fix: [background update] avoid multiple update on the same key

OVERVIEW.md

@@ -66,6 +66,7 @@ Redis and ARDB overview
 | paste_metadata:**item path** | super_father | **first url crawled** |
 | | father | **item father** |
 | | domain | **crawled domain**:**domain port** |
+| | screenshot | **screenshot hash** |
 
 ##### Set:
 | Key | Field |

bin/LAUNCH.sh

@@ -212,6 +212,8 @@ function launching_scripts {
     sleep 0.1
     screen -S "Script_AIL" -X screen -t "SentimentAnalysis" bash -c "cd ${AIL_BIN}; ${ENV_PY} ./SentimentAnalysis.py; read x"
     sleep 0.1
+    screen -S "Script_AIL" -X screen -t "SentimentAnalysis" bash -c "cd ${AIL_BIN}; ${ENV_PY} ./update-background.py; read x"
+    sleep 0.1
     screen -S "Script_AIL" -X screen -t "SubmitPaste" bash -c "cd ${AIL_BIN}; ${ENV_PY} ./submit_paste.py; read x"
 
 }

bin/update-background.py

@@ -5,7 +5,7 @@
 Update AIL
 ============================
 
-Update AIL clone and fork
+Update AIL in the background
 
 """
 
@@ -31,10 +31,7 @@ if __name__ == "__main__":
         db=cfg.getint("ARDB_DB", "db"),
         decode_responses=True)
 
-    ail_version = r_serv.get('ail:version')
+    if r_serv.exists('ail:update_v1.5'):
-    if ail_version is None:
-        sys.exit(0)
-    else:
         if ail_version == 'v1.5':
             onions_update_status = r_serv.get('v1.5:onions')
             if onions_update_status is None:

update/v1.4/Update-ARDB_Metadata.py

@@ -106,6 +106,7 @@ if __name__ == '__main__':
         # update domain port
         domain = r_serv_metadata.hget(new_item_metadata, 'domain')
         if domain:
+            if domain[-3:] ?= ':80':
                 r_serv_metadata.hset(new_item_metadata, 'domain', '{}:80'.format(domain))
         super_father = r_serv_metadata.hget(new_item_metadata, 'super_father')
         if super_father:
@@ -116,150 +117,18 @@ if __name__ == '__main__':
             if PASTES_FOLDER in father:
                 r_serv_metadata.hset(new_item_metadata, 'father', father.replace(PASTES_FOLDER, '', 1))
 
-
-
-
-    ######################################################################################################################
-    ######################################################################################################################
-    ######################################################################################################################
-    ######################################################################################################################
-    ######################################################################################################################
-    ######################################################################################################################
-    '''
-
-    string_keys_to_rename = ['misp_events:{}*'.format(PASTES_FOLDER), 'hive_cases:{}*'.format(PASTES_FOLDER)]
-    for key_to_rename in string_keys_to_rename:
-
-        keys_to_rename = []
-        for key in r_serv_metadata.scan_iter(key_to_rename):
-            new_key = key.replace(PASTES_FOLDER, '', 1)
-            keys_to_rename.append( (key, new_key) )
-            index = index + 1
-        for key, new_key in keys_to_rename:
-            r_serv_metadata.rename(key, new_key)
-
-    keys_to_rename = None
-
-    set_keys_to_rename = ['tag:{}*'.format(PASTES_FOLDER), 'paste_regular_external_links:{}*'.format(PASTES_FOLDER), 'paste_onion_external_links:{}*'.format(PASTES_FOLDER), 'paste_children:{}*'.format(PASTES_FOLDER)]
-    for key_to_rename in set_keys_to_rename:
-
-        keys_to_remove = []
-        keys_to_rename = []
-        for key in r_serv_metadata.scan_iter(key_to_rename):
-            new_key = key.replace(PASTES_FOLDER, '', 1)
-            # a set with this key already exist
-            if r_serv_metadata.exists(new_key):
-                # save data
-                for new_key_value in r_serv_metadata.smembers(key):
-                    r_serv_metadata.sadd(new_key, new_key_value)
-                    keys_to_remove.append(key)
-            else:
-                keys_to_rename.append( (key, new_key) )
-            index = index + 1
-        for key in keys_to_remove:
-            r_serv_metadata.delete(key)
-        for key, new_key in keys_to_rename:
-            r_serv_metadata.rename(key, new_key)
-
-    keys_to_remove = None
-    keys_to_rename = None
-
-
-    zset_keys_to_rename = ['nb_seen_hash:*', 'base64_hash:*', 'binary_hash:*']
-    for key_to_rename in zset_keys_to_rename:
-
-        keys_to_remove = []
-        zkeys_to_remove = []
-        keys_to_add = []
-        for key in r_serv_metadata.scan_iter(key_to_rename):
-            temp = []
-            for zset_key, value in r_serv_metadata.zscan_iter(key, '*{}*'.format(PASTES_FOLDER)):
-                new_key = zset_key.replace(PASTES_FOLDER, '', 1)
-                index = index +1
-                temp.append((key, zset_key))
-                keys_to_add.append((key, new_key, value))
-            if 0 < len(temp) < r_serv_metadata.zcard(key):
-                zkeys_to_remove.extend(temp)
-            else:
-                keys_to_remove.append(key)
-        for key in keys_to_remove:
-            r_serv_metadata.delete(key)
-        for key, zset_key in zkeys_to_remove:
-            r_serv_metadata.zrem(key, zset_key)
-        for key, new_key, value in keys_to_add:
-            r_serv_metadata.zincrby(key, new_key, int(value))
-    keys_to_remove = None
-    zkeys_to_remove = None
-    keys_to_add = None
-
-    set_keys_to_rename = ['paste_children:*']
-    for key_to_rename in set_keys_to_rename:
-        keys_to_remove = []
-        skeys_to_remove = []
-        keys_to_add = []
-        for key in r_serv_metadata.scan_iter(key_to_rename):
-            temp = []
-            for set_key in r_serv_metadata.sscan_iter(key, '*{}*'.format(PASTES_FOLDER)):
-                new_key = set_key.replace(PASTES_FOLDER, '', 1)
-                index = index +1
-                temp.append((key, set_key))
-                keys_to_add.append((key, new_key))
-            if 0 < len(temp) < r_serv_metadata.scard(key):
-                skeys_to_remove.extend(temp)
-            else:
-                keys_to_remove.append(key)
-        for key in keys_to_remove:
-            r_serv_metadata.delete(key)
-        for key, set_key in skeys_to_remove:
-            r_serv_metadata.srem(key, set_key)
-        for key, new_key in keys_to_add:
-            r_serv_metadata.sadd(key, new_key)
-    keys_to_remove = None
-    skeys_to_remove = None
-    keys_to_add = None
-
-    hset_keys_to_rename = ['paste_metadata:{}*'.format(PASTES_FOLDER)]
-    for key_to_rename in hset_keys_to_rename:
-
-        keys_to_rename = []
-        for key in r_serv_metadata.scan_iter(key_to_rename):
-            new_key = key.replace(PASTES_FOLDER, '', 1)
-            # a hset with this key already exist
-            if r_serv_metadata.exists(new_key):
-                pass
-            else:
-                keys_to_rename.append((key, new_key))
-                index = index + 1
-        for key, new_key in keys_to_rename:
-            r_serv_metadata.rename(key, new_key)
-    keys_to_rename = None
-
-    # to verify 120/100 try with scan
-    hset_keys_to_rename = ['paste_metadata:*']
-    for key_to_rename in hset_keys_to_rename:
-        for key in r_serv_metadata.scan_iter(key_to_rename):
-            father = r_serv_metadata.hget(key, 'father')
-            super_father = r_serv_metadata.hget(key, 'super_father')
-
-            if father:
-                if PASTES_FOLDER in father:
-                    index = index + 1
-                    r_serv_metadata.hdel(key, 'father')
-                    r_serv_metadata.hset(key, 'father', father.replace(PASTES_FOLDER, '', 1))
-
-            if super_father:
-                if PASTES_FOLDER in super_father:
-                    index = index + 1
-                    r_serv_metadata.hdel(key, 'super_father')
-                    r_serv_metadata.hset(key, 'super_father', super_father.replace(PASTES_FOLDER, '', 1))
-
-    keys_to_rename = None
-    '''
-
-
     end = time.time()
 
     print('Updating ARDB_Metadata Done => {} paths: {} s'.format(index, end - start))
     print()
 
     r_serv.set('v1.5:metadata', 1)
+
+    ##
+    #Key, Dynamic Update
+    ##
+    #paste_children
+    #nb_seen_hash, base64_hash, binary_hash
+    #paste_onion_external_links
+    #misp_events, hive_cases
+    ##

update/v1.4/Update-ARDB_Tags.py

@@ -7,14 +7,6 @@ import time
 import redis
 import configparser
 
-def tags_key_fusion(old_item_path_key, new_item_path_key):
-    print('fusion:')
-    print(old_item_path_key)
-    print(new_item_path_key)
-    for tag in r_serv_metadata.smembers(old_item_path_key):
-        r_serv_metadata.sadd(new_item_path_key, tag)
-        r_serv_metadata.srem(old_item_path_key, tag)
-
 if __name__ == '__main__':
 
     start_deb = time.time()
@@ -109,8 +101,12 @@ if __name__ == '__main__':
             # update metadata last_seen
             if item_date > tag_metadata[tag]['last_seen']:
                 tag_metadata[tag]['last_seen'] = item_date
+                last_seen_db = r_serv_tag.hget('tag_metadata:{}'.format(tag), 'last_seen')
+                if last_seen_db:
+                    if item_date > int(last_seen_db):
                         r_serv_tag.hset('tag_metadata:{}'.format(tag), 'last_seen', item_date)
-
+                    else:
+                        tag_metadata[tag]['last_seen'] = last_seen_db
 
             r_serv_tag.sadd('{}:{}'.format(tag, item_date), new_path)
             r_serv_tag.hincrby('daily_tags:{}'.format(item_date), tag, 1)

update/v1.4/Update.py

@@ -5,6 +5,7 @@ import os
 import sys
 import time
 import redis
+import datetime
 import configparser
 
 if __name__ == '__main__':
@@ -56,5 +57,9 @@ if __name__ == '__main__':
 
     #Set current ail version
     r_serv.set('ail:version', 'v1.5')
+    r_serv.set('ail:update_v1.5', 1)
+
+    #Set current ail version
+    r_serv.set('ail:update_date_v1.5', datetime.datetime.now().strftime("%Y%m%d"))
 
     print('Done in {} s'.format(end - start_deb))