Convert INSPIRE texkey to doi

maybe_texkey_to_doi.py

import json
import urllib.request

def maybe_get_insp_data(url, max_retries=3, sleep_ms=500):
    """Try to query an INSPIRE URL, with retries, and sleeping

    Parameters
    ----------
    url: string

    max_retries: int, optional [default: 3]

    sleep_ms: numeric, optional [default: 500]
      Length of sleep (in milliseconds) between HTTP 429 codes

    Returns
    -------
    response: bytes or None
      The positive response from INSPIRE; None if an error occurs, or
      if we failed the max number of times.
    """

    print(f"requesting {url}")
    n_retries = 0
    while n_retries < max_retries:
        try:
            req = urllib.request.urlopen(url)
        except urllib.error.HTTPError as e:
            print(e)
            if e.code == 429:
                retry_time = req.getheaders()["retry-in"]
                print(f'got 429 with "retry-in"={retry_time}')
                print(f"going to sleep for {sleep_ms}ms")
                sleep(sleep_ms / 1000.0)
                n_retries = n_retries + 1
                continue
            else:
                return None
        except urllib.error.URLError as e:
            print(e)
            return None
        else:  # Success
            return req.read()
    # maxed out on retries
    print(f"too many ({n_retries}) retries")
    return None

def maybe_texkey_to_doi(texkey):
    res = maybe_get_insp_data(f"https://inspirehep.net/api/literature/?format=json&q=texkey:{texkey}")
    try:
        resj = json.loads(res)
        return resj['hits']['hits'][0]['metadata']['dois'][0]['value']
    except:
        print("Something went wrong")
        return res