for a given collection, open a random sample of calisphere-test pages in a web browser (for qc)

calisphere_browser_qc.py

"""
This script takes a collection's CDL registry number and the number of items in a collection, and
opens calisphere-test tabs in a web browser representing a specified percentage of the collection's
total pages.
"""
import argparse
import math
import random
import sys
import webbrowser as wb

def main(args=None):
    """Parse command line arguments and pass their values to calisphere_page_qc().
    """
    parser = argparse.ArgumentParser(
        description='''within a given collection, use a web browser to open a random selection of
                       x percent of the collection's pages in calisphere-test''')

    parser.add_argument("collection_number", type=int, nargs=1,
                        help='''Collection's CDL registry number''')

    parser.add_argument("item_count", type=int, nargs=1,
                        help='''total number of digital objects in the collection''')

    parser.add_argument("qc_percent", type=int, nargs='?', default=10,
                        help='''percentage of pages to QC, as a whole number.
                             To check 15%%, use 15.
                             If no value supplied, default is 10.''')

    # print help if no args given
    if len(sys.argv) == 1:
        parser.print_help(sys.stderr)
        sys.exit(1)

    # parse
    if args is None:
        args = parser.parse_args()

    calisphere_page_qc(args.collection_number[0], args.item_count[0], args.qc_percent)

def calisphere_page_qc(collection_number, item_count, qc_percent):
    # how many pages does this collection have in Calisphere?
    # divide by 24 (default item count per page) and round up
    page_count = int(math.ceil(item_count/24))
    print("Collection contains", page_count, "pages")

    # how many pages do we need to check to check 1/10 of the pages? (round up)
    percent = int(math.ceil(qc_percent/100 * page_count))
    print("Opening", percent, "pages for QC.")

    # a sorted list of random pages within page_count with len(percent)
    check_pages = sorted(random.sample(range(1, page_count), percent))

    for page in check_pages:
        # generate "start" value for url
        start = (page - 1) * 24
        # make calisphere-test url
        url = "".join(["http://calisphere-test.cdlib.org/collections/", str(collection_number),
                       "?start=", str(start)])

        wb.open_new_tab(url)

if __name__ == "__main__":
    main()