find the most "fundamental" notes in the graph with PageRank (also with Katz centrality but I commented that out)

src/bin/build-index.rs

@@ -4,7 +4,7 @@ use std::{
     thread,
 };
 
-use ahash::AHashMap;
+use ahash::{AHashMap, AHashSet};
 use regex::Regex;
 use walkdir::WalkDir;
 
@@ -92,7 +92,7 @@ async fn main() {
 
         let index = note_name_unique_to_index
             .entry(note_name_unique.clone())
-            .or_insert_with(|| graph.add_node(note_name_unique));
+            .or_insert_with(|| graph.add_node(note_name_unique.clone()));
 
         let index = *index;
 
@@ -103,15 +103,21 @@ async fn main() {
 
         let wikilinks = wikilink_regex.captures_iter(&contents);
 
-        for wikilink in wikilinks {
-            tracing::info!(?wikilink);
+        let mut outgoing = AHashSet::new();
 
+        for wikilink in wikilinks {
             let link = wikilink.name("link").unwrap();
 
-            tracing::info!(?wikilink);
+            let link_as_string = link.as_str().to_string();
+
+            outgoing.insert(link_as_string);
+        }
+
+        for link in outgoing {
+            tracing::info!(note_name_unique, link, "linking");
 
             let outgoing_index = note_name_unique_to_index
-                .entry(link.as_str().to_string())
+                .entry(link)
                 .or_insert_with_key(|k| graph.add_node(k.to_owned()));
 
             graph.add_edge(index, *outgoing_index, ());

src/bin/fundamentals.rs

@@ -0,0 +1,51 @@
+use std::iter::zip;
+
+use petgraph::visit::EdgeRef;
+use rustworkx_core::centrality::katz_centrality;
+
+use simple_pagerank::Pagerank;
+
+use itertools::Itertools;
+
+#[tokio::main]
+async fn main() {
+    tracing_subscriber::fmt::init();
+
+    let file = std::fs::read("graph.postcard").unwrap();
+
+    let graph: petgraph::graph::DiGraph<String, ()> = postcard::from_bytes(&file).unwrap();
+
+    let mut pagerank = Pagerank::new();
+
+    for edge in graph.edge_references() {
+        let source = edge.source();
+        let target = edge.target();
+
+        pagerank.add_edge(source, target);
+    }
+
+    pagerank.calculate();
+    // pagerank.calculate_with_convergence(0.00000003);
+
+    for (node_index, rank) in pagerank.nodes().into_iter().take(50) {
+        let node_weight = graph.node_weight(*node_index);
+
+        tracing::info!(?node_weight, rank);
+    }
+
+    if false {
+        let katz_centralities =
+            katz_centrality(&graph, |_| Ok::<_, ()>(1.0), None, None, None, None, None);
+        let katz_centralities = katz_centralities.unwrap().unwrap();
+
+        let mut nodes_and_indices = zip(graph.node_indices(), katz_centralities).collect_vec();
+        nodes_and_indices.sort_by(|(idx_a, centrality_a), (idx_b, centrality_b)| {
+            centrality_a.partial_cmp(centrality_b).unwrap()
+        });
+
+        for (node_index, centrality) in nodes_and_indices {
+            let node = graph.node_weight(node_index);
+            tracing::info!(?node_index, ?node, centrality);
+        }
+    }
+}