openalex_counts.py

#!/usr/bin/env python3

"""
Run this program with an institution name and see the institutions and the count
of publications in OpenAlex.

$ ./openalex_counts "stanford"

Stanford University (I97018004): 430550
Stanford Medicine (I4210137306): 32576
Stanford Synchrotron Radiation Lightsource (I4210120900): 12939
Stanford Health Care (I4210105015): 4550
Stanford Blood Center (I4210133340): 510
SRI International (I1298353152): 18418
SLAC National Accelerator Laboratory (I2801935854): 23121
Stanford Cancer Institute (I4390039303): 0
Stanford SystemX Alliance (I4392738099): 0
Stanford Maternal and Child Health Research Institute (I4391767688): 0
Institute for Stem Cell Biology and Regenerative Medicine (I4394709089): 0
"""

import re
import sys
import time

import pyalex

pyalex.config.email = "you@example.com"


institution_name = sys.argv[1]

for inst in pyalex.Institutions().search(institution_name).get():
    inst_id = re.sub(r'https://.+/', '', inst['id'])
    count = pyalex.Works().filter(institutions={"id": inst_id}).count()
    print(f"{inst['display_name']} ({inst_id}): {count}")

    time.sleep(1)