def resolve_split(self, ary, args, kws):
out = SeriesType(types.List(string_type))
if (len(args) == 1 and isinstance(args[0], types.StringLiteral)
and len(args[0].literal_value) == 1):
out = SeriesType(types.List(string_type),
string_array_split_view_type)
return signature(out, *args)
def arr_to_series_type(arr):
series_type = None
if isinstance(arr, types.Array):
series_type = SeriesType(arr.dtype, arr)
elif arr == string_array_type:
# StringArray is readonly
series_type = SeriesType(string_type)
elif arr == list_string_array_type:
series_type = SeriesType(types.List(string_type))
return series_type
def generic(self, args, kws):
from sdc.str_arr_ext import is_str_arr_typ
assert not kws
[va, vb] = args
# if one of the inputs is string array
if is_str_series_typ(va) or is_str_series_typ(vb):
# inputs should be either string array or string
assert is_str_arr_typ(va) or va == string_type
assert is_str_arr_typ(vb) or vb == string_type
return signature(SeriesType(types.boolean), va, vb)
if ((is_dt64_series_typ(va) and vb == string_type)
or (is_dt64_series_typ(vb) and va == string_type)):
return signature(SeriesType(types.boolean), va, vb)
def _resolve_map_func(self, ary, args, kws):
dtype = ary.dtype
# getitem returns Timestamp for dt_index and series(dt64)
if dtype == types.NPDatetime('ns'):
dtype = pandas_timestamp_type
t = args[0].get_call_type(self.context, (dtype, ), {})
return signature(SeriesType(t.return_type), *args)
def resolve_sort_values(self, ary, args, kws):
# output will have permuted input index
out_index = ary.index
if out_index == types.none:
out_index = types.Array(types.intp, 1, 'C')
out = SeriesType(ary.dtype, ary.data, out_index)
return signature(out, *args)
def generic_expand_cumulative_series(self, args, kws):
# taken from arraydecl.py, replaced Array with Series
assert not args
assert not kws
assert isinstance(self.this, SeriesType)
return_type = SeriesType(_expand_integer(self.this.dtype))
return signature(return_type, recvr=self.this)
def _groupby_resolve_impl_func_type(series_dtype, method_name):
""" Used for typing result value of functions implementing groupby methods,
assuming that these implementation call method of Series.
"""
ty_series = SeriesType(series_dtype)
jitted_func = numba.njit(series_method_to_func[method_name])
return cpu_target.typing_context.resolve_function_type(
jitted_func, (ty_series, ), {})
def typer(val1, val2):
if is_dt64_series_typ(val1) and val2 == pandas_timestamp_type:
return SeriesType(types.NPTimedelta('ns'))
from sdc.hiframes.pd_index_ext import DatetimeIndexType
if isinstance(val1,
DatetimeIndexType) and val2 == pandas_timestamp_type:
from sdc.hiframes.pd_index_ext import TimedeltaIndexType
return TimedeltaIndexType(False)
def generic_resolve(self, s_str, func_name):
if sdc.config.config_pipeline_hpat_default and func_name in str2str_methods:
template_key = 'strmethod.' + func_name
out_typ = SeriesType(string_type)
class MethodTemplate(AbstractTemplate):
key = template_key
def generic(self, args, kws):
return signature(out_typ, *args)
return types.BoundFunction(MethodTemplate, s_str)
if func_name in str2str_methods_excluded:
return
raise NotImplementedError(
'Series.str.{} is not supported yet'.format(func_name))
def resolve_value_counts(self, ary, args, kws):
# output is int series with original data as index
out = SeriesType(types.int64, types.Array(types.int64, 1, 'C'),
ary.data)
return signature(out, *args)
def rolling_generic(self, args, kws):
# output is always float64
return signature(SeriesType(types.float64), *args)
def resolve_corr(self, ary, args, kws):
return signature(SeriesType(types.float64), *args)
def resolve_apply(self, ary, args, kws):
# result is always float64 (see Pandas window.pyx:roll_generic)
return signature(SeriesType(types.float64), *args)
def resolve_date_field(self, ary):
return SeriesType(types.int64)
def resolve_date(self, ary):
return SeriesType(datetime_date_type) # TODO: name, index
def resolve_contains(self, ary, args, kws):
return signature(SeriesType(types.bool_), *args)
def resolve_replace(self, ary, args, kws):
return signature(SeriesType(string_type), *args)
def resolve_get(self, ary, args, kws):
# XXX only list(list(str)) supported
return signature(SeriesType(string_type), *args)
def get_series_type(self, col_idx):
col_type = self.data[col_idx]
return SeriesType(col_type.dtype, col_type, self.index, is_named=True)