chg: [updater] refactor background updater + add v5.2 update

bin/lib/ail_updates.py

@@ -15,38 +15,15 @@ config_loader = ConfigLoader()
 r_db = config_loader.get_db_conn("Kvrocks_DB")
 config_loader = None
 
-BACKGROUND_UPDATES = {
-    'v1.5': {
-        'nb_updates': 5,
-        'message': 'Tags and Screenshots'
-    },
-    'v2.4': {
-        'nb_updates': 1,
-        'message': ' Domains Tags and Correlations'
-    },
-    'v2.6': {
-        'nb_updates': 1,
-        'message': 'Domains Tags and Correlations'
-    },
-    'v2.7': {
-        'nb_updates': 1,
-        'message': 'Domains Tags'
-    },
-    'v3.4': {
-        'nb_updates': 1,
-        'message': 'Domains Languages'
-    },
-    'v3.7': {
-        'nb_updates': 1,
-        'message': 'Trackers first_seen/last_seen'
-    }
-}
-
+# # # # # # # #
+#             #
+#   UPDATE    #
+#             #
+# # # # # # # #
 
 def get_ail_version():
     return r_db.get('ail:version')
 
-
 def get_ail_float_version():
     version = get_ail_version()
     if version:
@@ -55,6 +32,179 @@ def get_ail_float_version():
         version = 0
     return version
 
+# # # - - # # #
+
+# # # # # # # # # # # #
+#                     #
+#  UPDATE BACKGROUND  #
+#                     #
+# # # # # # # # # # # #
+
+
+BACKGROUND_UPDATES = {
+    'v5.2': {
+        'message': 'Compress HAR',
+        'scripts': ['compress_har.py']
+    },
+}
+
+class AILBackgroundUpdate:
+    """
+        AIL Background Update.
+    """
+
+    def __init__(self, version):
+        self.version = version
+
+    def _get_field(self, field):
+        return r_db.hget('ail:update:background', field)
+
+    def _set_field(self, field, value):
+        r_db.hset('ail:update:background', field, value)
+
+    def get_version(self):
+        return self.version
+
+    def get_message(self):
+        return BACKGROUND_UPDATES.get(self.version, {}).get('message', '')
+
+    def get_error(self):
+        return self._get_field('error')
+
+    def set_error(self, error):  # TODO ADD LOGS
+        self._set_field('error', error)
+
+    def get_nb_scripts(self):
+        return int(len(BACKGROUND_UPDATES.get(self.version, {}).get('scripts', [''])))
+
+    def get_scripts(self):
+        return BACKGROUND_UPDATES.get(self.version, {}).get('scripts', [])
+
+    def get_nb_scripts_done(self):
+        done = self._get_field('done')
+        try:
+            done = int(done)
+        except (TypeError, ValueError):
+            done = 0
+        return done
+
+    def inc_nb_scripts_done(self):
+        self._set_field('done', self.get_nb_scripts_done() + 1)
+
+    def get_script(self):
+        return self._get_field('script')
+
+    def get_script_path(self):
+        path = os.path.basename(self.get_script())
+        if path:
+            return os.path.join(os.environ['AIL_HOME'], 'update', self.version, path)
+
+    def get_nb_to_update(self):  # TODO use cache ?????
+        nb_to_update = self._get_field('nb_to_update')
+        if not nb_to_update:
+            nb_to_update = 1
+        return int(nb_to_update)
+
+    def set_nb_to_update(self, nb):
+        self._set_field('nb_to_update', int(nb))
+
+    def get_nb_updated(self):  # TODO use cache ?????
+        nb_updated = self._get_field('nb_updated')
+        if not nb_updated:
+            nb_updated = 0
+        return int(nb_updated)
+
+    def inc_nb_updated(self):  # TODO use cache ?????
+        r_db.hincrby('ail:update:background', 'nb_updated', 1)
+
+    def get_progress(self):  # TODO use cache ?????
+        return self._get_field('progress')
+
+    def set_progress(self, progress):
+        self._set_field('progress', progress)
+
+    def update_progress(self):
+        nb_updated = self.get_nb_updated()
+        nb_to_update = self.get_nb_to_update()
+        if nb_updated == nb_to_update:
+            progress = 100
+        elif nb_updated > nb_to_update:
+            progress = 99
+        else:
+            progress = int((nb_updated * 100) / nb_to_update)
+        self.set_progress(progress)
+        print(f'{nb_updated}/{nb_to_update}    updated    {progress}%')
+        return progress
+
+    def is_running(self):
+        return r_db.hget('ail:update:background', 'version') == self.version
+
+    def get_meta(self, options=set()):
+        meta = {'version': self.get_version(),
+                'error': self.get_error(),
+                'script': self.get_script(),
+                'script_progress': self.get_progress(),
+                'nb_update': self.get_nb_scripts(),
+                'nb_completed': self.get_nb_scripts_done()}
+        meta['progress'] = int(meta['nb_completed'] * 100 / meta['nb_update'])
+        if 'message' in options:
+            meta['message'] = self.get_message()
+        return meta
+
+    def start(self):
+        self._set_field('version', self.version)
+        r_db.hdel('ail:update:background', 'error')
+
+    def start_script(self, script):
+        self.clear()
+        self._set_field('script', script)
+        self.set_progress(0)
+
+    def end_script(self):
+        self.set_progress(100)
+        self.inc_nb_scripts_done()
+
+    def clear(self):
+        r_db.hdel('ail:update:background', 'error')
+        r_db.hdel('ail:update:background', 'progress')
+        r_db.hdel('ail:update:background', 'nb_updated')
+        r_db.hdel('ail:update:background', 'nb_to_update')
+
+    def end(self):
+        r_db.delete('ail:update:background')
+        r_db.srem('ail:updates:background', self.version)
+
+
+# To Add in update script
+def add_background_update(version):
+    r_db.sadd('ail:updates:background', version)
+
+def is_update_background_running():
+    return r_db.exists('ail:update:background')
+
+def get_update_background_version():
+    return r_db.hget('ail:update:background', 'version')
+
+def get_update_background_meta(options=set()):
+    version = get_update_background_version()
+    if version:
+        return AILBackgroundUpdate(version).get_meta(options=options)
+    else:
+        return {}
+
+def get_update_background_to_launch():
+    to_launch = []
+    updates = r_db.smembers('ail:updates:background')
+    for version in BACKGROUND_UPDATES:
+        if version in updates:
+            to_launch.append(version)
+    return to_launch
+
+# # # - - # # #
+
+##########################################################################################
+##########################################################################################
+##########################################################################################
 
 def get_ail_all_updates(date_separator='-'):
     dict_update = r_db.hgetall('ail:update_date')
@@ -87,111 +237,6 @@ def check_version(version):
     return True
 
 
-#### UPDATE BACKGROUND ####
-
-def exits_background_update_to_launch():
-    return r_db.scard('ail:update:to_update') != 0
-
-
-def is_version_in_background_update(version):
-    return r_db.sismember('ail:update:to_update', version)
-
-
-def get_all_background_updates_to_launch():
-    return r_db.smembers('ail:update:to_update')
-
-
-def get_current_background_update():
-    return r_db.get('ail:update:update_in_progress')
-
-
-def get_current_background_update_script():
-    return r_db.get('ail:update:current_background_script')
-
-
-def get_current_background_update_script_path(version, script_name):
-    return os.path.join(os.environ['AIL_HOME'], 'update', version, script_name)
-
-
-def get_current_background_nb_update_completed():
-    return r_db.scard('ail:update:update_in_progress:completed')
-
-
-def get_current_background_update_progress():
-    progress = r_db.get('ail:update:current_background_script_stat')
-    if not progress:
-        progress = 0
-    return int(progress)
-
-
-def get_background_update_error():
-    return r_db.get('ail:update:error')
-
-
-def add_background_updates_to_launch(version):
-    return r_db.sadd('ail:update:to_update', version)
-
-
-def start_background_update(version):
-    r_db.delete('ail:update:error')
-    r_db.set('ail:update:update_in_progress', version)
-
-
-def set_current_background_update_script(script_name):
-    r_db.set('ail:update:current_background_script', script_name)
-    r_db.set('ail:update:current_background_script_stat', 0)
-
-
-def set_current_background_update_progress(progress):
-    r_db.set('ail:update:current_background_script_stat', progress)
-
-
-def set_background_update_error(error):
-    r_db.set('ail:update:error', error)
-
-
-def end_background_update_script():
-    r_db.sadd('ail:update:update_in_progress:completed')
-
-
-def end_background_update(version):
-    r_db.delete('ail:update:update_in_progress')
-    r_db.delete('ail:update:current_background_script')
-    r_db.delete('ail:update:current_background_script_stat')
-    r_db.delete('ail:update:update_in_progress:completed')
-    r_db.srem('ail:update:to_update', version)
-
-
-def clear_background_update():
-    r_db.delete('ail:update:error')
-    r_db.delete('ail:update:update_in_progress')
-    r_db.delete('ail:update:current_background_script')
-    r_db.delete('ail:update:current_background_script_stat')
-    r_db.delete('ail:update:update_in_progress:completed')
-
-
-def get_update_background_message(version):
-    return BACKGROUND_UPDATES[version]['message']
-
-
-# TODO: Detect error in subprocess
-def get_update_background_metadata():
-    dict_update = {}
-    version = get_current_background_update()
-    if version:
-        dict_update['version'] = version
-        dict_update['script'] = get_current_background_update_script()
-        dict_update['script_progress'] = get_current_background_update_progress()
-        dict_update['nb_update'] = BACKGROUND_UPDATES[dict_update['version']]['nb_updates']
-        dict_update['nb_completed'] = get_current_background_nb_update_completed()
-        dict_update['progress'] = int(dict_update['nb_completed'] * 100 / dict_update['nb_update'])
-        dict_update['error'] = get_background_update_error()
-    return dict_update
-
-
-##-- UPDATE BACKGROUND --##
-
-
 if __name__ == '__main__':
     res = check_version('v3.1..1')
     print(res)

bin/lib/crawlers.py

@@ -366,8 +366,8 @@ def _reprocess_all_hars_etag():
             etag = Etags.create(etag_content)
             etag.add(date, domain)
 
-def _gzip_all_hars():
-    for har_id in get_all_har_ids():
+
+def _gzip_har(har_id):
     har_path = os.path.join(HAR_DIR, har_id)
     new_id = f'{har_path}.gz'
     if not har_id.endswith('.gz'):
@@ -382,6 +382,10 @@ def _gzip_all_hars():
         os.remove(har_path)
         print('delete:', har_path)
 
+def _gzip_all_hars():
+    for har_id in get_all_har_ids():
+        _gzip_har(har_id)
+
 # # # - - # # #
 
 ################################################################################

bin/update-background.py

@@ -10,6 +10,8 @@ Update AIL in the background
 """
 
 import os
+import logging
+import logging.config
 import sys
 import subprocess
 
@@ -19,35 +21,52 @@ sys.path.append(os.environ['AIL_BIN'])
 ##################################
 from lib import ail_updates
 
-def launch_background_upgrade(version, l_script_name):
-    if ail_updates.is_version_in_background_update(version):
-        ail_updates.start_background_update(version)
+logging.config.dictConfig(ail_logger.get_config(name='updates'))
+def launch_background_upgrade(version):
+    logger = logging.getLogger()
+    update.set_error('Error Updater Script')
+    print('launching background update', version)
+    update = ail_updates.AILBackgroundUpdate(version)
+    nb_done = update.get_nb_scripts_done()
+    update.start()
+    scripts = update.get_scripts()
+    scripts = scripts[nb_done:]
+    for script in scripts:
+        print('launching background script update', script)
+        # launch script
+        update.start_script(script)
+        script_path = update.get_script_path()
+        if script_path:
+            try:
+                process = subprocess.run(['python', script_path])
+                if process.returncode != 0:
+                    stderr = process.stderr
+                    if stderr:
+                        error = stderr.decode()
+                        logger.error(error)
+                        update.set_error(error)
+                    else:
+                        update.set_error('Error Updater Script')
+                        logger.error('Error Updater Script')
+                    sys.exit(0)
+            except Exception as e:
+                update.set_error(str(e))
+                logger.error(str(e))
+                sys.exit(0)
 
-        for script_name in l_script_name:
-            ail_updates.set_current_background_update_script(script_name)
-            update_file = ail_updates.get_current_background_update_script_path(version, script_name)
+        if not update.get_error():
+            update.end_script()
+        else:
+            logger.warning('Updater exited on error')
+            sys.exit(0)
 
-            # # TODO: Get error output
-            process = subprocess.run(['python', update_file])
-
-        update_progress = ail_updates.get_current_background_update_progress()
-        if update_progress == 100:
-            ail_updates.end_background_update_script()
-        # # TODO: Create Custom error
-        # 'Please relaunch the bin/update-background.py script'
-        # # TODO: Create Class background update
-
-        ail_updates.end_background_update(version)
+    update.end()
 
 
 if __name__ == "__main__":
-
-    if not ail_updates.exits_background_update_to_launch():
-        ail_updates.clear_background_update()
+    if ail_updates.is_update_background_running():
+        v = ail_updates.get_update_background_version()
+        launch_background_upgrade(v)
     else:
-        launch_background_upgrade('v1.5', ['Update-ARDB_Onions.py', 'Update-ARDB_Metadata.py', 'Update-ARDB_Tags.py',
-                                           'Update-ARDB_Tags_background.py', 'Update-ARDB_Onions_screenshots.py'])
-        launch_background_upgrade('v2.6', ['Update_screenshots.py'])
-        launch_background_upgrade('v2.7', ['Update_domain_tags.py'])
-        launch_background_upgrade('v3.4', ['Update_domain.py'])
-        launch_background_upgrade('v3.7', ['Update_trackers.py'])
+        for ver in ail_updates.get_update_background_to_launch():
+            launch_background_upgrade(ver)

update/bin/ail_updater.py

@@ -20,7 +20,7 @@ class AIL_Updater(object):
         self.start_time = time.time()
 
         self.config = ConfigLoader()
-        self.r_serv = self.config.get_redis_conn("Kvrocks_DB")
+        self.r_serv = self.config.get_db_conn("Kvrocks_DB")
 
         self.f_version = float(self.version[1:])
         self.current_f_version = ail_updates.get_ail_float_version()
@@ -35,7 +35,7 @@ class AIL_Updater(object):
         """
         Update DB version
         """
-        ail_updates.add_ail_update(version)
+        ail_updates.add_ail_update(self.version)
 
     def run_update(self):
         self.update()

update/v5.2/Update.py

@@ -0,0 +1,24 @@
+#!/usr/bin/env python3
+# -*-coding:UTF-8 -*
+
+import os
+import sys
+
+sys.path.append(os.environ['AIL_HOME'])
+##################################
+# Import Project packages
+##################################
+from update.bin.ail_updater import AIL_Updater
+from lib import ail_updates
+
+class Updater(AIL_Updater):
+    """default Updater."""
+
+    def __init__(self, version):
+        super(Updater, self).__init__(version)
+
+
+if __name__ == '__main__':
+    updater = Updater('v5.2')
+    updater.run_update()
+    ail_updates.add_background_update('v5.2')

update/v5.2/Update.sh

@@ -0,0 +1,33 @@
+#!/bin/bash
+
+[ -z "$AIL_HOME" ] && echo "Needs the env var AIL_HOME. Run the script from the virtual environment." && exit 1;
+[ -z "$AIL_REDIS" ] && echo "Needs the env var AIL_REDIS. Run the script from the virtual environment." && exit 1;
+[ -z "$AIL_ARDB" ] && echo "Needs the env var AIL_ARDB. Run the script from the virtual environment." && exit 1;
+[ -z "$AIL_BIN" ] && echo "Needs the env var AIL_ARDB. Run the script from the virtual environment." && exit 1;
+[ -z "$AIL_FLASK" ] && echo "Needs the env var AIL_FLASK. Run the script from the virtual environment." && exit 1;
+
+export PATH=$AIL_HOME:$PATH
+export PATH=$AIL_REDIS:$PATH
+export PATH=$AIL_ARDB:$PATH
+export PATH=$AIL_BIN:$PATH
+export PATH=$AIL_FLASK:$PATH
+
+GREEN="\\033[1;32m"
+DEFAULT="\\033[0;39m"
+
+echo -e $GREEN"Shutting down AIL ..."$DEFAULT
+bash ${AIL_BIN}/LAUNCH.sh -ks
+wait
+
+# SUBMODULES #
+git submodule update
+
+echo ""
+echo -e $GREEN"Updating AIL VERSION ..."$DEFAULT
+echo ""
+python ${AIL_HOME}/update/v5.2/Update.py
+wait
+echo ""
+echo ""
+
+exit 0

update/v5.2/compress_har.py

@@ -0,0 +1,27 @@
+#!/usr/bin/env python3
+# -*-coding:UTF-8 -*
+
+import gzip
+import os
+import sys
+
+sys.path.append(os.environ['AIL_BIN'])
+##################################
+# Import Project packages
+##################################
+from lib import ail_updates
+from lib import crawlers
+
+if __name__ == '__main__':
+    update = ail_updates.AILBackgroundUpdate('v5.2')
+    HAR_DIR = crawlers.HAR_DIR
+    hars_ids = crawlers.get_all_har_ids()
+    update.set_nb_to_update(len(hars_ids))
+    n = 0
+    for har_id in hars_ids:
+        crawlers._gzip_har(har_id)
+        update.inc_nb_updated()
+        if n % 100 == 0:
+            update.update_progress()
+
+    crawlers._gzip_all_hars()

var/www/blueprints/settings_b.py

@@ -48,7 +48,7 @@ def settings_page():
 @login_required
 @login_read_only
 def get_background_update_metadata_json():
-    return jsonify(ail_updates.get_update_background_metadata())
+    return jsonify(ail_updates.get_update_background_meta(options={}))
 
 @settings_b.route("/settings/modules", methods=['GET'])
 @login_required

var/www/modules/dashboard/Flask_dashboard.py

@@ -167,10 +167,9 @@ def index():
     # Check if update in progress
     background_update = False
     update_message = ''
-    if ail_updates.get_current_background_update():
+    if ail_updates.is_update_background_running():
         background_update = True
-        # update_message = ail_updates.get_update_background_message()
-        update_message = None
+        update_message = ail_updates.AILBackgroundUpdate(ail_updates.get_update_background_version()).get_message()
 
     return render_template("index.html", default_minute = default_minute,
                            threshold_stucked_module=threshold_stucked_module,

var/www/templates/settings/settings_index.html

@@ -141,6 +141,9 @@
 </body>
 
 <script>
+
+var to_refresh
+
 $(document).ready(function(){
   $("#nav_server_status").addClass("active");
 	$("#nav_server").removeClass("text-muted");
@@ -199,12 +202,12 @@ function refresh_update_status(){
   });
 
   if (to_refresh) {
-    setTimeout("refresh_crawler_status()", 10000);
+    setTimeout("refresh_update_status()", 10000);
 	}
 
 }
 
-update_progress();
+refresh_update_status();
 //Interval
 var progress_interval = setInterval(function(){
 	refresh_update_status()