from bs4 import BeautifulSoup
import pickle, sys
from collections import Counter
pages = pickle.load(open(sys.argv[1], 'rb'))

counts = Counter()

for url, src in pages.items():
    page = BeautifulSoup(src, "html.parser")
    for p in page.select('.display_name a'):
        if p.get('href'):
            counts[p['href']] += 1

ret = sorted([(c, url) for url,c in counts.items()])
for c, url in ret:
    print(url, c)