backend/ref.py

import faiss
import pandas as pd
from sentence_transformers import SentenceTransformer
import math
from keybert import KeyBERT
from kiwipiepy import Kiwi
import urllib.parse
import json

def safe_int(val):
    """val이 None 또는 NaN이면 '' 반환, 아니면 int로 변환"""
    if val is None:
        return ''
    if isinstance(val, float) and math.isnan(val):
        return ''
    return int(val)

def generate_dbpia_link(title):
    """논문 제목을 이용해 DBpia 검색 링크 생성"""
    base_url = "https://www.dbpia.co.kr/search/topSearch?searchOption=all&query="
    encoded_title = urllib.parse.quote(title)
    return base_url + encoded_title

def generate_reference(row):
    """

    row: 딕셔너리 형태의 논문 정보 (예: DataFrame의 한 행)

    반환값: 참고문헌 문자열

    """
    vol = safe_int(row.get('권'))
    issue = safe_int(row.get('호'))
    start_page = safe_int(row.get('시작페이지'))
    end_page = safe_int(row.get('끝페이지'))

    pages = f"{start_page}-{end_page}" if start_page != '' and end_page != '' else ''

    ref = f"{row.get('저자', '')}. ({safe_int(row.get('발행년'))}). {row.get('논문명(국문)', '')}. {row.get('학술지명(국문)', '')}"
    
    if vol != '' or issue != '':
        issue_str = f"({issue})" if issue != '' else ''
        ref += f", {vol}{issue_str}"
    
    if pages:
        ref += f", {pages}."
    else:
        ref += "."

    link = generate_dbpia_link(row.get('논문명(국문)', ''))
    return (ref,link)

def refRecommend(model,kw_model,kiwi,text,df,index):
    nouns_list = []
    for sentence in kiwi.analyze(text):
        nouns = [token.form for token in sentence[0] if token.tag.startswith('NN')]
        if nouns:
            nouns_list.extend(nouns)
    result_text = ' '.join(nouns_list)

    keywords = kw_model.extract_keywords(result_text, keyphrase_ngram_range=(1, 1), stop_words=None, top_n=5)

    query_vector = model.encode([keywords[0][0]+" "+keywords[1][0]])
    D, I = index.search(query_vector, k=3)  # k는 찾고 싶은 개수 (예: top-5)

    results = df.iloc[I[0]]  # I[0]은 top-k 결과의 인덱스 리스트

    name = []
    link = []
# 6. 예시 출력
    for i, row in results.iterrows():
        name_result,link_result = generate_reference(row)
        print(name_result)
        name.append(name_result)
        link.append(link_result)
    # print(f"{i+1}. 제목: {row['논문명(국문)']} / 키워드: {row['키워드(국문)']}")
    # print(f"{row['저자']}. ({row['발행년']}). {row['논문명(국문)']}. {row['학술지명(국문)']}, {int(row['권'])}({int(row['호'])}), {int(row['시작페이지'])}-{int(row['끝페이지'])}")
    return name,link