Overhaul graph reading and writing to support Multi/Di/Weighted graph…

…s naturally.

circles_graph.py

@@ -63,7 +63,7 @@ def main():
   graph = graph_tools.make_graph(nodes, edges)
   utils.log(f"Created graph with {graph.number_of_nodes():_} nodes and {graph.number_of_edges():_} edges")
 
-  graph_file = Path("results", "circles", "graph", f"{args.focus_id}.{args.num_circles}.adj.nx")
+  graph_file = Path("results", "circles", "graph", f"{args.focus_id}.{args.num_circles}")
   graph_file.parent.mkdir(parents=True, exist_ok=True)
   graph_tools.write_graph(graph, graph_file)
   utils.log(f"Wrote: {str(graph_file)}")

dump_build.sh

@@ -32,7 +32,7 @@ echo "(4) Building Graph"
 # 2m
 time python3 graph_make_family.py --version=${TIMESTAMP}
 # 10m
-time python3 graph_core.py ${VERSION_DIR}/graphs/family/all.graph.adj.nx
+time python3 graph_core.py ${VERSION_DIR}/graphs/family/all.adj.nx
 
 echo
 echo "(5) Convert to SQLite DB"

graph_core.py

@@ -168,8 +168,8 @@ def main():
   utils.log(f"Found main component:  {len(graph.nodes):_} Nodes / {len(graph.edges):_} Edges")
   utils.log(degree_distr_str(graph))
 
-  filename = graph_dir / "main.graph.adj.nx"
-  graph_tools.write_graph(graph, filename)
+  basename = graph_dir / "main"
+  filename = graph_tools.write_graph(graph, basename)
   utils.log(f"Saved main component to {str(filename)}")
 
   # Need to copy, because graph is frozen here (due to being a subgraph).
@@ -178,21 +178,20 @@ def main():
   utils.log(f"Shaved graph:  {len(graph.nodes):_} Nodes / {len(graph.edges):_} Edges")
   utils.log(degree_distr_str(graph))
 
-  filename = graph_dir / "shaved.graph.adj.nx"
-  graph_tools.write_graph(graph, filename)
+  basename = graph_dir / "shaved"
+  filename = graph_tools.write_graph(graph, basename)
   utils.log(f"Saved shaved main to {str(filename)}")
 
   # Map: core nodes -> nodes that collapse into this core node
   graph, rep_nodes = FindCore(graph)
   utils.log(f"Contracted graph:  {len(graph.nodes):_} Nodes / {len(graph.edges):_} Edges / {num_dup_edges(graph):_} Duplicate edges / {nx.number_of_selfloops(graph):_} Selfloops")
   utils.log(degree_distr_str(graph))
 
-  # Save as "Edgelist" to support edge weights.
-  filename = graph_dir / "core.graph.edgelist.nx"
-  graph_tools.write_graph(graph, filename)
-  utils.log(f"Saved core to {str(filename)}")
+  basename = graph_dir / "topo"
+  filename = graph_tools.write_graph(graph, basename)
+  utils.log(f"Saved topo to {str(filename)}")
 
-  filename = graph_dir / "core.collapse.csv"
+  filename = graph_dir / "topo.collapse.csv"
   with open(filename, "w") as f:
     csv_out = csv.DictWriter(f, ["core_node", "sub_node"])
     csv_out.writeheader()

graph_make_bipartite.py

@@ -109,8 +109,8 @@ def main():
   del df, child_edges, parent_edges
   utils.log(f"Built graph with {len(graph.nodes):_} Nodes / {len(graph.edges):_} Edges")
 
-  filename = Path(graph_dir, "all.graph.adj.nx")
-  graph_tools.write_graph(graph, filename)
+  basename = Path(graph_dir, "all")
+  filename = graph_tools.write_graph(graph, basename)
   utils.log(f"Saved graph to {str(filename)}")
 
   utils.log("Finished")

graph_make_family.py

@@ -86,8 +86,8 @@ def main():
   graph = nx.from_pandas_edgelist(df, "child_id", "parent_id")
   utils.log(f"Built graph with {len(graph.nodes):_} Nodes / {len(graph.edges):_} Edges")
 
-  filename = Path(graph_dir, "all.graph.adj.nx")
-  graph_tools.write_graph(graph, filename)
+  basename = Path(graph_dir, "all")
+  filename = graph_tools.write_graph(graph, basename)
   utils.log(f"Saved graph to {str(filename)}")
 
   utils.log("Finished")

graph_tools.py

@@ -64,36 +64,49 @@ def load_graph_nk(filename):
   else:
     raise Exception(f"Invalid graph filename: {filename}")
 
-def load_graph(filename):
+def load_graph(filename : Path) -> nx.Graph:
   """Load a graph from various formats depending on the extensions."""
   filename = Path(filename)
-  if ".adj" in filename.suffixes:
-    return nx.read_adjlist(filename)
 
-  elif ".edgelist" in filename.suffixes:
-    return nx.read_weighted_edgelist(filename)
+  if ".multi" in filename.suffixes:
+    if ".di" in filename.suffixes:
+      g_type = nx.MultiDiGraph
+    else:
+      g_type = nx.MultiGraph
+  else:
+    if ".di" in filename.suffixes:
+      g_type = nx.DiGraph
+    else:
+      g_type = nx.Graph
+
+  if ".adj" in filename.suffixes:
+    return nx.read_adjlist(filename, create_using=g_type)
 
-  elif ".gml" in filename.suffixes:
-    return nx.read_gml(filename)
+  elif ".edges" in filename.suffixes:
+    return nx.read_weighted_edgelist(filename, create_using=g_type)
 
   else:
     raise Exception(f"Invalid graph filename: {filename}")
 
-def write_graph(graph, filename):
-  """Write a graph into various formats depending on the extension."""
-  filename = Path(filename)
-  if ".adj" in filename.suffixes:
-    assert not is_weighted(graph), "Writing adjlist drops weights!"
-    nx.write_adjlist(graph, filename)
+def write_graph(graph : nx.Graph, basename : Path) -> Path:
+  """Write a graph into various formats depending on Type."""
+  basename = str(basename)
+
+  if graph.is_directed():
+    basename += ".di"
+
+  if graph.is_multigraph():
+    basename += ".multi"
 
-  elif ".edgelist" in filename.suffixes:
+  if is_weighted(graph):
+    filename = Path(basename + ".weight.edges.nx")
     nx.write_weighted_edgelist(graph, filename)
 
-  elif ".gml" in filename.suffixes:
-    nx.write_gml(graph, filename)
+  else:  # Non-weighted
+    filename = Path(basename + ".adj.nx")
+    nx.write_adjlist(graph, filename)
 
-  else:
-    raise Exception(f"Invalid graph filename: {filename}")
+  return filename
 
 def write_graph_nk(graph, names, filename):
   filename = Path(filename)

shapinsay_analyze.py

@@ -113,7 +113,7 @@ def build_bipartite_graph(user_nums, db):
     graph_info.add_person(user_num)
 
   graph = nx.Graph()
-  for id in graph_info.compute_node_ids():
+  for id in graph_info.compute_union_ids():
     graph.add_node(id)
   for (id1, id2) in graph_info.edge_ids:
     graph.add_edge(id1, id2)
@@ -152,8 +152,8 @@ def main():
   utils.log(f"Computed graph with {len(graph.nodes):_} nodes and "
             f"{len(graph.edges):_} edges")
 
-  graph_tools.write_graph(graph, "results/shapinsay/full.graph.adj.nx")
-  utils.log("Wrote graph to results/shapinsay/full.graph.adj.nx")
+  filename = graph_tools.write_graph(graph, "results/shapinsay/full")
+  utils.log(f"Wrote graph to {str(filename)}")
 
   components = sorted(nx.connected_components(graph), key = len, reverse=True)
 
@@ -191,8 +191,8 @@ def main():
 
   print()
   main = graph.subgraph(components[0])
-  graph_tools.write_graph(main, "results/shapinsay/main.graph.adj.nx")
-  utils.log("Wrote main component to results/shapinsay/main.graph.adj.nx")
+  filename = graph_tools.write_graph(main, "results/shapinsay/main")
+  utils.log(f"Wrote main component to {str(filename)}")
 
   bicomps = sorted(nx.biconnected_components(main), key = len, reverse=True)
 
@@ -230,8 +230,8 @@ def main():
 
   print()
   main_bi = graph.subgraph(bicomps[0])
-  graph_tools.write_graph(main_bi, "results/shapinsay/main_bi.graph.adj.nx")
-  utils.log("Wrote main bi-component to results/shapinsay/main_bi.graph.adj.nx")
+  filename = graph_tools.write_graph(main_bi, "results/shapinsay/main_bi")
+  utils.log(f"Wrote main bi-component to {str(filename)}")
 
   print()
   utils.log(f"Circle sizes around {most_central[0]}:")