Update pipelines/merge.py

pipelines/merge.py

@@ -7,7 +7,9 @@ def _period_key(item_text: str) -> str:
 
 def merge_normalized_records(records: List[Dict]) -> Dict:
     merged = {"work_experience": [], "education": [], "certifications": [], "skills": [], "raw_sections": {}}
-    seen_we, seen_edu, seen_cert, skill_set = set(), set(), set(), set()
+
+    seen_we, seen_edu, seen_cert = set(), set(), set()
+    skill_set = set()
 
     for r in records:
         for w in r.get("work_experience", []):
@@ -15,19 +17,23 @@ def merge_normalized_records(records: List[Dict]) -> Dict:
             if key not in seen_we:
                 seen_we.add(key)
                 merged["work_experience"].append(w)
+
         for e in r.get("education", []):
             k = e.get("text", "")
             if k and k not in seen_edu:
                 seen_edu.add(k)
                 merged["education"].append(e)
+
         for c in r.get("certifications", []):
             k = c.get("text", "")
             if k and k not in seen_cert:
                 seen_cert.add(k)
                 merged["certifications"].append(c)
+
         for s in r.get("skills", []):
             if s:
                 skill_set.add(s)
+
         for k, v in r.get("raw_sections", {}).items():
             merged["raw_sections"][k] = (merged["raw_sections"].get(k, "") + "\n" + v).strip()