Add followers count cache for future queries

migrations/002_followers_count_index.cypher

@@ -0,0 +1 @@
+CREATE INDEX FOR (u:User) ON (u.follower_count);

queries/README.md

@@ -0,0 +1,3 @@
+# Queries Directory
+
+This directory contains Neo4j Cypher queries intended for manual database changes, data inspection, and analysis. These queries can be used to make one-off updates, verify data integrity, and generate insights for reporting. As the project evolves, this collection will grow and be used for ongoing database analysis and reporting.

queries/find_spammers.cypher

@@ -0,0 +1,16 @@
+// Looks for users with more than 100 followers that have too many followers with very few followees
+
+MATCH (suspect:User)-[f:FOLLOWS]->(target:User)
+WITH target, COUNT(DISTINCT suspect) AS num_followers
+WHERE num_followers > 100
+WITH target, num_followers
+ORDER BY num_followers DESC
+LIMIT 10
+MATCH (suspect:User)-[f:FOLLOWS]->(target:User)
+MATCH (suspect)-[:FOLLOWS]->(others:User)
+WITH target, num_followers, suspect, COUNT(DISTINCT others) AS num_followees
+WHERE num_followees < 3
+WITH target, num_followers, COUNT(DISTINCT suspect) AS num_suspects
+ORDER BY num_suspects DESC
+RETURN target, num_followers, num_suspects
+LIMIT 10;

queries/set_null_followers_counts_values.cypher

@@ -0,0 +1,12 @@
+// This query selects up to 1000 users with a null follower count, counts their
+// followers, updates the follower count, and returns the user’s public key
+// along with the updated count.
+
+ MATCH (user:User)
+ WHERE user.follower_count IS NULL
+ WITH user
+ LIMIT 1000
+ OPTIONAL MATCH (follower:User)-[:FOLLOWS]->(user)
+ WITH user, COUNT(follower) AS follower_count
+ SET user.follower_count = follower_count
+ RETURN user.pubkey, user.follower_count;

src/repo.rs

@@ -166,12 +166,25 @@ impl RepoTrait for Repo {
     }
 
     async fn upsert_follow(&self, follow: &ContactListFollow) -> Result<(), RepoError> {
+        // Notice we only increment follower_count if it is not null.  This
+        // allows an external script to initialize the count with fresh,
+        // non-cached values.
         let statement = r#"
             MERGE (followee:User {pubkey: $followee_val})
+            ON CREATE SET followee.follower_count = 0
+
             MERGE (follower:User {pubkey: $follower_val})
+
             MERGE (follower)-[r:FOLLOWS]->(followee)
-            ON CREATE SET r.created_at = $updated_at, r.updated_at = $updated_at
-            ON MATCH SET r.updated_at = $updated_at
+            ON CREATE SET
+                r.created_at = $updated_at,
+                r.updated_at = $updated_at,
+                followee.follower_count = CASE
+                    WHEN followee.follower_count IS NOT NULL THEN followee.follower_count + 1
+                    ELSE followee.follower_count
+                END
+            ON MATCH SET
+                r.updated_at = $updated_at
             "#;
 
         let query = query(statement)
@@ -195,6 +208,11 @@ impl RepoTrait for Repo {
         let statement = r#"
             MATCH (follower:User {pubkey: $follower_val})-[r:FOLLOWS]->(followee:User {pubkey: $followee_val})
             DELETE r
+            WITH followee
+            SET followee.follower_count = CASE
+                WHEN followee.follower_count > 0 THEN followee.follower_count - 1
+                ELSE 0
+            END
             "#;
 
         let query = query(statement)