More changes needed for NetworkX plugin (#25)

* More changes needed for NetworkX plugin

- k-truss returns a normal Graph object
- Rename convert to convert_from_nx
- Add convert_to_nx method
- Make algorithms accept Graph, Matrix, or Matrix-variant (expression, transpose, etc)

* Use ensure_type

* Make name option in convert_??_nx functions

* Use networkx branch with changes so tests can pass

Author: jim22k

.github/workflows/test.yml

@@ -33,7 +33,7 @@ jobs:
           conda install -c conda-forge python-graphblas scipy pandas \
             pytest-cov pytest-randomly black flake8-comprehensions flake8-bugbear
             # matplotlib lxml pygraphviz pydot sympy  # Extra networkx deps we don't need yet
-          pip install git+https://github.com/mriduls/networkx.git@nx-sparse --no-deps
+          pip install git+https://github.com/jim22k/networkx.git@nx-sparse --no-deps
           pip install -e . --no-deps
       - name: Style checks
         run: |

graphblas_algorithms/classes/_utils.py

@@ -2,6 +2,7 @@
 import numpy as np
 from graphblas import Matrix, Vector, binary
 from graphblas.core.matrix import TransposedMatrix
+from graphblas.core.utils import ensure_type
 
 ################
 # Classmethods #
@@ -19,7 +20,8 @@ def from_networkx(cls, G, weight=None, dtype=None):
 
 
 def from_graphblas(cls, A, *, key_to_id=None):
-    # Does not copy!
+    # Does not copy if A is a Matrix!
+    A = ensure_type(A, Matrix)
     if A.nrows != A.ncols:
         raise ValueError(f"Adjacency matrix must be square; got {A.nrows} x {A.ncols}")
     rv = cls()

graphblas_algorithms/classes/digraph.py

@@ -414,8 +414,10 @@ def to_directed_graph(G, weight=None, dtype=None):
     # We should do some sanity checks here to ensure we're returning a valid directed graph
     if isinstance(G, DiGraph):
         return G
-    if isinstance(G, Matrix):
+    try:
         return DiGraph.from_graphblas(G)
+    except TypeError:
+        pass
 
     try:
         import networkx as nx
@@ -431,9 +433,11 @@ def to_directed_graph(G, weight=None, dtype=None):
 def to_graph(G, weight=None, dtype=None):
     if isinstance(G, (DiGraph, ga.Graph)):
         return G
-    if isinstance(G, Matrix):
+    try:
         # Should we check if it can be undirected?
         return DiGraph.from_graphblas(G)
+    except TypeError:
+        pass
 
     try:
         import networkx as nx

graphblas_algorithms/classes/graph.py

@@ -152,8 +152,10 @@ def to_undirected_graph(G, weight=None, dtype=None):
     # We should do some sanity checks here to ensure we're returning a valid undirected graph
     if isinstance(G, Graph):
         return G
-    if isinstance(G, Matrix):
+    try:
         return Graph.from_graphblas(G)
+    except TypeError:
+        pass
 
     try:
         import networkx as nx

graphblas_algorithms/interface.py

@@ -68,7 +68,7 @@ class Dispatcher:
     is_triad = nxapi.triads.is_triad
 
     @staticmethod
-    def convert(graph, weight=None):
+    def convert_from_nx(graph, weight=None, *, name=None):
         import networkx as nx
 
         from .classes import DiGraph, Graph, MultiDiGraph, MultiGraph
@@ -83,6 +83,14 @@ def convert(graph, weight=None):
             return Graph.from_networkx(graph, weight=weight)
         raise TypeError(f"Unsupported type of graph: {type(graph)}")
 
+    @staticmethod
+    def convert_to_nx(obj, *, name=None):
+        from .classes import Graph
+
+        if isinstance(obj, Graph):
+            obj = obj.to_networkx()
+        return obj
+
     @staticmethod
     def on_start_tests(items):
         skip = [

graphblas_algorithms/nxapi/core.py

@@ -10,6 +10,4 @@
 def k_truss(G, k):
     G = to_undirected_graph(G, dtype=bool)
     result = algorithms.k_truss(G, k)
-    # TODO: don't convert to networkx graph
-    # We want to be able to pass networkx tests, so we need to improve our graph objects
-    return result.to_networkx()
+    return result