nx-cugraph: automatically generate trees in README.md

.gitignore

@@ -78,6 +78,8 @@ datasets/*
 !datasets/karate-disjoint.csv
 !datasets/netscience.csv
 
+# nx-python side effects
+python/nx-cugraph/objects.inv
 
 .pydevproject
 

.pre-commit-config.yaml

@@ -17,6 +17,7 @@ repos:
         language_version: python3
         args: [--target-version=py38]
         files: ^(python/.*|benchmarks/.*)$
+        exclude: ^python/nx-cugraph/
   - repo: https://github.com/PyCQA/flake8
     rev: 6.0.0
     hooks:

ci/test_python.sh

@@ -149,6 +149,13 @@ python -m nx_cugraph.scripts.print_tree --dispatch-name --plc --incomplete --dif
 python -m nx_cugraph.scripts.print_table
 popd
 
+rapids-logger "ensure nx-cugraph autogenerated files are up to date"
+pushd python/nx-cugraph
+make || true
+git diff --exit-code .
+git checkout .
+popd
+
 rapids-logger "pytest cugraph-service (single GPU)"
 pushd python/cugraph-service
 pytest \

python/nx-cugraph/Makefile

@@ -1,12 +1,12 @@
-# Copyright (c) 2023, NVIDIA CORPORATION.
+# Copyright (c) 2023-2024, NVIDIA CORPORATION.
 SHELL= /bin/bash
 
 .PHONY: all
-all: plugin-info lint
+all: plugin-info lint readme
 
 .PHONY: lint
 lint:
-	git ls-files | xargs pre-commit run --config lint.yaml --files
+	git ls-files | xargs pre-commit run --config lint.yaml --files || true
 
 .PHONY: lint-update
 lint-update:
@@ -15,3 +15,10 @@ lint-update:
 .PHONY: plugin-info
 plugin-info:
 	python _nx_cugraph/__init__.py
+
+objects.inv:
+	wget https://networkx.org/documentation/stable/objects.inv
+
+.PHONY: readme
+readme: objects.inv
+	python scripts/update_readme.py README.md objects.inv

python/nx-cugraph/lint.yaml

@@ -45,12 +45,12 @@ repos:
       - id: pyupgrade
         args: [--py39-plus]
   - repo: https://github.com/psf/black
-    rev: 23.12.1
+    rev: 24.1.1
     hooks:
       - id: black
       # - id: black-jupyter
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.1.14
+    rev: v0.2.1
     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==24.1.17
+          - flake8-bugbear==24.2.6
           - flake8-simplify==0.21.0
   - repo: https://github.com/asottile/yesqa
     rev: v1.5.0
@@ -77,7 +77,7 @@ repos:
         additional_dependencies: [tomli]
         files: ^(nx_cugraph|docs)/
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.1.14
+    rev: v0.2.1
     hooks:
       - id: ruff
   - repo: https://github.com/pre-commit/pre-commit-hooks

python/nx-cugraph/nx_cugraph/algorithms/link_analysis/hits_alg.py

@@ -66,9 +66,9 @@ def hits(
             resource_handle=plc.ResourceHandle(),
             graph=G._get_plc_graph(weight, 1, dtype, store_transposed=True),
             tol=tol,
-            initial_hubs_guess_vertices=None
-            if nstart is None
-            else cp.arange(N, dtype=index_dtype),
+            initial_hubs_guess_vertices=(
+                None if nstart is None else cp.arange(N, dtype=index_dtype)
+            ),
             initial_hubs_guess_values=nstart,
             max_iter=max_iter,
             normalized=normalized,

python/nx-cugraph/nx_cugraph/algorithms/link_analysis/pagerank_alg.py

@@ -78,9 +78,9 @@ def pagerank(
         "graph": G._get_plc_graph(weight, 1, dtype, store_transposed=True),
         "precomputed_vertex_out_weight_vertices": None,
         "precomputed_vertex_out_weight_sums": None,
-        "initial_guess_vertices": None
-        if nstart is None
-        else cp.arange(N, dtype=index_dtype),
+        "initial_guess_vertices": (
+            None if nstart is None else cp.arange(N, dtype=index_dtype)
+        ),
         "initial_guess_values": nstart,
         "alpha": alpha,
         "epsilon": N * tol,

python/nx-cugraph/nx_cugraph/classes/graph.py

@@ -634,8 +634,7 @@ def _get_plc_graph(
                             "pylibcugraph only supports float16 and float32 dtypes."
                         )
                 elif (
-                    edge_array.dtype == np.uint64
-                    and edge_array.max().tolist() > 2**53
+                    edge_array.dtype == np.uint64 and edge_array.max().tolist() > 2**53
                 ):
                     raise ValueError(
                         f"Integer value of value is too large (> 2**53): {val}; "

python/nx-cugraph/nx_cugraph/classes/multigraph.py

@@ -360,9 +360,7 @@ def get_edge_data(
                 if k not in self.edge_masks or self.edge_masks[k][index]
             }
         return {
-            edge_keys[index]
-            if edge_keys is not None
-            else index: {
+            edge_keys[index] if edge_keys is not None else index: {
                 k: v[index].tolist()
                 for k, v in self.edge_values.items()
                 if k not in self.edge_masks or self.edge_masks[k][index]

python/nx-cugraph/nx_cugraph/interface.py

@@ -220,9 +220,9 @@ def key(testpath):
             )
             if sys.version_info[:2] == (3, 9):
                 # This test is sensitive to RNG, which depends on Python version
-                xfail[
-                    key("test_louvain.py:test_threshold")
-                ] = "Louvain does not support seed parameter"
+                xfail[key("test_louvain.py:test_threshold")] = (
+                    "Louvain does not support seed parameter"
+                )
             if nxver.major == 3 and nxver.minor >= 2:
                 xfail.update(
                     {

python/nx-cugraph/nx_cugraph/scripts/print_table.py

@@ -61,7 +61,8 @@ def main(path_to_info=None, *, file=sys.stdout):
     lines = ["networkx_path,dispatch_name,version_added,plc,is_incomplete,is_different"]
     lines.extend(",".join(map(str, info)) for info in path_to_info.values())
     text = "\n".join(lines)
-    print(text, file=file)
+    if file is not None:
+        print(text, file=file)
     return text
 
 

python/nx-cugraph/nx_cugraph/scripts/print_tree.py

@@ -12,29 +12,58 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import argparse
-import re
 import sys
 
-import networkx as nx
+from nx_cugraph.scripts.print_table import Info, get_path_to_info
 
-from nx_cugraph.scripts.print_table import get_path_to_info
 
+def assoc_in(d, keys, value):
+    """Like Clojure's assoc-in, but modifies d in-place."""
+    inner = d
+    keys = iter(keys)
+    key = next(keys)
+    for next_key in keys:
+        if key not in inner:
+            inner[key] = {}
+        inner = inner[key]
+        key = next_key
+    inner[key] = value
+    return d
 
-def add_branch(G, path, extra="", *, skip=0):
-    branch = path.split(".")
-    prev = ".".join(branch[: skip + 1])
-    for i in range(skip + 2, len(branch)):
-        cur = ".".join(branch[:i])
-        G.add_edge(prev, cur)
-        prev = cur
-    if extra:
-        if not isinstance(extra, str):
-            extra = ", ".join(extra)
-        path += f" ({extra})"
-    G.add_edge(prev, path)
+
+def default_get_payload_internal(keys):
+    return keys[-1]
+
+
+def tree_lines(
+    tree,
+    parents=(),
+    are_levels_closing=(),
+    get_payload_internal=default_get_payload_internal,
+):
+    pre = "".join(
+        "    " if is_level_closing else " │  "
+        for is_level_closing in are_levels_closing
+    )
+    c = "├"
+    are_levels_closing += (False,)
+    for i, (key, val) in enumerate(tree.items(), 1):
+        if i == len(tree):  # Last item
+            c = "└"
+            are_levels_closing = are_levels_closing[:-1] + (True,)
+        if isinstance(val, str):
+            yield pre + f" {c}─ " + val
+        else:
+            yield pre + f" {c}─ " + get_payload_internal((*parents, key))
+            yield from tree_lines(
+                val,
+                (*parents, key),
+                are_levels_closing,
+                get_payload_internal=get_payload_internal,
+            )
 
 
-def get_extra(
+def get_payload(
     info,
     *,
     networkx_path=False,
@@ -64,7 +93,10 @@ def get_extra(
         extra.append("is-incomplete")
     if different and info.is_different:
         extra.append("is-different")
-    return extra
+    extra = ", ".join(extra)
+    if extra:
+        extra = f" ({extra})"
+    return info.networkx_path.rsplit(".", 1)[-1] + extra
 
 
 def create_tree(
@@ -80,20 +112,28 @@ def create_tree(
     incomplete=False,
     different=False,
     prefix="",
+    strip_networkx=True,
+    get_payload=get_payload,
 ):
     if path_to_info is None:
         path_to_info = get_path_to_info()
+    if strip_networkx:
+        path_to_info = {
+            key: Info(info.networkx_path.replace("networkx.", "", 1), *info[1:])
+            for key, info in path_to_info.items()
+        }
     if isinstance(by, str):
         by = [by]
-    G = nx.DiGraph()
+    # We rely on the fact that dicts maintain order
+    tree = {}
     for info in sorted(
         path_to_info.values(),
         key=lambda x: (*(getattr(x, b) for b in by), x.networkx_path),
     ):
         if not all(getattr(info, b) for b in by):
             continue
         path = prefix + ".".join(getattr(info, b) for b in by)
-        extra = get_extra(
+        payload = get_payload(
             info,
             networkx_path=networkx_path,
             dispatch_name=dispatch_name,
@@ -103,8 +143,8 @@ def create_tree(
             incomplete=incomplete,
             different=different,
         )
-        add_branch(G, path, extra=extra, skip=skip)
-    return G
+        assoc_in(tree, path.split("."), payload)
+    return tree
 
 
 def main(
@@ -132,45 +172,33 @@ def main(
         "different": different,
     }
     if by == "networkx_path":
-        G = create_tree(path_to_info, by="networkx_path", **kwargs)
-        text = re.sub(
-            r" [A-Za-z_\./]+\.", " ", ("\n".join(nx.generate_network_text(G)))
-        )
+        tree = create_tree(path_to_info, by="networkx_path", **kwargs)
+        text = "\n".join(tree_lines(tree))
     elif by == "plc":
-        G = create_tree(
-            path_to_info, by=["plc", "networkx_path"], prefix="plc-", **kwargs
-        )
-        text = re.sub(
-            "plc-",
-            "plc.",
-            re.sub(
-                r" plc-[A-Za-z_\./]*\.",
-                " ",
-                "\n".join(nx.generate_network_text(G)),
-            ),
+        tree = create_tree(
+            path_to_info,
+            by=["plc", "networkx_path"],
+            prefix="plc-",
+            **kwargs,
         )
+        text = "\n".join(tree_lines(tree)).replace("plc-", "plc.")
     elif by == "version_added":
-        G = create_tree(
+        tree = create_tree(
             path_to_info,
             by=["version_added", "networkx_path"],
             prefix="version_added-",
             **kwargs,
         )
-        text = re.sub(
-            "version_added-",
-            "version: ",
-            re.sub(
-                r" version_added-[-0-9A-Za-z_\./]*\.",
-                " ",
-                "\n".join(nx.generate_network_text(G)),
-            ),
-        ).replace("-", ".")
+        text = "\n".join(tree_lines(tree)).replace("version_added-", "version: ")
+        for digit in "0123456789":
+            text = text.replace(f"2{digit}-", f"2{digit}.")
     else:
         raise ValueError(
             "`by` argument should be one of {'networkx_path', 'plc', 'version_added' "
             f"got: {by}"
         )
-    print(text, file=file)
+    if file is not None:
+        print(text, file=file)
     return text
 
 

python/nx-cugraph/pyproject.toml

@@ -149,8 +149,10 @@ exclude_lines = [
 # https://github.com/charliermarsh/ruff/
 line-length = 88
 target-version = "py39"
+[tool.ruff.lint]
 unfixable = [
     "F841",  # unused-variable (Note: can leave useless expression)
+    "B905",  # zip-without-explicit-strict (Note: prefer `zip(x, y, strict=True)`)
 ]
 select = [
     "ALL",
@@ -172,7 +174,6 @@ ignore = [
     # "SIM401",  # Use dict.get ... instead of if-else-block (Note: if-else better for coverage and sometimes clearer)
     # "TRY004",  # Prefer `TypeError` exception for invalid type (Note: good advice, but not worth the nuisance)
     "B904",  # Bare `raise` inside exception clause (like TRY200; sometimes okay)
-    "TRY200",  # Use `raise from` to specify exception cause (Note: sometimes okay to raise original exception)
 
     # Intentionally ignored
     "A003",  # Class attribute ... is shadowing a python builtin
@@ -224,22 +225,22 @@ ignore = [
     "PD",  # pandas-vet (Intended for scripts that use pandas, not libraries)
 ]
 
-[tool.ruff.per-file-ignores]
+[tool.ruff.lint.per-file-ignores]
 "__init__.py" = ["F401"]  # Allow unused imports (w/o defining `__all__`)
 # Allow assert, print, RNG, and no docstring
 "nx_cugraph/**/tests/*py" = ["S101", "S311", "T201", "D103", "D100"]
 "_nx_cugraph/__init__.py" = ["E501"]
 "nx_cugraph/algorithms/**/*py" = ["D205", "D401"]  # Allow flexible docstrings for algorithms
 
-[tool.ruff.flake8-annotations]
+[tool.ruff.lint.flake8-annotations]
 mypy-init-return = true
 
-[tool.ruff.flake8-builtins]
+[tool.ruff.lint.flake8-builtins]
 builtins-ignorelist = ["copyright"]
 
-[tool.ruff.flake8-pytest-style]
+[tool.ruff.lint.flake8-pytest-style]
 fixture-parentheses = false
 mark-parentheses = false
 
-[tool.ruff.pydocstyle]
+[tool.ruff.lint.pydocstyle]
 convention = "numpy"