Merge pull request #165 from GulSam00/fix/kyVerify

Fix/ky verify

Author: GulSam00

.github/workflows/update_ky_youtube.yml

@@ -1,6 +1,5 @@
 name: Update ky by Youtube
 
-# 실행 일시 중지
 on:
   schedule:
     - cron: "0 14 * * *" # 한국 시간 23:00 실행 (UTC+9 → UTC 14:00)

.github/workflows/verify_ky_youtube.yml

@@ -1,6 +1,8 @@
 name: Verify ky by Youtube
 
 on:
+  schedule:
+    - cron: "0 14 * * *" # 한국 시간 23:00 실행 (UTC+9 → UTC 14:00)
   workflow_dispatch:
 
 permissions:

apps/web/src/constants/krToJpnArtist.ts

@@ -2,7 +2,6 @@ export const krToJpnArtistSort = {
   // A-Z
   Ado: 'Ado',
   Aimer: 'Aimer',
-  'ASIAN KUNG-FU GENERATION': '아시안 쿵푸 제너레이션',
 
   DECO27: 'DECO*27',
   Eve: 'Eve',
@@ -39,6 +38,7 @@ export const krToJpnArtistSort = {
   // 아
   아라시: '嵐',
   아마자라시: 'amazarashi',
+  '아시안 쿵푸 제너레이션': 'ASIAN KUNG-FU GENERATION',
   아이묭: 'あいみょん',
   엘레가든: 'ELLEGARDEN',
   '오이시 마사요시': 'オーイシマサヨシ',

packages/crawling/src/crawling/crawlYoutubeVerify.ts

@@ -45,6 +45,8 @@ for (const song of data) {
 
   index++;
   console.log('crawlYoutubeVerify : ', index);
+
+  if (index >= 2000) break;
 }
 
 browser.close();

packages/crawling/src/utils/validateSongMatch.ts

@@ -25,22 +25,35 @@ export const validateSongMatch = async (
     messages: [
       {
         role: 'system',
-        content:
-          'Decide if two (title, artist) pairs refer to the same song. Allow spelling variants (spaces, en/kr, case). Return JSON: {"isValid":boolean}',
+        content: `
+        You are a music database expert. 
+        Decide if two (title, artist) pairs refer to the same song recording.
+        
+        Rules:
+        1. Ignore additional info in parentheses like "(Original Artist Name)", "(Movie OST)", or "Remake".
+        2. Allow spelling variants, spaces, case, and Language mix (KR/EN/JP).
+        3. If the song title and the PERFORMING artist are the same, it is a MATCH, even if the original composer/artist is mentioned in the found title.
+        
+        Return JSON: {"isValid": boolean}
+      `,
       },
       {
         role: 'user',
-        content: `"${inputTitle}"(${inputArtist}) vs "${foundTitle}"(${foundArtist})`,
+        content: `Pair A: "${inputTitle}" by "${inputArtist}"\nPair B: "${foundTitle}" by "${foundArtist}"`,
       },
     ],
     response_format: { type: 'json_object' },
     temperature: 0,
-    max_tokens: 20,
+    max_tokens: 50,
   });
 
   const content = response.choices[0].message.content;
   if (!content) return false;
 
-  const result: { isValid: boolean } = JSON.parse(content);
-  return result.isValid;
+  try {
+    const result = JSON.parse(content);
+    return result.isValid === true;
+  } catch {
+    return false;
+  }
 };