fix edge generation, serve graph in cytoscape format

backend/lib/backend/crawler/crawler.ex

@@ -80,7 +80,7 @@ defmodule Backend.Crawler do
 
   # Save the state (after crawling) to the database.
   defp save(%Crawler{domain: domain, result: result, found_api?: true, error: nil}) do
-    now = NaiveDateTime.truncate(NaiveDateTime.utc_now(), :second)
+    now = get_now()
 
     ## Update the instance we crawled ##
     Repo.insert!(

backend/lib/backend/edge.ex

@@ -13,6 +13,8 @@ defmodule Backend.Edge do
       type: :string,
       foreign_key: :target_domain
 
+    field :weight, :float
+
     timestamps()
   end
 

backend/lib/backend/instance_peer.ex

@@ -13,8 +13,6 @@ defmodule Backend.InstancePeer do
       type: :string,
       foreign_key: :target_domain
 
-    field :weight, :float, default: 0.0
-
     timestamps()
   end
 

backend/lib/backend/scheduler.ex

@@ -5,7 +5,8 @@ defmodule Backend.Scheduler do
 
   use Quantum.Scheduler, otp_app: :backend
 
-  alias Backend.{Crawl, Edge, Interaction, Instance, Repo}
+  alias Backend.{Crawl, Edge, CrawlInteraction, Instance, Repo}
+  import Backend.Util
   import Ecto.Query
   require Logger
 
@@ -29,88 +30,136 @@ defmodule Backend.Scheduler do
     Logger.info("Pruned #{deleted_num} old crawls.")
   end
 
+  @doc """
+  Calculates every instance's "insularity score" -- that is, the percentage of mentions that are among users on the
+  instance, rather than at other instances.
+  """
+  def generate_insularity_scores() do
+    now = get_now()
+
+    crawls_subquery =
+      Crawl
+      |> select([c], %{
+        instance_domain: c.instance_domain,
+        interactions_seen: sum(c.interactions_seen)
+      })
+      |> where([c], is_nil(c.error))
+      |> group_by([c], c.instance_domain)
+
+    scores =
+      CrawlInteraction
+      |> join(:left, [ci], c in subquery(crawls_subquery),
+        on: ci.source_domain == c.instance_domain
+      )
+      |> where([ci], ci.source_domain == ci.target_domain)
+      |> group_by([ci], ci.source_domain)
+      |> select([ci, c], %{
+        domain: ci.source_domain,
+        mentions: sum(ci.mentions),
+        # we can take min() because every row is the same
+        interactions: min(c.interactions_seen)
+      })
+      |> Repo.all()
+      |> (fn o ->
+            Logger.info(inspect(o))
+            o
+          end).()
+      |> Enum.map(fn %{domain: domain, mentions: mentions, interactions: interactions} ->
+        %{
+          domain: domain,
+          insularity: mentions / interactions,
+          inserted_at: now,
+          updated_at: now
+        }
+      end)
+
+    Instance
+    |> Repo.insert_all(scores,
+      on_conflict: {:replace, [:insularity, :updated_at]},
+      conflict_target: :domain
+    )
+  end
+
   @doc """
   This function aggregates statistics from the interactions in the database.
   It calculates the strength of edges between nodes.
-
-  TODO: generate edge weights. The weight of an edge between two instances will be
-  (number of mentions of each other) / (total number of statuses crawled).
-  This requires us to keep track of how many statuses we've seen.
   """
   def generate_edges() do
+    now = get_now()
+
+    crawls_subquery =
+      Crawl
+      |> select([c], %{
+        instance_domain: c.instance_domain,
+        statuses_seen: sum(c.statuses_seen)
+      })
+      |> where([c], is_nil(c.error))
+      |> group_by([c], c.instance_domain)
+
     interactions =
-      Interaction
-      |> select([inter], {inter.source_domain, inter.target_domain})
-      |> join(:left, [inter], i_source in Instance, on: inter.source_domain == i_source.domain)
-      |> join(:left, [inter], i_target in Instance, on: inter.target_domain == i_target.domain)
-      |> where(
-        [inter, i_source, i_target],
-        not is_nil(i_source.last_crawl_timestamp) and not is_nil(i_target.last_crawl_timestamp)
+      CrawlInteraction
+      |> join(:left, [ci], c_source in subquery(crawls_subquery),
+        on: ci.source_domain == c_source.instance_domain
       )
-      # Repo.all() returns a tuple like {"mastodon.social", "cursed.technology"}
+      |> join(:left, [ci], c_target in subquery(crawls_subquery),
+        on: ci.target_domain == c_target.instance_domain
+      )
+      |> group_by([ci], [ci.source_domain, ci.target_domain])
+      |> select([ci, c_source, c_target], %{
+        source_domain: ci.source_domain,
+        target_domain: ci.target_domain,
+        mentions: sum(ci.mentions),
+        # we can take min() because every row is the same
+        source_statuses_seen: min(c_source.statuses_seen),
+        target_statuses_seen: min(c_target.statuses_seen)
+      })
       |> Repo.all()
-      # Create a map of %{source_domain => [target_domains]}
-      |> Enum.group_by(fn tuple -> Kernel.elem(tuple, 0) end, fn tuple ->
-        Kernel.elem(tuple, 1)
-      end)
 
-    # Calculate insularity score
+    # Get edges and their weights
     Repo.transaction(fn ->
-      interactions
-      |> Enum.each(fn {source, targets} ->
-        total_mentions = length(targets)
-        self_mentions = Enum.count(targets, fn t -> t == source end)
-
-        insularity = self_mentions / total_mentions
-
-        Repo.insert!(
-          %Instance{
-            domain: source,
-            insularity: insularity
-          },
-          on_conflict: [set: [insularity: insularity]],
-          conflict_target: :domain
-        )
-      end)
-
-      # Get edges
-      edges = MapSet.new()
-
-      interactions
-      |> Enum.each(fn {source, targets} ->
-        targets
-        |> Enum.each(fn target ->
-          [key_a, key_b] = Enum.sort([source, target])
-
-          edge = %Edge{
-            source_domain: key_a,
-            target_domain: key_b
-          }
-
-          MapSet.put(edges, edge)
-          Logger.debug(inspect(edges))
-        end)
-      end)
-
-      Logger.debug(inspect(edges))
-
-      now = NaiveDateTime.truncate(NaiveDateTime.utc_now(), :second)
-
-      Repo.delete_all(Edge)
+      Edge
+      |> Repo.delete_all()
 
       edges =
-        edges
-        |> MapSet.to_list()
-        |> Enum.map(fn %{source_domain: source_domain, target_domain: target_domain} ->
-          %Edge{
+        interactions
+        # Get a map of %{{source, target} => {total_mention_count, total_statuses_seen}}
+        |> Enum.reduce(%{}, fn
+          %{
             source_domain: source_domain,
             target_domain: target_domain,
-            updated_at: now,
-            inserted_at: now
+            mentions: mentions,
+            source_statuses_seen: source_statuses_seen,
+            target_statuses_seen: target_statuses_seen
+          } = x,
+          acc ->
+            Logger.info(inspect(x))
+            key = get_interaction_key(source_domain, target_domain)
+
+            # target_statuses_seen might be nil if that instance was never crawled. default to 0.
+            target_statuses_seen =
+              case target_statuses_seen do
+                nil -> 0
+                _ -> target_statuses_seen
+              end
+
+            statuses_seen = source_statuses_seen + target_statuses_seen
+
+            Map.update(acc, key, {mentions, statuses_seen}, fn {curr_mentions, curr_statuses_seen} ->
+              {curr_mentions + mentions, curr_statuses_seen}
+            end)
+        end)
+        |> Enum.map(fn {{source_domain, target_domain}, {mention_count, statuses_seen}} ->
+          %{
+            source_domain: source_domain,
+            target_domain: target_domain,
+            weight: mention_count / statuses_seen,
+            inserted_at: now,
+            updated_at: now
           }
         end)
 
-      Repo.insert_all(Edge, edges)
+      Edge
+      |> Repo.insert_all(edges)
     end)
   end
 end

backend/lib/backend_web/views/graph_view.ex

@@ -16,21 +16,28 @@ defmodule BackendWeb.GraphView do
         false -> 1
       end
 
+    # This is the format that cytoscape.js expects.
     %{
-      id: node.domain,
-      label: node.domain,
-      size: size,
-      x: node.x,
-      y: node.y
+      data: %{
+        id: node.domain,
+        label: node.domain,
+        size: size
+      },
+      position: %{
+        x: node.x,
+        y: node.y
+      }
     }
   end
 
   def render("edge.json", %{graph: edge}) do
     %{
-      id: edge.id,
-      source: edge.source_domain,
-      target: edge.target_domain,
-      size: edge.weight
+      data: %{
+        id: edge.id,
+        source: edge.source_domain,
+        target: edge.target_domain,
+        weight: edge.weight
+      }
     }
   end
 end

backend/priv/repo/migrations/20190624090436_create_instances.exs

@@ -10,6 +10,9 @@ defmodule Backend.Repo.Migrations.CreateInstances do
       add :version, :string
       add :insularity, :float
 
+      add :x, :float
+      add :y, :float
+
       timestamps()
     end
 
@@ -19,8 +22,6 @@ defmodule Backend.Repo.Migrations.CreateInstances do
       add :source_domain, references(:instances, column: :domain, type: :string)
       add :target_domain, references(:instances, column: :domain, type: :string)
 
-      add :weight, :float
-
       timestamps()
     end
 

backend/priv/repo/migrations/20190710133755_create_edges.exs

@@ -6,6 +6,8 @@ defmodule Backend.Repo.Migrations.CreateEdges do
       add :source_domain, references(:instances, column: :domain, type: :string), null: false
       add :target_domain, references(:instances, column: :domain, type: :string), null: false
 
+      add :weight, :float, null: false
+
       timestamps()
     end
 

backend/priv/repo/migrations/20190712133009_add_instance_coords.exs

@@ -1,10 +0,0 @@
-defmodule Backend.Repo.Migrations.AddInstanceCoords do
-  use Ecto.Migration
-
-  def change do
-    alter table(:instances) do
-      add :x, :float
-      add :y, :float
-    end
-  end
-end

frontend/src/components/CytoscapeGraph.tsx

@@ -69,11 +69,12 @@ class GraphImpl extends React.Component<IGraphProps, IGraphState> {
 
     // Check that all nodes have size & coordinates; otherwise the graph will look messed up
     const lengthBeforeFilter = graph.nodes.length;
-    graph = { ...graph, nodes: graph.nodes.filter(n => n.size && n.x && n.y) };
+    graph = { ...graph, nodes: graph.nodes.filter(n => n.data.size && n.position.x && n.position.y) };
     if (graph.nodes.length !== lengthBeforeFilter) {
       // tslint:disable-next-line:no-console
       console.error(
-        "Some nodes were missing details: " + graph.nodes.filter(n => !n.size || !n.x || !n.y).map(n => n.label)
+        "Some nodes were missing details: " +
+          graph.nodes.filter(n => !n.data.size || !n.position.x || !n.position.y).map(n => n.data.label)
       );
       this.setState({ didError: true });
     }
@@ -125,29 +126,9 @@ class GraphImpl extends React.Component<IGraphProps, IGraphState> {
       return;
     }
     this.cy = cytoscape({
-      autoungrabify: true,
+      autoungrabify: false,
       container: this.cytoscapeDiv.current,
-      elements: {
-        edges: graph.edges.map(edge => ({
-          data: {
-            id: edge.id || `${edge.source}${edge.target}`,
-            source: edge.source,
-            target: edge.target,
-            weight: edge.size
-          },
-          group: "edges" as "edges"
-        })),
-        nodes: graph.nodes.map(node => ({
-          data: {
-            id: node.id
-          },
-          group: "nodes" as "nodes",
-          position: {
-            x: node.x,
-            y: node.y
-          }
-        }))
-      },
+      elements: graph,
       layout: {
         name: "preset"
       },

frontend/src/components/Sidebar.tsx

@@ -83,7 +83,7 @@ class SidebarImpl extends React.Component<ISidebarProps, ISidebarState> {
     } else if (
       this.props.graph &&
       this.props.instanceName &&
-      this.props.graph.nodes.map(n => n.id).indexOf(this.props.instanceName) < 0
+      this.props.graph.nodes.map(n => n.data.id).indexOf(this.props.instanceName) < 0
     ) {
       return this.renderQuietInstanceState();
     }
@@ -177,13 +177,15 @@ class SidebarImpl extends React.Component<ISidebarProps, ISidebarState> {
     if (!this.props.graph || !this.props.instanceName) {
       return;
     }
-    const edges = this.props.graph.edges.filter(e => [e.source, e.target].indexOf(this.props.instanceName!) > -1);
+    const edges = this.props.graph.edges.filter(
+      e => [e.data.source, e.data.target].indexOf(this.props.instanceName!) > -1
+    );
     const neighbors: any[] = [];
     edges.forEach(e => {
-      if (e.source === this.props.instanceName) {
-        neighbors.push({ neighbor: e.target, weight: e.size });
+      if (e.data.source === this.props.instanceName) {
+        neighbors.push({ neighbor: e.data.target, weight: e.data.weight });
       } else {
-        neighbors.push({ neighbor: e.source, weight: e.size });
+        neighbors.push({ neighbor: e.data.source, weight: e.data.weight });
       }
     });
     const neighborRows = orderBy(neighbors, ["weight"], ["desc"]).map((neighborDetails: any, idx: number) => (

frontend/src/redux/types.ts

@@ -33,19 +33,24 @@ export interface IInstanceDetails {
 }
 
 interface IGraphNode {
-  id: string;
-  label: string;
-  x: number;
-  y: number;
-  size?: number;
-  color?: string;
+  data: {
+    id: string;
+    label: string;
+    size: number;
+  };
+  position: {
+    x: number;
+    y: number;
+  };
 }
 
 interface IGraphEdge {
-  source: string;
-  target: string;
-  id?: string;
-  size?: number;
+  data: {
+    source: string;
+    target: string;
+    id: string;
+    weight: number;
+  };
 }
 
 export interface IGraph {

gephi/src/main/java/space/fediverse/graph/GraphBuilder.java

@@ -24,7 +24,6 @@ import java.io.IOException;
 import java.sql.Connection;
 import java.sql.PreparedStatement;
 import java.sql.SQLException;
-import java.util.Arrays;
 import java.util.concurrent.TimeUnit;
 
 public class GraphBuilder {