Python 3 support

.travis.yml

@@ -1,8 +1,10 @@
 language: python
-python:
- - "2.6"
- - "2.7"
- - "pypy"
-install: python setup.py install
+before_install:
+    - sudo apt-get update -qq -y
+    - sudo apt-get install python3.4-dev
+
+install:
+    - pip install tox six
+
 before_script: mysql -e 'create database mysqldb_test charset utf8;'
-script: TESTDB=travis.cnf nosetests
+script: TESTDB=travis.cnf tox

MANIFEST.in

@@ -6,7 +6,6 @@ include HISTORY
 include INSTALL
 include README.md
 include GPL-2.0
-include pymemcompat.h
 include metadata.cfg
 include site.cfg
 include setup_common.py

MySQLdb/__init__.py

@@ -73,7 +73,7 @@ def test_DBAPISet_set_inequality_membership():
     assert FIELD_TYPE.DATE != STRING
 
 def Binary(x):
-    return str(x)
+    return bytes(x)
 
 def Connect(*args, **kwargs):
     """Factory function for connections.Connection."""

MySQLdb/compat.py

@@ -0,0 +1,12 @@
+import sys
+
+if sys.version_info[0] == 3:
+    PY2 = False
+    unicode = str
+    unichr = chr
+    long = int
+else:
+    PY2 = True
+    unicode = unicode
+    unichr = unichr
+    long = long

MySQLdb/connections.py

@@ -7,10 +7,11 @@
 
 """
 from MySQLdb import cursors
+from MySQLdb.compat import unicode, PY2
 from _mysql_exceptions import Warning, Error, InterfaceError, DataError, \
      DatabaseError, OperationalError, IntegrityError, InternalError, \
      NotSupportedError, ProgrammingError
-import types, _mysql
+import _mysql
 import re
 
 
@@ -33,7 +34,11 @@ def defaulterrorhandler(connection, cursor, errorclass, errorvalue):
         connection.messages.append(error)
     del cursor
     del connection
-    raise errorclass, errorvalue
+    if errorclass is not None:    
+        raise errorclass(errorvalue)
+    else:
+        raise Exception(errorvalue)
+
 
 re_numeric_part = re.compile(r"^(\d+)")
 
@@ -115,6 +120,7 @@ class object, used to create cursors (keyword only)
           columns are returned as strings.  columns are returned as
           normal strings. Unicode objects will always be encoded to
           the connection's character set regardless of this setting.
+          Default to False on Python 2 and True on Python 3.
 
         charset
           If supplied, the connection character set will be changed
@@ -170,7 +176,7 @@ class object, used to create cursors (keyword only)
         cursorclass = kwargs2.pop('cursorclass', self.default_cursor)
         charset = kwargs2.pop('charset', '')
 
-        if charset:
+        if charset or not PY2:
             use_unicode = True
         else:
             use_unicode = False
@@ -199,13 +205,20 @@ class object, used to create cursors (keyword only)
 
         db = proxy(self)
         def _get_string_literal():
+            # Note: string_literal() is called for bytes object on Python 3.
             def string_literal(obj, dummy=None):
                 return db.string_literal(obj)
             return string_literal
 
         def _get_unicode_literal():
-            def unicode_literal(u, dummy=None):
-                return db.literal(u.encode(unicode_literal.charset))
+            if PY2:
+                # unicode_literal is called for only unicode object.
+                def unicode_literal(u, dummy=None):
+                    return db.literal(u.encode(unicode_literal.charset))
+            else:
+                # unicode_literal() is called for arbitrary object.
+                def unicode_literal(u, dummy=None):
+                    return db.literal(str(u).encode(unicode_literal.charset))
             return unicode_literal
 
         def _get_string_decoder():
@@ -229,8 +242,8 @@ def string_decoder(s):
             self.converter[FIELD_TYPE.VARCHAR].append((None, string_decoder))
             self.converter[FIELD_TYPE.BLOB].append((None, string_decoder))
 
-        self.encoders[types.StringType] = string_literal
-        self.encoders[types.UnicodeType] = unicode_literal
+        self.encoders[bytes] = string_literal
+        self.encoders[unicode] = unicode_literal
         self._transactional = self.server_capabilities & CLIENT.TRANSACTIONS
         if self._transactional:
             if autocommit is not None:
@@ -275,7 +288,15 @@ def literal(self, o):
         applications.
 
         """
-        return self.escape(o, self.encoders)
+        s = self.escape(o, self.encoders)
+        # Python 3 doesn't support % operation for bytes object.
+        # We should decode it before using %.
+        # Decoding with ascii and surrogateescape allows convert arbitrary
+        # bytes to unicode and back again.
+        # See http://python.org/dev/peps/pep-0383/
+        if not PY2 and isinstance(s, bytes):
+            return s.decode('ascii', 'surrogateescape')
+        return s
 
     def begin(self):
         """Explicitly begin a connection. Non-standard.

MySQLdb/converters.py

@@ -35,16 +35,9 @@
 from _mysql import string_literal, escape_sequence, escape_dict, escape, NULL
 from MySQLdb.constants import FIELD_TYPE, FLAG
 from MySQLdb.times import *
+from MySQLdb.compat import PY2, long
 
-try:
-    from types import IntType, LongType, FloatType, NoneType, TupleType, ListType, DictType, InstanceType, \
-        StringType, UnicodeType, ObjectType, BooleanType, ClassType, TypeType
-except ImportError:
-    # Python 3
-    long = int
-    IntType, LongType, FloatType, NoneType = int, long, float, type(None)
-    TupleType, ListType, DictType, InstanceType = tuple, list, dict, None
-    StringType, UnicodeType, ObjectType, BooleanType = bytes, str, object, bool
+NoneType = type(None)
 
 import array
 
@@ -76,8 +69,6 @@ def Unicode2Str(s, d):
     is connection-dependent."""
     return s.encode()
 
-Long2Int = Thing2Str
-
 def Float2Str(o, d):
     return '%.15g' % o
 
@@ -86,43 +77,13 @@ def None2NULL(o, d):
     return NULL # duh
 
 def Thing2Literal(o, d):
-
     """Convert something into a SQL string literal.  If using
     MySQL-3.23 or newer, string_literal() is a method of the
     _mysql.MYSQL object, and this function will be overridden with
     that method when the connection is created."""
-
     return string_literal(o, d)
 
 
-def Instance2Str(o, d):
-
-    """
-
-    Convert an Instance to a string representation.  If the __str__()
-    method produces acceptable output, then you don't need to add the
-    class to conversions; it will be handled by the default
-    converter. If the exact class is not found in d, it will use the
-    first class it can find for which o is an instance.
-
-    """
-
-    if o.__class__ in d:
-        return d[o.__class__](o, d)
-    cl = filter(lambda x,o=o:
-                type(x) is ClassType
-                and isinstance(o, x), d.keys())
-    if not cl:
-        cl = filter(lambda x,o=o:
-                    type(x) is TypeType
-                    and isinstance(o, x)
-                    and d[x] is not Instance2Str,
-                    d.keys())
-    if not cl:
-        return d[StringType](o,d)
-    d[o.__class__] = d[cl[0]]
-    return d[cl[0]](o, d)
-
 def char_array(s):
     return array.array('c', s)
 
@@ -133,21 +94,19 @@ def quote_tuple(t, d):
     return "(%s)" % (','.join(escape_sequence(t, d)))
 
 conversions = {
-    IntType: Thing2Str,
-    LongType: Long2Int,
-    FloatType: Float2Str,
+    int: Thing2Str,
+    long: Thing2Str,
+    float: Float2Str,
     NoneType: None2NULL,
-    TupleType: quote_tuple,
-    ListType: quote_tuple,
-    DictType: escape_dict,
-    InstanceType: Instance2Str,
+    tuple: quote_tuple,
+    list: quote_tuple,
+    dict: escape_dict,
     ArrayType: array2Str,
-    StringType: Thing2Literal, # default
-    UnicodeType: Unicode2Str,
-    ObjectType: Instance2Str,
-    BooleanType: Bool2Str,
+    bool: Bool2Str,
+    Date: Thing2Literal,
     DateTimeType: DateTime2literal,
     DateTimeDeltaType: DateTimeDelta2literal,
+    str: Thing2Literal,  # default
     set: Set2Str,
     FIELD_TYPE.TINY: int,
     FIELD_TYPE.SHORT: int,
@@ -165,25 +124,26 @@ def quote_tuple(t, d):
     FIELD_TYPE.TIME: TimeDelta_or_None,
     FIELD_TYPE.DATE: Date_or_None,
     FIELD_TYPE.BLOB: [
-        (FLAG.BINARY, str),
+        (FLAG.BINARY, bytes),
         ],
     FIELD_TYPE.STRING: [
-        (FLAG.BINARY, str),
+        (FLAG.BINARY, bytes),
         ],
     FIELD_TYPE.VAR_STRING: [
-        (FLAG.BINARY, str),
+        (FLAG.BINARY, bytes),
         ],
     FIELD_TYPE.VARCHAR: [
-        (FLAG.BINARY, str),
+        (FLAG.BINARY, bytes),
         ],
     }
+if PY2:
+    conversions[unicode] = Unicode2Str
+else:
+    conversions[bytes] = Thing2Literal
 
 try:
     from decimal import Decimal
     conversions[FIELD_TYPE.DECIMAL] = Decimal
     conversions[FIELD_TYPE.NEWDECIMAL] = Decimal
 except ImportError:
     pass
-
-
-