Introduce and use decorators for graph sanitizers

netrd/distance/communicability_jsd.py

@@ -21,12 +21,14 @@
 import networkx as nx
 import numpy as np
 from .base import BaseDistance
-from ..utilities import entropy, ensure_undirected, ensure_unweighted
+from ..utilities import entropy, undirected, unweighted
 
 
 class CommunicabilityJSD(BaseDistance):
     """Jensen-Shannon divergence between communicability sequences."""
 
+    @undirected
+    @unweighted
     def dist(self, G1, G2):
         r"""Compares the communicability matrix of two graphs.
 
@@ -83,12 +85,6 @@ def dist(self, G1, G2):
 
         """
 
-        G1 = ensure_undirected(G1)
-        G2 = ensure_undirected(G2)
-
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
-
         N1 = G1.number_of_nodes()
         N2 = G2.number_of_nodes()
 

netrd/distance/degree_divergence.py

@@ -15,12 +15,13 @@
 import numpy as np
 import networkx as nx
 from .base import BaseDistance
-from ..utilities import entropy, ensure_unweighted
+from ..utilities import entropy, unweighted
 
 
 class DegreeDivergence(BaseDistance):
     """Compare two degree distributions."""
 
+    @unweighted
     def dist(self, G1, G2):
         """Jenson-Shannon divergence between degree distributions.
 
@@ -40,9 +41,6 @@ def dist(self, G1, G2):
 
         """
 
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
-
         def degree_vector_histogram(graph):
             """Return the degrees in both formats.
 

netrd/distance/deltacon.py

@@ -18,12 +18,13 @@
 import numpy as np
 import networkx as nx
 from .base import BaseDistance
-from ..utilities import ensure_undirected
+from ..utilities import undirected
 
 
 class DeltaCon(BaseDistance):
     """Compare matrices related to Fast Belief Propagation."""
 
+    @undirected
     def dist(self, G1, G2, exact=True, g=None):
         """DeltaCon is based on the Matsusita between matrices created from fast
         belief propagation (FBP) on graphs G1 and G2.
@@ -70,9 +71,6 @@ def dist(self, G1, G2, exact=True, g=None):
         if not exact and g is None:
             g = N
 
-        G1 = ensure_undirected(G1)
-        G2 = ensure_undirected(G2)
-
         A1 = nx.to_numpy_array(G1)
         L1 = nx.laplacian_matrix(G1).toarray()
         D1 = L1 + A1

netrd/distance/distributional_nbd.py

@@ -10,7 +10,7 @@
 import networkx as nx
 import scipy.sparse as sp
 from scipy.spatial.distance import euclidean, chebyshev
-from ..utilities.graph import ensure_unweighted
+from ..utilities.graph import unweighted
 
 from .base import BaseDistance
 
@@ -31,6 +31,7 @@ class DistributionalNBD(BaseDistance):
 
     VECTOR_DISTANCES = {'euclidean': euclidean, 'chebyshev': chebyshev}
 
+    @unweighted
     def dist(
         self,
         G1,
@@ -78,9 +79,6 @@ def dist(
             The distance between `G1` and `G2`
 
         """
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
-
         B1 = reduced_hashimoto(G1, shave=shave, sparse=sparse, **kwargs)
         B2 = reduced_hashimoto(G2, shave=shave, sparse=sparse, **kwargs)
 

netrd/distance/dk_series.py

@@ -16,12 +16,14 @@
 from scipy.sparse import coo_matrix
 from collections import defaultdict
 from .base import BaseDistance
-from ..utilities import entropy, ensure_undirected, ensure_unweighted
+from ..utilities import entropy, undirected, unweighted
 
 
 class dkSeries(BaseDistance):
     """Compare graphs based on their :math:`dk`-series."""
 
+    @unweighted
+    @undirected
     def dist(self, G1, G2, d=2):
         r"""Compute the distance between two graphs by using the Jensen-Shannon
         divergence between the :math:`dk`-series of the graphs.
@@ -59,11 +61,6 @@ def dist(self, G1, G2, d=2):
 
         """
 
-        G1 = ensure_undirected(G1)
-        G2 = ensure_undirected(G2)
-
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
         N = max(len(G1), len(G2))
 
         if d == 1:

netrd/distance/dmeasure.py

@@ -23,12 +23,13 @@
 from scipy.stats import entropy
 from .base import BaseDistance
 from ..utilities.entropy import js_divergence
-from ..utilities import ensure_undirected
+from ..utilities import undirected
 
 
 class DMeasure(BaseDistance):
     """Compare two graphs by their network node dispersion."""
 
+    @undirected
     def dist(self, G1, G2, w1=0.45, w2=0.45, w3=0.10, niter=50):
         r"""The D-Measure is a comparison of structural dissimilarities between graphs.
 
@@ -104,9 +105,6 @@ def dist(self, G1, G2, w1=0.45, w2=0.45, w3=0.10, niter=50):
         if sum([w1, w2, w3]) != 1:
             raise ValueError("Weights must sum to one.")
 
-        G1 = ensure_undirected(G1)
-        G2 = ensure_undirected(G2)
-
         first_term = 0
         second_term = 0
         third_term = 0

netrd/distance/frobenius.py

@@ -9,12 +9,13 @@
 import numpy as np
 import networkx as nx
 from .base import BaseDistance
-from ..utilities.graph import ensure_unweighted
+from ..utilities.graph import unweighted
 
 
 class Frobenius(BaseDistance):
     """The Frobenius distance between their adjacency matrices."""
 
+    @unweighted
     def dist(self, G1, G2):
         r"""Frobenius distance between two graphs.
 
@@ -45,9 +46,6 @@ def dist(self, G1, G2):
 
         """
 
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
-
         adj1 = nx.to_numpy_array(G1)
         adj2 = nx.to_numpy_array(G2)
         dist = np.linalg.norm((adj1 - adj2))

netrd/distance/hamming.py

@@ -11,12 +11,13 @@
 import numpy as np
 import networkx as nx
 from .base import BaseDistance
-from ..utilities.graph import ensure_unweighted
+from ..utilities import unweighted
 
 
 class Hamming(BaseDistance):
     """Entry-wise disagreement between adjacency matrices."""
 
+    @unweighted
     def dist(self, G1, G2):
         r"""The proportion of disagreeing nodes between the flattened adjacency
         matrices.
@@ -57,9 +58,6 @@ def dist(self, G1, G2):
 
         """
 
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
-
         if G1.number_of_nodes() == G2.number_of_nodes():
             N = G1.number_of_nodes()
         else:

netrd/distance/hamming_ipsen_mikhailov.py

@@ -18,12 +18,13 @@
 from .base import BaseDistance
 from scipy.optimize import fsolve
 from .ipsen_mikhailov import _im_distance
-from ..utilities.graph import ensure_unweighted
+from ..utilities.graph import unweighted
 
 
 class HammingIpsenMikhailov(BaseDistance):
     """Combination of Hamming and Ipsen-Mikhailov distances."""
 
+    @unweighted
     def dist(self, G1, G2, combination_factor=1):
         """Graph distance combining local and global distances.
 
@@ -80,9 +81,6 @@ def dist(self, G1, G2, combination_factor=1):
         """
         N = len(G1)
 
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
-
         # get the adjacency matrices
         adj1 = nx.to_numpy_array(G1)
         adj2 = nx.to_numpy_array(G2)

netrd/distance/ipsen_mikhailov.py

@@ -19,12 +19,13 @@
 from scipy.sparse.csgraph import laplacian
 from scipy.linalg import eigh
 from scipy.integrate import quad
-from ..utilities.graph import ensure_unweighted
+from ..utilities.graph import unweighted
 
 
 class IpsenMikhailov(BaseDistance):
     """Compares the spectrum of the Laplacian matrices."""
 
+    @unweighted
     def dist(self, G1, G2, hwhm=0.08):
         """Compare the spectrum ot the associated Laplacian matrices.
 
@@ -58,8 +59,6 @@ def dist(self, G1, G2, hwhm=0.08):
 
         """
         N = len(G1)
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
 
         # get the adjacency matrices
         adj1 = nx.to_numpy_array(G1)

netrd/distance/jaccard_distance.py

@@ -11,14 +11,13 @@
 """
 
 from .base import BaseDistance
-import networkx as nx
-import numpy as np
-from ..utilities import ensure_unweighted
+from ..utilities import unweighted
 
 
 class JaccardDistance(BaseDistance):
     """Jaccard distance between edge sets."""
 
+    @unweighted
     def dist(self, G1, G2):
         r"""Compute the Jaccard index between two graphs.
 
@@ -50,9 +49,6 @@ def dist(self, G1, G2):
 
         """
 
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
-
         e1 = set(G1.edges)
         e2 = set(G2.edges)
         cup = set.union(e1, e2)

netrd/distance/laplacian_spectral_method.py

@@ -15,7 +15,7 @@
 import numpy as np
 import networkx as nx
 from .base import BaseDistance
-from ..utilities.graph import ensure_unweighted
+from ..utilities.graph import unweighted
 from scipy.special import erf
 from scipy.integrate import quad
 from scipy.linalg import eigvalsh
@@ -27,6 +27,7 @@
 class LaplacianSpectral(BaseDistance):
     """Flexible distance able to compare the spectrum of the Laplacian in many ways."""
 
+    @unweighted
     def dist(
         self,
         G1,
@@ -113,8 +114,6 @@ def dist(
         .. [2] https://ieeexplore.ieee.org/abstract/document/7344816.
 
         """
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
         adj1 = nx.to_numpy_array(G1)
         adj2 = nx.to_numpy_array(G2)
         self.results['adjacency_matrices'] = adj1, adj2

netrd/distance/nbd.py

@@ -12,7 +12,7 @@
 from .base import BaseDistance
 from collections import defaultdict, Counter
 from ortools.linear_solver import pywraplp
-from ..utilities.graph import ensure_unweighted
+from ..utilities import unweighted
 
 
 class NonBacktrackingSpectral(BaseDistance):
@@ -23,6 +23,7 @@ class NonBacktrackingSpectral(BaseDistance):
 
     """
 
+    @unweighted
     def dist(self, G1, G2, topk='automatic', batch=100, tol=1e-5):
         """Non-Backtracking Distance between two graphs.
 
@@ -53,9 +54,6 @@ def dist(self, G1, G2, topk='automatic', batch=100, tol=1e-5):
 
         """
 
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
-
         vals1 = nbvals(G1, topk, batch, tol)
         vals2 = nbvals(G2, topk, batch, tol)
 

netrd/distance/netlsd.py

@@ -13,12 +13,14 @@
 import scipy.linalg as spl
 
 from .base import BaseDistance
-from ..utilities import ensure_undirected, ensure_unweighted
+from ..utilities import undirected, unweighted
 
 
 class NetLSD(BaseDistance):
     """Compares spectral node signature distributions."""
 
+    @undirected
+    @unweighted
     def dist(self, G1, G2, normalization=None, timescales=None):
         """NetLSD: Hearing the Shape of a Graph.
 
@@ -62,12 +64,6 @@ def dist(self, G1, G2, normalization=None, timescales=None):
             normalization, str
         ), 'Normalization parameter must be of string type'
 
-        G1 = ensure_undirected(G1)
-        G2 = ensure_undirected(G2)
-
-        G1 = ensure_unweighted(G1)
-        G2 = ensure_unweighted(G2)
-
         lap1 = nx.normalized_laplacian_matrix(G1)
         lap2 = nx.normalized_laplacian_matrix(G2)