테스트

app.py

@@ -103,11 +103,7 @@ def query_select_summarize_stream(results, query, ai, min_similarity: float = 0.
             context_parts.append(f"[유사도: {sim:.3f}]\n{doc}")
         context = "\n\n".join(context_parts)
     else :
-        print('일반', results.get('documents'))
+        context = results
-        context_parts = [f'검색된 사용자 수는 {len(results.get('ids'))}']
-        docs = [f"[{d[d.find('[이름]'): d.find('[', d.find('[이름]')+1)]} {d[d.find('[부서]'): d.find('[', d.find('[부서]')+1)]}]" for d in results.get('documents')]
-        context = ",".join(context_parts + docs)
-    print(context)
 
     # 모델 로드
     model, tokenizer = get_qwen_model()
@@ -145,7 +141,7 @@ def query_select_summarize_stream(results, query, ai, min_similarity: float = 0.
     generation_kwargs = dict(
         **model_inputs,
         streamer=streamer,
-        max_new_tokens=600,
+        max_new_tokens=400,
         do_sample=True,
         temperature=0.3,
         top_p=0.9,
@@ -258,19 +254,40 @@ async def set_data(query: Item):
     클라이언트로부터 받은 인사 데이터를 자연어 문장으로 변환하여 벡터 DB에 저장합니다.
     기존 세션 데이터는 삭제 후 재생성됩니다.
     """
-    # 기존 데이터 삭제
+    # # 기존 데이터 삭제
-    collection.delete(
+    # collection.delete(
-        where={"sessionId": query.sessionId}
+    #     where={"sessionId": query.sessionId}
-    )
+    # )
-    
+    #
-    # 삭제 확인 (디버깅용)
+    # # 삭제 확인 (디버깅용)
-    remaining_count = collection.get(where={"sessionId": query.sessionId})
+    # remaining_count = collection.get(where={"sessionId": query.sessionId})
-    print(f"남은 데이터 수: {len(remaining_count['ids'])}")
+    # print(f"남은 데이터 수: {len(remaining_count['ids'])}")
-    
+
+    depts = []
+    users = []
+    for n in query.context :
+        if n['userYn'] == False :
+            depts.append(n)
+        else :
+            users.append(n)
+
+    print(len(depts), len(users))
 
     store_data[query.sessionId] = query.context
-    init(query.sessionId, query.context)
 
+    results = []
+    for d in depts :
+        subStr = f'[부서명] : {d['deptNm']} [상위부서명] : {d['pDeptNm']} [부서장] : {d['chiefDisplayNm']}\n'
+        inUsers = list(filter(lambda u : u['deptCd'] == d['deptCd'], users ))
+        reUsers = []
+        for iu in inUsers :
+            reUsers.append(f'** 이름 : {iu['name']} 사번 : {iu['sabun']} 직급 : {iu['gradeNm']} 직위 : {iu['ptsnNm']} **')
+        subStr += f'[직원수] : {len(inUsers)}\n'
+        subStr += f'[직원명단] : {', '.join(reUsers)}'
+        results.append(subStr)
+    # init(query.sessionId, query.context)
+
+    store_data[query.sessionId] = results
     return {"status": "success", "message": f"{len(query.context)}건의 데이터가 로드되었습니다."}
 
 
@@ -279,16 +296,20 @@ def question(sessionId: str, query: str):
     """
     질의응답 API 엔드포인트
     """
-    type = query_summarize_simple(query=query)
+    # type = query_summarize_simple(query=query)
-    print(type)
+    # print(type)
-    if(type == '99') :
+    # if(type == '99') :
-        results, keyword = query_select(sessionId, query, False)
+    #     results, keyword = query_select(sessionId, query, False)
-        print(f'단순질문 AI : {len(results)}')
+    #     print(f'단순질문 AI : {len(results)}')
-        generate = query_select_summarize_stream(results, query=keyword, ai=True)
+    #     generate = query_select_summarize_stream(results, query=keyword, ai=True)
-    # 부서 총직원수
+    # # 부서 총직원수
-    else :
+    # else :
-        results, keyword = query_select(sessionId, query, True)
+    #     results, keyword = query_select(sessionId, query, True)
-        generate = query_select_summarize_stream(results, query=keyword, ai=False)
+    sessionData = store_data.get(sessionId)
+
+    results = '\n-------------------------------------------\n'.join(sessionData)
+    print(sessionId, query)
+    generate = query_select_summarize_stream(results, query, ai=False)
 
 
     return StreamingResponse(generate(), media_type="application/x-ndjson")

org.html

@@ -42,7 +42,7 @@
 <div class="container">
     <div class="chat-wrap">
         <div class="chat"></div>
-        <input type="text" class="prompt" value="2직급 총 몇명이야?" placeholder="질문을 입력하고 Enter를 누르세요">
+        <input type="text" class="prompt" value="발전처 직원수?" placeholder="질문을 입력하고 Enter를 누르세요">
 <!--        <input type="text" class="prompt" value="재난관리부 부서장은 누구야?" placeholder="질문을 입력하고 Enter를 누르세요">-->
     </div>
 </div>
@@ -74,7 +74,7 @@
 
             const xhr = new XMLHttpRequest();
 
-            xhr.open('GET', `http://127.0.0.1:8000/?sessionId=20033011&query=${encodeURIComponent(query)}`);
+            xhr.open('GET', `http://127.0.0.1:8000/?sessionId=psn14020&query=${encodeURIComponent(query)}`);
 
             let seenBytes = 0;
             let buffer = '';