update slugging and check for existing companies in script

Author: AmarTrebinjac

.infra/Pulumi.prod.yaml

@@ -196,3 +196,5 @@ config:
       secure: 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
   gcp:project: devkit-prod
   gcp:region: us-central1
+  api:anthropicApiKey:
+    secure: AAABAHEzcSWbWl8xVDhOLgPCopvQnihNX6MIzyG/JbaYeGA3p1qrJ3UEQhOvBg7kMoIbx9u+0CRC102IldakBAlxx0l9l9kCDSE/JqqfCfjiLT5mjdLISRE5q2dQz+MaqqVzqm0MzaIQQkWBmOxeJAxRxQ/+/dWdla8RMWKG+Q7PnrH8vS8PeNKASRQ=

__tests__/schema/autocompletes.ts

@@ -1251,4 +1251,45 @@ describe('query autocompleteCompany', () => {
       },
     ]);
   });
+
+  it('should return results when searching by non-Latin characters (Korean)', async () => {
+    const res = await client.query(QUERY, {
+      variables: { query: '삼성' },
+    });
+
+    expect(res.errors).toBeFalsy();
+    expect(res.data.autocompleteCompany).toMatchObject([
+      {
+        id: 'samsung',
+        name: 'Samsung Electronics',
+        image: 'https://example.com/samsung.png',
+      },
+    ]);
+  });
+
+  it('should return results when searching by non-Latin characters (Japanese)', async () => {
+    const res = await client.query(QUERY, {
+      variables: { query: 'トヨタ' },
+    });
+
+    expect(res.errors).toBeFalsy();
+    expect(res.data.autocompleteCompany).toMatchObject([
+      {
+        id: 'toyota',
+        name: 'Toyota Motor Corporation',
+        image: 'https://example.com/toyota.png',
+      },
+    ]);
+  });
+
+  it('should not return unrelated companies when searching with non-Latin characters', async () => {
+    const res = await client.query(QUERY, {
+      variables: { query: '삼성' },
+    });
+
+    expect(res.errors).toBeFalsy();
+    // Should only return Samsung, not other companies with non-Latin altNames
+    expect(res.data.autocompleteCompany.length).toBe(1);
+    expect(res.data.autocompleteCompany[0].id).toBe('samsung');
+  });
 });

bin/enrichExistingCompanies.ts

@@ -3,10 +3,12 @@ import '../src/config';
 import createOrGetConnection from '../src/db';
 import { UserExperience } from '../src/entity/user/experiences/UserExperience';
 import { UserExperienceType } from '../src/entity/user/experiences/types';
+import { Company } from '../src/entity/Company';
 import {
   enrichCompanyForExperience,
   EnrichmentLogger,
 } from '../src/common/companyEnrichment';
+import { DataSource } from 'typeorm';
 
 const QUEUE_CONCURRENCY = 1;
 
@@ -16,6 +18,20 @@ interface ExperienceData {
   type: string;
 }
 
+async function findExactCompanyMatch(
+  con: DataSource,
+  customCompanyName: string,
+): Promise<Company | null> {
+  return con
+    .getRepository(Company)
+    .createQueryBuilder('company')
+    .where('LOWER(company.name) = LOWER(:name)', { name: customCompanyName })
+    .orWhere('LOWER(company.altName) = LOWER(:name)', {
+      name: customCompanyName,
+    })
+    .getOne();
+}
+
 const silentLogger: EnrichmentLogger = {
   info: () => {},
   warn: () => {},
@@ -47,6 +63,7 @@ const silentLogger: EnrichmentLogger = {
   );
 
   let enriched = 0;
+  let exactMatches = 0;
   let processedCount = 0;
 
   const builder = con
@@ -67,6 +84,23 @@ const silentLogger: EnrichmentLogger = {
   const stream = await builder.stream();
 
   const enrichQueue = fastq.promise(async (experience: ExperienceData) => {
+    // First check for exact name match to avoid API calls
+    const exactMatch = await findExactCompanyMatch(
+      con,
+      experience.customCompanyName,
+    );
+
+    if (exactMatch) {
+      await con
+        .getRepository(UserExperience)
+        .update({ id: experience.id }, { companyId: exactMatch.id });
+
+      processedCount++;
+      exactMatches++;
+      return;
+    }
+
+    // No exact match, use enrichment API
     const result = await enrichCompanyForExperience(
       con,
       {
@@ -95,7 +129,7 @@ const silentLogger: EnrichmentLogger = {
   await enrichQueue.drained();
 
   console.log(
-    `Completed. Processed ${processedCount} experiences, enriched ${enriched} (offset ${offset} to ${offset + processedCount - 1}).`,
+    `Completed. Processed ${processedCount} experiences: ${exactMatches} exact matches, ${enriched} enriched (offset ${offset} to ${offset + processedCount - 1}).`,
   );
   console.log(
     `Next batch command: npx ts-node bin/enrichExistingCompanies.ts ${limit} ${offset + limit}`,

src/schema/autocompletes.ts

@@ -1,7 +1,7 @@
 import { Keyword, KeywordStatus } from '../entity';
 import { AutocompleteType, Autocomplete } from '../entity/Autocomplete';
 import { traceResolvers } from './trace';
-import { ILike, Raw } from 'typeorm';
+import { FindOptionsWhere, ILike, Raw } from 'typeorm';
 import { AuthContext, BaseContext } from '../Context';
 import { textToSlug, toGQLEnum, type GQLCompany } from '../common';
 import { queryReadReplica } from '../common/queryReadReplica';
@@ -161,20 +161,30 @@ export const resolvers = traceResolvers<unknown, BaseContext>({
       ctx: AuthContext,
     ): Promise<GQLCompany[]> => {
       const { type, query, limit } = autocompleteCompanySchema.parse(payload);
-      const slugQuery = Raw((alias) => `slugify(${alias}) = :slug`, {
-        slug: textToSlug(query),
-      });
+      const slug = textToSlug(query);
+
+      const whereConditions: FindOptionsWhere<Company>[] = [
+        { type, name: ILike(`%${query}%`) },
+        { type, altName: ILike(`%${query}%`) },
+      ];
+
+      // Only add slug-based search if the slug is non-empty
+      // (non-Latin characters like Korean get stripped by slugify, causing false matches)
+      if (slug) {
+        const slugQuery = Raw((alias) => `slugify(${alias}) = :slug`, {
+          slug,
+        });
+        whereConditions.unshift(
+          { type, name: slugQuery },
+          { type, altName: slugQuery },
+        );
+      }
 
       return await queryReadReplica(ctx.con, ({ queryRunner }) =>
         queryRunner.manager.getRepository(Company).find({
           take: limit,
           order: { name: 'ASC' },
-          where: [
-            { type, name: slugQuery },
-            { type, name: ILike(`%${query}%`) },
-            { type, altName: slugQuery },
-            { type, altName: ILike(`%${query}%`) },
-          ],
+          where: whereConditions,
         }),
       );
     },