publications.py

# Based on: https://github.com/allenai/s2-folks/tree/main/examples/python
#
#

import os
import traceback

import requests

S2_API_KEY = os.getenv("S2_API_KEY")
result_limit = 10


def find_basis_paper(query, result_limit=10):
    papers = None
    info = """Search results: 

"""
    while not papers:
        if not query:
            continue

        try:
            if len(S2_API_KEY):
                print("Using S2_API_KEY %s..." % S2_API_KEY[:4])
                '''
                headers={"x-api-key": S2_API_KEY} if S2_API_KEY else {},'''
            rsp = requests.get(
                "https://api.semanticscholar.org/graph/v1/paper/search",
                params={
                    "query": query,
                    "limit": 100,
                    "fields": "title,url,authors,year,tldr,journal,citationCount",
                    "sort": "citationCount:desc",
                },
            )
            rsp.raise_for_status()
            results = rsp.json()
            total = results["total"]
            if not total:
                return "No matches found. Please try another query."

            print(f"Found {total} results. Showing up to {result_limit}.")
            papers = results["data"]
            info += format_paper_list(papers, int(result_limit))

        except Exception as e:
            print(traceback.format_exc())
            info += "There was a problem...\n\n%s" % e

    print(info)
    return info


def get_element(paper, part):
    simple_parts = ["year", "url", "title", "citationCount"]
    if part in simple_parts:
        if part in paper and paper[part] is not None:
            v = paper[part]
            return v.strip() if isinstance(v, str) else v
        else:
            return "???"

    elif part == "journal":
        if part in paper and paper[part] is not None:
            return paper[part]["name"].strip()
        else:
            return "???"
    elif part == "authors":
        if part in paper and paper[part] is not None:
            return paper[part][0]["name"].strip()
        else:
            return "???"
    elif part == "tldr":
        if part in paper and paper[part] is not None:
            return paper[part]["text"].strip()
        else:
            return "???"
    else:
        return "<unknown field %s>" % part


def format_paper_list(all_papers, result_limit):
    list = ""

    sorted_papers = sorted(all_papers, key=lambda d: d["citationCount"], reverse=True)[
        :result_limit
    ]

    print(sorted_papers)
    for idx, paper in enumerate(sorted_papers):
        list += f"{idx}: **{get_element(paper, 'authors')} et al. {get_element(paper, 'year')}**, [{get_element(paper, 'journal')}]({get_element(paper, 'url')}) {get_element(paper, 'title')}. Cited by {get_element(paper, 'citationCount')}\n\n"
        if paper["tldr"] and paper["tldr"]["text"]:
            list += f"> _{paper['tldr']['text'].strip()}_ \n\n"

    return list


if __name__ == "__main__":
    find_basis_paper("NeuroML")