[BUG] Fix Calls to cudf.DataFrame/Series.unique that relied on old be…

…havior (#3616) Updates the code that relied on `unique()` to return values in sorted order by explicitly sorting the unique values. Closes #3615 Authors: - Alex Barghi (https://github.com/alexbarghi-nv) - Brad Rees (https://github.com/BradReesWork) Approvers: - Rick Ratzel (https://github.com/rlratzel) - Vibhu Jawa (https://github.com/VibhuJawa) URL: #3616
rapidsai · May 27, 2023 · 1e6da2b · 1e6da2b
1 parent aa00704
commit 1e6da2b
Show file tree

Hide file tree

Showing 7 changed files with 21 additions and 15 deletions.
diff --git a/python/cugraph-service/tests/test_remote_graph.py b/python/cugraph-service/tests/test_remote_graph.py
@@ -392,9 +392,11 @@ def test_extract_subgraph(
 
     assert remote_sg.get_num_vertices() == sg.number_of_vertices()
 
-    expected_vertex_ids = cudf.concat(
-        [sg.edgelist.edgelist_df["src"], sg.edgelist.edgelist_df["dst"]]
-    ).unique()
+    expected_vertex_ids = (
+        cudf.concat([sg.edgelist.edgelist_df["src"], sg.edgelist.edgelist_df["dst"]])
+        .unique()
+        .sort_values()
+    )
     if renumber:
         expected_vertex_ids = sg.unrenumber(
             cudf.DataFrame({"v": expected_vertex_ids}), "v"

diff --git a/python/cugraph/cugraph/components/connectivity.py b/python/cugraph/cugraph/components/connectivity.py
@@ -84,7 +84,7 @@ def _convert_df_to_output_type(df, input_type, return_labels):
         #       The number of connected components (number of unique labels).
         #   labels: ndarray
         #       The length-N array of labels of the connected components.
-        n_components = len(df["labels"].unique())
+        n_components = df["labels"].nunique()
         sorted_df = df.sort_values("vertex")
         if return_labels:
             if is_cp_matrix_type(input_type):

diff --git a/python/cugraph/cugraph/dask/structure/mg_property_graph.py b/python/cugraph/cugraph/dask/structure/mg_property_graph.py
@@ -379,7 +379,11 @@ def get_vertices(self, selection=None):
         vert_sers = self.__get_all_vertices_series()
         if vert_sers:
             if self.__series_type is dask_cudf.Series:
-                return dask_cudf.concat(vert_sers, ignore_index=True).unique()
+                return (
+                    dask_cudf.concat(vert_sers, ignore_index=True)
+                    .unique()
+                    .sort_values()
+                )
             else:
                 raise TypeError("dataframe must be a CUDF Dask dataframe.")
         return self.__series_type()

diff --git a/python/cugraph/cugraph/structure/graph_implementation/simpleGraph.py b/python/cugraph/cugraph/structure/graph_implementation/simpleGraph.py
@@ -397,7 +397,7 @@ def view_edge_list(self):
             df[dst] : cudf.Series
                 contains the destination index for each edge
 
-            df[weight] : cusd.Series
+            df[weight] : cudf.Series
                 Column is only present for weighted Graph,
                 then containing the weight value for each edge
         """
@@ -1179,7 +1179,8 @@ def edges(self):
 
     def nodes(self):
         """
-        Returns all the nodes in the graph as a cudf.Series.
+        Returns all the nodes in the graph as a cudf.Series, in order of appearance
+        in the edgelist (source column first, then destination column).
         If multi columns vertices, return a cudf.DataFrame.
         """
         if self.edgelist is not None:

diff --git a/python/cugraph/cugraph/structure/hypergraph.py b/python/cugraph/cugraph/structure/hypergraph.py
@@ -1,4 +1,4 @@
-# Copyright (c) 2020-2022, NVIDIA CORPORATION.
+# Copyright (c) 2020-2023, NVIDIA CORPORATION.
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
 # You may obtain a copy of the License at
@@ -325,7 +325,7 @@ def _create_entity_nodes(
 
     for key, col in events[columns].items():
         cat = categories.get(key, key)
-        col = col.unique()
+        col = col.unique().sort_values()
         col = col.nans_to_nulls().dropna() if dropna else col
         if len(col) == 0:
             continue

diff --git a/python/cugraph/cugraph/structure/property_graph.py b/python/cugraph/cugraph/structure/property_graph.py
@@ -488,7 +488,7 @@ def get_num_edges(self, type=None):
     def get_vertices(self, selection=None):
         """
         Return a Series containing the unique vertex IDs contained in both
-        the vertex and edge property data.
+        the vertex and edge property data in ascending order.
         Selection is not yet supported.
 
         Parameters
@@ -530,12 +530,11 @@ def get_vertices(self, selection=None):
         if vert_sers:
             if self.__series_type is cudf.Series:
                 return self.__series_type(
-                    cudf.concat(vert_sers, ignore_index=True).unique()
+                    cudf.concat(vert_sers, ignore_index=True).unique().sort_values()
                 )
             else:
-                return self.__series_type(
-                    pd.concat(vert_sers, ignore_index=True).unique()
-                )
+                x = pd.Series(pd.concat(vert_sers, ignore_index=True).unique())
+                return self.__series_type(x.sort_values())
         return self.__series_type()
 
     def vertices_ids(self):

diff --git a/python/cugraph/cugraph/tests/structure/test_graph.py b/python/cugraph/cugraph/tests/structure/test_graph.py
@@ -646,7 +646,7 @@ def test_bipartite_api(graph_file):
     # This test only tests the functionality of adding set of nodes and
     # retrieving them. The datasets currently used are not truly bipartite.
     cu_M = utils.read_csv_file(graph_file)
-    nodes = cudf.concat([cu_M["0"], cu_M["1"]]).unique()
+    nodes = cudf.concat([cu_M["0"], cu_M["1"]]).unique().sort_values()
 
     # Create set of nodes for partition
     set1_exp = cudf.Series(nodes[0 : int(len(nodes) / 2)])