script.py

import arxiv, os, json, datetime as dt
from db import insert_paper          # 自己封的写入 PG 函数
from quality import quality_score    # 质量打分模块

KEYWORDS = ['diffusion model', 'chain of thought', 'multimodal']
MAX_DAILY = 200                      # 每天最多抓 200 篇

def fetch():
    for kw in KEYWORDS:
        search = arxiv.Search(query=kw,
                              max_results=MAX_DAILY//len(KEYWORDS),
                              sort_by=arxiv.SortCriterion.SubmittedDate)
        for r in search.results():
            meta = {
                'arxivid': r.entry_id.split('/')[-1],
                'title': r.title,
                'abstract': r.summary,
                'authors': [a.name for a in r.authors],
                'pdf': r.pdf_url,
                'submitted': r.published,
                'kw': kw
            }
            meta['quality'] = quality_score(meta)
            insert_paper(meta)

if __name__ == '__main__':
    fetch()