Removes APIs that have been deprecated and have exceeded the grace pe…

…riod, improves *_warning_wrapper performance (#4176) * Removes APIs that have been deprecated >1 releases ago. * Updates tests that assumed the deprecated APIs were in place. * Changes the `*_warning_wrapper` utilities to accept a namespace name to use in the warning messages instead of making an expensive `inspect.stack()` call. `inspect.stack()` is still used if a namespace name was not provided. This reduces overall `import` time, especially when used with `cudf.pandas` (attn: @shwina). * Removes the redundant `cugraph.utilities.api_tools` module in favor of just using the equivalent `pylibcugraph` module directly. * Fixes jaccard notebook to remove deprecated API calls that are no longer present and updates it for clarity and to match the current implementation. * Cleans up jaccard docstrings. Authors: - Rick Ratzel (https://github.com/rlratzel) Approvers: - Naim (https://github.com/naimnv) - Joseph Nke (https://github.com/jnke2016) - Don Acosta (https://github.com/acostadon) - Brad Rees (https://github.com/BradReesWork) URL: #4176
rapidsai · Feb 23, 2024 · 2c478fb · 2c478fb
1 parent 3499f28
commit 2c478fb
Show file tree

Hide file tree

Showing 20 changed files with 161 additions and 1,096 deletions.
diff --git a/notebooks/algorithms/link_prediction/Jaccard-Similarity.ipynb b/notebooks/algorithms/link_prediction/Jaccard-Similarity.ipynb
diff --git a/notebooks/img/karate_similarity.png b/notebooks/img/karate_similarity.png
diff --git a/python/cugraph/cugraph/__init__.py b/python/cugraph/cugraph/__init__.py
@@ -1,4 +1,4 @@
-# Copyright (c) 2019-2023, NVIDIA CORPORATION.
+# Copyright (c) 2019-2024, 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
@@ -80,9 +80,6 @@
     overlap_coefficient,
     sorensen,
     sorensen_coefficient,
-    jaccard_w,
-    overlap_w,
-    sorensen_w,
 )
 
 from cugraph.traversal import (
@@ -100,9 +97,6 @@
 
 from cugraph.utilities import utils
 
-from cugraph.experimental import strong_connected_component
-from cugraph.experimental import find_bicliques
-
 from cugraph.linear_assignment import hungarian, dense_hungarian
 from cugraph.layout import force_atlas2
 

diff --git a/python/cugraph/cugraph/dask/link_prediction/jaccard.py b/python/cugraph/cugraph/dask/link_prediction/jaccard.py
@@ -1,4 +1,4 @@
-# Copyright (c) 2022-2023, NVIDIA CORPORATION.
+# Copyright (c) 2022-2024, NVIDIA CORPORATION.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -66,38 +66,13 @@ def jaccard(input_graph, vertex_pair=None, use_weight=False):
     of their intersection divided by the volume of their union. In the context
     of graphs, the neighborhood of a vertex is seen as a set. The Jaccard
     similarity weight of each edge represents the strength of connection
-    between vertices based on the relative similarity of their neighbors. If
-    first is specified but second is not, or vice versa, an exception will be
-    thrown.
-
-    NOTE: If the vertex_pair parameter is not specified then the behavior
-    of cugraph.jaccard is different from the behavior of
-    networkx.jaccard_coefficient.
+    between vertices based on the relative similarity of their neighbors.
 
     cugraph.dask.jaccard, in the absence of a specified vertex pair list, will
     compute the two_hop_neighbors of the entire graph to construct a vertex pair
     list and will return the jaccard coefficient for those vertex pairs. This is
     not advisable as the vertex_pairs can grow exponentially with respect to the
-    size of the datasets
-
-    networkx.jaccard_coefficient, in the absence of a specified vertex
-    pair list, will return an upper triangular dense matrix, excluding
-    the diagonal as well as vertex pairs that are directly connected
-    by an edge in the graph, of jaccard coefficients.  Technically, networkx
-    returns a lazy iterator across this upper triangular matrix where
-    the actual jaccard coefficient is computed when the iterator is
-    dereferenced.  Computing a dense matrix of results is not feasible
-    if the number of vertices in the graph is large (100,000 vertices
-    would result in 4.9 billion values in that iterator).
-
-    If your graph is small enough (or you have enough memory and patience)
-    you can get the interesting (non-zero) values that are part of the networkx
-    solution by doing the following:
-
-    But please remember that cugraph will fill the dataframe with the entire
-    solution you request, so you'll need enough memory to store the 2-hop
-    neighborhood dataframe.
-
+    size of the datasets.
 
     Parameters
     ----------

diff --git a/python/cugraph/cugraph/experimental/__init__.py b/python/cugraph/cugraph/experimental/__init__.py
@@ -11,59 +11,41 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from cugraph.utilities.api_tools import experimental_warning_wrapper
-from cugraph.utilities.api_tools import deprecated_warning_wrapper
-from cugraph.utilities.api_tools import promoted_experimental_warning_wrapper
+from pylibcugraph.utilities.api_tools import (
+    experimental_warning_wrapper,
+    promoted_experimental_warning_wrapper,
+)
+
+# Passing in the namespace name of this module to the *_wrapper functions
+# allows them to bypass the expensive inspect.stack() lookup.
+_ns_name = __name__
 
 from cugraph.structure.property_graph import EXPERIMENTAL__PropertyGraph
 
-PropertyGraph = experimental_warning_wrapper(EXPERIMENTAL__PropertyGraph)
+PropertyGraph = experimental_warning_wrapper(EXPERIMENTAL__PropertyGraph, _ns_name)
 
 from cugraph.structure.property_graph import EXPERIMENTAL__PropertySelection
 
-PropertySelection = experimental_warning_wrapper(EXPERIMENTAL__PropertySelection)
+PropertySelection = experimental_warning_wrapper(
+    EXPERIMENTAL__PropertySelection, _ns_name
+)
 
 from cugraph.dask.structure.mg_property_graph import EXPERIMENTAL__MGPropertyGraph
 
-MGPropertyGraph = experimental_warning_wrapper(EXPERIMENTAL__MGPropertyGraph)
+MGPropertyGraph = experimental_warning_wrapper(EXPERIMENTAL__MGPropertyGraph, _ns_name)
 
 from cugraph.dask.structure.mg_property_graph import EXPERIMENTAL__MGPropertySelection
 
-MGPropertySelection = experimental_warning_wrapper(EXPERIMENTAL__MGPropertySelection)
-
-# FIXME: Remove experimental.triangle_count next release
-from cugraph.community.triangle_count import triangle_count
-
-triangle_count = promoted_experimental_warning_wrapper(triangle_count)
+MGPropertySelection = experimental_warning_wrapper(
+    EXPERIMENTAL__MGPropertySelection, _ns_name
+)
 
 from cugraph.experimental.components.scc import EXPERIMENTAL__strong_connected_component
 
 strong_connected_component = experimental_warning_wrapper(
-    EXPERIMENTAL__strong_connected_component
-)
-
-from cugraph.experimental.structure.bicliques import EXPERIMENTAL__find_bicliques
-
-find_bicliques = deprecated_warning_wrapper(
-    experimental_warning_wrapper(EXPERIMENTAL__find_bicliques)
+    EXPERIMENTAL__strong_connected_component, _ns_name
 )
 
 from cugraph.gnn.data_loading import BulkSampler
 
-BulkSampler = promoted_experimental_warning_wrapper(BulkSampler)
-
-
-from cugraph.link_prediction.jaccard import jaccard, jaccard_coefficient
-
-jaccard = promoted_experimental_warning_wrapper(jaccard)
-jaccard_coefficient = promoted_experimental_warning_wrapper(jaccard_coefficient)
-
-from cugraph.link_prediction.sorensen import sorensen, sorensen_coefficient
-
-sorensen = promoted_experimental_warning_wrapper(sorensen)
-sorensen_coefficient = promoted_experimental_warning_wrapper(sorensen_coefficient)
-
-from cugraph.link_prediction.overlap import overlap, overlap_coefficient
-
-overlap = promoted_experimental_warning_wrapper(overlap)
-overlap_coefficient = promoted_experimental_warning_wrapper(overlap_coefficient)
+BulkSampler = promoted_experimental_warning_wrapper(BulkSampler, _ns_name)
diff --git a/python/cugraph/cugraph/experimental/gnn/__init__.py b/python/cugraph/cugraph/experimental/gnn/__init__.py
@@ -1,4 +1,4 @@
-# Copyright (c) 2023, NVIDIA CORPORATION.
+# Copyright (c) 2023-2024, 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
@@ -12,6 +12,10 @@
 # limitations under the License.
 
 from cugraph.gnn.data_loading import BulkSampler
-from cugraph.utilities.api_tools import promoted_experimental_warning_wrapper
+from pylibcugraph.utilities.api_tools import promoted_experimental_warning_wrapper
 
-BulkSampler = promoted_experimental_warning_wrapper(BulkSampler)
+# Passing in the namespace name of this module to the *_wrapper functions
+# allows them to bypass the expensive inspect.stack() lookup.
+_ns_name = __name__
+
+BulkSampler = promoted_experimental_warning_wrapper(BulkSampler, _ns_name)
diff --git a/python/cugraph/cugraph/link_prediction/__init__.py b/python/cugraph/cugraph/link_prediction/__init__.py
@@ -1,4 +1,4 @@
-# Copyright (c) 2019-2023, NVIDIA CORPORATION.
+# Copyright (c) 2019-2024, 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
@@ -11,26 +11,9 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-
-from cugraph.utilities.api_tools import deprecated_warning_wrapper
 from cugraph.link_prediction.jaccard import jaccard
 from cugraph.link_prediction.jaccard import jaccard_coefficient
-
 from cugraph.link_prediction.sorensen import sorensen
 from cugraph.link_prediction.sorensen import sorensen_coefficient
-
 from cugraph.link_prediction.overlap import overlap
 from cugraph.link_prediction.overlap import overlap_coefficient
-
-# To be deprecated
-from cugraph.link_prediction.wjaccard import jaccard_w
-
-jaccard_w = deprecated_warning_wrapper(jaccard_w)
-
-from cugraph.link_prediction.woverlap import overlap_w
-
-overlap_w = deprecated_warning_wrapper(overlap_w)
-
-from cugraph.link_prediction.wsorensen import sorensen_w
-
-sorensen_w = deprecated_warning_wrapper(sorensen_w)
diff --git a/python/cugraph/cugraph/link_prediction/jaccard.py b/python/cugraph/cugraph/link_prediction/jaccard.py
@@ -1,4 +1,4 @@
-# Copyright (c) 2019-2023, NVIDIA CORPORATION.
+# Copyright (c) 2019-2024, 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
@@ -56,7 +56,6 @@ def ensure_valid_dtype(input_graph, vertex_pair):
 def jaccard(
     input_graph: Graph,
     vertex_pair: cudf.DataFrame = None,
-    do_expensive_check: bool = False,  # deprecated
     use_weight: bool = False,
 ):
     """
@@ -66,43 +65,13 @@ def jaccard(
     of their intersection divided by the volume of their union. In the context
     of graphs, the neighborhood of a vertex is seen as a set. The Jaccard
     similarity weight of each edge represents the strength of connection
-    between vertices based on the relative similarity of their neighbors. If
-    first is specified but second is not, or vice versa, an exception will be
-    thrown.
-
-    NOTE: If the vertex_pair parameter is not specified then the behavior
-    of cugraph.jaccard is different from the behavior of
-    networkx.jaccard_coefficient.
+    between vertices based on the relative similarity of their neighbors.
 
     cugraph.jaccard, in the absence of a specified vertex pair list, will
     compute the two_hop_neighbors of the entire graph to construct a vertex pair
     list and will return the jaccard coefficient for those vertex pairs. This is
     not advisable as the vertex_pairs can grow exponentially with respect to the
-    size of the datasets
-
-    networkx.jaccard_coefficient, in the absence of a specified vertex
-    pair list, will return an upper triangular dense matrix, excluding
-    the diagonal as well as vertex pairs that are directly connected
-    by an edge in the graph, of jaccard coefficients.  Technically, networkx
-    returns a lazy iterator across this upper triangular matrix where
-    the actual jaccard coefficient is computed when the iterator is
-    dereferenced.  Computing a dense matrix of results is not feasible
-    if the number of vertices in the graph is large (100,000 vertices
-    would result in 4.9 billion values in that iterator).
-
-    If your graph is small enough (or you have enough memory and patience)
-    you can get the interesting (non-zero) values that are part of the networkx
-    solution by doing the following:
-
-    >>> from cugraph.datasets import karate
-    >>> input_graph = karate.get_graph(download=True, ignore_weights=True)
-    >>> pairs = input_graph.get_two_hop_neighbors()
-    >>> df = cugraph.jaccard(input_graph, pairs)
-
-    But please remember that cugraph will fill the dataframe with the entire
-    solution you request, so you'll need enough memory to store the 2-hop
-    neighborhood dataframe.
-
+    size of the datasets.
 
     Parameters
     ----------
@@ -121,21 +90,11 @@ def jaccard(
         current implementation computes the jaccard coefficient for all
         adjacent vertices in the graph.
 
-    do_expensive_check : bool, optional (default=False)
-        Deprecated.
-
-        This option added a check to ensure integer vertex IDs are sequential
-        values from 0 to V-1. That check is now redundant because cugraph
-        unconditionally renumbers and un-renumbers integer vertex IDs for
-        optimal performance, therefore this option is deprecated and will be
-        removed in a future version.
-
     use_weight : bool, optional (default=False)
         Flag to indicate whether to compute weighted jaccard (if use_weight==True)
         or un-weighted jaccard (if use_weight==False).
         'input_graph' must be weighted if 'use_weight=True'.
 
-
     Returns
     -------
     df  : cudf.DataFrame
@@ -161,13 +120,6 @@ def jaccard(
     >>> df = jaccard(input_graph)
 
     """
-    if do_expensive_check:
-        warnings.warn(
-            "do_expensive_check is deprecated since vertex IDs are no longer "
-            "required to be consecutively numbered",
-            FutureWarning,
-        )
-
     if input_graph.is_directed():
         raise ValueError("Input must be an undirected Graph.")
 
@@ -220,7 +172,6 @@ def jaccard(
 def jaccard_coefficient(
     G: Union[Graph, "networkx.Graph"],
     ebunch: Union[cudf.DataFrame, Iterable[Union[int, str, float]]] = None,
-    do_expensive_check: bool = False,  # deprecated
 ):
     """
     For NetworkX Compatability.  See `jaccard`
@@ -244,14 +195,6 @@ def jaccard_coefficient(
         pairs. Otherwise, the current implementation computes the overlap
         coefficient for all adjacent vertices in the graph.
 
-    do_expensive_check : bool, optional (default=False)
-        Deprecated.
-                This option added a check to ensure integer vertex IDs are sequential
-        values from 0 to V-1. That check is now redundant because cugraph
-        unconditionally renumbers and un-renumbers integer vertex IDs for
-        optimal performance, therefore this option is deprecated and will be
-        removed in a future version.
-
     Returns
     -------
     df  : cudf.DataFrame
@@ -277,13 +220,6 @@ def jaccard_coefficient(
     >>> df = jaccard_coefficient(G)
 
     """
-    if do_expensive_check:
-        warnings.warn(
-            "do_expensive_check is deprecated since vertex IDs are no longer "
-            "required to be consecutively numbered",
-            FutureWarning,
-        )
-
     vertex_pair = None
 
     G, isNx = ensure_cugraph_obj_for_nx(G)