README.md

@@ -1,6 +1,3 @@
 # mpri-webdam
 
-Generate realistic fake browsing histories for borderline and/or activists
+Générer tout plein de faux historiques. Parce qu'il faut bien valider ce cours.
-users, to hide real traffic from global surveillance.
-
-Lacks proper documentation at the moment `:(`

data/search_engine.json

@@ -13,13 +13,6 @@
             "query_pattern":"?q={}"
         }
     },
-    {
-		"searchengine": {
-            "name":"Duckduckgo Lite",
-            "url":"https://duckduckgo.com/lite/",
-            "query_pattern":"?q={}"
-        }
-    },
     {
 		"searchengine": {
             "name":"Qwant",

histories/management/commands/gen_history.py

@@ -1,16 +0,0 @@
-from django.core.management.base import BaseCommand
-from profiles import models as profiles
-from histories.models import generate_history
-from datetime import datetime
-
-
-class Command(BaseCommand):
-    ''' Generates an history and prints the related XML '''
-
-    def add_arguments(self, parser):
-        pass
-
-    def handle(self, *args, **kwargs):
-        prof = profiles.Profile.objects.all()[0]
-        history = generate_history(prof, datetime.now())
-        print(history.to_xml_string())

histories/models.py

@@ -5,7 +5,6 @@ interests, keywords...
 
 from collections import namedtuple
 import random
-import asyncio
 from math import floor
 from xml.etree import ElementTree as ET
 from datetime import datetime
@@ -44,9 +43,9 @@ class HistoryEntry(models.Model):
     def to_xml(self, xml_root):
         entry = ET.Element('history')
         entry_url = ET.Element('url')
-        entry_url.text = str(self.search)
+        entry_url.text = self.search
         entry_ts = ET.Element('timestamp')
-        entry_ts.text = str(self.timestamp.timestamp())
+        entry_ts.text = self.timestamp.timestamp()
         entry.append(entry_url)
         entry.append(entry_ts)
         xml_root.append(entry)
@@ -108,18 +107,13 @@ class History(models.Model):
             output += str(entry) + '\n'
         return output
 
-    async def _handler(self):
+    async def play_histories(self):
-        runner = await TorInstance.create(self.return_history(), self.user.browser_fingerprint.serialize_headers())
-        await runner.run()
-        self.played = True
-        self.save()
-
-    def play_histories(self):
         """ Actually plays the history.
         """
-        loop = asyncio.new_event_loop()
+        self.played = True
-        asyncio.set_event_loop(loop)
+        runner = await TorInstance.create(self.return_history(), self.user.browser_fingerprint.serialize_headers())
-        loop.run_until_complete(asyncio.wait([self._handler()]))
+        runner.run()
+        self.save()
 
     def to_xml(self, xml_root=None):
         ''' Exports the current history to xml '''
@@ -129,9 +123,9 @@ class History(models.Model):
             xml_root = ET.Element('root')
 
         hist_node = ET.Element("history", attrib={
-            'start-ts': str(self.start_ts),
+            'start-ts': self.start_ts,
-            'played': '1' if self.played else '0',
+            'played': 1 if self.played else 0,
-            'user': str(self.user.pk),
+            'user': self.user.pk,
         })
         xml_root.append(hist_node)
         for entry in self.historyentry_set.all():
@@ -140,10 +134,6 @@ class History(models.Model):
         if standalone:
             return xml_root
 
-    def to_xml_string(self):
-        xml = self.to_xml()
-        return ET.tostring(xml)
-
     @staticmethod
     def from_xml(xml_root):
         ''' Loads an history from an XML file '''
@@ -187,6 +177,7 @@ def generate_partial_history(user, t_start):
     timestamp = t_start
     result = []
     basis = generate_first_url(user)
+    result.append(PartialHistoryEntry(basis, timestamp))
     t_start += 5 * random.weibullvariate(1, 1.5)
     crawler = crawl.CrawlingThread(basis)
     crawler.start()

histories/tor_runner.py

@@ -58,9 +58,7 @@ class TorInstance():
     async def run(self):
         """ Runs the Tor Instance on the history.
         """
-        while (self.history) and (dt.datetime.combine(self.history[0][1],
+        while (self.history[0][1] - dt.datetime.now()).total_seconds >= 10:
-                                                      dt.datetime.min.time()) -
-                                  dt.datetime.now()).total_seconds() >= 10:
             print("Sleeping")
             sleep(10)
         while self.history:
@@ -68,9 +66,8 @@ class TorInstance():
             async with async_timeout.timeout(30):
                 await(self.query(item[0]))
                 now = dt.datetime.now()
-                print(self.history[0])
+                if now <= self.history[0][1]:
-                if now <= dt.datetime.combine(self.history[0][1], dt.datetime.min.time()):
+                    sleep((self.history[0][1] - now).total_seconds())
-                    sleep((dt.datetime.combine(self.history[0][1], dt.datetime.min.time()) - now).total_seconds())
 
 
     def create_session(self):

requirements.txt

@@ -16,4 +16,3 @@ stem==1.6.0
 pycurl==7.43.0.1
 rdflib==4.2.2
 git+https://github.com/tobast/RDFSerializer.git
-aiosocks==0.2.6