[26 - Stack 3/5] 경험 순차 중복 판정 추가

myeongsung/app/services/experience_merge_service.py

@@ -406,3 +406,32 @@ def apply_merge_results_to_step2(
         experience.merge_similarity = result.similarity
 
     return step2_experiences
+
+
+def apply_sequential_merge_results_to_step2(
+    step2_experiences: List[dict],
+    existing_experiences: List[MergeExperiencePayload],
+    threshold: Optional[float] = None,
+    embedding_client: Optional[Any] = None,
+) -> List[dict]:
+    accepted_candidates: List[Any] = list(existing_experiences)
+
+    for index, experience in enumerate(step2_experiences):
+        merge_response = check_merge_candidates(
+            targets=[experience],
+            existing_experiences=accepted_candidates,
+            threshold=threshold,
+            top_k=1,
+            embedding_client=embedding_client,
+        )
+        result = merge_response.results[0]
+        experience["needs_merge"] = result.needs_merge
+        experience["merge_candidate_id"] = result.merge_candidate_id
+        experience["merge_similarity"] = result.similarity
+
+        if not result.needs_merge:
+            accepted_candidate = dict(experience)
+            accepted_candidate["id"] = f"batch:{index}"
+            accepted_candidates.append(accepted_candidate)
+
+    return step2_experiences

myeongsung/tests/test_experience_merge_service.py

@@ -2,7 +2,11 @@
 import unittest
 
 from app.schemas.resume_dto import MergeExperiencePayload, Step2ExtractedExperience
-from app.services.experience_merge_service import build_embedding_text, check_merge_candidates
+from app.services.experience_merge_service import (
+    apply_sequential_merge_results_to_step2,
+    build_embedding_text,
+    check_merge_candidates,
+)
 
 
 class FakeEmbeddings:
@@ -34,6 +38,20 @@ class LowSimilarityOpenAI:
     embeddings = LowSimilarityEmbeddings()
 
 
+class SameEmbeddings:
+    def create(self, model, input):
+        return SimpleNamespace(
+            data=[
+                SimpleNamespace(embedding=[1.0, 0.0])
+                for _ in input
+            ]
+        )
+
+
+class SameOpenAI:
+    embeddings = SameEmbeddings()
+
+
 class ExperienceMergeServiceTest(unittest.TestCase):
 
     def test_check_merge_candidates_marks_similar_target(self):
@@ -185,6 +203,35 @@ def test_low_embedding_with_two_key_field_matches_marks_merge(self):
         self.assertTrue(response.results[0].needs_merge)
         self.assertEqual("exp-1", response.results[0].merge_candidate_id)
 
+    def test_sequential_merge_uses_first_non_duplicate_as_batch_candidate(self):
+        experiences = [
+            {
+                "experience_name": "캡스톤 프로젝트",
+                "experience_group": "상세 서술형",
+                "experience_type": "프로젝트",
+                "basic_info": {"project_name": "캡스톤 프로젝트"},
+                "experience_content": "추천 모델을 개발했습니다.",
+            },
+            {
+                "experience_name": "캡스톤 프로젝트",
+                "experience_group": "상세 서술형",
+                "experience_type": "프로젝트",
+                "basic_info": {"project_name": "캡스톤 프로젝트"},
+                "experience_content": "추천 모델을 개발했습니다.",
+            },
+        ]
+
+        result = apply_sequential_merge_results_to_step2(
+            experiences,
+            [],
+            threshold=0.86,
+            embedding_client=SameOpenAI(),
+        )
+
+        self.assertFalse(result[0]["needs_merge"])
+        self.assertTrue(result[1]["needs_merge"])
+        self.assertEqual("batch:0", result[1]["merge_candidate_id"])
+
 
 if __name__ == "__main__":
     unittest.main()