chg: [language messages] add nb languages stats by chat/subchannel objects

This commit is contained in:
terrtia 2024-03-25 14:13:16 +01:00
parent 59ca8c5d31
commit b9c37167ad
No known key found for this signature in database
GPG key ID: 1E1B1F50D84613D0
6 changed files with 87 additions and 48 deletions

View file

@ -324,24 +324,32 @@ 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) delete_obj_translation(obj_global_id, language)
@ -353,27 +361,33 @@ def remove_obj_language(language, obj_type, obj_subtype, obj_id):
if r_lang.scard(f'objs:langs:{obj_type}') <= 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)
if rem:
for global_id in objs_containers:
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 # TODO handle fields
def detect_obj_language(obj_type, obj_subtype, obj_id, content): def detect_obj_language(obj_type, obj_subtype, obj_id, content, objs_containers=set()):
detector = LanguagesDetector(nb_langs=1) detector = LanguagesDetector(nb_langs=1)
language = detector.detect(content) language = detector.detect(content)
if language: if language:
language = language[0] language = language[0]
previous_lang = get_obj_languages(obj_type, obj_subtype, obj_id) previous_lang = get_obj_languages(obj_type, obj_subtype, obj_id)
if previous_lang: if previous_lang:
previous_lang = previous_lang[0] previous_lang = previous_lang.pop()
if language != previous_lang: if language != previous_lang:
remove_obj_language(language, obj_type, obj_subtype, obj_id) 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) add_obj_language(language, obj_type, obj_subtype, obj_id, objs_containers=objs_containers)
else: else:
add_obj_language(language, obj_type, obj_subtype, obj_id) add_obj_language(language, obj_type, obj_subtype, obj_id, objs_containers=objs_containers)
return language 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
""" """
@ -385,17 +399,20 @@ def get_obj_translation(obj_global_id, language, source=None, content=None, fiel
# 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:
source, translation = LanguageTranslator().translate(content, source=source, target=language) source, translation = LanguageTranslator().translate(content, source=source, target=language)
if source and translation: if source and translation:
obj_type, subtype, obj_id = obj_global_id.split(':', 2) obj_type, subtype, obj_id = obj_global_id.split(':', 2)
add_obj_language(source, obj_type, subtype, obj_id) 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 ????

View file

@ -480,7 +480,6 @@ def api_manually_translate_message(message_id, source, translation_target, trans
return {"status": "error", "reason": "Max Size reached"}, 400 return {"status": "error", "reason": "Max Size reached"}, 400
all_languages = Language.get_translation_languages() all_languages = Language.get_translation_languages()
if source not in all_languages: if source not in all_languages:
print(source)
return {"status": "error", "reason": "Unknown source Language"}, 400 return {"status": "error", "reason": "Unknown source Language"}, 400
message_language = message.get_language() message_language = message.get_language()
if message_language != source: if message_language != source:

View file

@ -103,8 +103,17 @@ 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'user-account:{subchannel["chat-subchannel"].pop()}'
def get_thread(self): def get_thread(self):
for child in self.get_childrens(): for child in self.get_childrens():
@ -183,25 +192,6 @@ class Message(AbstractObject):
else: else:
return None return None
def get_translation(self, content=None, source=None, target='fr'):
"""
Returns translated content
"""
# return self._get_field('translated')
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):
""" """
Set translated content Set translated content
@ -305,6 +295,8 @@ class Message(AbstractObject):
else: else:
source = None source = None
meta['translation'] = self.translate(content=meta.get('content'), source=source, target=translation_target) 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
@ -318,11 +310,29 @@ 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_thread() # TODO 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

@ -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, detect_obj_language, get_obj_translation, set_obj_translation, delete_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'))
@ -302,26 +302,33 @@ 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): def edit_language(self, old_language, new_language):
if old_language:
self.remove_language(old_language) self.remove_language(old_language)
self.add_language(new_language) self.add_language(new_language)
def detect_language(self, field=''): def detect_language(self, field=''):
return detect_obj_language(self.type, self.get_subtype(r_str=True), self.id, self.get_content()) 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)
@ -333,7 +340,7 @@ class AbstractObject(ABC):
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

@ -252,6 +252,9 @@ def objects_message_translate():
resp = chats_viewer.api_manually_translate_message(message_id, source, 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:
if request.referrer:
return redirect(request.referrer)
else: else:
return redirect(url_for('chats_explorer.objects_message', id=message_id, target=target)) return redirect(url_for('chats_explorer.objects_message', id=message_id, target=target))
@ -264,6 +267,9 @@ def objects_message_detect_language():
resp = chats_viewer.api_message_detect_language(message_id) resp = chats_viewer.api_message_detect_language(message_id)
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:
if request.referrer:
return redirect(request.referrer)
else: else:
return redirect(url_for('chats_explorer.objects_message', id=message_id, target=target)) return redirect(url_for('chats_explorer.objects_message', id=message_id, target=target))

View file

@ -102,7 +102,7 @@
</span> </span>
<div class="collapse" id="collapseTrans{{ mess_id_escape }}"> <div class="collapse" id="collapseTrans{{ mess_id_escape }}">
<div class="card card-body"> <div class="card card-body">
<form method="post" action="{{ url_for('chats_explorer.objects_message_translate') }}" target="_blank"> <form method="post" action="{{ url_for('chats_explorer.objects_message_translate') }}">
<input type="text" id="id" name="id" value="{{message['id']}}" hidden> <input type="text" id="id" name="id" value="{{message['id']}}" hidden>
<span class="badge badge-primary">Source:</span> <span class="badge badge-primary">Source:</span>
<span class=""> <span class="">