Merge branch 'branch-24.06' into run_nx_example

ci/release/update-version.sh

@@ -103,5 +103,9 @@ sed_runner "s/branch-.*/branch-${NEXT_SHORT_TAG}/g" python/nx-cugraph/README.md
 find .devcontainer/ -type f -name devcontainer.json -print0 | while IFS= read -r -d '' filename; do
  sed_runner "s@rapidsai/devcontainers:[0-9.]*@rapidsai/devcontainers:${NEXT_SHORT_TAG}@g" "${filename}"
  sed_runner "s@rapidsai/devcontainers/features/ucx:[0-9.]*@rapidsai/devcontainers/features/ucx:${NEXT_SHORT_TAG_PEP440}@" "${filename}"
+ sed_runner "s@rapidsai/devcontainers/features/cuda:[0-9.]*@rapidsai/devcontainers/features/cuda:${NEXT_SHORT_TAG_PEP440}@" "${filename}"
  sed_runner "s@rapidsai/devcontainers/features/rapids-build-utils:[0-9.]*@rapidsai/devcontainers/features/rapids-build-utils:${NEXT_SHORT_TAG_PEP440}@" "${filename}"
 done
+
+sed_runner "s/:[0-9][0-9]\.[0-9][0-9]/:${NEXT_SHORT_TAG}/" ./notebooks/README.md
+sed_runner "s/branch-[0-9][0-9].[0-9][0-9]/branch-${NEXT_SHORT_TAG}/" ./docs/cugraph/source/nx_cugraph/nx_cugraph.md

conda/recipes/cugraph-pyg/conda_build_config.yaml

@@ -12,5 +12,8 @@ cuda_compiler:
 cmake_version:
  - ">=3.26.4"
 
-sysroot_version:
+c_stdlib:
+ - sysroot
+
+c_stdlib_version:
  - "2.17"

conda/recipes/cugraph-pyg/meta.yaml

@@ -20,7 +20,7 @@ build:
 
 requirements:
  build:
- - sysroot_{{ target_platform }} {{ sysroot_version }}
+ - {{ stdlib("c") }}
  host:
  - cython >=3.0.0
  - python

conda/recipes/cugraph/conda_build_config.yaml

@@ -13,7 +13,10 @@ cuda11_compiler:
 cmake_version:
  - ">=3.26.4"
 
-sysroot_version:
+c_stdlib:
+ - sysroot
+
+c_stdlib_version:
  - "2.17"
 
 ucx_py_version:

conda/recipes/cugraph/meta.yaml

@@ -53,7 +53,7 @@ requirements:
  - cuda-version ={{ cuda_version }}
  - cmake {{ cmake_version }}
  - ninja
- - sysroot_{{ target_platform }} {{ sysroot_version }}
+ - {{ stdlib("c") }}
  host:
  - cuda-version ={{ cuda_version }}
  {% if cuda_major == "11" %}

conda/recipes/libcugraph/conda_build_config.yaml

@@ -19,7 +19,10 @@ doxygen_version:
 nccl_version:
  - ">=2.9.9"
 
-sysroot_version:
+c_stdlib:
+ - sysroot
+
+c_stdlib_version:
  - "2.17"
 
 # The CTK libraries below are missing from the conda-forge::cudatoolkit

conda/recipes/libcugraph/meta.yaml

@@ -43,7 +43,7 @@ requirements:
  - cmake {{ cmake_version }}
  - ninja
  - openmpi # Required for building cpp-mgtests (multi-GPU tests)
- - sysroot_{{ target_platform }} {{ sysroot_version }}
+ - {{ stdlib("c") }}
  host:
  {% if cuda_major == "11" %}
  - cudatoolkit

conda/recipes/pylibcugraph/conda_build_config.yaml

@@ -13,7 +13,10 @@ cuda11_compiler:
 cmake_version:
  - ">=3.26.4"
 
-sysroot_version:
+c_stdlib:
+ - sysroot
+
+c_stdlib_version:
  - "2.17"
 
 ucx_py_version:

conda/recipes/pylibcugraph/meta.yaml

@@ -53,7 +53,7 @@ requirements:
  - cuda-version ={{ cuda_version }}
  - cmake {{ cmake_version }}
  - ninja
- - sysroot_{{ target_platform }} {{ sysroot_version }}
+ - {{ stdlib("c") }}
  host:
  - cuda-version ={{ cuda_version }}
  {% if cuda_major == "11" %}

cpp/src/community/leiden_impl.cuh

@@ -561,6 +561,13 @@ std::pair<std::unique_ptr<Dendrogram<vertex_t>>, weight_t> leiden(
  louvain_of_refined_graph.size(),
  false);
  }
+ } else {
+ // Reset dendrogram.
+ // FIXME: Revisit how dendrogram is populated
+ detail::sequence_fill(handle.get_stream(),
+ dendrogram->current_level_begin(),
+ dendrogram->current_level_size(),
+ current_graph_view.local_vertex_partition_range_first());
  }
 
  copied_louvain_partition.resize(0, handle.get_stream());

python/cugraph/cugraph/datasets/dataset.py

@@ -174,7 +174,7 @@ def get_edgelist(self, download=False, reader="cudf"):
  reader : 'cudf' or 'pandas' (default='cudf')
  The library used to read a CSV and return an edgelist DataFrame.
  """
- if self._edgelist is None:
+ if self._edgelist is None or not isinstance(self._edgelist, cudf.DataFrame):
  full_path = self.get_path()
  if not full_path.is_file():
  if download:
@@ -223,7 +223,9 @@ def get_dask_edgelist(self, download=False):
  Automatically download the dataset from the 'url' location within
  the YAML file.
  """
- if self._edgelist is None:
+ if self._edgelist is None or not isinstance(
+ self._edgelist, dask_cudf.DataFrame
+ ):
  full_path = self.get_path()
  if not full_path.is_file():
  if download:
@@ -286,7 +288,7 @@ def get_graph(
  for certain algorithms, such as pagerank.
  """
  if self._edgelist is None:
- self.get_edgelist(download)
+ self.get_edgelist(download=download)
 
  if create_using is None:
  G = Graph()
@@ -351,7 +353,7 @@ def get_dask_graph(
  for certain algorithms.
  """
  if self._edgelist is None:
- self.get_dask_edgelist(download)
+ self.get_dask_edgelist(download=download)
 
  if create_using is None:
  G = Graph()
@@ -367,7 +369,7 @@ def get_dask_graph(
  f"{type(create_using)}"
  )
 
- if len(self.metadata["col_names"]) > 2 and not (ignore_weights):
+ if len(self.metadata["col_names"]) > 2 and not ignore_weights:
  G.from_dask_cudf_edgelist(
  self._edgelist,
  source=self.metadata["col_names"][0],

python/cugraph/cugraph/tests/utils/test_dataset.py

@@ -20,7 +20,11 @@
 import pytest
 
 import cudf
+import cugraph
 import dask_cudf
+from cugraph import datasets
+from cugraph.dask.common.mg_utils import is_single_gpu
+from cugraph.datasets import karate
 from cugraph.structure import Graph
 from cugraph.testing import (
  RAPIDS_DATASET_ROOT_DIR_PATH,
@@ -29,15 +33,15 @@
  SMALL_DATASETS,
  BENCHMARKING_DATASETS,
 )
-from cugraph import datasets
-from cugraph.dask.common.mg_utils import is_single_gpu
+
 
 # Add the sg marker to all tests in this module.
 pytestmark = pytest.mark.sg
 
 
 ###############################################################################
 # Fixtures
+###############################################################################
 
 
 # module fixture - called once for this module
@@ -201,21 +205,26 @@ def test_reader_dask(dask_client, dataset):
 @pytest.mark.parametrize("dataset", ALL_DATASETS)
 def test_get_edgelist(dataset):
  E = dataset.get_edgelist(download=True)
+
  assert E is not None
+ assert isinstance(E, cudf.DataFrame)
 
 
 @pytest.mark.skipif(is_single_gpu(), reason="skipping MG testing on Single GPU system")
 @pytest.mark.skip(reason="MG not supported on CI")
 @pytest.mark.parametrize("dataset", ALL_DATASETS)
 def test_get_dask_edgelist(dask_client, dataset):
  E = dataset.get_dask_edgelist(download=True)
+
  assert E is not None
+ assert isinstance(E, dask_cudf.DataFrame)
 
 
 @pytest.mark.parametrize("dataset", ALL_DATASETS)
 def test_get_graph(dataset):
  G = dataset.get_graph(download=True)
  assert G is not None
+ assert isinstance(G, cugraph.Graph)
 
 
 @pytest.mark.skipif(is_single_gpu(), reason="skipping MG testing on Single GPU system")
@@ -224,12 +233,14 @@ def test_get_graph(dataset):
 def test_get_dask_graph(dask_client, dataset):
  G = dataset.get_dask_graph(download=True)
  assert G is not None
+ # TODO Check G is a DistributedGraph
 
 
 @pytest.mark.parametrize("dataset", ALL_DATASETS)
 def test_metadata(dataset):
  M = dataset.metadata
  assert M is not None
+ assert isinstance(M, dict)
 
 
 @pytest.mark.parametrize("dataset", ALL_DATASETS)
@@ -346,32 +357,19 @@ def test_ctor_with_datafile():
  assert ds.get_path() == karate_csv
 
 
-def test_unload():
- email_csv = RAPIDS_DATASET_ROOT_DIR_PATH / "email-Eu-core.csv"
+@pytest.mark.parametrize("dataset", [karate])
+def test_unload(dataset):
+ assert dataset._edgelist is None
 
- ds = datasets.Dataset(
- csv_file=email_csv.as_posix(),
- csv_col_names=["src", "dst", "wgt"],
- csv_col_dtypes=["int32", "int32", "float32"],
- )
+ dataset.get_edgelist()
+ assert dataset._edgelist is not None
+ dataset.unload()
+ assert dataset._edgelist is None
 
- # FIXME: another (better?) test would be to check free memory and assert
- # the memory use increases after get_*(), then returns to the pre-get_*()
- # level after unload(). However, that type of test may fail for several
- # reasons (the device being monitored is accidentally also being used by
- # another process, and the use of memory pools to name two). Instead, just
- # test that the internal members get cleared on unload().
- assert ds._edgelist is None
-
- ds.get_edgelist()
- assert ds._edgelist is not None
- ds.unload()
- assert ds._edgelist is None
-
- ds.get_graph()
- assert ds._edgelist is not None
- ds.unload()
- assert ds._edgelist is None
+ dataset.get_graph()
+ assert dataset._edgelist is not None
+ dataset.unload()
+ assert dataset._edgelist is None
 
 
 @pytest.mark.parametrize("dataset", ALL_DATASETS)

python/nx-cugraph/_nx_cugraph/__init__.py

@@ -166,10 +166,6 @@
  "eigenvector_centrality": "`nstart` parameter is not used, but it is checked for validity.",
  "from_pandas_edgelist": "cudf.DataFrame inputs also supported; value columns with str is unsuppported.",
  "generic_bfs_edges": "`neighbors` and `sort_neighbors` parameters are not yet supported.",
- "k_truss": (
- "Currently raises `NotImplementedError` for graphs with more than one connected\n"
- "component when k >= 3. We expect to fix this soon."
- ),
  "katz_centrality": "`nstart` isn't used (but is checked), and `normalized=False` is not supported.",
  "louvain_communities": "`seed` parameter is currently ignored, and self-loops are not yet supported.",
  "pagerank": "`dangling` parameter is not supported, but it is checked for validity.",

python/nx-cugraph/lint.yaml

@@ -11,7 +11,7 @@ default_language_version:
  python: python3
 repos:
  - repo: https://github.com/pre-commit/pre-commit-hooks
- rev: v4.5.0
+ rev: v4.6.0
  hooks:
  - id: check-added-large-files
  - id: check-case-conflict
@@ -31,7 +31,7 @@ repos:
  - id: validate-pyproject
  name: Validate pyproject.toml
  - repo: https://github.com/PyCQA/autoflake
- rev: v2.3.0
+ rev: v2.3.1
  hooks:
  - id: autoflake
  args: [--in-place]
@@ -40,17 +40,17 @@ repos:
  hooks:
  - id: isort
  - repo: https://github.com/asottile/pyupgrade
- rev: v3.15.1
+ rev: v3.15.2
  hooks:
  - id: pyupgrade
  args: [--py39-plus]
  - repo: https://github.com/psf/black
- rev: 24.2.0
+ rev: 24.4.2
  hooks:
  - id: black
  # - id: black-jupyter
  - repo: https://github.com/astral-sh/ruff-pre-commit
- rev: v0.3.2
+ rev: v0.4.2
  hooks:
  - id: ruff
  args: [--fix-only, --show-fixes] # --unsafe-fixes]
@@ -62,7 +62,7 @@ repos:
  additional_dependencies: &flake8_dependencies
  # These versions need updated manually
  - flake8==7.0.0
- - flake8-bugbear==24.2.6
+ - flake8-bugbear==24.4.21
  - flake8-simplify==0.21.0
  - repo: https://github.com/asottile/yesqa
  rev: v1.5.0
@@ -77,11 +77,11 @@ repos:
  additional_dependencies: [tomli]
  files: ^(nx_cugraph|docs)/
  - repo: https://github.com/astral-sh/ruff-pre-commit
- rev: v0.3.2
+ rev: v0.4.2
  hooks:
  - id: ruff
  - repo: https://github.com/pre-commit/pre-commit-hooks
- rev: v4.5.0
+ rev: v4.6.0
  hooks:
  - id: no-commit-to-branch
  args: [-p, "^branch-2....$"]

python/nx-cugraph/nx_cugraph/algorithms/bipartite/generators.py

@@ -48,13 +48,17 @@ def complete_bipartite_graph(n1, n2, create_using=None):
  nodes.extend(range(n2) if nodes2 is None else nodes2)
  if len(set(nodes)) != len(nodes):
  raise nx.NetworkXError("Inputs n1 and n2 must contain distinct nodes")
+ if nx.__version__[:3] <= "3.3":
+ name = f"complete_bipartite_graph({orig_n1}, {orig_n2})"
+ else:
+ name = f"complete_bipartite_graph({n1}, {n2})"
  G = graph_class.from_coo(
  n1 + n2,
  src_indices,
  dst_indices,
  node_values={"bipartite": bipartite},
  id_to_key=nodes,
- name=f"complete_bipartite_graph({orig_n1}, {orig_n2})",
+ name=name,
  )
  if inplace:
  return create_using._become(G)

python/nx-cugraph/nx_cugraph/algorithms/centrality/degree_alg.py

@@ -26,6 +26,11 @@ def degree_centrality(G):
  return G._nodearray_to_dict(centrality)
 
 
+@degree_centrality._should_run
+def _(G):
+ return "Fast algorithm; not worth converting."
+
+
 @not_implemented_for("undirected")
 @networkx_algorithm(version_added="23.12")
 def in_degree_centrality(G):
@@ -37,6 +42,11 @@ def in_degree_centrality(G):
  return G._nodearray_to_dict(centrality)
 
 
+@in_degree_centrality._should_run
+def _(G):
+ return "Fast algorithm; not worth converting."
+
+
 @not_implemented_for("undirected")
 @networkx_algorithm(version_added="23.12")
 def out_degree_centrality(G):
@@ -46,3 +56,8 @@ def out_degree_centrality(G):
  deg = G._out_degrees_array()
  centrality = deg * (1 / (len(G) - 1))
  return G._nodearray_to_dict(centrality)
+
+
+@out_degree_centrality._should_run
+def _(G):
+ return "Fast algorithm; not worth converting."