_UNKNOWN:
+            if util.py3k:
+                raise exc.InvalidRequestError(
+                    "RETURNS_UNKNOWN is unsupported in Python 3"
+                )
+            self.returns_unicode_strings = self._check_unicode_returns(
+                connection
+            )
 
         if (
             self.description_encoding is not None
@@ -415,9 +422,13 @@ class DefaultDialect(interfaces.Dialect):
         results = {check_unicode(test) for test in tests}
 
         if results.issuperset([True, False]):
-            return "conditional"
+            return sqltypes.String.RETURNS_CONDITIONAL
         else:
-            return results == {True}
+            return (
+                sqltypes.String.RETURNS_UNICODE
+                if results == {True}
+                else sqltypes.String.RETURNS_BYTES
+            )
 
     def _check_unicode_description(self, connection):
         # all DBAPIs on Py2K return cursor.description as encoded
diff --git a/lib/sqlalchemy/sql/sqltypes.py b/lib/sqlalchemy/sql/sqltypes.py
index a65989e93..8684a7922 100644
--- a/lib/sqlalchemy/sql/sqltypes.py
+++ b/lib/sqlalchemy/sql/sqltypes.py
@@ -135,6 +135,67 @@ class String(Concatenable, TypeEngine):
 
     __visit_name__ = "string"
 
+    RETURNS_UNICODE = util.symbol(
+        "RETURNS_UNICODE",
+        """Indicates that the DBAPI returns Python Unicode for VARCHAR,
+        NVARCHAR, and other character-based datatypes in all cases.
+
+        This is the default value for
+        :attr:`.DefaultDialect.returns_unicode_strings` under Python 3.
+
+        .. versionadded:: 1.4
+
+        """,
+    )
+
+    RETURNS_BYTES = util.symbol(
+        "RETURNS_BYTES",
+        """Indicates that the DBAPI returns byte objects under Python 3
+        or non-Unicode string objects under Python 2 for VARCHAR, NVARCHAR,
+        and other character-based datatypes in all cases.
+
+        This may be applied to the
+        :attr:`.DefaultDialect.returns_unicode_strings` attribute.
+
+        .. versionadded:: 1.4
+
+        """,
+    )
+
+    RETURNS_CONDITIONAL = util.symbol(
+        "RETURNS_CONDITIONAL",
+        """Indicates that the DBAPI may return Unicode or bytestrings for
+        VARCHAR, NVARCHAR, and other character-based datatypes, and that
+        SQLAlchemy's default String datatype will need to test on a per-row
+        basis for Unicode or bytes.
+
+        This may be applied to the
+        :attr:`.DefaultDialect.returns_unicode_strings` attribute.
+
+        .. versionadded:: 1.4
+
+        """,
+    )
+
+    RETURNS_UNKNOWN = util.symbol(
+        "RETURNS_UNKNOWN",
+        """Indicates that the dialect should test on first connect what the
+        string-returning behavior of character-based datatypes is.
+
+        This is the default value for DefaultDialect.unicode_returns under
+        Python 2.
+
+        This may be applied to the
+        :attr:`.DefaultDialect.returns_unicode_strings` attribute under
+        Python 2 only.   The value is disallowed under Python 3.
+
+        .. versionadded:: 1.4
+
+        .. deprecated:: 1.4  This value will be removed in SQLAlchemy 2.0.
+
+        """,
+    )
+
     @util.deprecated_params(
         convert_unicode=(
             "1.3",
@@ -293,12 +354,13 @@ class String(Concatenable, TypeEngine):
     def result_processor(self, dialect, coltype):
         wants_unicode = self._expect_unicode or dialect.convert_unicode
         needs_convert = wants_unicode and (
-            dialect.returns_unicode_strings is not True
+            dialect.returns_unicode_strings is not String.RETURNS_UNICODE
             or self._expect_unicode in ("force", "force_nocheck")
         )
         needs_isinstance = (
             needs_convert
             and dialect.returns_unicode_strings
+            in (String.RETURNS_CONDITIONAL, String.RETURNS_UNICODE,)
             and self._expect_unicode != "force_nocheck"
         )
         if needs_convert:
diff --git a/test/base/test_tutorials.py b/test/base/test_tutorials.py
index 3ac7913f9..69f9f7e90 100644
--- a/test/base/test_tutorials.py
+++ b/test/base/test_tutorials.py
@@ -12,6 +12,8 @@ from sqlalchemy.testing import fixtures
 
 
 class DocTest(fixtures.TestBase):
+    __requires__ = ("python3",)
+
     def _setup_logger(self):
         rootlogger = logging.getLogger("sqlalchemy.engine.Engine")
 
diff --git a/test/engine/test_execute.py b/test/engine/test_execute.py
index a8d7b397e..285dd2acf 100644
--- a/test/engine/test_execute.py
+++ b/test/engine/test_execute.py
@@ -648,6 +648,32 @@ class ExecuteTest(fixtures.TablesTest):
         eng2 = eng.execution_options(foo="bar")
         assert eng2._has_events
 
+    def test_works_after_dispose(self):
+        eng = create_engine(testing.db.url)
+        for i in range(3):
+            eq_(eng.scalar(select([1])), 1)
+            eng.dispose()
+
+    def test_works_after_dispose_testing_engine(self):
+        eng = engines.testing_engine()
+        for i in range(3):
+            eq_(eng.scalar(select([1])), 1)
+            eng.dispose()
+
+
+class UnicodeReturnsTest(fixtures.TestBase):
+    @testing.requires.python3
+    def test_unicode_test_not_in_python3(self):
+        eng = engines.testing_engine()
+        eng.dialect.returns_unicode_strings = String.RETURNS_UNKNOWN
+
+        assert_raises_message(
+            tsa.exc.InvalidRequestError,
+            "RETURNS_UNKNOWN is unsupported in Python 3",
+            eng.connect,
+        )
+
+    @testing.requires.python2
     def test_unicode_test_fails_warning(self):
         class MockCursor(engines.DBAPIProxyCursor):
             def execute(self, stmt, params=None, **kw):
@@ -663,21 +689,9 @@ class ExecuteTest(fixtures.TablesTest):
             eng.connect()
 
         # because plain varchar passed, we don't know the correct answer
-        eq_(eng.dialect.returns_unicode_strings, "conditional")
+        eq_(eng.dialect.returns_unicode_strings, String.RETURNS_CONDITIONAL)
         eng.dispose()
 
-    def test_works_after_dispose(self):
-        eng = create_engine(testing.db.url)
-        for i in range(3):
-            eq_(eng.scalar(select([1])), 1)
-            eng.dispose()
-
-    def test_works_after_dispose_testing_engine(self):
-        eng = engines.testing_engine()
-        for i in range(3):
-            eq_(eng.scalar(select([1])), 1)
-            eng.dispose()
-
 
 class ConvenienceExecuteTest(fixtures.TablesTest):
     __backend__ = True
diff --git a/test/engine/test_reconnect.py b/test/engine/test_reconnect.py
index f0d0a9b2f..48e324806 100644
--- a/test/engine/test_reconnect.py
+++ b/test/engine/test_reconnect.py
@@ -950,6 +950,16 @@ class CursorErrTest(fixtures.TestBase):
             url, options=dict(module=dbapi, _initialize=initialize)
         )
         eng.pool.logger = Mock()
+
+        def get_default_schema_name(connection):
+            try:
+                cursor = connection.connection.cursor()
+                connection._cursor_execute(cursor, "statement", {})
+                cursor.close()
+            except exc.DBAPIError:
+                util.warn("Exception attempting to detect")
+
+        eng.dialect._get_default_schema_name = get_default_schema_name
         return eng
 
     def test_cursor_explode(self):
diff --git a/test/sql/test_types.py b/test/sql/test_types.py
index e0c1359b4..a0c4ee022 100644
--- a/test/sql/test_types.py
+++ b/test/sql/test_types.py
@@ -86,6 +86,7 @@ from sqlalchemy.testing.schema import Table
 from sqlalchemy.testing.util import picklers
 from sqlalchemy.testing.util import round_decimal
 from sqlalchemy.util import OrderedDict
+from sqlalchemy.util import u
 
 
 def _all_dialect_modules():
@@ -763,6 +764,50 @@ class UserDefinedTest(
         eq_(a.dialect_specific_args["bar"], "bar")
 
 
+class StringConvertUnicodeTest(fixtures.TestBase):
+    @testing.combinations((Unicode,), (String,), argnames="datatype")
+    @testing.combinations((True,), (False,), argnames="convert_unicode")
+    @testing.combinations(
+        (String.RETURNS_CONDITIONAL,),
+        (String.RETURNS_BYTES,),
+        (String.RETURNS_UNICODE),
+        argnames="returns_unicode_strings",
+    )
+    def test_convert_unicode(
+        self, datatype, convert_unicode, returns_unicode_strings
+    ):
+        s1 = datatype()
+        dialect = mock.Mock(
+            returns_unicode_strings=returns_unicode_strings,
+            encoding="utf-8",
+            convert_unicode=convert_unicode,
+        )
+
+        proc = s1.result_processor(dialect, None)
+
+        string = u("méil")
+        bytestring = string.encode("utf-8")
+
+        if (
+            datatype is Unicode or convert_unicode
+        ) and returns_unicode_strings in (
+            String.RETURNS_CONDITIONAL,
+            String.RETURNS_BYTES,
+        ):
+            eq_(proc(bytestring), string)
+
+            if returns_unicode_strings is String.RETURNS_CONDITIONAL:
+                eq_(proc(string), string)
+            else:
+                if util.py3k:
+                    # trying to decode a unicode
+                    assert_raises(TypeError, proc, string)
+                else:
+                    assert_raises(UnicodeEncodeError, proc, string)
+        else:
+            is_(proc, None)
+
+
 class TypeCoerceCastTest(fixtures.TablesTest):
     __backend__ = True