Fix multiple tracking domains binary tree lookup

This commit is contained in:
relikd
2020-10-02 11:40:39 +02:00
parent 9d241dbed1
commit 568a620811
2 changed files with 9 additions and 2 deletions

View File

@@ -135,7 +135,11 @@ def combine_all(changes=['_']):
continue # added by exodus, not a tracker per se
res.add(dom[::-1]) # reverse for bintree lookup
with open(final, 'w') as fp:
prev_dom = ''
for domain in sorted(res):
if domain.startswith(prev_dom + '.'):
continue # parent is tracker already
prev_dom = domain
fp.write(domain + '\n')

View File

@@ -65,8 +65,8 @@ def gen_html_top_10(path, subset, total, title):
def gen_html_trinity(idx_dir, app_count, json, title, symlink):
list1 = [(dom, len(ids) - 1) for dom, ids in json['subdom'].items()]
list2 = [(dom, len(ids) - 1) for dom, ids in json['pardom'].items()]
list1 = [(dom, len(ids)) for dom, ids in json['subdom'].items()]
list2 = [(dom, len(ids)) for dom, ids in json['pardom'].items()]
def write_index(fname, title, button):
HTML.write(idx_dir, '<h2>{}</h2>{}{}'.format(
@@ -132,6 +132,9 @@ def process():
title='Subdomain Lookup')
print(' All Domains')
for key in ['subdom', 'pardom']:
for x in json[key].keys():
json[key][x] = json[key][x][1:]
gen_html_trinity(mylib.path_out('index', 'domains', 'all'), app_count,
json=json, title='Requested Domains',
symlink=index_domains.fname_all())