[BUG]: Broken graph methods

[jnke2016]

Version

23.08

Which installation method(s) does this occur on?

Docker, Conda, Pip, Source

Describe the bug.

Several graph methods in both the SG and MG API are broken, some due to the migration away from cython.cu renumbering which include:

1. view_edge_list() and edges() returns internal column names for SG.
2. view_edge_list() doesn't return the upper triangular part when creating and MG graph and return internal column names when the vertices were renumbered.
3. number_of_nodes() and number_of_vertices() for MG when the graph is un-renumbered or has string vertices.
4. number_of_edges() for MG when the graph is undirected.

Minimum reproducible example

def get_graph_0(directed):
      df = cudf.DataFrame()
      df["source"] = cudf.Series(["a", "b", "x", "f", "b", "g"], dtype=str)
      df["target"] = cudf.Series(["f", "g", "a", "a", "v", "h"], dtype=str)
      df["value"] = cudf.Series([1.0, 1.0, 1.0, 1.0, 1.0, 1.0], dtype="float32")
  
      ddf = dask_cudf.from_cudf(df, npartitions=2)
  
      mG = cugraph.Graph(directed=directed)
  
      mG.from_dask_cudf_edgelist(ddf, source='source', destination='target', weight='value', renumber=True)
      return mG

def get_graph_1(directed):
      df = cudf.DataFrame()
      df["src_0"] = [1, 2, 3, 4, 5]
      df["src_1"] = [2, 1, 5, 7, 8]
      df["dst_0"] = [4, 1, 5, 6, 7]
      df["dst_1"] = [9, 4, 10, 5, 7]
      df["value"] = [1.0, 1.0, 1.0, 1.0, 1.0]
  
      ddf = dask_cudf.from_cudf(df, npartitions=2)
  
      sG = cugraph.Graph(directed=directed)
      mG = cugraph.Graph(directed=directed)

      sG.from_cudf_edgelist(df, source=["src_0", "src_1"], destination=["dst_0", "dst_1"], weight='value', renumber=True)
      mG.from_dask_cudf_edgelist(ddf, source=["src_0", "src_1"], destination=["dst_0", "dst_1"], weight='value', renumber=True)
   
       return sG, mG

if __name__ == "__main__":
    directed = False
    mG = get_graph(directed)
    # Reproducer 1: Getting the number of nodes from a graph having string vertices
    mG.number_of_nodes()
    # Reproducer 2: The number of edges must be 5 because only the upper triangular part must be 
    # returned. Instead, it returns 10
    assert mG.number_of_edges() == 5


if __name__ == "__main__":
    directed = False
    sG, mG = get_graph(directed)
    # Reproducer 3: Retrieving the edge list view from the SG graph
    sG.view_edge_list()
    # Reproducer 4: Retrieving the edge list view from the MG graph
    mG.view_edge_list() #  Returns an edge list with the renumbered vertices instead of the user input dataframe
    

Relevant log output

Reproducer 1 log
File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/cugraph/structure/graph_implementation/simpleDistributedGraph.py", line 431, in number_of_nodes
    return self.number_of_vertices()
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/cugraph/structure/graph_implementation/simpleDistributedGraph.py", line 422, in number_of_vertices
    self.properties.node_count = ddf.max().max().compute() + 1
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/cudf/utils/utils.py", line 232, in __getattr__
    raise AttributeError(
AttributeError: DataFrame object has no attribute infer_objects
Exception ignored in: <function Comms.__del__ at 0x7f89a0ed1990>


Reproducer 3 log
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/cugraph/structure/graph_implementation/simpleGraph.py", line 421, in view_edge_list
    edgelist_df[simpleGraphImpl.srcCol]
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/nvtx/nvtx.py", line 101, in inner
    result = func(*args, **kwargs)
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/cudf/core/dataframe.py", line 1159, in __getitem__
    return self._get_columns_by_label(arg, downcast=True)
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/nvtx/nvtx.py", line 101, in inner
    result = func(*args, **kwargs)
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/cudf/core/dataframe.py", line 1792, in _get_columns_by_label
    new_data = super()._get_columns_by_label(labels, downcast)
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/nvtx/nvtx.py", line 101, in inner
    result = func(*args, **kwargs)
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/cudf/core/frame.py", line 425, in _get_columns_by_label
    return self._data.select_by_label(labels)
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/cudf/core/column_accessor.py", line 357, in select_by_label
    return self._select_by_label_grouped(key)
  File "/home/nfs/jnke/miniconda3/envs/branch23.08_/lib/python3.10/site-packages/cudf/core/column_accessor.py", line 512, in _select_by_label_grouped
    result = self._grouped_data[key]
KeyError: 'src'

Code of Conduct

• I agree to follow cuGraph's Code of Conduct
• I have searched the open bugs and have found no duplicates for this bug report