#
# title: uniprot_fetch.py
# author: Nick Fitzkee (nfitzkee@chemistry.msstate.edu)
# date:   June 20, 2022
# summary: Fetch a list of sequences given UniProtKB IDs
#
# To use $> python3 ./uniprot_fetch.py.txt uniprot_ids.txt | tee uniprot.fasta
# the above command will make the file "uniprot.fasta"


import urllib.request

def open_list(fname):
    result = []

    with open(fname) as uplist:
        l = uplist.readline()

        while l:
            l = l.strip()
            if not l or l[0] == '#':
                l = uplist.readline()
                continue

            result.append(l)
            l = uplist.readline()

    return result

def main(fname):
    uplist = open_list(fname)
        
    for up in uplist:
        url = 'https://www.uniprot.org/uniprot/%s.fasta' % up
        with urllib.request.urlopen(url) as page:
            print(page.read().decode('utf-8'))

    
if __name__ == '__main__':
    import os, sys

    try:
        fn = sys.argv[1]
    except:
        print('usage: %s <uniprot list file>' % os.path.split(sys.argv[0])[1])
        sys.exit(1)

    main(fn)