Add blacklist functionality while ingesting stats (#37)

* add blacklist functionality while ingesting stats

* minor fix

app/ingester/loginsIngester.py

@@ -1,4 +1,5 @@
 from app.logger import log
+from app.utils import configParser
 from app.utils.ipDatabase import geoip2Database
 from sqlalchemy.exc import NoResultFound
 from .utilsIngester import utilsIngester
@@ -7,6 +8,7 @@
 class LoginDataIngester:
     logger = log.get_logger("LoginDataIngester")
 
+
     @classmethod
     def getIdpId(cls, entityid, idpName, tenenvId, session):
         # Check if IdP exists
@@ -152,8 +154,20 @@ def getCountryFromIP(cls, ipAddress, session):
         return countryId
 
     @classmethod
-    def ingestLoginDataPerTenenv(cls, tenenvId, session):
+    def ingestLoginDataPerTenenv(cls, tenenv, session):
+        tenenvId = tenenv['id']
+        tenant_name = tenenv['tenant_name']
+        environment_name = tenenv['environment_name']
+        hashed_user_ids = []
+
+        config_file = f'config.{tenant_name.lower()}.{environment_name.lower()}.py'
 
+        if (configParser.getConfig('user_id_blacklist', config_file) is not False and
+             'user_ids' in configParser.getConfig('user_id_blacklist', config_file)):
+            user_ids = configParser.getConfig('user_id_blacklist', config_file)['user_ids'].split('\n')
+            # Hash each value using SHA-256
+            hashed_user_ids = [hashlib.md5(value.strip().encode()).hexdigest() for value in user_ids]
+
         # get dates not mapped for logi5ns data
         datesNotMapped = utilsIngester.getDatesNotMapped(
             "statistics_country_hashed",
@@ -175,7 +189,14 @@ def ingestLoginDataPerTenenv(cls, tenenvId, session):
                 AND tenenv_id={0} {1}
         """.format(tenenvId, between)).all()
         loginMappedItems = 0
+
         for login in loginsNotMapped:
+
+            if (login[0]['voPersonId'] in hashed_user_ids):
+                cls.logger.info("""Ignore this user with
+                    hash {0} as he is at the blacklist""". format(login[0]['voPersonId']))
+                continue
+
             if (not login[0]['failedLogin']
                 and utilsIngester.validateTenenv(login[0]['tenenvId'], session)
                 and 'voPersonId' in login[0]
@@ -197,7 +218,7 @@ def ingestLoginDataPerTenenv(cls, tenenvId, session):
                                           login[0]['spName'],
                                           login[0]['tenenvId'],
                                           session)
-                
+
                 if ('countryCode' in login[0] and 'countryName' in login[0]):
                     # find countryId
                     countryId = LoginDataIngester.getCountryFromCountryCode([login[0]['countryCode'], login[0]['countryName']], session)
@@ -222,6 +243,12 @@ def ingestLoginDataPerTenenv(cls, tenenvId, session):
 
     @classmethod
     def ingestLoginData(cls, session):
-        tenenvIds = session.exec("""SELECT id FROM tenenv_info""").all()
-        for tenenvId in tenenvIds:
-            LoginDataIngester.ingestLoginDataPerTenenv(tenenvId[0], session)
+        tenenvIds = session.exec("""SELECT tenenv_info.id,
+                                  tenant_info.name AS tenant_name,
+                                  environment_info.name AS environment_name
+                                 FROM tenenv_info 
+                                 JOIN tenant_info ON tenant_id=tenant_info.id
+                                 JOIN environment_info ON env_id=environment_info.id
+                                 """).all()
+        for tenenv in tenenvIds:
+            LoginDataIngester.ingestLoginDataPerTenenv(tenenv, session)

app/ingester/membeshipIngester.py

@@ -1,6 +1,8 @@
 from app.logger import log
+from app.utils import configParser
 from sqlalchemy.exc import NoResultFound
 from .utilsIngester import utilsIngester
+import hashlib
 
 
 class MembershipDataIngester:
@@ -27,7 +29,20 @@ def getCommunityId(cls, communityName, tenenvId, session):
         return communityId
 
     @classmethod
-    def ingestMembershipDataPerTenenv(cls, tenenvId, session):
+    def ingestMembershipDataPerTenenv(cls, tenenv, session):
+        tenenvId = tenenv['id']
+        tenant_name = tenenv['tenant_name']
+        environment_name = tenenv['environment_name']
+        hashed_user_ids = []
+
+        config_file = f'config.{tenant_name.lower()}.{environment_name.lower()}.py'
+
+        if (configParser.getConfig('user_id_blacklist', config_file) is not False and
+           'user_ids' in configParser.getConfig('user_id_blacklist', config_file)):
+            user_ids = configParser.getConfig('user_id_blacklist', config_file)['user_ids'].split('\n')
+            # Hash each value using SHA-256
+            hashed_user_ids = [hashlib.md5(value.strip().encode()).hexdigest() for value in user_ids]
+
         # get dates not mapped for users data
         datesNotMapped = utilsIngester.getDatesNotMapped(
             "members",
@@ -54,6 +69,10 @@ def ingestMembershipDataPerTenenv(cls, tenenvId, session):
                 cls.logger.error("""
                     VO name '{0}' not found """.format(membership[0]['voName']))
                 continue
+            if (membership[0]['voPersonId'] in hashed_user_ids):
+                cls.logger.info("""Ignore this user with
+                    hash {0} as he is at the blacklist""". format(membership[0]['voPersonId']))
+                continue
             session.exec("""INSERT INTO members(community_id,
                 hasheduserid, status, tenenv_id, created, updated)
                 VALUES ('{0}','{1}','{2}', {3}, '{4}', '{4}')
@@ -69,9 +88,15 @@ def ingestMembershipDataPerTenenv(cls, tenenvId, session):
 
     @classmethod
     def ingestMembershipData(cls, session):
-        tenenvIds = session.exec("""SELECT id FROM tenenv_info""").all()
+        tenenvIds = session.exec("""SELECT tenenv_info.id,
+                                  tenant_info.name AS tenant_name,
+                                  environment_info.name AS environment_name
+                                 FROM tenenv_info 
+                                 JOIN tenant_info ON tenant_id=tenant_info.id
+                                 JOIN environment_info ON env_id=environment_info.id
+                                 """).all()
         # for each tenenv on database try to ingest UserData
         # from statistics_raw table
-        for tenenvId in tenenvIds:
+        for tenenv in tenenvIds:
             MembershipDataIngester.ingestMembershipDataPerTenenv(
-                tenenvId[0], session)
+                tenenv, session)

app/ingester/usersIngester.py

@@ -1,12 +1,26 @@
 from app.logger import log
+from app.utils import configParser
 from .utilsIngester import utilsIngester
-
+import hashlib
 
 class UserDataIngester:
     logger = log.get_logger("UserDataIngester")
 
     @classmethod
-    def ingestUserDataPerTenenv(cls, tenenvId, session):
+    def ingestUserDataPerTenenv(cls, tenenv, session):
+        tenenvId = tenenv['id']
+        tenant_name = tenenv['tenant_name']
+        environment_name = tenenv['environment_name']
+        hashed_user_ids = []
+
+        config_file = f'config.{tenant_name.lower()}.{environment_name.lower()}.py'
+
+        if (configParser.getConfig('user_id_blacklist', config_file) is not False and
+           'user_ids' in configParser.getConfig('user_id_blacklist', config_file)):
+            user_ids = configParser.getConfig('user_id_blacklist', config_file)['user_ids'].split('\n')
+            # Hash each value using SHA-256
+            hashed_user_ids = [hashlib.md5(value.strip().encode()).hexdigest() for value in user_ids]
+
         # get dates not mapped for users data
         datesNotMapped = utilsIngester.getDatesNotMapped(
             "users",
@@ -34,6 +48,10 @@ def ingestUserDataPerTenenv(cls, tenenvId, session):
                 cls.logger.error("""
                     user status '{0}' is not valid """.format(user[0]['status']))
                 continue
+            if (user[0]['voPersonId'] in hashed_user_ids):
+                cls.logger.info("""Ignore this user with
+                    hash {0} as he is at the blacklist""". format(user[0]['voPersonId']))
+                continue
             session.exec("""INSERT INTO users(hasheduserid, created, updated, status, tenenv_id)
                 VALUES ('{0}','{1}','{1}', '{2}', {3})
                 ON CONFLICT(hasheduserid, tenenv_id)
@@ -48,8 +66,14 @@ def ingestUserDataPerTenenv(cls, tenenvId, session):
 
     @classmethod
     def ingestUserData(cls, session):
-        tenenvIds = session.exec("""SELECT id FROM tenenv_info""").all()
+        tenenvIds = session.exec("""SELECT tenenv_info.id,
+                                  tenant_info.name AS tenant_name,
+                                  environment_info.name AS environment_name
+                                 FROM tenenv_info 
+                                 JOIN tenant_info ON tenant_id=tenant_info.id
+                                 JOIN environment_info ON env_id=environment_info.id
+                                 """).all()
         # for each tenenv on database try to ingest UserData
         # from statistics_raw table
-        for tenenvId in tenenvIds:
-            UserDataIngester.ingestUserDataPerTenenv(tenenvId[0], session)
+        for tenenv in tenenvIds:
+            UserDataIngester.ingestUserDataPerTenenv(tenenv, session)

app/utils/configParser.py

@@ -30,7 +30,6 @@ def getConfig(section='source_database', config_file='config.py'):
       config[param[0]] = param[1]
 
   else:
-    raise Exception(
-      'Section {0} not found in the {1} file'.format(section, config_file))
+    return False
 
   return config