batches author lookups in imports

server/search/queries.ts

@@ -18,3 +18,43 @@ export const getSearchUsers = async (searchString: string, limit = 5) => {
 		limit,
 	});
 };
+
+/**
+ * Batch-search for multiple author names in a single query, returning a map
+ * from each searched name to its matching users (up to `limit` per name).
+ */
+export const batchSearchUsers = async (names: string[], limit = 5) => {
+	const nonEmpty = names.filter((n) => n.length > 0);
+	if (nonEmpty.length === 0) {
+		return new Map<string, any[]>();
+	}
+
+	const users = await User.findAll({
+		where: {
+			[Op.or]: nonEmpty.flatMap((name) => [
+				{ fullName: { [Op.iLike]: `%${name}%` } },
+				{ slug: { [Op.iLike]: `%${name}%` } },
+				{ email: { [Op.iLike]: name } },
+			]),
+		},
+		attributes: ['id', 'slug', 'fullName', 'initials', 'avatar'],
+	});
+
+	const resultMap = new Map<string, any[]>(nonEmpty.map((name) => [name, []]));
+	const allUsers = users.map((u) => u.toJSON());
+
+	for (const name of nonEmpty) {
+		const lowerName = name.toLowerCase();
+		const matches = allUsers
+			.filter(
+				(u: any) =>
+					u.fullName?.toLowerCase().includes(lowerName) ||
+					u.slug?.toLowerCase().includes(lowerName) ||
+					u.email?.toLowerCase() === lowerName,
+			)
+			.slice(0, limit);
+		resultMap.set(name, matches);
+	}
+
+	return resultMap;
+};

workers/queue.ts

@@ -19,6 +19,7 @@ let currentWorkerThreads = 0;
 /** Nice to be able to run certain tasks longer than the default timeout */
 const customTimeouts = {
 	archive: 14_400, // 4 hours
+	import: 300, // 5 minutes
 } satisfies Partial<Record<TaskType, number>>;
 
 if (env.NODE_ENV === 'production') {

workers/tasks/import/metadata.ts

@@ -3,7 +3,7 @@ import type { Falsy } from 'types';
 import { metaValueToJsonSerializable, metaValueToString } from '@pubpub/prosemirror-pandoc';
 import unidecode from 'unidecode';
 
-import { getSearchUsers } from 'server/search/queries';
+import { batchSearchUsers } from 'server/search/queries';
 import { isValidDate } from 'utils/dates';
 
 const getAuthorsArray = (author) => {
@@ -28,15 +28,19 @@ const getAttributions = async (author) => {
 	if (author) {
 		const authorsArray = getAuthorsArray(author);
 		const authorEntries = authorsArray.map(metaValueToJsonSerializable) as any[];
-		const attributions = await Promise.all(
-			authorEntries.map(async (authorEntry: string) => {
-				if (typeof authorEntry === 'string') {
-					const users = await getSearchUsers(authorEntry);
-					return { name: authorEntry, users: users.map((user) => user.toJSON()) };
-				}
-				return authorEntry;
-			}),
+		const stringEntries = authorEntries.filter(
+			(entry): entry is string => typeof entry === 'string',
 		);
+		const usersByName =
+			stringEntries.length > 0
+				? await batchSearchUsers(stringEntries)
+				: new Map<string, any[]>();
+		const attributions = authorEntries.map((authorEntry) => {
+			if (typeof authorEntry === 'string') {
+				return { name: authorEntry, users: usersByName.get(authorEntry) ?? [] };
+			}
+			return authorEntry;
+		});
 		return attributions;
 	}
 	return null;