From 74b7fdf7542b773fa4b5dea6ee193165d9627831 Mon Sep 17 00:00:00 2001
From: Daniel Holmgren <dtholmgren@gmail.com>
Date: Tue, 14 Nov 2023 18:14:08 -0600
Subject: [PATCH] Randomize suggestions (#1844)

* randomize suggestions

* fix snap

* cursor fix

* pr feedback
---
 .../src/api/app/bsky/actor/getSuggestions.ts  | 66 +++++++++++++------
 packages/bsky/tests/views/suggestions.test.ts | 22 +++++--
 .../proxied/__snapshots__/views.test.ts.snap  |  2 +-
 3 files changed, 61 insertions(+), 29 deletions(-)

diff --git a/packages/bsky/src/api/app/bsky/actor/getSuggestions.ts b/packages/bsky/src/api/app/bsky/actor/getSuggestions.ts
index 18ab99debe2..f68ba68eb66 100644
--- a/packages/bsky/src/api/app/bsky/actor/getSuggestions.ts
+++ b/packages/bsky/src/api/app/bsky/actor/getSuggestions.ts
@@ -42,12 +42,12 @@ const skeleton = async (
   ctx: Context,
 ): Promise<SkeletonState> => {
   const { db } = ctx
-  const { limit, cursor, viewer } = params
+  const { viewer } = params
+  const alreadyIncluded = parseCursor(params.cursor)
   const { ref } = db.db.dynamic
-  let suggestionsQb = db.db
+  const suggestions = await db.db
     .selectFrom('suggested_follow')
     .innerJoin('actor', 'actor.did', 'suggested_follow.did')
-    .innerJoin('profile_agg', 'profile_agg.did', 'actor.did')
     .where(notSoftDeletedClause(ref('actor')))
     .where('suggested_follow.did', '!=', viewer ?? '')
     .whereNotExists((qb) =>
@@ -57,27 +57,30 @@ const skeleton = async (
         .where('creator', '=', viewer ?? '')
         .whereRef('subjectDid', '=', ref('actor.did')),
     )
+    .if(alreadyIncluded.length > 0, (qb) =>
+      qb.where('suggested_follow.order', 'not in', alreadyIncluded),
+    )
     .selectAll()
-    .select('profile_agg.postsCount as postsCount')
-    .limit(limit)
     .orderBy('suggested_follow.order', 'asc')
+    .execute()
 
-  if (cursor) {
-    const cursorRow = await db.db
-      .selectFrom('suggested_follow')
-      .where('did', '=', cursor)
-      .selectAll()
-      .executeTakeFirst()
-    if (cursorRow) {
-      suggestionsQb = suggestionsQb.where(
-        'suggested_follow.order',
-        '>',
-        cursorRow.order,
-      )
-    }
-  }
-  const suggestions = await suggestionsQb.execute()
-  return { params, suggestions, cursor: suggestions.at(-1)?.did }
+  // always include first two
+  const firstTwo = suggestions.filter(
+    (row) => row.order === 1 || row.order === 2,
+  )
+  const rest = suggestions.filter((row) => row.order !== 1 && row.order !== 2)
+  const limited = firstTwo.concat(shuffle(rest)).slice(0, params.limit)
+
+  // if the result set ends up getting larger, consider using a seed included in the cursor for for the randomized shuffle
+  const cursor =
+    limited.length > 0
+      ? limited
+          .map((row) => row.order.toString())
+          .concat(alreadyIncluded.map((id) => id.toString()))
+          .join(':')
+      : undefined
+
+  return { params, suggestions: limited, cursor }
 }
 
 const hydration = async (state: SkeletonState, ctx: Context) => {
@@ -110,6 +113,27 @@ const presentation = (state: HydrationState) => {
   return { actors: suggestedActors, cursor }
 }
 
+const parseCursor = (cursor?: string): number[] => {
+  if (!cursor) {
+    return []
+  }
+  try {
+    return cursor
+      .split(':')
+      .map((id) => parseInt(id, 10))
+      .filter((id) => !isNaN(id))
+  } catch {
+    return []
+  }
+}
+
+const shuffle = <T>(arr: T[]): T[] => {
+  return arr
+    .map((value) => ({ value, sort: Math.random() }))
+    .sort((a, b) => a.sort - b.sort)
+    .map(({ value }) => value)
+}
+
 type Context = {
   db: Database
   actorService: ActorService
diff --git a/packages/bsky/tests/views/suggestions.test.ts b/packages/bsky/tests/views/suggestions.test.ts
index 2dcadf9e6ad..4253f528b13 100644
--- a/packages/bsky/tests/views/suggestions.test.ts
+++ b/packages/bsky/tests/views/suggestions.test.ts
@@ -19,10 +19,12 @@ describe('pds user search views', () => {
     await network.bsky.processAll()
 
     const suggestions = [
-      { did: sc.dids.bob, order: 1 },
-      { did: sc.dids.carol, order: 2 },
-      { did: sc.dids.dan, order: 3 },
+      { did: sc.dids.alice, order: 1 },
+      { did: sc.dids.bob, order: 2 },
+      { did: sc.dids.carol, order: 3 },
+      { did: sc.dids.dan, order: 4 },
     ]
+
     await network.bsky.ctx.db
       .getPrimary()
       .db.insertInto('suggested_follow')
@@ -63,16 +65,22 @@ describe('pds user search views', () => {
       { limit: 1 },
       { headers: await network.serviceHeaders(sc.dids.carol) },
     )
+    expect(result1.data.actors.length).toBe(1)
+    expect(result1.data.actors[0].handle).toEqual('bob.test')
+
     const result2 = await agent.api.app.bsky.actor.getSuggestions(
       { limit: 1, cursor: result1.data.cursor },
       { headers: await network.serviceHeaders(sc.dids.carol) },
     )
-
-    expect(result1.data.actors.length).toBe(1)
-    expect(result1.data.actors[0].handle).toEqual('bob.test')
-
     expect(result2.data.actors.length).toBe(1)
     expect(result2.data.actors[0].handle).toEqual('dan.test')
+
+    const result3 = await agent.api.app.bsky.actor.getSuggestions(
+      { limit: 1, cursor: result2.data.cursor },
+      { headers: await network.serviceHeaders(sc.dids.carol) },
+    )
+    expect(result3.data.actors.length).toBe(0)
+    expect(result3.data.cursor).toBeUndefined()
   })
 
   it('fetches suggestions unauthed', async () => {
diff --git a/packages/pds/tests/proxied/__snapshots__/views.test.ts.snap b/packages/pds/tests/proxied/__snapshots__/views.test.ts.snap
index fcf1063954c..0dbe9b5498d 100644
--- a/packages/pds/tests/proxied/__snapshots__/views.test.ts.snap
+++ b/packages/pds/tests/proxied/__snapshots__/views.test.ts.snap
@@ -114,7 +114,7 @@ Object {
       },
     },
   ],
-  "cursor": "user(2)",
+  "cursor": "1:3",
 }
 `;