def from_series(self, src_series, dst_series=None, store_transposed=False):
"""
Populate the numbering map with vertices from the specified
pair of series objects, one for the source and one for
the destination
Parameters
----------
src_series: cudf.Series or dask_cudf.Series
Contains a list of external vertex identifiers that will be
numbered by the NumberMap class.
dst_series: cudf.Series or dask_cudf.Series
Contains a list of external vertex identifiers that will be
numbered by the NumberMap class.
store_transposed : bool
Identify how the graph adjacency will be used.
If True, the graph will be organized by destination.
If False, the graph will be organized by source
"""
if self.implementation is not None:
raise Exception("NumberMap is already populated")
if dst_series is not None and type(src_series) != type(dst_series):
raise Exception("src_series and dst_series must have same type")
if type(src_series) is cudf.Series:
dst_series_list = None
df = cudf.DataFrame()
df["s"] = src_series
if dst_series is not None:
df["d"] = dst_series
dst_series_list = ["d"]
self.implementation = NumberMap.SingleGPU(df, ["s"],
dst_series_list,
self.id_type,
store_transposed)
elif type(src_series) is dask_cudf.Series:
dst_series_list = None
df = dask_cudf.DataFrame()
df["s"] = src_series
if dst_series is not None:
df["d"] = dst_series
dst_series_list = ["d"]
self.implementation = NumberMap.MultiGPU(df, ["s"],
dst_series_list,
self.id_type,
store_transposed)
else:
raise Exception("src_series must be cudf.Series or "
"dask_cudf.Series")
self.implementation.compute()
def to_internal_vertex_id(self, df, col_names=None):
"""
Given a collection of external vertex ids, return the internal
vertex ids
Parameters
----------
df: cudf.DataFrame, cudf.Series, dask_cudf.DataFrame, dask_cudf.Series
Contains a list of external vertex identifiers that will be
converted into internal vertex identifiers
col_names: (optional) list of strings
This list of 1 or more strings contain the names
of the columns that uniquely identify an external
vertex identifier
Returns
---------
vertex_ids : cudf.Series or dask_cudf.Series
The vertex identifiers. Note that to_internal_vertex_id
does not guarantee order or partitioning (in the case of
dask_cudf) of vertex ids. If order matters use
add_internal_vertex_id
"""
tmp_df = None
tmp_col_names = None
if type(df) is cudf.Series:
tmp_df = cudf.DataFrame()
tmp_df["0"] = df
tmp_col_names = ["0"]
elif type(df) is dask_cudf.Series:
tmp_df = dask_cudf.DataFrame()
tmp_df["0"] = df
tmp_col_names = ["0"]
else:
tmp_df = df
tmp_col_names = col_names
reply = self.implementation.to_internal_vertex_id(
tmp_df, tmp_col_names)
if type(df) in [cudf.DataFrame, dask_cudf.DataFrame]:
return reply["0"]
else:
return reply