ail-framework/bin/importer/feeders/abstract_chats_feeder.py

382 lines
12 KiB
Python
Raw Normal View History

2023-11-02 15:28:33 +00:00
#!/usr/bin/env python3
# -*-coding:UTF-8 -*
"""
Abstract Chat JSON Feeder Importer Module
================
Process Feeder Json (example: Twitter feeder)
"""
import datetime
import os
import sys
from abc import abstractmethod, ABC
sys.path.append(os.environ['AIL_BIN'])
##################################
# Import Project packages
##################################
from importer.feeders.Default import DefaultFeeder
from lib.objects.Chats import Chat
from lib.objects import ChatSubChannels
2023-11-29 15:28:25 +00:00
from lib.objects import ChatThreads
from lib.objects import Images
2023-11-02 15:28:33 +00:00
from lib.objects import Messages
from lib.objects import FilesNames
# from lib.objects import Files
2023-11-02 15:28:33 +00:00
from lib.objects import UsersAccount
from lib.objects.Usernames import Username
from lib import chats_viewer
import base64
import io
import gzip
# TODO remove compression ???
def _gunzip_bytes_obj(bytes_obj):
gunzipped_bytes_obj = None
try:
in_ = io.BytesIO()
in_.write(bytes_obj)
in_.seek(0)
with gzip.GzipFile(fileobj=in_, mode='rb') as fo:
gunzipped_bytes_obj = fo.read()
except Exception as e:
print(f'Global; Invalid Gzip file: {e}')
return gunzipped_bytes_obj
class AbstractChatFeeder(DefaultFeeder, ABC):
def __init__(self, name, json_data):
super().__init__(json_data)
self.obj = None
self.name = name
def get_chat_protocol(self): # TODO # # # # # # # # # # # # #
return self.name
def get_chat_network(self):
self.json_data['meta'].get('network', None)
def get_chat_address(self):
self.json_data['meta'].get('address', None)
def get_chat_instance_uuid(self):
chat_instance_uuid = chats_viewer.create_chat_service_instance(self.get_chat_protocol(),
network=self.get_chat_network(),
address=self.get_chat_address())
# TODO SET
return chat_instance_uuid
def get_chat_id(self): # TODO RAISE ERROR IF NONE
return self.json_data['meta']['chat']['id']
def get_subchannel_id(self):
2023-11-29 15:28:25 +00:00
return self.json_data['meta']['chat'].get('subchannel', {}).get('id')
2023-11-02 15:28:33 +00:00
def get_subchannels(self):
pass
def get_thread_id(self):
2023-11-29 15:28:25 +00:00
return self.json_data['meta'].get('thread', {}).get('id')
2023-11-02 15:28:33 +00:00
2023-11-07 10:24:24 +00:00
def get_message_id(self):
return self.json_data['meta']['id']
def get_media_name(self):
return self.json_data['meta'].get('media', {}).get('name')
def get_reactions(self):
return self.json_data['meta'].get('reactions', [])
2023-11-02 15:28:33 +00:00
def get_message_timestamp(self):
return self.json_data['meta']['date']['timestamp'] # TODO CREATE DEFAULT TIMESTAMP
# if self.json_data['meta'].get('date'):
# date = datetime.datetime.fromtimestamp( self.json_data['meta']['date']['timestamp'])
# date = date.strftime('%Y/%m/%d')
# else:
# date = datetime.date.today().strftime("%Y/%m/%d")
def get_message_date_timestamp(self):
timestamp = self.get_message_timestamp()
date = datetime.datetime.fromtimestamp(timestamp)
date = date.strftime('%Y%m%d')
return date, timestamp
def get_message_sender_id(self):
return self.json_data['meta']['sender']['id']
def get_message_reply(self):
return self.json_data['meta'].get('reply_to') # TODO change to reply ???
def get_message_reply_id(self):
2023-11-29 15:28:25 +00:00
return self.json_data['meta'].get('reply_to', {}).get('message_id')
2023-11-02 15:28:33 +00:00
def get_message_content(self):
decoded = base64.standard_b64decode(self.json_data['data'])
return _gunzip_bytes_obj(decoded)
def get_obj(self): # TODO handle others objects -> images, pdf, ...
#### TIMESTAMP ####
timestamp = self.get_message_timestamp()
#### Create Object ID ####
chat_id = self.get_chat_id()
message_id = self.get_message_id()
2023-11-29 15:28:25 +00:00
thread_id = self.get_thread_id()
2023-11-02 15:28:33 +00:00
# channel id
# thread id
# TODO sanitize obj type
obj_type = self.get_obj_type()
if obj_type == 'image':
self.obj = Images.Image(self.json_data['data-sha256'])
else:
2023-11-29 15:28:25 +00:00
obj_id = Messages.create_obj_id(self.get_chat_instance_uuid(), chat_id, message_id, timestamp, thread_id=thread_id)
self.obj = Messages.Message(obj_id)
2023-11-02 15:28:33 +00:00
return self.obj
2023-11-29 15:28:25 +00:00
def process_chat(self, new_objs, obj, date, timestamp, reply_id=None):
meta = self.json_data['meta']['chat'] # todo replace me by function
2023-11-02 15:28:33 +00:00
chat = Chat(self.get_chat_id(), self.get_chat_instance_uuid())
# date stat + correlation
chat.add(date, obj)
2023-11-02 15:28:33 +00:00
if meta.get('name'):
chat.set_name(meta['name'])
if meta.get('info'):
chat.set_info(meta['info'])
2023-11-02 15:28:33 +00:00
if meta.get('date'): # TODO check if already exists
chat.set_created_at(int(meta['date']['timestamp']))
if meta.get('icon'):
img = Images.create(meta['icon'], b64=True)
img.add(date, chat)
chat.set_icon(img.get_global_id())
new_objs.add(img)
2023-11-02 15:28:33 +00:00
if meta.get('username'):
username = Username(meta['username'], self.get_chat_protocol())
chat.update_username_timeline(username.get_global_id(), timestamp)
if meta.get('subchannel'):
subchannel = self.process_subchannel(obj, date, timestamp, reply_id=reply_id)
2023-11-02 15:28:33 +00:00
chat.add_children(obj_global_id=subchannel.get_global_id())
else:
if obj.type == 'message':
2023-11-29 15:28:25 +00:00
if self.get_thread_id():
self.process_thread(obj, chat, date, timestamp, reply_id=reply_id)
else:
chat.add_message(obj.get_global_id(), self.get_message_id(), timestamp, reply_id=reply_id)
2023-11-02 15:28:33 +00:00
# if meta.get('subchannels'): # TODO Update icon + names
return chat
def process_subchannel(self, obj, date, timestamp, reply_id=None): # TODO CREATE DATE
2023-11-02 15:28:33 +00:00
meta = self.json_data['meta']['chat']['subchannel']
subchannel = ChatSubChannels.ChatSubChannel(f'{self.get_chat_id()}/{meta["id"]}', self.get_chat_instance_uuid())
# TODO correlation with obj = message/image
2023-11-02 15:28:33 +00:00
subchannel.add(date)
if meta.get('date'): # TODO check if already exists
subchannel.set_created_at(int(meta['date']['timestamp']))
if meta.get('name'):
subchannel.set_name(meta['name'])
# subchannel.update_name(meta['name'], timestamp) # TODO #################
if meta.get('info'):
subchannel.set_info(meta['info'])
if obj.type == 'message':
2023-11-29 15:28:25 +00:00
if self.get_thread_id():
self.process_thread(obj, subchannel, date, timestamp, reply_id=reply_id)
else:
subchannel.add_message(obj.get_global_id(), self.get_message_id(), timestamp, reply_id=reply_id)
2023-11-02 15:28:33 +00:00
return subchannel
2023-11-29 15:28:25 +00:00
def process_thread(self, obj, obj_chat, date, timestamp, reply_id=None):
meta = self.json_data['meta']['thread']
thread_id = self.get_thread_id()
p_chat_id = meta['parent'].get('chat')
p_subchannel_id = meta['parent'].get('subchannel')
p_message_id = meta['parent'].get('message')
# print(thread_id, p_chat_id, p_subchannel_id, p_message_id)
2023-11-29 15:28:25 +00:00
if p_chat_id == self.get_chat_id() and p_subchannel_id == self.get_subchannel_id():
thread = ChatThreads.create(thread_id, self.get_chat_instance_uuid(), p_chat_id, p_subchannel_id, p_message_id, obj_chat)
thread.add(date, obj)
thread.add_message(obj.get_global_id(), self.get_message_id(), timestamp, reply_id=reply_id)
# TODO OTHERS CORRELATIONS TO ADD
if meta.get('name'):
thread.set_name(meta['name'])
2023-11-29 15:28:25 +00:00
return thread
# TODO
# else:
# # ADD NEW MESSAGE REF (used by discord)
2023-11-24 15:35:45 +00:00
def process_sender(self, new_objs, obj, date, timestamp):
2023-11-02 15:28:33 +00:00
meta = self.json_data['meta']['sender']
user_account = UsersAccount.UserAccount(meta['id'], self.get_chat_instance_uuid())
# date stat + correlation
user_account.add(date, obj)
2023-11-02 15:28:33 +00:00
if meta.get('username'):
username = Username(meta['username'], self.get_chat_protocol())
# TODO timeline or/and correlation ????
user_account.add_correlation(username.type, username.get_subtype(r_str=True), username.id)
user_account.update_username_timeline(username.get_global_id(), timestamp)
# Username---Message
username.add(date) # TODO # correlation message ???
# ADDITIONAL METAS
if meta.get('firstname'):
user_account.set_first_name(meta['firstname'])
if meta.get('lastname'):
user_account.set_last_name(meta['lastname'])
if meta.get('phone'):
user_account.set_phone(meta['phone'])
2023-11-24 15:35:45 +00:00
if meta.get('icon'):
img = Images.create(meta['icon'], b64=True)
img.add(date, user_account)
user_account.set_icon(img.get_global_id())
new_objs.add(img)
if meta.get('info'):
user_account.set_info(meta['info'])
2023-11-02 15:28:33 +00:00
return user_account
# Create abstract class: -> new API endpoint ??? => force field, check if already imported ?
# 1) Create/Get MessageInstance - # TODO uuidv5 + text like discord and telegram for default
# 2) Create/Get CHAT ID - Done
# 3) Create/Get Channel IF is in channel
# 4) Create/Get Thread IF is in thread
# 5) Create/Update Username and User-account - Done
def process_meta(self): # TODO CHECK MANDATORY FIELDS
"""
Process JSON meta filed.
"""
# meta = self.get_json_meta()
objs = set()
if self.obj:
objs.add(self.obj)
new_objs = set()
2023-11-02 15:28:33 +00:00
date, timestamp = self.get_message_date_timestamp()
# REPLY
reply_id = self.get_message_reply_id()
# TODO Translation
print(self.obj.type)
# TODO FILES + FILES REF
# get object by meta object type
if self.obj.type == 'message':
# Content
obj = Messages.create(self.obj.id, self.get_message_content()) # TODO translation
# FILENAME
media_name = self.get_media_name()
if media_name:
print(media_name)
FilesNames.FilesNames().create(media_name, date, obj)
for reaction in self.get_reactions():
obj.add_reaction(reaction['reaction'], int(reaction['count']))
else:
chat_id = self.get_chat_id()
thread_id = self.get_thread_id()
channel_id = self.get_subchannel_id()
message_id = self.get_message_id()
message_id = Messages.create_obj_id(self.get_chat_instance_uuid(), chat_id, message_id, timestamp, channel_id=channel_id, thread_id=thread_id)
message = Messages.Message(message_id)
# create empty message if message don't exists
if not message.exists():
message.create('')
objs.add(message)
if message.exists():
obj = Images.create(self.get_message_content())
obj.add(date, message)
obj.set_parent(obj_global_id=message.get_global_id())
# FILENAME
media_name = self.get_media_name()
if media_name:
FilesNames.FilesNames().create(media_name, date, message, file_obj=obj)
2023-11-29 15:28:25 +00:00
for reaction in self.get_reactions():
message.add_reaction(reaction['reaction'], int(reaction['count']))
for obj in objs: # TODO PERF avoid parsing metas multiple times
# CHAT
chat = self.process_chat(new_objs, obj, date, timestamp, reply_id=reply_id)
2023-11-02 15:28:33 +00:00
# SENDER # TODO HANDLE NULL SENDER
2023-11-24 15:35:45 +00:00
user_account = self.process_sender(new_objs, obj, date, timestamp)
2023-11-02 15:28:33 +00:00
# UserAccount---Chat
user_account.add_correlation(chat.type, chat.get_subtype(r_str=True), chat.id)
2023-11-02 15:28:33 +00:00
# if chat: # TODO Chat---Username correlation ???
# # Chat---Username => need to handle members and participants
# chat.add_correlation(username.type, username.get_subtype(r_str=True), username.id)
2023-11-02 15:28:33 +00:00
# TODO Sender image -> correlation
# image
# -> subchannel ?
# -> thread id ?
2023-11-02 15:28:33 +00:00
return new_objs | objs
2023-11-02 15:28:33 +00:00