You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
95 lines
3.6 KiB
95 lines
3.6 KiB
from threading import Thread
|
|
import time
|
|
import requests
|
|
import hashlib
|
|
import os
|
|
|
|
# Hmm Polymorphism datastore, thread, etc
|
|
class perform_site_check(Thread):
|
|
def __init__(self, *args, uuid=False, datastore, **kwargs):
|
|
super().__init__(*args, **kwargs)
|
|
self.timestamp = int(time.time()) # used for storage etc too
|
|
self.uuid = uuid
|
|
self.datastore = datastore
|
|
self.url = datastore.get_val(uuid, 'url')
|
|
self.current_md5 = datastore.get_val(uuid, 'previous_md5')
|
|
self.output_path = "/datastore/{}".format(self.uuid)
|
|
|
|
def save_firefox_screenshot(self, uuid, output):
|
|
# @todo call selenium or whatever
|
|
return
|
|
|
|
def ensure_output_path(self):
|
|
|
|
try:
|
|
os.stat(self.output_path)
|
|
except:
|
|
os.mkdir(self.output_path)
|
|
|
|
def save_response_html_output(self, output):
|
|
# @todo maybe record a history.json, [timestamp, md5, filename]
|
|
|
|
|
|
with open("{}/{}.txt".format(self.output_path, self.timestamp), 'w') as f:
|
|
f.write(output)
|
|
f.close()
|
|
|
|
|
|
def save_response_stripped_output(self, output):
|
|
# @todo maybe record a history.json, [timestamp, md5, filename]
|
|
|
|
with open("{}/{}.stripped.txt".format(self.output_path, self.timestamp), 'w') as f:
|
|
f.write(output)
|
|
f.close()
|
|
|
|
def run(self):
|
|
|
|
headers = {
|
|
'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/87.0.4280.66 Safari/537.36',
|
|
'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.9',
|
|
'Accept-Encoding': 'gzip, deflate, br',
|
|
'Accept-Language': 'en-GB,en-US;q=0.9,en;q=0.8,cs;q=0.7'
|
|
}
|
|
|
|
|
|
print("Checking", self.url)
|
|
import html2text
|
|
|
|
try:
|
|
r = requests.get(self.url, headers=headers, timeout=15, verify=False)
|
|
stripped_text_from_html = html2text.html2text(r.content.decode('utf-8'))
|
|
self.save_response_stripped_output(stripped_text_from_html)
|
|
|
|
# Usually from networkIO/requests level
|
|
except (requests.exceptions.ConnectionError,requests.exceptions.ReadTimeout) as e:
|
|
self.datastore.update_watch(self.uuid, 'last_error', str(e))
|
|
print(str(e))
|
|
|
|
# Usually from html2text level
|
|
except UnicodeDecodeError as e:
|
|
self.datastore.update_watch(self.uuid, 'last_error', str(e))
|
|
print(str(e))
|
|
# figure out how to deal with this cleaner..
|
|
# 'utf-8' codec can't decode byte 0xe9 in position 480: invalid continuation byte
|
|
|
|
else:
|
|
|
|
# We rely on the actual text in the html output.. many sites have random script vars etc
|
|
self.datastore.update_watch(self.uuid, 'last_error', False)
|
|
self.datastore.update_watch(self.uuid, 'last_check_status', r.status_code)
|
|
|
|
fetched_md5 = hashlib.md5(stripped_text_from_html.encode('utf-8')).hexdigest()
|
|
|
|
if self.current_md5 != fetched_md5:
|
|
|
|
# Dont confuse people by putting last-changed, when it actually just changed from nothing..
|
|
if self.datastore.get_val(self.uuid, 'previous_md5') is not None:
|
|
self.datastore.update_watch(self.uuid, 'last_changed', self.timestamp)
|
|
|
|
self.datastore.update_watch(self.uuid, 'previous_md5', fetched_md5)
|
|
self.save_response_html_output(r.text)
|
|
|
|
|
|
self.datastore.update_watch(self.uuid, 'last_checked', int(time.time()))
|
|
pass
|