@ -2,10 +2,8 @@ import asyncio
import logging
import os
import pathlib
from collections import defaultdict
from datetime import datetime , timedelta
from functools import partial
from typing import Dict , List , Optional
from typing import Optional
import discord
from chatterbot import ChatBot
@ -17,9 +15,6 @@ from redbot.core.commands import Cog
from redbot . core . data_manager import cog_data_path
from redbot . core . utils . predicates import MessagePredicate
from chatter . trainers import MovieTrainer , TwitterCorpusTrainer , UbuntuCorpusTrainer2
chatterbot_log = logging . getLogger ( " red.fox_v3.chatterbot " )
log = logging . getLogger ( " red.fox_v3.chatter " )
@ -30,12 +25,6 @@ def my_local_get_prefix(prefixes, content):
return None
class ENG_TRF :
ISO_639_1 = " en_core_web_trf "
ISO_639 = " eng "
ENGLISH_NAME = " English "
class ENG_LG :
ISO_639_1 = " en_core_web_lg "
ISO_639 = " eng "
@ -59,77 +48,50 @@ class Chatter(Cog):
This cog trains a chatbot that will talk like members of your Guild
"""
models = [ ENG_SM , ENG_MD , ENG_LG , ENG_TRF ]
algos = [ SpacySimilarity , JaccardSimilarity , LevenshteinDistance ]
def __init__ ( self , bot ) :
super ( ) . __init__ ( )
self . bot = bot
self . config = Config . get_conf ( self , identifier = 6710497116116101114 )
default_global = { " learning " : True , " model_number " : 0 , " algo_number " : 0 , " threshold " : 0.90 }
self . default_guild = {
" whitelist " : None ,
" days " : 1 ,
" convo_delta " : 15 ,
" chatchannel " : None ,
" reply " : True ,
}
default_global = { }
default_guild = { " whitelist " : None , " days " : 1 , " convo_delta " : 15 , " chatchannel " : None }
path : pathlib . Path = cog_data_path ( self )
self . data_path = path / " database.sqlite3 "
# TODO: Move training_model and similarity_algo to config
# TODO: Add an option to see current settings
self . tagger_language = ENG_ S M
self . tagger_language = ENG_ MD
self . similarity_algo = SpacySimilarity
self . similarity_threshold = 0.90
self . chatbot = None
self . chatbot = self . _create_chatbot ( )
# self.chatbot.set_trainer(ListTrainer)
# self.trainer = ListTrainer(self.chatbot)
self . config . register_global ( * * default_global )
self . config . register_guild ( * * self . default_guild )
self . config . register_guild ( * * default_guild )
self . loop = asyncio . get_event_loop ( )
self . _guild_cache = defaultdict ( dict )
self . _global_cache = { }
self . _last_message_per_channel : Dict [ Optional [ discord . Message ] ] = defaultdict ( lambda : None )
async def red_delete_data_for_user ( self , * * kwargs ) :
""" Nothing to delete """
return
async def initialize ( self ) :
all_config = dict ( self . config . defaults [ " GLOBAL " ] )
all_config . update ( await self . config . all ( ) )
model_number = all_config [ " model_number " ]
algo_number = all_config [ " algo_number " ]
threshold = all_config [ " threshold " ]
self . tagger_language = self . models [ model_number ]
self . similarity_algo = self . algos [ algo_number ]
self . similarity_threshold = threshold
self . chatbot = self . _create_chatbot ( )
def _create_chatbot ( self ) :
return ChatBot (
" ChatterBot " ,
# storage_adapter="chatterbot.storage.SQLStorageAdapter",
storage_adapter = " chatter.storage_adapters.MyDumbSQLStorageAdapter " ,
storage_adapter = " chatterbot.storage.SQLStorageAdapter " ,
database_uri = " sqlite:/// " + str ( self . data_path ) ,
statement_comparison_function = self . similarity_algo ,
response_selection_method = get_random_response ,
logic_adapters = [ " chatterbot.logic.BestMatch " ] ,
maximum_similarity_threshold = self . similarity_threshold ,
tagger_language = self . tagger_language ,
logger = chatterbot_ log,
logger = log ,
)
async def _get_conversation ( self , ctx , in_channel s: List [ discord . TextChannel ] ) :
async def _get_conversation ( self , ctx , in_channel : discord . TextChannel = None ) :
"""
Compiles all conversation in the Guild this bot can get it ' s hands on
Currently takes a stupid long time
@ -143,12 +105,20 @@ class Chatter(Cog):
return msg . clean_content
def new_conversation ( msg , sent , out_in , delta ) :
# Should always be positive numbers
# if sent is None:
# return False
# Don't do "too short" processing here. Sometimes people don't respond.
# if len(out_in) < 2:
# return False
# print(msg.created_at - sent)
return msg . created_at - sent > = delta
for channel in in_channels :
# if in_channel:
# channel = in_channel
for channel in ctx. guild . text _channels:
if in_channel
await ctx . maybe_send_embed ( " Gathering {} " . format ( channel . mention ) )
user = None
i = 0
@ -183,16 +153,11 @@ class Chatter(Cog):
except discord . HTTPException :
pass
# if in_channel
#
if in_channel
return out
def _train_twitter ( self , * args , * * kwargs ) :
trainer = TwitterCorpusTrainer ( self . chatbot )
trainer . train ( * args , * * kwargs )
return True
def _train_ubuntu ( self ) :
trainer = UbuntuCorpusTrainer (
self . chatbot , ubuntu_corpus_data_directory = cog_data_path ( self ) / " ubuntu_data "
@ -200,30 +165,6 @@ class Chatter(Cog):
trainer . train ( )
return True
async def _train_movies ( self ) :
trainer = MovieTrainer ( self . chatbot , cog_data_path ( self ) )
return await trainer . asynctrain ( )
async def _train_ubuntu2 ( self , intensity ) :
train_kwarg = { }
if intensity == 196 :
train_kwarg [ " train_dialogue " ] = False
train_kwarg [ " train_196 " ] = True
elif intensity == 301 :
train_kwarg [ " train_dialogue " ] = False
train_kwarg [ " train_301 " ] = True
elif intensity == 497 :
train_kwarg [ " train_dialogue " ] = False
train_kwarg [ " train_196 " ] = True
train_kwarg [ " train_301 " ] = True
elif intensity > = 9000 : # NOT 9000!
train_kwarg [ " train_dialogue " ] = True
train_kwarg [ " train_196 " ] = True
train_kwarg [ " train_301 " ] = True
trainer = UbuntuCorpusTrainer2 ( self . chatbot , cog_data_path ( self ) )
return await trainer . asynctrain ( * * train_kwarg )
def _train_english ( self ) :
trainer = ChatterBotCorpusTrainer ( self . chatbot )
# try:
@ -235,10 +176,13 @@ class Chatter(Cog):
def _train ( self , data ) :
trainer = ListTrainer ( self . chatbot )
total = len ( data )
# try:
for c , convo in enumerate ( data , 1 ) :
log . info ( f " { c } / { total } " )
if len ( convo ) > 1 : # TODO: Toggleable skipping short conversations
print ( f " { c } / { total } " )
trainer . train ( convo )
# except:
# return False
return True
@commands.group ( invoke_without_command = False )
@ -246,10 +190,10 @@ class Chatter(Cog):
"""
Base command for this cog . Check help for the commands list .
"""
self . _guild_cache [ ctx . guild . id ] = { } # Clear cache when modifying values
self . _global_cache = { }
if ctx . invoked_subcommand is None :
pass
@c ommand s.admin( )
@c heck s.admin( )
@chatter.command ( name = " channel " )
async def chatter_channel (
self , ctx : commands . Context , channel : Optional [ discord . TextChannel ] = None
@ -269,55 +213,13 @@ class Chatter(Cog):
await self . config . guild ( ctx . guild ) . chatchannel . set ( channel . id )
await ctx . maybe_send_embed ( f " Chat channel is now { channel . mention } " )
@commands.admin ( )
@chatter.command ( name = " reply " )
async def chatter_reply ( self , ctx : commands . Context , toggle : Optional [ bool ] = None ) :
"""
Toggle bot reply to messages if conversation continuity is not present
"""
reply = await self . config . guild ( ctx . guild ) . reply ( )
if toggle is None :
toggle = not reply
await self . config . guild ( ctx . guild ) . reply . set ( toggle )
if toggle :
await ctx . maybe_send_embed (
" I will now respond to you if conversation continuity is not present "
)
else :
await ctx . maybe_send_embed (
" I will not reply to your message if conversation continuity is not present, anymore "
)
@commands.is_owner ( )
@chatter.command ( name = " learning " )
async def chatter_learning ( self , ctx : commands . Context , toggle : Optional [ bool ] = None ) :
"""
Toggle the bot learning from its conversations .
This is a global setting .
This is on by default .
"""
learning = await self . config . learning ( )
if toggle is None :
toggle = not learning
await self . config . learning . set ( toggle )
if toggle :
await ctx . maybe_send_embed ( " I will now learn from conversations. " )
else :
await ctx . maybe_send_embed ( " I will no longer learn from conversations. " )
@commands.is_owner ( )
@checks.is_owner ( )
@chatter.command ( name = " cleardata " )
async def chatter_cleardata ( self , ctx : commands . Context , confirm : bool = False ) :
"""
This command will erase all training data and reset your configuration settings .
This command will erase all training data and reset your configuration settings
This applies to all guilds .
Use ` [ p ] chatter cleardata True ` to confirm .
Use ` [ p ] chatter cleardata True `
"""
if not confirm :
@ -344,18 +246,21 @@ class Chatter(Cog):
await ctx . tick ( )
@c ommand s.is_owner( )
@c heck s.is_owner( )
@chatter.command ( name = " algorithm " , aliases = [ " algo " ] )
async def chatter_algorithm (
self , ctx : commands . Context , algo_number : int , threshold : float = None
) :
"""
Switch the active logic algorithm to one of the three . Default is Spacy
Switch the active logic algorithm to one of the three . Default after reload is Spacy
0 : Spacy
1 : Jaccard
2 : Levenshtein
"""
algos = [ SpacySimilarity , JaccardSimilarity , LevenshteinDistance ]
if algo_number < 0 or algo_number > 2 :
await ctx . send_help ( )
return
@ -368,32 +273,31 @@ class Chatter(Cog):
return
else :
self . similarity_threshold = threshold
await self . config . threshold . set ( self . similarity_threshold )
self . similarity_algo = self . algos [ algo_number ]
await self . config . algo_number . set ( algo_number )
self . similarity_algo = algos [ algo_number ]
async with ctx . typing ( ) :
self . chatbot = self . _create_chatbot ( )
await ctx . tick ( )
@c ommand s.is_owner( )
@c heck s.is_owner( )
@chatter.command ( name = " model " )
async def chatter_model ( self , ctx : commands . Context , model_number : int ) :
"""
Switch the active model to one of the three . Default is Small
Switch the active model to one of the three . Default after reload is Medium
0 : Small
1 : Medium ( Requires additional setup )
1 : Medium
2 : Large ( Requires additional setup )
3. Accurate ( Requires additional setup )
"""
if model_number < 0 or model_number > 3 :
models = [ ENG_SM , ENG_MD , ENG_LG ]
if model_number < 0 or model_number > 2 :
await ctx . send_help ( )
return
if model_number >= 0 :
if model_number == 2 :
await ctx . maybe_send_embed (
" Additional requirements needed. See guide before continuing. \n " " Continue? "
)
@ -406,8 +310,7 @@ class Chatter(Cog):
if not pred . result :
return
self . tagger_language = self . models [ model_number ]
await self . config . model_number . set ( model_number )
self . tagger_language = models [ model_number ]
async with ctx . typing ( ) :
self . chatbot = self . _create_chatbot ( )
@ -415,14 +318,8 @@ class Chatter(Cog):
f " Model has been switched to { self . tagger_language . ISO_639_1 } "
)
@commands.is_owner ( )
@chatter.group ( name = " trainset " )
async def chatter_trainset ( self , ctx : commands . Context ) :
""" Commands for configuring training """
pass
@commands.is_owner ( )
@chatter_trainset.command ( name = " minutes " )
@checks.is_owner ( )
@chatter.command ( name = " minutes " )
async def minutes ( self , ctx : commands . Context , minutes : int ) :
"""
Sets the number of minutes the bot will consider a break in a conversation during training
@ -433,12 +330,12 @@ class Chatter(Cog):
await ctx . send_help ( )
return
await self . config . guild ( ctx . guild ) . convo_ delta . set ( minutes )
await self . config . guild ( ctx . guild ) . convo_ length . set ( minutes )
await ctx . tick ( )
@c ommand s.is_owner( )
@chatter _trainset .command( name = " age " )
@c heck s.is_owner( )
@chatter .command( name = " age " )
async def age ( self , ctx : commands . Context , days : int ) :
"""
Sets the number of days to look back
@ -452,16 +349,7 @@ class Chatter(Cog):
await self . config . guild ( ctx . guild ) . days . set ( days )
await ctx . tick ( )
@commands.is_owner ( )
@chatter.command ( name = " kaggle " )
async def chatter_kaggle ( self , ctx : commands . Context ) :
""" Register with the kaggle API to download additional datasets for training """
if not await self . check_for_kaggle ( ) :
await ctx . maybe_send_embed (
" [Click here for instructions to setup the kaggle api](https://github.com/Kaggle/kaggle-api#api-credentials) "
)
@commands.is_owner ( )
@checks.is_owner ( )
@chatter.command ( name = " backup " )
async def backup ( self , ctx , backupname ) :
"""
@ -483,71 +371,8 @@ class Chatter(Cog):
else :
await ctx . maybe_send_embed ( " Error occurred :( " )
@commands.is_owner ( )
@chatter.group ( name = " train " )
async def chatter_train ( self , ctx : commands . Context ) :
""" Commands for training the bot """
pass
@chatter_train.group ( name = " kaggle " )
async def chatter_train_kaggle ( self , ctx : commands . Context ) :
"""
Base command for kaggle training sets .
See ` [ p ] chatter kaggle ` for details on how to enable this option
"""
pass
@chatter_train_kaggle.command ( name = " ubuntu " )
async def chatter_train_kaggle_ubuntu (
self , ctx : commands . Context , confirmation : bool = False , intensity = 0
) :
"""
WARNING : Large Download ! Trains the bot using * NEW * Ubuntu Dialog Corpus data .
"""
if not confirmation :
await ctx . maybe_send_embed (
" Warning: This command downloads ~800MB and is CPU intensive during training \n "
" If you ' re sure you want to continue, run `[p]chatter train kaggle ubuntu True` "
)
return
async with ctx . typing ( ) :
future = await self . _train_ubuntu2 ( intensity )
if future :
await ctx . maybe_send_embed ( " Training successful! " )
else :
await ctx . maybe_send_embed ( " Error occurred :( " )
@chatter_train_kaggle.command ( name = " movies " )
async def chatter_train_kaggle_movies ( self , ctx : commands . Context , confirmation : bool = False ) :
"""
WARNING : Language ! Trains the bot using Cornell University ' s " Movie Dialog Corpus " .
This training set contains dialog from a spread of movies with different MPAA .
This dialog includes racism , sexism , and any number of sensitive topics .
Use at your own risk .
"""
if not confirmation :
await ctx . maybe_send_embed (
" Warning: This command downloads ~29MB and is CPU intensive during training \n "
" If you ' re sure you want to continue, run `[p]chatter train kaggle movies True` "
)
return
async with ctx . typing ( ) :
future = await self . _train_movies ( )
if future :
await ctx . maybe_send_embed ( " Training successful! " )
else :
await ctx . maybe_send_embed ( " Error occurred :( " )
@chatter_train.command ( name = " ubuntu " )
@checks.is_owner ( )
@chatter.command ( name = " trainubuntu " )
async def chatter_train_ubuntu ( self , ctx : commands . Context , confirmation : bool = False ) :
"""
WARNING : Large Download ! Trains the bot using Ubuntu Dialog Corpus data .
@ -555,7 +380,7 @@ class Chatter(Cog):
if not confirmation :
await ctx . maybe_send_embed (
" Warning: This command downloads ~500MB and is CPU intensive during training\n "
" Warning: This command downloads ~500MB then eats your CPU for training\n "
" If you ' re sure you want to continue, run `[p]chatter trainubuntu True` "
)
return
@ -564,11 +389,12 @@ class Chatter(Cog):
future = await self . loop . run_in_executor ( None , self . _train_ubuntu )
if future :
await ctx . maybe_ send_embe d( " Training successful! " )
await ctx . ( " Training successful! " )
else :
await ctx . maybe_ send_embe d( " Error occurred :( " )
await ctx . ( " Error occurred :( " )
@chatter_train.command ( name = " english " )
@checks.is_owner ( )
@chatter.command ( name = " trainenglish " )
async def chatter_train_english ( self , ctx : commands . Context ) :
"""
Trains the bot in english
@ -581,32 +407,12 @@ class Chatter(Cog):
else :
await ctx . maybe_send_embed ( " Error occurred :( " )
@chatter_train.command ( name = " list " )
async def chatter_train_list ( self , ctx : commands . Context ) :
""" Trains the bot based on an uploaded list.
Must be a file in the format of a python list : [ ' prompt ' , ' response1 ' , ' response2 ' ]
@checks.is_owner ( )
@chatter.command ( )
async def train ( self , ctx : commands . Context , channel : discord . TextChannel ) :
"""
if not ctx . message . attachments :
await ctx . maybe_send_embed ( " You must upload a file when using this command " )
return
attachment : discord . Attachment = ctx . message . attachments [ 0 ]
a_bytes = await attachment . read ( )
await ctx . send ( " Not yet implemented " )
@chatter_train.command ( name = " channel " )
async def chatter_train_channel (
self , ctx : commands . Context , channels : commands . Greedy [ discord . TextChannel ]
) :
"""
Trains the bot based on language in this guild .
Trains the bot based on language in this guild
"""
if not channels :
await ctx . send_help ( )
return
await ctx . maybe_send_embed (
" Warning: The cog may use significant RAM or CPU if trained on large data sets. \n "
@ -615,7 +421,7 @@ class Chatter(Cog):
)
async with ctx . typing ( ) :
conversation = await self . _get_conversation ( ctx , channel s )
conversation = await self . _get_conversation ( ctx , channel )
if not conversation :
await ctx . maybe_send_embed ( " Failed to gather training data " )
@ -657,7 +463,7 @@ class Chatter(Cog):
guild : discord . Guild = getattr ( message , " guild " , None )
if guild is None or await self . bot . cog_disabled_in_guild ( self , guild ) :
if await self . bot . cog_disabled_in_guild ( self , guild ) :
return
ctx : commands . Context = await self . bot . get_context ( message )
@ -669,18 +475,7 @@ class Chatter(Cog):
# Thank you Cog-Creators
channel : discord . TextChannel = message . channel
if not self . _guild_cache [ guild . id ] :
self . _guild_cache [ guild . id ] = await self . config . guild ( guild ) . all ( )
is_reply = False # this is only useful with in_response_to
if (
message . reference is not None
and isinstance ( message . reference . resolved , discord . Message )
and message . reference . resolved . author . id == self . bot . user . id
) :
is_reply = True # this is only useful with in_response_to
pass # this is a reply to the bot, good to go
elif guild is not None and channel . id == self . _guild_cache [ guild . id ] [ " chatchannel " ] :
if guild is not None and channel . id == await self . config . guild ( guild ) . chatchannel ( ) :
pass # good to go
else :
when_mentionables = commands . when_mentioned ( self . bot , message )
@ -695,57 +490,10 @@ class Chatter(Cog):
text = message . clean_content
async with ctx . typing ( ) :
if is_reply :
in_response_to = message . reference . resolved . content
elif self . _last_message_per_channel [ ctx . channel . id ] is not None :
last_m : discord . Message = self . _last_message_per_channel [ ctx . channel . id ]
minutes = self . _guild_cache [ ctx . guild . id ] [ " convo_delta " ]
if ( datetime . utcnow ( ) - last_m . created_at ) . seconds > minutes * 60 :
in_response_to = None
else :
in_response_to = last_m . content
else :
in_response_to = None
# Always use generate reponse
# Chatterbot tries to learn based on the result it comes up with, which is dumb
log . debug ( " Generating response " )
Statement = self . chatbot . storage . get_object ( " statement " )
future = await self . loop . run_in_executor (
None , self . chatbot . generate_response , Statement ( text )
)
if not self . _global_cache :
self . _global_cache = await self . config . all ( )
if in_response_to is not None and self . _global_cache [ " learning " ] :
log . debug ( " learning response " )
await self . loop . run_in_executor (
None ,
partial (
self . chatbot . learn_response ,
Statement ( text ) ,
previous_statement = in_response_to ,
) ,
)
replying = None
if (
" reply " not in self . _guild_cache [ guild . id ] and self . default_guild [ " reply " ]
) or self . _guild_cache [ guild . id ] [ " reply " ] :
if message != ctx . channel . last_message :
replying = message
async with channel . typing ( ) :
future = await self . loop . run_in_executor ( None , self . chatbot . get_response , text )
if future and str ( future ) :
self . _last_message_per_channel [ ctx . channel . id ] = await channel . send (
str ( future ) , reference = replying
)
await channel . send ( str ( future ) )
else :
await ctx . send ( " :thinking: " )
async def check_for_kaggle ( self ) :
""" Check whether Kaggle is installed and configured properly """
# TODO: This
return False
await channel . send ( " :thinking: " )
bobloy