From 767ee054acbfb3f663b30eb09b9df6d68506b137 Mon Sep 17 00:00:00 2001 From: Eric Bailey Date: Tue, 12 Sep 2023 15:25:27 -0500 Subject: [PATCH] clean up into pipeline steps --- .../bsky/graph/getSuggestedFollowsByActor.ts | 197 ++++++++++-------- 1 file changed, 107 insertions(+), 90 deletions(-) diff --git a/packages/bsky/src/api/app/bsky/graph/getSuggestedFollowsByActor.ts b/packages/bsky/src/api/app/bsky/graph/getSuggestedFollowsByActor.ts index cc86a832346..be42ce2b959 100644 --- a/packages/bsky/src/api/app/bsky/graph/getSuggestedFollowsByActor.ts +++ b/packages/bsky/src/api/app/bsky/graph/getSuggestedFollowsByActor.ts @@ -2,9 +2,10 @@ import { sql } from 'kysely' import { Server } from '../../../../lexicon' import AppContext from '../../../../context' import { InvalidRequestError } from '@atproto/xrpc-server' +import { Database } from '../../../../db' +import { ActorService } from '../../../../services/actor' -const MAX_RESULTS_LENGTH = 10 -const RESULT_OVERFETCH = 20 +const RESULT_LENGTH = 10 export default function (server: Server, ctx: AppContext) { server.app.bsky.graph.getSuggestedFollowsByActor({ @@ -21,101 +22,117 @@ export default function (server: Server, ctx: AppContext) { throw new InvalidRequestError('Actor not found') } - const actorsViewerFollows = db.db - .selectFrom('follow') - .where('creator', '=', viewer) - .select('subjectDid') - const mostLikedAccounts = await db.db - .selectFrom( - db.db - .selectFrom('like') - .where('creator', '=', actorDid) - .select(sql`split_part(subject, '/', 3)`.as('subjectDid')) - .limit(1000) // limit to 1000 - .as('likes'), - ) - .select('likes.subjectDid as did') - .select((qb) => qb.fn.count('likes.subjectDid').as('count')) - .where('likes.subjectDid', 'not in', actorsViewerFollows) - .where('likes.subjectDid', 'not in', [actorDid, viewer]) - .groupBy('likes.subjectDid') - .orderBy('count', 'desc') - .limit(RESULT_OVERFETCH) - .execute() - const resultDids = mostLikedAccounts.map((a) => ({ did: a.did })) as { - did: string - }[] - - if (resultDids.length < MAX_RESULTS_LENGTH) { - // backfill with popular accounts followed by actor - const mostPopularAccountsActorFollows = await db.db - .selectFrom('follow') - .innerJoin('profile_agg', 'follow.subjectDid', 'profile_agg.did') - .select('follow.subjectDid as did') - .where('follow.creator', '=', actorDid) - .where('follow.subjectDid', '!=', viewer) - .where('follow.subjectDid', 'not in', actorsViewerFollows) - .if(resultDids.length > 0, (qb) => - qb.where( - 'subjectDid', - 'not in', - resultDids.map((a) => a.did), - ), - ) - .orderBy('profile_agg.followersCount', 'desc') - .limit(RESULT_OVERFETCH) - .execute() - - resultDids.push(...mostPopularAccountsActorFollows) - } - - if (resultDids.length < MAX_RESULTS_LENGTH) { - // backfill with suggested_follow table - const additional = await db.db - .selectFrom('suggested_follow') - .where( - 'did', - 'not in', - // exclude any we already have - resultDids.map((a) => a.did).concat([actorDid, viewer]), - ) - // and aren't already followed by viewer - .where('did', 'not in', actorsViewerFollows) - .selectAll() - .execute() - - resultDids.push(...additional) - } - - const actors = await db.db - .selectFrom('actor') - .selectAll() - .where( - 'did', - 'in', - resultDids.map((a) => a.did), - ) - .limit(RESULT_OVERFETCH) - .execute() - - // resolve all profiles, this handles blocks/mutes etc - const suggestions = ( - await actorService.views.hydrateProfiles(actors, viewer) - ).filter((account) => { + const skeleton = await getSkeleton( + { + actor: actorDid, + viewer, + }, + { + db, + actorService, + }, + ) + const hydrationState = await actorService.views.profileDetailHydration( + skeleton.map((a) => a.did), + { viewer }, + ) + const presentationState = actorService.views.profileDetailPresentation( + skeleton.map((a) => a.did), + hydrationState, + { viewer }, + ) + const suggestions = Object.values(presentationState).filter((profile) => { return ( - !account.viewer?.muted && - !account.viewer?.mutedByList && - !account.viewer?.blocking && - !account.viewer?.blockedBy + !profile.viewer?.muted && + !profile.viewer?.mutedByList && + !profile.viewer?.blocking && + !profile.viewer?.blockedBy ) }) return { encoding: 'application/json', - body: { - suggestions: suggestions.slice(0, MAX_RESULTS_LENGTH), - }, + body: { suggestions }, } }, }) } + +async function getSkeleton( + params: { + actor: string + viewer: string + }, + ctx: { + db: Database + actorService: ActorService + }, +): Promise<{ did: string }[]> { + const actorsViewerFollows = ctx.db.db + .selectFrom('follow') + .where('creator', '=', params.viewer) + .select('subjectDid') + const mostLikedAccounts = await ctx.db.db + .selectFrom( + ctx.db.db + .selectFrom('like') + .where('creator', '=', params.actor) + .select(sql`split_part(subject, '/', 3)`.as('subjectDid')) + .limit(1000) // limit to 1000 + .as('likes'), + ) + .select('likes.subjectDid as did') + .select((qb) => qb.fn.count('likes.subjectDid').as('count')) + .where('likes.subjectDid', 'not in', actorsViewerFollows) + .where('likes.subjectDid', 'not in', [params.actor, params.viewer]) + .groupBy('likes.subjectDid') + .orderBy('count', 'desc') + .limit(RESULT_LENGTH) + .execute() + const resultDids = mostLikedAccounts.map((a) => ({ did: a.did })) as { + did: string + }[] + + if (resultDids.length < RESULT_LENGTH) { + // backfill with popular accounts followed by actor + const mostPopularAccountsActorFollows = await ctx.db.db + .selectFrom('follow') + .innerJoin('profile_agg', 'follow.subjectDid', 'profile_agg.did') + .select('follow.subjectDid as did') + .where('follow.creator', '=', params.actor) + .where('follow.subjectDid', '!=', params.viewer) + .where('follow.subjectDid', 'not in', actorsViewerFollows) + .if(resultDids.length > 0, (qb) => + qb.where( + 'subjectDid', + 'not in', + resultDids.map((a) => a.did), + ), + ) + .orderBy('profile_agg.followersCount', 'desc') + .limit(RESULT_LENGTH) + .execute() + + resultDids.push(...mostPopularAccountsActorFollows) + } + + if (resultDids.length < RESULT_LENGTH) { + // backfill with suggested_follow table + const additional = await ctx.db.db + .selectFrom('suggested_follow') + .where( + 'did', + 'not in', + // exclude any we already have + resultDids.map((a) => a.did).concat([params.actor, params.viewer]), + ) + // and aren't already followed by viewer + .where('did', 'not in', actorsViewerFollows) + .selectAll() + .execute() + + resultDids.push(...additional) + } + + return resultDids +}