Merge takluyver/tseries-py3

Author: wesm

pandas/core/common.py

@@ -13,6 +13,13 @@
     from cStringIO import StringIO as BytesIO
 import itertools
 
+try:
+    next
+except NameError:  # pragma: no cover
+    # Python < 2.6
+    def next(x):
+        return x.next()
+
 from cStringIO import StringIO
 
 from numpy.lib.format import read_array, write_array
@@ -519,7 +526,7 @@ def iterpairs(seq):
     # input may not be sliceable
     seq_it = iter(seq)
     seq_it_next = iter(seq)
-    _ = seq_it_next.next()
+    _ = next(seq_it_next)
 
     return itertools.izip(seq_it, seq_it_next)
 

pandas/core/index.py

@@ -8,6 +8,7 @@
 import numpy as np
 
 from pandas.util.decorators import cache_readonly
+from pandas.util import py3compat
 import pandas.core.common as com
 import pandas._tseries as lib
 import pandas._engines as _gin

pandas/io/parsers.py

@@ -6,6 +6,13 @@
 from itertools import izip
 from urlparse import urlparse
 
+try:
+    next
+except NameError:  # pragma: no cover
+    # Python < 2.6
+    def next(x):
+        return x.next()
+
 import numpy as np
 
 from pandas.core.index import Index, MultiIndex
@@ -489,9 +496,9 @@ def _next_line(self):
                 raise StopIteration
         else:
             while self.pos in self.skiprows:
-                self.data.next()
+                next(self.data)
                 self.pos += 1
-            line = self.data.next()
+            line = next(self.data)
         self.pos += 1
         self.buf.append(line)
 
@@ -682,10 +689,10 @@ def _get_lines(self, rows=None):
             try:
                 if rows is not None:
                     for _ in xrange(rows):
-                        lines.append(source.next())
+                        lines.append(next(source))
                 else:
                     while True:
-                        lines.append(source.next())
+                        lines.append(next(source))
             except StopIteration:
                 if len(lines) == 0:
                     raise
@@ -757,10 +764,13 @@ def __init__(self, f, colspecs, filler):
             assert isinstance(colspec[1], int)
 
     def next(self):
-        line = self.f.next()
+        line = next(self.f)
         # Note: 'colspecs' is a sequence of half-open intervals.
         return [line[fromm:to].strip(self.filler or ' ')
                 for (fromm, to) in self.colspecs]
+    
+    # Iterator protocol in Python 3 uses __next__()
+    __next__ = next
 
 
 class FixedWidthFieldParser(TextParser):

pandas/src/datetime.pyx

@@ -34,6 +34,12 @@ ctypedef enum time_res:
     r_max = 98
     r_invalid = 99
 
+try:
+    basestring
+except NameError: # py3
+    basestring = str
+
+
 # Python front end to C extension type _Timestamp
 # This serves as the box for datetime64
 class Timestamp(_Timestamp):
@@ -1107,12 +1113,12 @@ cpdef int64_t period_asfreq(int64_t period_ordinal, int base1, int64_t mult1,
     cdef:
         int64_t retval
 
-    if relation not in ('S', 'E'):
+    if relation not in (b'S', b'E'):
         raise ValueError('relation argument must be one of S or E')
 
     period_ordinal = remove_mult(period_ordinal, mult1)
 
-    if mult1 != 1 and relation == 'E':
+    if mult1 != 1 and relation == b'E':
         period_ordinal += (mult1 - 1)
 
     retval = asfreq(period_ordinal, base1, base2, (<char*>relation)[0])
@@ -1130,7 +1136,7 @@ def period_asfreq_arr(ndarray[int64_t] arr, int base1, int64_t mult1, int base2,
         ndarray[int64_t] new_arr
         Py_ssize_t i, sz
 
-    if relation not in ('S', 'E'):
+    if relation not in (b'S', b'E'):
         raise ValueError('relation argument must be one of S or E')
 
     sz = len(arr)

pandas/src/np_datetime.c

@@ -11,6 +11,23 @@
 #include <numpy/ndarrayobject.h>
 #include "np_datetime.h"
 
+#if PY_MAJOR_VERSION >= 3
+  #define PyIntObject                  PyLongObject
+  #define PyInt_Type                   PyLong_Type
+  #define PyInt_Check(op)              PyLong_Check(op)
+  #define PyInt_CheckExact(op)         PyLong_CheckExact(op)
+  #define PyInt_FromString             PyLong_FromString
+  #define PyInt_FromUnicode            PyLong_FromUnicode
+  #define PyInt_FromLong               PyLong_FromLong
+  #define PyInt_FromSize_t             PyLong_FromSize_t
+  #define PyInt_FromSsize_t            PyLong_FromSsize_t
+  #define PyInt_AsLong                 PyLong_AsLong
+  #define PyInt_AS_LONG                PyLong_AS_LONG
+  #define PyInt_AsSsize_t              PyLong_AsSsize_t
+  #define PyInt_AsUnsignedLongMask     PyLong_AsUnsignedLongMask
+  #define PyInt_AsUnsignedLongLongMask PyLong_AsUnsignedLongLongMask
+#endif
+
 /*
  * Returns 1 if the given year is a leap year, 0 otherwise.
  */

pandas/tseries/offsets.py

@@ -938,6 +938,11 @@ def __eq__(self, other):
         else:
             return DateOffset.__eq__(self, other)
 
+    # This is identical to DateOffset.__hash__, but has to be redefined here
+    # for Python 3, because we've redefined __eq__.
+    def __hash__(self):
+        return hash(self._params())
+
     def __ne__(self, other):
         if isinstance(other, basestring):
             from pandas.tseries.frequencies import to_offset

pandas/tseries/period.py

@@ -8,6 +8,7 @@
 
 import pandas.core.common as com
 import pandas.core.datetools as datetools
+from pandas.util import py3compat
 
 from pandas._tseries import Timestamp
 import pandas._tseries as lib
@@ -185,7 +186,7 @@ def asfreq(self, freq=None, how='E'):
         base2, mult2 = _gfc(freq)
 
         new_ordinal = lib.period_asfreq(self.ordinal, base1, mult1,
-                                        base2, mult2, how)
+                                        base2, mult2, py3compat.str_to_bytes(how))
 
         return Period(new_ordinal, (base2, mult2))
 
@@ -514,7 +515,7 @@ def __new__(cls, data=None,
                     base1, mult1 = _gfc(data.freq)
                     base2, mult2 = _gfc(freq)
                     data = lib.period_asfreq_arr(data.values, base1, mult1,
-                                                 base2, mult2, 'E')
+                                                 base2, mult2, b'E')
             else:
                 if freq is None:
                     raise ValueError('freq cannot be none')
@@ -558,7 +559,7 @@ def asfreq(self, freq=None, how='E'):
 
         new_data = lib.period_asfreq_arr(self.values,
                                          base1, mult1,
-                                         base2, mult2, how)
+                                         base2, mult2, py3compat.str_to_bytes(how))
 
         return PeriodIndex(new_data, freq=freq)
 

pandas/util/py3compat.py

@@ -5,10 +5,16 @@
 if PY3:
     def isidentifier(s):
         return s.isidentifier()
+        
+    def str_to_bytes(s, encoding='ascii'):
+        return s.encode(encoding)
 
 else:
     # Python 2
     import re
     _name_re = re.compile(r"[a-zA-Z_][a-zA-Z0-9_]*$")
     def isidentifier(s, dotted=False):
         return bool(_name_re.match(s))
+    
+    def str_to_bytes(s, encoding='ascii'):
+        return s

setup.py

@@ -40,6 +40,8 @@
                          'zip_safe': False,
                          'install_requires': ['python-dateutil >= 2',
                                               'numpy >= 1.4'],
+                         'use_2to3_exclude_fixers': ['lib2to3.fixes.fix_next',
+                                                    ],
                         }
     if not _have_setuptools:
         sys.exit("need setuptools/distribute for Py3k"