Fetch author citation benchmarks from Google Scholar

fetch_benchmarks.py

#!/usr/bin/python3 -OO
import sys
import operator
import scholarly
import urllib.parse

DEFAULT_ATTRIBUTES = (
	'citedby', 'citedby5y',
	'hindex', 'hindex5y',
	'i10index', 'i10index5y'
)


def main(attributes=None):
	if attributes is None:
		attributes = sys.argv[1:]
	if not attributes:
		attributes = DEFAULT_ATTRIBUTES

	print('Name', *attributes, sep='\t')
	attrgetter = operator.attrgetter(*attributes)

	for record in sys.stdin:
		record = record.rstrip('\n').split('\t')
		if len(record) >= 2 and record[1] and record[1] != '-':
			if '=' in record[1]:
				gs_url = urllib.parse.urlparse(record[1])
				assert gs_url.path == '/citations', 'Unexpected URL: ' + str(gs_url)
				gs_id = urllib.parse.parse_qs(
					gs_url.query, strict_parsing=True, errors='strict')['user'][0]
			else:
				gs_id = record[1]
			author = scholarly.Author(gs_id).fill()
			record[1:] = attrgetter(author)
		else:
			del record[1:]

		print(*record, sep='\t')


if __name__ == '__main__':
	main()