exclude_badExits/lookupdns.py
2022-11-14 12:00:23 +00:00

80 lines
2.2 KiB
Python

#!/usr/local/bin/python3.sh
# -*-mode: python; indent-tabs-mode: nil; py-indent-offset: 4; coding: utf-8 -*
import sys
import os
import traceback
from phantompy import Render
global LOG
import logging
import warnings
warnings.filterwarnings('ignore')
LOG = logging.getLogger()
class LookFor(Render):
def __init__(self, url, outfile, jsfile=None):
self.uri = url
Render.__init__(self, url, outfile, jsfile)
def ilookfor(self, html):
import json
marker = '<pre style="word-wrap: break-word; white-space: pre-wrap;">'
if marker not in html: return ''
i = html.find(marker) + len(marker)
html = html[i:]
assert html[0] == '{', html
i = html.find('</pre')
html = html[:i]
assert html[-1] == '}', html
LOG.debug(f"Found {len(html)} json")
o = json.loads(html)
if "Answer" not in o.keys() or type(o["Answer"]) != list:
LOG.warn(f"FAIL {self.uri}")
return 1
for elt in o["Answer"]:
assert type(elt) == dict, elt
assert 'type' in elt, elt
if elt['type'] != 16: continue
assert 'data' in elt, elt
if elt['data'] == 'we-run-this-tor-relay':
LOG.info(f"OK {self.uri}")
return 0
LOG.warn(f"BAD {self.uri}")
return 2
def _html_callback(self, *args):
"""print(self, QPrinter, Callable[[bool], None])"""
if type(args[0]) is str:
self._save(args[0])
i = self.ilookfor(args[0])
self._exit(i)
def _save(self, html):
sfile = self.outfile.replace('.pdf','.out')
# CompleteHtmlSaveFormat SingleHtmlSaveFormat MimeHtmlSaveFormat
with open(sfile, 'wt') as ofd:
ofd.write(html)
LOG.debug(f"Saved {sfile}")
def _loadFinished(self, result):
LOG.debug("phantom.py: Loading finished!")
self.toHtml(self._html_callback)
def main():
if (len(sys.argv) < 3):
LOG.info("USAGE: lookupdns.py <url> <pdf-file> [<javascript-file>]")
else:
url = sys.argv[1]
outfile = sys.argv[2]
jsfile = sys.argv[3] if len(sys.argv) > 3 else None
r = LookFor(url, outfile, jsfile)
sys.exit(0)
if __name__ == "__main__":
main()