def _align_frame(self, indexer, df):
is_frame = self.obj.ndim == 2
is_panel = self.obj.ndim >= 3
if isinstance(indexer, tuple):
idx, cols = None, None
sindexers = []
for i, ix in enumerate(indexer):
ax = self.obj.axes[i]
if com._is_sequence(ix) or isinstance(ix, slice):
if idx is None:
idx = ax[ix].ravel()
elif cols is None:
cols = ax[ix].ravel()
else:
break
else:
sindexers.append(i)
# panel
if is_panel:
if len(sindexers) == 1 and idx is None and cols is None:
if sindexers[0] == 0:
df = df.T
return self.obj.conform(df, axis=sindexers[0])
df = df.T
if idx is not None and cols is not None:
if df.index.equals(idx) and df.columns.equals(cols):
val = df.copy().values
else:
val = df.reindex(idx, columns=cols).values
return val
elif (isinstance(indexer, slice) or com.is_list_like(indexer)) and is_frame:
ax = self.obj.index[indexer]
if df.index.equals(ax):
val = df.copy().values
else:
val = df.reindex(ax).values
return val
elif np.isscalar(indexer) and not is_frame:
idx = self.obj.axes[1]
cols = self.obj.axes[2]
# by definition we are indexing on the 0th axis
if is_panel:
df = df.T
if idx.equals(df.index) and cols.equals(df.columns):
return df.copy().values
# a passed in dataframe which is actually a transpose
# of what is needed
elif idx.equals(df.columns) and cols.equals(df.index):
return df.T.copy().values
return df.reindex(idx, columns=cols).values
raise ValueError("Incompatible indexer with DataFrame")
def _convert_to_list_like(list_like):
if hasattr(list_like, "dtype"):
return list_like
if isinstance(list_like, list):
return list_like
if (com._is_sequence(list_like) or isinstance(list_like, tuple)
or isinstance(list_like, types.GeneratorType)):
return list(list_like)
elif np.isscalar(list_like):
return [list_like]
else:
# is this reached?
return [list_like]
def _convert_to_list_like(list_like):
if hasattr(list_like, "dtype"):
return list_like
if isinstance(list_like, list):
return list_like
if (com._is_sequence(list_like) or isinstance(list_like, tuple)
or isinstance(list_like, types.GeneratorType)):
return list(list_like)
elif np.isscalar(list_like):
return [list_like]
else:
# is this reached?
return [list_like]
def _align_frame(self, indexer, df):
from pandas import DataFrame
is_frame = isinstance(self.obj, DataFrame)
if not is_frame:
df = df.T
if isinstance(indexer, tuple):
idx, cols = None, None
for i, ix in enumerate(indexer):
ax = self.obj.axes[i]
if com._is_sequence(ix) or isinstance(ix, slice):
if idx is None:
idx = ax[ix].ravel()
elif cols is None:
cols = ax[ix].ravel()
else:
break
if idx is not None and cols is not None:
if df.index.equals(idx) and df.columns.equals(cols):
val = df.copy().values
else:
val = df.reindex(idx, columns=cols).values
return val
elif ((isinstance(indexer, slice) or com.is_list_like(indexer))
and is_frame):
ax = self.obj.index[indexer]
if df.index.equals(ax):
val = df.copy().values
else:
val = df.reindex(ax).values
return val
elif np.isscalar(indexer) and not is_frame:
idx = self.obj.axes[1]
cols = self.obj.axes[2]
if idx.equals(df.index) and cols.equals(df.columns):
return df.copy().values
return df.reindex(idx, columns=cols).values
raise ValueError('Incompatible indexer with DataFrame')
def _align_frame(self, indexer, df):
from pandas import DataFrame
is_frame = isinstance(self.obj, DataFrame)
if not is_frame:
df = df.T
if isinstance(indexer, tuple):
idx, cols = None, None
for i, ix in enumerate(indexer):
ax = self.obj.axes[i]
if com._is_sequence(ix) or isinstance(ix, slice):
if idx is None:
idx = ax[ix]
elif cols is None:
cols = ax[ix]
else:
break
if idx is not None and cols is not None:
if df.index.equals(idx) and df.columns.equals(cols):
val = df.copy().values
else:
val = df.reindex(idx, columns=cols).values
return val
elif ((isinstance(indexer, slice) or com.is_list_like(indexer))
and is_frame):
ax = self.obj.index[indexer]
if df.index.equals(ax):
val = df.copy().values
else:
val = df.reindex(ax).values
return val
elif np.isscalar(indexer) and not is_frame:
idx = self.obj.axes[1]
cols = self.obj.axes[2]
if idx.equals(df.index) and cols.equals(df.columns):
return df.copy().values
return df.reindex(idx, columns=cols).values
raise ValueError('Incompatible indexer with DataFrame')
def _align_series(self, indexer, ser):
# indexer to assign Series can be tuple or scalar
if isinstance(indexer, tuple):
for i, idx in enumerate(indexer):
ax = self.obj.axes[i]
if com._is_sequence(idx) or isinstance(idx, slice):
new_ix = ax[idx]
if ser.index.equals(new_ix):
return ser.values.copy()
return ser.reindex(new_ix).values
elif np.isscalar(indexer):
ax = self.obj._get_axis(1)
if ser.index.equals(ax):
return ser.values.copy()
return ser.reindex(ax).values
raise ValueError('Incompatible indexer with Series')
def _align_series(self, indexer, ser):
# indexer to assign Series can be tuple or scalar
if isinstance(indexer, tuple):
for i, idx in enumerate(indexer):
ax = self.obj.axes[i]
if com._is_sequence(idx) or isinstance(idx, slice):
new_ix = ax[idx]
if ser.index.equals(new_ix):
return ser.values.copy()
return ser.reindex(new_ix).values
elif np.isscalar(indexer):
ax = self.obj._get_axis(1)
if ser.index.equals(ax):
return ser.values.copy()
return ser.reindex(ax).values
raise ValueError('Incompatible indexer with Series')
def makeCustomIndex(nentries, nlevels, prefix='#', names=False, ndupe_l=None,
idx_type=None):
"""Create an index/multindex with given dimensions, levels, names, etc'
nentries - number of entries in index
nlevels - number of levels (> 1 produces multindex)
prefix - a string prefix for labels
names - (Optional), bool or list of strings. if True will use default names,
if false will use no names, if a list is given, the name of each level
in the index will be taken from the list.
ndupe_l - (Optional), list of ints, the number of rows for which the
label will repeated at the corresponding level, you can specify just
the first few, the rest will use the default ndupe_l of 1.
len(ndupe_l) <= nlevels.
idx_type - "i"/"f"/"s"/"u"/"dt/"p".
If idx_type is not None, `idx_nlevels` must be 1.
"i"/"f" creates an integer/float index,
"s"/"u" creates a string/unicode index
"dt" create a datetime index.
if unspecified, string labels will be generated.
"""
if ndupe_l is None:
ndupe_l = [1] * nlevels
assert (_is_sequence(ndupe_l) and len(ndupe_l) <= nlevels)
assert (names is None or names is False
or names is True or len(names) is nlevels)
assert idx_type is None or \
(idx_type in ('i', 'f', 's', 'u', 'dt', 'p') and nlevels == 1)
if names is True:
# build default names
names = [prefix + str(i) for i in range(nlevels)]
if names is False:
# pass None to index constructor for no name
names = None
# make singelton case uniform
if isinstance(names, compat.string_types) and nlevels == 1:
names = [names]
# specific 1D index type requested?
idx_func = dict(i=makeIntIndex, f=makeFloatIndex, s=makeStringIndex,
u=makeUnicodeIndex, dt=makeDateIndex, p=makePeriodIndex).get(idx_type)
if idx_func:
idx = idx_func(nentries)
# but we need to fill in the name
if names:
idx.name = names[0]
return idx
elif idx_type is not None:
raise ValueError('"%s" is not a legal value for `idx_type`, use '
'"i"/"f"/"s"/"u"/"dt/"p".' % idx_type)
if len(ndupe_l) < nlevels:
ndupe_l.extend([1] * (nlevels - len(ndupe_l)))
assert len(ndupe_l) == nlevels
assert all([x > 0 for x in ndupe_l])
tuples = []
for i in range(nlevels):
def keyfunc(x):
import re
numeric_tuple = re.sub("[^\d_]_?", "", x).split("_")
return lmap(int, numeric_tuple)
# build a list of lists to create the index from
div_factor = nentries // ndupe_l[i] + 1
cnt = Counter()
for j in range(div_factor):
label = prefix + '_l%d_g' % i + str(j)
cnt[label] = ndupe_l[i]
# cute Counter trick
result = list(sorted(cnt.elements(), key=keyfunc))[:nentries]
tuples.append(result)
tuples = lzip(*tuples)
# convert tuples to index
if nentries == 1:
index = Index(tuples[0], name=names[0])
else:
index = MultiIndex.from_tuples(tuples, names=names)
return index
def makeCustomIndex(nentries,
nlevels,
prefix='#',
names=False,
ndupe_l=None,
idx_type=None):
"""Create an index/multindex with given dimensions, levels, names, etc'
nentries - number of entries in index
nlevels - number of levels (> 1 produces multindex)
prefix - a string prefix for labels
names - (Optional), bool or list of strings. if True will use default names,
if false will use no names, if a list is given, the name of each level
in the index will be taken from the list.
ndupe_l - (Optional), list of ints, the number of rows for which the
label will repeated at the corresponding level, you can specify just
the first few, the rest will use the default ndupe_l of 1.
len(ndupe_l) <= nlevels.
idx_type - "i"/"f"/"s"/"u"/"dt/"p".
If idx_type is not None, `idx_nlevels` must be 1.
"i"/"f" creates an integer/float index,
"s"/"u" creates a string/unicode index
"dt" create a datetime index.
if unspecified, string labels will be generated.
"""
if ndupe_l is None:
ndupe_l = [1] * nlevels
assert (_is_sequence(ndupe_l) and len(ndupe_l) <= nlevels)
assert (names is None or names is False or names is True
or len(names) is nlevels)
assert idx_type is None or \
(idx_type in ('i', 'f', 's', 'u', 'dt', 'p') and nlevels == 1)
if names is True:
# build default names
names = [prefix + str(i) for i in range(nlevels)]
if names is False:
# pass None to index constructor for no name
names = None
# make singelton case uniform
if isinstance(names, compat.string_types) and nlevels == 1:
names = [names]
# specific 1D index type requested?
idx_func = dict(i=makeIntIndex,
f=makeFloatIndex,
s=makeStringIndex,
u=makeUnicodeIndex,
dt=makeDateIndex,
p=makePeriodIndex).get(idx_type)
if idx_func:
idx = idx_func(nentries)
# but we need to fill in the name
if names:
idx.name = names[0]
return idx
elif idx_type is not None:
raise ValueError('"%s" is not a legal value for `idx_type`, use '
'"i"/"f"/"s"/"u"/"dt/"p".' % idx_type)
if len(ndupe_l) < nlevels:
ndupe_l.extend([1] * (nlevels - len(ndupe_l)))
assert len(ndupe_l) == nlevels
assert all([x > 0 for x in ndupe_l])
tuples = []
for i in range(nlevels):
def keyfunc(x):
import re
numeric_tuple = re.sub("[^\d_]_?", "", x).split("_")
return lmap(int, numeric_tuple)
# build a list of lists to create the index from
div_factor = nentries // ndupe_l[i] + 1
cnt = Counter()
for j in range(div_factor):
label = prefix + '_l%d_g' % i + str(j)
cnt[label] = ndupe_l[i]
# cute Counter trick
result = list(sorted(cnt.elements(), key=keyfunc))[:nentries]
tuples.append(result)
tuples = lzip(*tuples)
# convert tuples to index
if nentries == 1:
index = Index(tuples[0], name=names[0])
else:
index = MultiIndex.from_tuples(tuples, names=names)
return index
def _align_series(self, indexer, ser):
# indexer to assign Series can be tuple or scalar
if isinstance(indexer, tuple):
aligners = [ not _is_null_slice(idx) for idx in indexer ]
sum_aligners = sum(aligners)
single_aligner = sum_aligners == 1
is_frame = self.obj.ndim == 2
is_panel = self.obj.ndim >= 3
obj = self.obj
# are we a single alignable value on a non-primary
# dim (e.g. panel: 1,2, or frame: 0) ?
# hence need to align to a single axis dimension
# rather that find all valid dims
# frame
if is_frame:
single_aligner = single_aligner and aligners[0]
# panel
elif is_panel:
single_aligner = single_aligner and (aligners[1] or aligners[2])
# we have a frame, with multiple indexers on both axes; and a series,
# so need to broadcast (see GH5206)
if sum_aligners == self.ndim and all([ com._is_sequence(_) for _ in indexer ]):
ser = ser.reindex(obj.axes[0][indexer[0].ravel()],copy=True).values
l = len(indexer[1].ravel())
ser = np.tile(ser,l).reshape(l,-1).T
return ser
for i, idx in enumerate(indexer):
ax = obj.axes[i]
# multiple aligners (or null slices)
if com._is_sequence(idx) or isinstance(idx, slice):
if single_aligner and _is_null_slice(idx):
continue
new_ix = ax[idx]
if not is_list_like(new_ix):
new_ix = Index([new_ix])
else:
new_ix = Index(new_ix.ravel())
if ser.index.equals(new_ix):
return ser.values.copy()
return ser.reindex(new_ix).values
# 2 dims
elif single_aligner and is_frame:
# reindex along index
ax = self.obj.axes[1]
if ser.index.equals(ax):
return ser.values.copy()
return ser.reindex(ax).values
# >2 dims
elif single_aligner:
broadcast = []
for n, labels in enumerate(self.obj._get_plane_axes(i)):
# reindex along the matching dimensions
if len(labels & ser.index):
ser = ser.reindex(labels)
else:
broadcast.append((n,len(labels)))
# broadcast along other dims
ser = ser.values.copy()
for (axis,l) in broadcast:
shape = [ -1 ] * (len(broadcast)+1)
shape[axis] = l
ser = np.tile(ser,l).reshape(shape)
if self.obj.ndim == 3:
ser = ser.T
return ser
elif np.isscalar(indexer):
ax = self.obj._get_axis(1)
if ser.index.equals(ax):
return ser.values.copy()
return ser.reindex(ax).values
raise ValueError('Incompatible indexer with Series')
def _align_frame(self, indexer, df):
is_frame = self.obj.ndim == 2
is_panel = self.obj.ndim >= 3
if isinstance(indexer, tuple):
idx, cols = None, None
sindexers = []
for i, ix in enumerate(indexer):
ax = self.obj.axes[i]
if com._is_sequence(ix) or isinstance(ix, slice):
if idx is None:
idx = ax[ix].ravel()
elif cols is None:
cols = ax[ix].ravel()
else:
break
else:
sindexers.append(i)
# panel
if is_panel:
if len(sindexers) == 1 and idx is None and cols is None:
if sindexers[0] == 0:
df = df.T
return self.obj.conform(df, axis=sindexers[0])
df = df.T
if idx is not None and cols is not None:
if df.index.equals(idx) and df.columns.equals(cols):
val = df.copy().values
else:
val = df.reindex(idx, columns=cols).values
return val
elif ((isinstance(indexer, slice) or com.is_list_like(indexer))
and is_frame):
ax = self.obj.index[indexer]
if df.index.equals(ax):
val = df.copy().values
else:
val = df.reindex(ax).values
return val
elif np.isscalar(indexer) and not is_frame:
idx = self.obj.axes[1]
cols = self.obj.axes[2]
# by definition we are indexing on the 0th axis
if is_panel:
df = df.T
if idx.equals(df.index) and cols.equals(df.columns):
return df.copy().values
# a passed in dataframe which is actually a transpose
# of what is needed
elif idx.equals(df.columns) and cols.equals(df.index):
return df.T.copy().values
return df.reindex(idx, columns=cols).values
raise ValueError('Incompatible indexer with DataFrame')
def _align_series(self, indexer, ser):
# indexer to assign Series can be tuple or scalar
if isinstance(indexer, tuple):
aligners = [not _is_null_slice(idx) for idx in indexer]
single_aligner = sum(aligners) == 1
is_frame = self.obj.ndim == 2
is_panel = self.obj.ndim >= 3
# are we a single alignable value on a non-primary
# dim (e.g. panel: 1,2, or frame: 0) ?
# hence need to align to a single axis dimension
# rather that find all valid dims
# frame
if is_frame:
single_aligner = single_aligner and aligners[0]
# panel
elif is_panel:
single_aligner = single_aligner and (aligners[1]
or aligners[2])
obj = self.obj
for i, idx in enumerate(indexer):
ax = obj.axes[i]
# multiple aligners (or null slices)
if com._is_sequence(idx) or isinstance(idx, slice):
if single_aligner and _is_null_slice(idx):
continue
new_ix = ax[idx]
if not is_list_like(new_ix):
new_ix = Index([new_ix])
if ser.index.equals(new_ix):
return ser.values.copy()
return ser.reindex(new_ix).values
# 2 dims
elif single_aligner and is_frame:
# reindex along index
ax = self.obj.axes[1]
if ser.index.equals(ax):
return ser.values.copy()
return ser.reindex(ax).values
# >2 dims
elif single_aligner:
broadcast = []
for n, labels in enumerate(self.obj._get_plane_axes(i)):
# reindex along the matching dimensions
if len(labels & ser.index):
ser = ser.reindex(labels)
else:
broadcast.append((n, len(labels)))
# broadcast along other dims
ser = ser.values.copy()
for (axis, l) in broadcast:
shape = [-1] * (len(broadcast) + 1)
shape[axis] = l
ser = np.tile(ser, l).reshape(shape)
if self.obj.ndim == 3:
ser = ser.T
return ser
elif np.isscalar(indexer):
ax = self.obj._get_axis(1)
if ser.index.equals(ax):
return ser.values.copy()
return ser.reindex(ax).values
raise ValueError('Incompatible indexer with Series')