merge lang

ocr v5.4
terrtia 2024-03-25 16:58:21 +01:00
commit 9d481bd0b0
No known key found for this signature in database
GPG Key ID: 1E1B1F50D84613D0
13 changed files with 364 additions and 96 deletions

View File

@ -206,8 +206,7 @@ class AbstractChatFeeder(DefaultFeeder, ABC):
subchannel = ChatSubChannels.ChatSubChannel(f'{self.get_chat_id()}/{meta["id"]}', self.get_chat_instance_uuid()) subchannel = ChatSubChannels.ChatSubChannel(f'{self.get_chat_id()}/{meta["id"]}', self.get_chat_instance_uuid())
thread = None thread = None
# TODO correlation with obj = message/image subchannel.add(date, obj)
subchannel.add(date)
if meta.get('date'): # TODO check if already exists if meta.get('date'): # TODO check if already exists
subchannel.set_created_at(int(meta['date']['timestamp'])) subchannel.set_created_at(int(meta['date']['timestamp']))
@ -358,7 +357,58 @@ class AbstractChatFeeder(DefaultFeeder, ABC):
# CHAT # CHAT
chat_objs = self.process_chat(new_objs, obj, date, timestamp, reply_id=reply_id) chat_objs = self.process_chat(new_objs, obj, date, timestamp, reply_id=reply_id)
# Message forward # # TODO HANDLE OTHERS OBJECT TYPE
# # TODO MAKE IT GENERIC FOR OTHERS CHATS !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# # Message forward + Discussion
# if self.get_json_meta().get('forward'):
# discussion_id = self.get_json_meta().get('discussion')
# forward_from = self.get_message_forward()
#
# if discussion_id: # TODO HANDLE FORWARDED MESSAGES FROM EXTERNAL CHANNELS
# chat_forward_id = forward_from['from']['id']
# message_forward_id = forward_from['from']['channel_post']
#
# # if chat_forward_id == discussion_id:
# # linked_chat = Chat(chat_forward_id, self.get_chat_instance_uuid())
# # if linked_chat.exists():
# # # create thread
# # # add message replies for each childrens
#
# # TODO HANDLE THREAD
# # TODO Change FORWARD META FIELDS
# # meta['forward'] = {}
# # # CHAT ID
# # # SUBCHANNEL ID -> can be None
# # # Message ID
#
# # meta['forward']['origin']
# # # same as 'forward'
#
# if self.get_json_meta().get('forward'):
# forward = self.get_message_forward()
# f_chat = forward['chat']
# f_subchannel = forward.get('subchannel')
# f_id = forward.get('id')
# if not f_subchannel:
# chat_forward = Chat(f_chat, self.get_chat_instance_uuid())
# if chat_forward.exists():
# for chat_obj in chat_objs:
# if chat_obj.type == 'chat':
# chat_forward.add_relationship(chat_obj.get_global_id(), 'forward')
# # TODO LIST FORWARDED MESSAGES
#
#
# # Discord -> serverID + subchannel ID + message ID
# # Telegram -> chat ID + Message ID
# # + ORIGIN IDs
#
#
#
# # TODO create relationships graph
#
#
# # TODO REMOVE ME
# # Message forward # TODO handle subchannel + message ID
# if self.get_json_meta().get('forward'): # if self.get_json_meta().get('forward'):
# forward_from = self.get_message_forward() # forward_from = self.get_message_forward()
# print('-----------------------------------------------------------') # print('-----------------------------------------------------------')

View File

@ -7,6 +7,7 @@ import sys
import html2text import html2text
import gcld3 import gcld3
from lexilang.detector import detect as lexilang_detect
from libretranslatepy import LibreTranslateAPI from libretranslatepy import LibreTranslateAPI
sys.path.append(os.environ['AIL_BIN']) sys.path.append(os.environ['AIL_BIN'])
@ -264,7 +265,10 @@ def _get_html2text(content, ignore_links=False):
h = html2text.HTML2Text() h = html2text.HTML2Text()
h.ignore_links = ignore_links h.ignore_links = ignore_links
h.ignore_images = ignore_links h.ignore_images = ignore_links
return h.handle(content) content = h.handle(content)
if content == '\n\n':
content = ''
return content
def _clean_text_to_translate(content, html=False, keys_blocks=True): def _clean_text_to_translate(content, html=False, keys_blocks=True):
if html: if html:
@ -323,67 +327,105 @@ def get_objs_languages(obj_type, obj_subtype=''):
def get_obj_languages(obj_type, obj_subtype, obj_id): def get_obj_languages(obj_type, obj_subtype, obj_id):
return r_lang.smembers(f'obj:lang:{obj_type}:{obj_subtype}:{obj_id}') return r_lang.smembers(f'obj:lang:{obj_type}:{obj_subtype}:{obj_id}')
def get_obj_language_stats(obj_type, obj_subtype, obj_id):
return r_lang.zrange(f'obj:langs:stat:{obj_type}:{obj_subtype}:{obj_id}', 0, -1, withscores=True)
# TODO ADD language to CHAT GLOBAL SET # TODO ADD language to CHAT GLOBAL SET
def add_obj_language(language, obj_type, obj_subtype, obj_id): # (s) def add_obj_language(language, obj_type, obj_subtype, obj_id, objs_containers=set()): # (s)
if not obj_subtype: if not obj_subtype:
obj_subtype = '' obj_subtype = ''
obj_global_id = f'{obj_type}:{obj_subtype}:{obj_id}' obj_global_id = f'{obj_type}:{obj_subtype}:{obj_id}'
r_lang.sadd(f'objs:langs:{obj_type}', language) r_lang.sadd(f'objs:langs:{obj_type}', language)
r_lang.sadd(f'objs:lang:{obj_type}:{obj_subtype}', language) r_lang.sadd(f'objs:lang:{obj_type}:{obj_subtype}', language)
r_lang.sadd(f'obj:lang:{obj_global_id}', language) new = r_lang.sadd(f'obj:lang:{obj_global_id}', language)
r_lang.sadd(f'languages:{language}', f'{obj_type}:{obj_subtype}') ################### REMOVE ME ??? r_lang.sadd(f'languages:{language}', f'{obj_type}:{obj_subtype}') ################### REMOVE ME ???
r_lang.sadd(f'langs:{obj_type}:{obj_subtype}:{language}', obj_global_id) r_lang.sadd(f'langs:{obj_type}:{obj_subtype}:{language}', obj_global_id)
def remove_obj_language(language, obj_type, obj_subtype, obj_id): if new:
for global_id in objs_containers:
r_lang.zincrby(f'obj:langs:stat:{global_id}', 1, language)
def remove_obj_language(language, obj_type, obj_subtype, obj_id, objs_containers=set()):
if not obj_subtype: if not obj_subtype:
obj_subtype = '' obj_subtype = ''
obj_global_id = f'{obj_type}:{obj_subtype}:{obj_id}' obj_global_id = f'{obj_type}:{obj_subtype}:{obj_id}'
r_lang.srem(f'obj:lang:{obj_global_id}', language) rem = r_lang.srem(f'obj:lang:{obj_global_id}', language)
delete_obj_translation(obj_global_id, language)
r_lang.srem(f'langs:{obj_type}:{obj_subtype}:{language}', obj_global_id) r_lang.srem(f'langs:{obj_type}:{obj_subtype}:{language}', obj_global_id)
if not r_lang.exists(f'langs:{obj_type}:{obj_subtype}:{language}'): if not r_lang.exists(f'langs:{obj_type}:{obj_subtype}:{language}'):
r_lang.srem(f'objs:lang:{obj_type}:{obj_subtype}', language) r_lang.srem(f'objs:lang:{obj_type}:{obj_subtype}', language)
r_lang.srem(f'languages:{language}', f'{obj_type}:{obj_subtype}') r_lang.srem(f'languages:{language}', f'{obj_type}:{obj_subtype}')
if not r_lang.exists(f'objs:lang:{obj_type}:{obj_subtype}'): if not r_lang.exists(f'objs:lang:{obj_type}:{obj_subtype}'):
if r_lang.scard(f'objs:langs:{obj_type}', language) <= 1: if r_lang.scard(f'objs:langs:{obj_type}') <= 1:
r_lang.srem(f'objs:langs:{obj_type}', language) r_lang.srem(f'objs:langs:{obj_type}', language)
def edit_obj_language(language, obj_type, obj_subtype, obj_id): if rem:
remove_obj_language(language, obj_type, obj_subtype, obj_id) for global_id in objs_containers:
add_obj_language(language, obj_type, obj_subtype, obj_id) r = r_lang.zincrby(f'obj:langs:stat:{global_id}', -1, language)
if r < 1:
r_lang.zrem(f'obj:langs:stat:{global_id}', language)
# TODO handle fields
def detect_obj_language(obj_type, obj_subtype, obj_id, content, objs_containers=set()):
detector = LanguagesDetector(nb_langs=1)
language = detector.detect(content)
if language:
language = language[0]
previous_lang = get_obj_languages(obj_type, obj_subtype, obj_id)
if previous_lang:
previous_lang = previous_lang.pop()
if language != previous_lang:
remove_obj_language(previous_lang, obj_type, obj_subtype, obj_id, objs_containers=objs_containers)
add_obj_language(language, obj_type, obj_subtype, obj_id, objs_containers=objs_containers)
else:
add_obj_language(language, obj_type, obj_subtype, obj_id, objs_containers=objs_containers)
return language
## Translation ## Translation
def _get_obj_translation(obj_global_id, language, field=''): def r_get_obj_translation(obj_global_id, language, field=''):
return r_lang.hget(f'tr:{obj_global_id}:{field}', language) return r_lang.hget(f'tr:{obj_global_id}:{field}', language)
def get_obj_translation(obj_global_id, language, source=None, content=None, field=''): def _get_obj_translation(obj_global_id, language, source=None, content=None, field='', objs_containers=set()):
""" """
Returns translated content Returns translated content
""" """
translation = r_cache.get(f'translation:{language}:{obj_global_id}:{field}') translation = r_cache.get(f'translation:{language}:{obj_global_id}:{field}')
r_cache.expire(f'translation:{language}:{obj_global_id}:{field}', 0)
if translation: if translation:
# DEBUG # DEBUG
# print('cache') # print('cache')
# r_cache.expire(f'translation:{language}:{obj_global_id}:{field}', 0) # r_cache.expire(f'translation:{language}:{obj_global_id}:{field}', 0)
return translation return translation
# TODO HANDLE FIELDS TRANSLATION # TODO HANDLE FIELDS TRANSLATION
translation = _get_obj_translation(obj_global_id, language, field=field) translation = r_get_obj_translation(obj_global_id, language, field=field)
if not translation: if not translation:
translation = LanguageTranslator().translate(content, source=source, target=language) source, translation = LanguageTranslator().translate(content, source=source, target=language)
if source:
obj_type, subtype, obj_id = obj_global_id.split(':', 2)
add_obj_language(source, obj_type, subtype, obj_id, objs_containers=objs_containers)
if translation: if translation:
r_cache.set(f'translation:{language}:{obj_global_id}:{field}', translation) r_cache.set(f'translation:{language}:{obj_global_id}:{field}', translation)
r_cache.expire(f'translation:{language}:{obj_global_id}:{field}', 300) r_cache.expire(f'translation:{language}:{obj_global_id}:{field}', 300)
return translation return translation
def get_obj_translation(obj_global_id, language, source=None, content=None, field='', objs_containers=set()):
return _get_obj_translation(obj_global_id, language, source=source, content=content, field=field, objs_containers=objs_containers)
# TODO Force to edit ???? # TODO Force to edit ????
def set_obj_translation(obj_global_id, language, translation, field=''): def set_obj_translation(obj_global_id, language, translation, field=''):
r_cache.delete(f'translation:{language}:{obj_global_id}:') r_cache.delete(f'translation:{language}:{obj_global_id}:')
return r_lang.hset(f'tr:{obj_global_id}:{field}', language, translation) return r_lang.hset(f'tr:{obj_global_id}:{field}', language, translation)
def delete_obj_translation(obj_global_id, language, field=''):
r_cache.delete(f'translation:{language}:{obj_global_id}:')
r_lang.hdel(f'tr:{obj_global_id}:{field}', language)
## --LANGUAGE ENGINE-- ## ## --LANGUAGE ENGINE-- ##
@ -410,11 +452,22 @@ class LanguagesDetector:
if self.min_len > 0: if self.min_len > 0:
if len(content) < self.min_len: if len(content) < self.min_len:
return languages return languages
# p = self.detector.FindTopNMostFreqLangs(content, num_langs=3)
# for lang in p:
# print(lang.language, lang.probability, lang.proportion, lang.is_reliable)
# print('------------------------------------------------')
for lang in self.detector.FindTopNMostFreqLangs(content, num_langs=self.nb_langs): for lang in self.detector.FindTopNMostFreqLangs(content, num_langs=self.nb_langs):
if lang.proportion >= self.min_proportion and lang.probability >= self.min_probability and lang.is_reliable: if lang.proportion >= self.min_proportion and lang.probability >= self.min_probability and lang.is_reliable:
languages.append(lang.language) languages.append(lang.language)
return languages return languages
def detect_lexilang(self, content): # TODO clean text ??? - TODO REMOVE SEPARATOR
language, prob = lexilang_detect(content)
if prob > 0:
return [language]
else:
return []
def detect_libretranslate(self, content): def detect_libretranslate(self, content):
languages = [] languages = []
try: try:
@ -431,19 +484,35 @@ class LanguagesDetector:
languages.append(language) languages.append(language)
return languages return languages
def detect(self, content, force_gcld3=False): def detect(self, content, force_gcld3=False): # TODO detect length between 20-200 ????
if not content:
return None
content = _clean_text_to_translate(content, html=True)
if not content:
return None
# DEBUG
# print('-------------------------------------------------------')
# print(content)
# print(len(content))
# lexilang
if len(content) < 150:
# print('lexilang')
languages = self.detect_lexilang(content)
# gcld3 # gcld3
if len(content) >= 200 or not self.lt or force_gcld3:
language = self.detect_gcld3(content)
# libretranslate
else: else:
language = self.detect_libretranslate(content) # if len(content) >= 200 or not self.lt or force_gcld3:
return language # print('gcld3')
languages = self.detect_gcld3(content)
# libretranslate
# else:
# languages = self.detect_libretranslate(content)
return languages
class LanguageTranslator: class LanguageTranslator:
def __init__(self): def __init__(self):
self.lt = LibreTranslateAPI(get_translator_instance()) self.lt = LibreTranslateAPI(get_translator_instance())
self.ld = LanguagesDetector(nb_langs=1)
def languages(self): def languages(self):
languages = [] languages = []
@ -473,13 +542,13 @@ class LanguageTranslator:
return language[0].get('language') return language[0].get('language')
def detect(self, content): def detect(self, content):
# gcld3 # print('++++++++++++++++++++++++++++++++++++++++++++++++++++++')
if len(content) >= 200: # print(content)
language = self.detect_gcld3(content) language = self.ld.detect(content)
# libretranslate if language:
else: # print(language[0])
language = self.detect_libretranslate(content) # print('##############################################################')
return language return language[0]
def translate(self, content, source=None, target="en"): # TODO source target def translate(self, content, source=None, target="en"): # TODO source target
if target not in get_translation_languages(): if target not in get_translation_languages():
@ -498,9 +567,9 @@ class LanguageTranslator:
translation = None translation = None
# TODO LOG and display error # TODO LOG and display error
if translation == content: if translation == content:
print('EQUAL') # print('EQUAL')
translation = None translation = None
return translation return source, translation
LIST_LANGUAGES = {} LIST_LANGUAGES = {}

View File

@ -323,7 +323,6 @@ def get_username_meta_from_global_id(username_global_id):
username = Usernames.Username(username_id, instance_uuid) username = Usernames.Username(username_id, instance_uuid)
return username.get_meta() return username.get_meta()
# TODO Filter # TODO Filter
## Instance type ## Instance type
## Chats IDS ## Chats IDS
@ -380,6 +379,22 @@ def get_nb_messages_iterator(filters={}):
nb_messages += chat.get_nb_messages() nb_messages += chat.get_nb_messages()
return nb_messages return nb_messages
#### FIX ####
def fix_correlations_subchannel_message():
for instance_uuid in get_chat_service_instances():
for chat_id in ChatServiceInstance(instance_uuid).get_chats():
chat = Chats.Chat(chat_id, instance_uuid)
# subchannels
for subchannel_gid in chat.get_subchannels():
_, _, subchannel_id = subchannel_gid.split(':', 2)
subchannel = ChatSubChannels.ChatSubChannel(subchannel_id, instance_uuid)
messages, _ = subchannel._get_messages(nb=-1)
for mess in messages:
_, _, message_id = mess[0].split(':', )
subchannel.add_correlation('message', '', message_id)
#### API #### #### API ####
def api_get_chat_service_instance(chat_instance_uuid): def api_get_chat_service_instance(chat_instance_uuid):
@ -392,6 +407,7 @@ def api_get_chat(chat_id, chat_instance_uuid, translation_target=None, nb=-1, pa
chat = Chats.Chat(chat_id, chat_instance_uuid) chat = Chats.Chat(chat_id, chat_instance_uuid)
if not chat.exists(): if not chat.exists():
return {"status": "error", "reason": "Unknown chat"}, 404 return {"status": "error", "reason": "Unknown chat"}, 404
# print(chat.get_obj_language_stats())
meta = chat.get_meta({'created_at', 'icon', 'info', 'nb_participants', 'subchannels', 'threads', 'translation', 'username'}, translation_target=translation_target) meta = chat.get_meta({'created_at', 'icon', 'info', 'nb_participants', 'subchannels', 'threads', 'translation', 'username'}, translation_target=translation_target)
if meta['username']: if meta['username']:
meta['username'] = get_username_meta_from_global_id(meta['username']) meta['username'] = get_username_meta_from_global_id(meta['username'])
@ -437,6 +453,7 @@ def api_get_subchannel(chat_id, chat_instance_uuid, translation_target=None, nb=
subchannel = ChatSubChannels.ChatSubChannel(chat_id, chat_instance_uuid) subchannel = ChatSubChannels.ChatSubChannel(chat_id, chat_instance_uuid)
if not subchannel.exists(): if not subchannel.exists():
return {"status": "error", "reason": "Unknown subchannel"}, 404 return {"status": "error", "reason": "Unknown subchannel"}, 404
# print(subchannel.get_obj_language_stats())
meta = subchannel.get_meta({'chat', 'created_at', 'icon', 'nb_messages', 'nb_participants', 'threads', 'translation'}, translation_target=translation_target) meta = subchannel.get_meta({'chat', 'created_at', 'icon', 'nb_messages', 'nb_participants', 'threads', 'translation'}, translation_target=translation_target)
if meta['chat']: if meta['chat']:
meta['chat'] = get_chat_meta_from_global_id(meta['chat']) meta['chat'] = get_chat_meta_from_global_id(meta['chat'])
@ -451,6 +468,7 @@ def api_get_thread(thread_id, thread_instance_uuid, translation_target=None, nb=
thread = ChatThreads.ChatThread(thread_id, thread_instance_uuid) thread = ChatThreads.ChatThread(thread_id, thread_instance_uuid)
if not thread.exists(): if not thread.exists():
return {"status": "error", "reason": "Unknown thread"}, 404 return {"status": "error", "reason": "Unknown thread"}, 404
# print(thread.get_obj_language_stats())
meta = thread.get_meta({'chat', 'nb_messages', 'nb_participants'}) meta = thread.get_meta({'chat', 'nb_messages', 'nb_participants'})
# if meta['chat']: # if meta['chat']:
# meta['chat'] = get_chat_meta_from_global_id(meta['chat']) # meta['chat'] = get_chat_meta_from_global_id(meta['chat'])
@ -461,18 +479,32 @@ def api_get_message(message_id, translation_target=None):
message = Messages.Message(message_id) message = Messages.Message(message_id)
if not message.exists(): if not message.exists():
return {"status": "error", "reason": "Unknown uuid"}, 404 return {"status": "error", "reason": "Unknown uuid"}, 404
meta = message.get_meta({'chat', 'content', 'files-names', 'icon', 'images', 'link', 'parent', 'parent_meta', 'reactions', 'thread', 'translation', 'user-account'}, translation_target=translation_target) meta = message.get_meta({'chat', 'content', 'files-names', 'icon', 'images', 'language', 'link', 'parent', 'parent_meta', 'reactions', 'thread', 'translation', 'user-account'}, translation_target=translation_target)
return meta, 200 return meta, 200
def api_manually_translate_message(message_id, translation_target, translation): def api_message_detect_language(message_id):
message = Messages.Message(message_id)
if not message.exists():
return {"status": "error", "reason": "Unknown uuid"}, 404
lang = message.detect_language()
return {"language": lang}, 200
def api_manually_translate_message(message_id, source, translation_target, translation):
message = Messages.Message(message_id) message = Messages.Message(message_id)
if not message.exists(): if not message.exists():
return {"status": "error", "reason": "Unknown uuid"}, 404 return {"status": "error", "reason": "Unknown uuid"}, 404
if len(translation) > 200000: # TODO REVIEW LIMIT
return {"status": "error", "reason": "Max Size reached"}, 400
if translation_target not in Language.get_translation_languages():
return {"status": "error", "reason": "Unknown Language"}, 400
if translation: if translation:
if len(translation) > 200000: # TODO REVIEW LIMIT
return {"status": "error", "reason": "Max Size reached"}, 400
all_languages = Language.get_translation_languages()
if source not in all_languages:
return {"status": "error", "reason": "Unknown source Language"}, 400
message_language = message.get_language()
if message_language != source:
message.edit_language(message_language, source)
if translation:
if translation_target not in all_languages:
return {"status": "error", "reason": "Unknown target Language"}, 400
message.set_translation(translation_target, translation) message.set_translation(translation_target, translation)
# TODO SANITYZE translation # TODO SANITYZE translation
return None, 200 return None, 200

View File

@ -107,9 +107,24 @@ class Message(AbstractObject):
return message_id return message_id
def get_chat_id(self): # TODO optimize -> use me to tag Chat def get_chat_id(self): # TODO optimize -> use me to tag Chat
chat_id = self.get_basename().rsplit('_', 1)[0] c_id = self.id.split('/')
return chat_id return c_id[2]
def get_chat(self):
c_id = self.id.split('/')
return f'chat:{c_id[0]}:{c_id[2]}'
def get_subchannel(self):
subchannel = self.get_correlation('chat-subchannel')
if subchannel.get('chat-subchannel'):
return f'chat-subchannel:{subchannel["chat-subchannel"].pop()}'
def get_current_thread(self):
subchannel = self.get_correlation('chat-thread')
if subchannel.get('chat-thread'):
return f'chat-thread:{subchannel["chat-thread"].pop()}'
# children thread
def get_thread(self): def get_thread(self):
for child in self.get_childrens(): for child in self.get_childrens():
obj_type, obj_subtype, obj_id = child.split(':', 2) obj_type, obj_subtype, obj_id = child.split(':', 2)
@ -180,24 +195,12 @@ class Message(AbstractObject):
# message media # message media
# flag is deleted -> event or missing from feeder pass ??? # flag is deleted -> event or missing from feeder pass ???
def get_translation(self, content=None, source=None, target='fr'): def get_language(self):
""" languages = self.get_languages()
Returns translated content if languages:
""" return languages.pop()
else:
# return self._get_field('translated') return None
global_id = self.get_global_id()
translation = r_cache.get(f'translation:{target}:{global_id}')
r_cache.expire(f'translation:{target}:{global_id}', 0)
if translation:
return translation
if not content:
content = self.get_content()
translation = Language.LanguageTranslator().translate(content, source=source, target=target)
if translation:
r_cache.set(f'translation:{target}:{global_id}', translation)
r_cache.expire(f'translation:{target}:{global_id}', 300)
return translation
def _set_translation(self, translation): def _set_translation(self, translation):
""" """
@ -296,8 +299,16 @@ class Message(AbstractObject):
meta['files-names'] = self.get_files_names() meta['files-names'] = self.get_files_names()
if 'reactions' in options: if 'reactions' in options:
meta['reactions'] = self.get_reactions() meta['reactions'] = self.get_reactions()
if 'language' in options:
meta['language'] = self.get_language()
if 'translation' in options and translation_target: if 'translation' in options and translation_target:
meta['translation'] = self.translate(content=meta.get('content'), target=translation_target) if meta.get('language'):
source = meta['language']
else:
source = None
meta['translation'] = self.translate(content=meta.get('content'), source=source, target=translation_target)
if 'language' in options:
meta['language'] = self.get_language()
# meta['encoding'] = None # meta['encoding'] = None
return meta return meta
@ -311,11 +322,29 @@ class Message(AbstractObject):
# self._set_translation(translated) # self._set_translation(translated)
# return translated # return translated
def create(self, content, translation=None, tags=[]): ## Language ##
def get_objs_container(self):
objs_containers = set()
# chat
objs_containers.add(self.get_chat())
subchannel = self.get_subchannel()
if subchannel:
objs_containers.add(subchannel)
thread = self.get_current_thread()
if thread:
objs_containers.add(thread)
return objs_containers
#- Language -#
def create(self, content, language=None, translation=None, tags=[]):
self._set_field('content', content) self._set_field('content', content)
# r_content.get(f'content:{self.type}:{self.get_subtype(r_str=True)}:{self.id}', content) if not language and content:
if translation: language = self.detect_language()
if translation and content:
self._set_translation(translation) self._set_translation(translation)
self.set_translation(language, translation)
for tag in tags: for tag in tags:
self.add_tag(tag) self.add_tag(tag)

View File

@ -226,7 +226,7 @@ class AbstractChatObject(AbstractSubtypeObject, ABC):
def get_message_meta(self, message, timestamp=None, translation_target='', options=None): # TODO handle file message def get_message_meta(self, message, timestamp=None, translation_target='', options=None): # TODO handle file message
message = Messages.Message(message[9:]) message = Messages.Message(message[9:])
if not options: if not options:
options = {'content', 'files-names', 'images', 'link', 'parent', 'parent_meta', 'reactions', 'thread', 'translation', 'user-account'} options = {'content', 'files-names', 'images', 'language', 'link', 'parent', 'parent_meta', 'reactions', 'thread', 'translation', 'user-account'}
meta = message.get_meta(options=options, timestamp=timestamp, translation_target=translation_target) meta = message.get_meta(options=options, timestamp=timestamp, translation_target=translation_target)
return meta return meta

View File

@ -25,7 +25,7 @@ from lib import Duplicate
from lib.correlations_engine import get_nb_correlations, get_correlations, add_obj_correlation, delete_obj_correlation, delete_obj_correlations, exists_obj_correlation, is_obj_correlated, get_nb_correlation_by_correl_type, get_obj_inter_correlation from lib.correlations_engine import get_nb_correlations, get_correlations, add_obj_correlation, delete_obj_correlation, delete_obj_correlations, exists_obj_correlation, is_obj_correlated, get_nb_correlation_by_correl_type, get_obj_inter_correlation
from lib.Investigations import is_object_investigated, get_obj_investigations, delete_obj_investigations from lib.Investigations import is_object_investigated, get_obj_investigations, delete_obj_investigations
from lib.relationships_engine import get_obj_nb_relationships, add_obj_relationship from lib.relationships_engine import get_obj_nb_relationships, add_obj_relationship
from lib.Language import get_obj_languages, add_obj_language, remove_obj_language, get_obj_translation, set_obj_translation from lib.Language import get_obj_languages, add_obj_language, remove_obj_language, detect_obj_language, get_obj_language_stats, get_obj_translation, set_obj_translation, delete_obj_translation
from lib.Tracker import is_obj_tracked, get_obj_trackers, delete_obj_trackers from lib.Tracker import is_obj_tracked, get_obj_trackers, delete_obj_trackers
logging.config.dictConfig(ail_logger.get_config(name='ail')) logging.config.dictConfig(ail_logger.get_config(name='ail'))
@ -305,28 +305,45 @@ class AbstractObject(ABC):
## -Relationship- ## ## -Relationship- ##
def get_objs_container(self):
return set()
## Language ## ## Language ##
def get_languages(self): def get_languages(self):
return get_obj_languages(self.type, self.get_subtype(r_str=True), self.id) return get_obj_languages(self.type, self.get_subtype(r_str=True), self.id)
def add_language(self, language): def add_language(self, language):
return add_obj_language(language, self.type, self.get_subtype(r_str=True), self.id) return add_obj_language(language, self.type, self.get_subtype(r_str=True), self.id, objs_containers=self.get_objs_container())
def remove_language(self, language): def remove_language(self, language):
return remove_obj_language(language, self.type, self.get_subtype(r_str=True), self.id) return remove_obj_language(language, self.type, self.get_subtype(r_str=True), self.id, objs_containers=self.get_objs_container())
def edit_language(self, old_language, new_language):
if old_language:
self.remove_language(old_language)
self.add_language(new_language)
def detect_language(self, field=''):
return detect_obj_language(self.type, self.get_subtype(r_str=True), self.id, self.get_content(), objs_containers=self.get_objs_container())
def get_obj_language_stats(self):
return get_obj_language_stats(self.type, self.get_subtype(r_str=True), self.id)
def get_translation(self, language, field=''): def get_translation(self, language, field=''):
return get_obj_translation(self.get_global_id(), language, field=field) return get_obj_translation(self.get_global_id(), language, field=field, objs_containers=self.get_objs_container())
def set_translation(self, language, translation, field=''): def set_translation(self, language, translation, field=''):
return set_obj_translation(self.get_global_id(), language, translation, field=field) return set_obj_translation(self.get_global_id(), language, translation, field=field)
def delete_translation(self, language, field=''):
return delete_obj_translation(self.get_global_id(), language, field=field)
def translate(self, content=None, field='', source=None, target='en'): def translate(self, content=None, field='', source=None, target='en'):
global_id = self.get_global_id() global_id = self.get_global_id()
if not content: if not content:
content = self.get_content() content = self.get_content()
translation = get_obj_translation(global_id, target, source=source, content=content, field=field) translation = get_obj_translation(global_id, target, source=source, content=content, field=field, objs_containers=self.get_objs_container())
return translation return translation
## -Language- ## ## -Language- ##

View File

@ -33,6 +33,10 @@ class Languages(AbstractModule):
for lang in obj.get_languages(min_probability=0.8, force_gcld3=True): for lang in obj.get_languages(min_probability=0.8, force_gcld3=True):
print(lang) print(lang)
domain.add_language(lang) domain.add_language(lang)
# Detect Chat Message Language
# elif obj.type == 'message':
# lang = obj.detect_language()
# print(self.obj.id, lang)
if __name__ == '__main__': if __name__ == '__main__':

View File

@ -20,17 +20,39 @@ from lib.ail_core import is_object_type
from lib import ail_queues from lib import ail_queues
from lib.objects import ail_objects from lib.objects import ail_objects
def reprocess_message_objects(object_type): # from modules.ApiKey import ApiKey
queue = ail_queues.AILQueue('FeederModuleImporter', -1) # from modules.Categ import Categ
for obj in ail_objects.obj_iterator(object_type, filters={}): # from modules.CreditCards import CreditCards
queue.send_message(obj.get_global_id(), message='reprocess') # from modules.DomClassifier import DomClassifier
queue.end() # from modules.Global import Global
# from modules.Keys import Keys
# from modules.Onion import Onion
# from modules.Telegram import Telegram
from modules.Languages import Languages
MODULES = {
'Languages': Languages
}
def reprocess_message_objects(object_type, module_name=None):
if module_name:
module = MODULES[module_name]()
for obj in ail_objects.obj_iterator(object_type, filters={}):
module.obj = obj
module.compute(None)
else:
queue = ail_queues.AILQueue('FeederModuleImporter', -1)
for obj in ail_objects.obj_iterator(object_type, filters={}):
queue.send_message(obj.get_global_id(), message='reprocess')
queue.end()
if __name__ == "__main__": if __name__ == "__main__":
parser = argparse.ArgumentParser(description='Reprocess AIL Objects') parser = argparse.ArgumentParser(description='Reprocess AIL Objects')
parser.add_argument('-t', '--type', type=str, help='AIL Object Type', required=True) parser.add_argument('-t', '--type', type=str, help='AIL Object Type', required=True)
parser.add_argument('-m', '--module', type=str, help='AIL Module Name')
args = parser.parse_args() args = parser.parse_args()
if not args.type: if not args.type:
@ -43,4 +65,7 @@ if __name__ == "__main__":
if obj_type not in ['item', 'message']: # TODO image if obj_type not in ['item', 'message']: # TODO image
raise Exception(f'Currently not supported Object Type: {obj_type}') raise Exception(f'Currently not supported Object Type: {obj_type}')
reprocess_message_objects(obj_type) modulename = args.module
if modulename not in MODULES:
raise Exception(f'Currently not supported Module: {modulename}')
reprocess_message_objects(obj_type, module_name=modulename)

View File

@ -10,6 +10,7 @@ sys.path.append(os.environ['AIL_HOME'])
################################## ##################################
from update.bin.ail_updater import AIL_Updater from update.bin.ail_updater import AIL_Updater
from lib import ail_updates from lib import ail_updates
from lib import chats_viewer
class Updater(AIL_Updater): class Updater(AIL_Updater):
"""default Updater.""" """default Updater."""
@ -19,6 +20,7 @@ class Updater(AIL_Updater):
if __name__ == '__main__': if __name__ == '__main__':
chats_viewer.fix_correlations_subchannel_message()
updater = Updater('v5.4') updater = Updater('v5.4')
updater.run_update() updater.run_update()

View File

@ -23,6 +23,7 @@ git submodule update
echo "" echo ""
echo -e $GREEN"Updating python packages ..."$DEFAULT echo -e $GREEN"Updating python packages ..."$DEFAULT
echo "" echo ""
pip install -U pyail
pip install -U pylacus pip install -U pylacus
pip install -U git+https://github.com/ail-project/demoji pip install -U git+https://github.com/ail-project/demoji
pip install -U lexilang pip install -U lexilang

View File

@ -244,15 +244,34 @@ def objects_message():
@login_read_only @login_read_only
def objects_message_translate(): def objects_message_translate():
message_id = request.form.get('id') message_id = request.form.get('id')
source = request.form.get('language_target')
target = request.form.get('target') target = request.form.get('target')
translation = request.form.get('translation') translation = request.form.get('translation')
if target == "Don't Translate": if target == "Don't Translate":
target = None target = None
resp = chats_viewer.api_manually_translate_message(message_id, target, translation) resp = chats_viewer.api_manually_translate_message(message_id, source, target, translation)
if resp[1] != 200: if resp[1] != 200:
return create_json_response(resp[0], resp[1]) return create_json_response(resp[0], resp[1])
else: else:
return redirect(url_for('chats_explorer.objects_message', id=message_id, target=target)) if request.referrer:
return redirect(request.referrer)
else:
return redirect(url_for('chats_explorer.objects_message', id=message_id, target=target))
@chats_explorer.route("/objects/message/detect/language", methods=['GET'])
@login_required
@login_read_only
def objects_message_detect_language():
message_id = request.args.get('id')
target = request.args.get('target')
resp = chats_viewer.api_message_detect_language(message_id)
if resp[1] != 200:
return create_json_response(resp[0], resp[1])
else:
if request.referrer:
return redirect(request.referrer)
else:
return redirect(url_for('chats_explorer.objects_message', id=message_id, target=target))
@chats_explorer.route("/objects/user-account", methods=['GET']) @chats_explorer.route("/objects/user-account", methods=['GET'])
@login_required @login_required

View File

@ -81,24 +81,6 @@
<hr class="m-1"> <hr class="m-1">
<pre class="my-0 text-secondary">{{ message['translation'] }}</pre> <pre class="my-0 text-secondary">{{ message['translation'] }}</pre>
{% set mess_id_escape= message['id'] | replace("/", "_") %}
<button class="btn btn-light p-0" type="button" data-toggle="collapse" data-target="#collapseTrans{{ mess_id_escape }}" aria-expanded="false" aria-controls="collapseTrans{{ mess_id_escape }}">
<i class="fas fa-language"></i>
</button>
<div class="collapse" id="collapseTrans{{ mess_id_escape }}">
<div class="card card-body">
<form method="post" action="{{ url_for('chats_explorer.objects_message_translate') }}" target="_blank">
<input type="text" id="id" name="id" value="{{message['id']}}" hidden>
<input type="text" id="target" name="target" value="{{translation_target}}" hidden>
<span>{{translation_target}}:</span>
<textarea class="form-control" id="translation" name="translation">{{ message['translation'] }}</textarea>
<button class="btn btn-primary" type="submit">
<i class="fas fa-pen-alt"> Manual Translation</i>
</button>
</form>
</div>
</div>
{% endif %} {% endif %}
{% for reaction in message['reactions'] %} {% for reaction in message['reactions'] %}
<span class="border rounded px-1">{{ reaction }} {{ message['reactions'][reaction] }}</span> <span class="border rounded px-1">{{ reaction }} {{ message['reactions'][reaction] }}</span>
@ -113,10 +95,47 @@
<span class="badge badge-{{ bootstrap_label[loop.index0 % 5] }}">{{ tag }}</span> <span class="badge badge-{{ bootstrap_label[loop.index0 % 5] }}">{{ tag }}</span>
{% endfor %} {% endfor %}
<div class=""> <div class="">
{% set mess_id_escape= message['id'] | replace("/", "_") %}
<span class="btn btn-outline-dark p-0 px-1" type="button" data-toggle="collapse" data-target="#collapseTrans{{ mess_id_escape }}" aria-expanded="false" aria-controls="collapseTrans{{ mess_id_escape }}">
<i class="fas fa-language"></i> {% if message['language'] %}{{ message['language'] }}{% endif %}
</span>
<div class="collapse" id="collapseTrans{{ mess_id_escape }}">
<div class="card card-body">
<form method="post" action="{{ url_for('chats_explorer.objects_message_translate') }}">
<input type="text" id="id" name="id" value="{{message['id']}}" hidden>
<span class="badge badge-primary">Source:</span>
<span class="">
<select id="language_target" name="language_target" class="form-select" aria-label="Message Language" onchange="$('#translation').val('');">
<option selected value="{{ message['language'] }}">{{ message['language'] }}</option>
{% for language in translation_languages %}
<option value="{{ language }}">{{ translation_languages[language] }}</option>
{% endfor %}
</select>
</span>
{% if translation_target %}
<input type="text" id="target" name="target" value="{{translation_target}}" hidden>
&nbsp;&nbsp;&nbsp;&nbsp;<span class="badge badge-primary">Target:</span><span>{{translation_target}}</span>
<textarea class="form-control" id="translation" name="translation">{{ message['translation'] }}</textarea>
<button class="btn btn-dark" type="submit">
<i class="fas fa-pen-alt"> Update Language or Translation</i>
</button>
{% else %}
<button class="btn btn-dark" type="submit">
<i class="fas fa-pen-alt"> Update Language</i>
</button>
{% endif %}
</form>
<div>
<a class="btn btn-primary" href="{{ url_for('chats_explorer.objects_message_detect_language')}}?id={{ message['id'] }}">
<i class="fas fa-redo"></i> Detect Language
</a>
</div>
</div>
</div>
<a class="btn btn-light btn-sm text-secondary px-1" href="{{ url_for('correlation.show_correlation')}}?type={{ message['type'] }}&subtype={{ message['subtype'] }}&id={{ message['id'] }}"><i class="fas fa-project-diagram"></i></a> <a class="btn btn-light btn-sm text-secondary px-1" href="{{ url_for('correlation.show_correlation')}}?type={{ message['type'] }}&subtype={{ message['subtype'] }}&id={{ message['id'] }}"><i class="fas fa-project-diagram"></i></a>
<a class="btn btn-light btn-sm text-secondary px-1" href="{{ message['link'] }}"><i class="fas fa-eye"></i></a> <a class="btn btn-light btn-sm text-secondary px-1" href="{{ message['link'] }}"><i class="fas fa-eye"></i></a>
</div> </div>
</div> </div>
</div> </div>

View File

@ -10,6 +10,7 @@
<option selected>Don't Translate</option> <option selected>Don't Translate</option>
{% else %} {% else %}
<option selected value="{{ translation_target }}">{{ translation_target }}</option> <option selected value="{{ translation_target }}">{{ translation_target }}</option>
<option>Don't Translate</option>
{% endif %} {% endif %}
{% for language in translation_languages %} {% for language in translation_languages %}
<option value="{{ language }}">{{ translation_languages[language] }}</option> <option value="{{ language }}">{{ translation_languages[language] }}</option>