Parent domain incl subdomains

This commit is contained in:
relikd
2020-09-28 16:22:25 +02:00
parent 8f88f70977
commit fc73635cb6
3 changed files with 75 additions and 25 deletions

View File

@@ -1,22 +1,30 @@
function lookup_domain_js(fname_a, fname_b, id1, id2, id3) { function lookup_domain_js(fname_doms, fname_apps, fname_subs) {
let dom = window.location.hash.substr(1); let dom = window.location.hash.substr(1); // domain name
document.getElementById(id1).innerHTML = dom; // domain name document.getElementById('name').innerHTML = dom;
let dom_app_list = document.getElementById(id3); // apps list let dom_num_apps = document.getElementById('num-apps');
let dom_app_list = document.getElementById('app-toc');
let dom_sub_doms = document.getElementById('subdoms');
let dom_known_trkr = document.getElementById('known');
let template = dom_app_list.firstElementChild; let template = dom_app_list.firstElementChild;
dom_app_list.innerHTML = 'loading…'; dom_app_list.innerHTML = 'loading…';
// load reverse domains json // load reverse domains json
loadJSON(fname_a, function(response) { loadJSON(fname_doms, function(response) {
let elem = JSON.parse(response)[dom]; let elem = JSON.parse(response)[dom];
if (!elem || elem.length == 0) { let count = elem.length - 1;
document.getElementById(id2).innerHTML = '0 applications'; if (!elem || count < 1) {
dom_num_apps.innerHTML = '0 applications';
dom_app_list.innerHTML = ' None '; dom_app_list.innerHTML = ' None ';
return; return;
} else if (count == 1) {
dom_num_apps.innerHTML = '1 application';
} else {
dom_num_apps.innerHTML = elem.length - 1 + ' applications';
} }
document.getElementById(id2).innerHTML = elem.length + ' applications'; dom_known_trkr.innerHTML = elem[0] ? 'Yes' : 'No';
// load app name json // load app name json
loadJSON(fname_b, function(response) { loadJSON(fname_apps, function(response) {
let name_list = JSON.parse(response); let name_list = JSON.parse(response);
var apps = []; var apps = [];
for (var i = elem.length - 1; i >= 0; i--) { for (var i = elem.length - 1; i >= 0; i--) {
@@ -39,6 +47,23 @@ function lookup_domain_js(fname_a, fname_b, id1, id2, id3) {
dom_app_list.appendChild(item); dom_app_list.appendChild(item);
} }
const observer = lozad(); observer.observe(); const observer = lozad(); observer.observe();
if (!dom_sub_doms) { return }
loadJSON(fname_subs, function(response) {
let subdomains_list = JSON.parse(response)[dom];
if (subdomains_list) {
var src = '';
for (var i = 0; i < subdomains_list.length; i++) {
let sub = subdomains_list[i];
let full = sub ? sub + '.' + dom : dom;
let lnk = '<a href="/subdomain/#' + full + '">' + sub + '.</a> ';
src += lnk;
}
dom_sub_doms.innerHTML = src;
} else {
dom_sub_doms.innerHTML = ' None ';
}
});
}); });
}); });
} }

View File

@@ -89,24 +89,23 @@ def gen_html_trinity(idx_dir, app_count, json, title, symlink):
mylib.symlink(symlink, mylib.path_out(idx_dir, 'data.json')) mylib.symlink(symlink, mylib.path_out(idx_dir, 'data.json'))
def gen_lookup(html_dir, doms_dict, names_dict, title): def gen_lookup(html_dir, doms_dict, flag, title):
header = HTML.a_path([('All Domains', '/index/domains/all/')], HTML.write(html_dir, f'''
'<span id="name"></span>') <h2>{ HTML.a_path([('All Domains', '/index/domains/all/')],
HTML.write(html_dir, ''' '<span id="name"></span>') }</h2>
<h2>{}</h2> <p>Known Tracker: <b id="known">?</b></p>
<p>Present in: <b id="num-apps">… applications</b></p> <p>Present in: <b id="num-apps">… applications</b></p>
{ '<h3>Subdomains:</h3><div id="subdoms" class="tags"></div>' if flag else '' }
<h3>Apps containing this domain:</h3> <h3>Apps containing this domain:</h3>
<div id="app-toc" class="no-ul-all"> <div id="app-toc" class="no-ul-all">
{} { HTML.app_tile_template() }
</div> </div>
<script type="text/javascript" src="/static/lookup-domain.js"></script> <script type="text/javascript" src="/static/lookup-domain.js?2"></script>
<script type="text/javascript" src="/static/lozad.js"></script> <script type="text/javascript" src="/static/lozad.js"></script>
<script type="text/javascript"> <script type="text/javascript">
lookup_domain_js('doms.json', 'apps.json', 'name', 'num-apps', 'app-toc'); lookup_domain_js('doms.json', '/results/lookup-apps.json', '/results/subdoms.json');
</script> </script>
'''.format(header, HTML.app_tile_template()), title=title) ''', title=title)
# after html write which will create the dir
mylib.json_write(mylib.path_add(html_dir, 'apps.json'), names_dict)
mylib.json_write(mylib.path_add(html_dir, 'doms.json'), doms_dict) mylib.json_write(mylib.path_add(html_dir, 'doms.json'), doms_dict)
@@ -117,14 +116,21 @@ def process():
app_count = index_domains.number_of_apps(json) app_count = index_domains.number_of_apps(json)
dom_count = len(json['subdom']) dom_count = len(json['subdom'])
print(' Lookup') # Prepare for lookup
names = [[x, index_app_names.get_name(x)] for x in json['bundle']] names = [[x, index_app_names.get_name(x)] for x in json['bundle']]
gen_lookup(mylib.path_out('domain'), json['pardom'], names, dest_dir = mylib.path_out('results')
title='Domain Lookup') mylib.mkdir(dest_dir)
gen_lookup(mylib.path_out('subdomain'), json['subdom'], names, mylib.json_write(mylib.path_add(dest_dir, 'lookup-apps.json'), names)
title='Subdomain Lookup') mylib.symlink(index_domains.fname_dom_subdoms(),
mylib.path_add(dest_dir, 'subdoms.json'))
names = None names = None
print(' Lookup')
gen_lookup(mylib.path_out('domain'), json['pardom'], True,
title='Domain Lookup')
gen_lookup(mylib.path_out('subdomain'), json['subdom'], False,
title='Subdomain Lookup')
print(' All Domains') print(' All Domains')
gen_html_trinity(mylib.path_out('index', 'domains', 'all'), app_count, gen_html_trinity(mylib.path_out('index', 'domains', 'all'), app_count,
json=json, title='Requested Domains', json=json, title='Requested Domains',

View File

@@ -18,6 +18,10 @@ def fname_no_tracker():
return mylib.path_data_index('domains_no_tracker.json') return mylib.path_data_index('domains_no_tracker.json')
def fname_dom_subdoms():
return mylib.path_data_index('domains_subdomains.json')
def load_json_from_disk(index_file): def load_json_from_disk(index_file):
return mylib.json_safe_read( return mylib.json_safe_read(
index_file, fallback={'bundle': [], 'pardom': {}, 'subdom': {}}) index_file, fallback={'bundle': [], 'pardom': {}, 'subdom': {}})
@@ -110,6 +114,18 @@ def filter_list_at_least(index, min_count):
index['pardom'] = par index['pardom'] = par
def dict_dom_subdomains(index):
ret = {}
for subdomain in index['subdom'].keys():
pardom = mylib.parent_domain(subdomain)
host = subdomain[:-len(pardom) - 1] # - '.'
try:
ret[pardom].append(host)
except KeyError:
ret[pardom] = [host]
return ret
def number_of_apps(index): def number_of_apps(index):
return sum(1 for x in index['bundle'] if x != '_') return sum(1 for x in index['bundle'] if x != '_')
@@ -149,6 +165,9 @@ def process(bundle_ids, deleteOnly=False):
mylib.json_write(fname_tracker(), dict_trkr, pretty=False) mylib.json_write(fname_tracker(), dict_trkr, pretty=False)
filter_list_at_least(dict_no_trkr, 5) # or 0.1 * len(ids) filter_list_at_least(dict_no_trkr, 5) # or 0.1 * len(ids)
mylib.json_write(fname_no_tracker(), dict_no_trkr, pretty=False) mylib.json_write(fname_no_tracker(), dict_no_trkr, pretty=False)
mylib.json_write(fname_dom_subdoms(), dict_dom_subdomains(index),
pretty=False)
else: else:
print(' no change') print(' no change')
print('') print('')