File size: 9,472 Bytes
8d1c4bc 3340b1c 4e9f5d4 8d1c4bc 879e7eb 3340b1c 8d1c4bc 6debdfd 9ff5f83 3340b1c 6c9f8ba 3340b1c bc0f75e 879e7eb 8d1c4bc 4e9f5d4 5f0c401 f547fd9 5f0c401 2d11842 5f0c401 acb6ed9 3340b1c acb6ed9 3340b1c acb6ed9 bc0f75e 5f0c401 35c1a65 32d6875 5f0c401 32d6875 5f0c401 32d6875 5f0c401 32d6875 5f0c401 32d6875 fbfe5d9 32d6875 5f33166 32d6875 08860c0 a5501b7 bc0f75e 8d1c4bc 32d6875 5f33166 32d6875 bc0f75e 3340b1c 5f33166 32d6875 3340b1c 9a485d3 8d1c4bc 5910af2 3340b1c f547fd9 4e9f5d4 8d1c4bc 3340b1c 4e9f5d4 9a485d3 3340b1c 9a485d3 73dbfa8 50587b1 8910a27 9a485d3 4e9f5d4 2d11842 4e9f5d4 a5501b7 9fa3a58 d388d30 32d6875 6293d43 4e9f5d4 1d11933 6293d43 9a485d3 f547fd9 9e7c830 9a485d3 5910af2 9a485d3 fbfe5d9 bda1f51 32d6875 879e7eb bc0f75e 879e7eb bc0f75e 2d11842 bc0f75e 5f33166 bc0f75e 8910a27 bc0f75e 73dbfa8 bc0f75e 73dbfa8 bc0f75e 32d6875 9f4a9e1 bc0f75e 8d1c4bc bc0f75e 6895902 bc0f75e 3bac2a0 bc0f75e 3bac2a0 8d1c4bc 3bac2a0 bc0f75e f547fd9 bc0f75e f547fd9 bc0f75e 2d11842 f547fd9 bc0f75e f547fd9 bc0f75e 4e9f5d4 bc0f75e 2d11842 a5501b7 3340b1c 4e9f5d4 f547fd9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 |
import os
import json
import pickle
import random
import tempfile
import asyncio
import numpy as np
import redis
import uvicorn
import nltk
from nltk.stem import WordNetLemmatizer
from tensorflow.keras import Sequential
from tensorflow.keras.layers import Dense, Dropout, Input
from tensorflow.keras.optimizers import SGD
from tensorflow.keras.models import load_model, save_model
from fastapi import FastAPI
from fastapi.responses import HTMLResponse
from pydantic import BaseModel
from dotenv import load_dotenv
from datetime import datetime
import logging
from tqdm import tqdm # Importar tqdm
# Configuración de logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
load_dotenv()
app = FastAPI()
lemmatizer = WordNetLemmatizer()
redis_password = os.getenv("REDIS_PASSWORD")
r = redis.Redis(host=os.getenv("REDIS_HOST"), port=int(os.getenv("REDIS_PORT")), password=redis_password)
def create_intents_json():
intents = {
"intents": [
{
"tag": "greeting",
"patterns": ["Hola", "¿Cómo estás?", "Buenos días"],
"responses": ["¡Hola!", "¿Cómo puedo ayudarte?"],
"date": "2021-01-01"
},
{
"tag": "goodbye",
"patterns": ["Adiós", "Hasta luego", "Nos vemos"],
"responses": ["¡Hasta luego!", "Cuídate!"],
"date": "2021-01-01"
}
]
}
with open('intents.json', 'w') as f:
json.dump(intents, f, ensure_ascii=False, indent=4)
def load_and_filter_data():
with open("intents.json") as file:
intents = json.load(file)
filtered_intents = {"intents": []}
for intent in intents['intents']:
if "date" in intent:
intent_date = datetime.strptime(intent["date"], "%Y-%m-%d")
if intent_date.year >= 2000 and intent_date <= datetime.now():
filtered_intents['intents'].append(intent)
return filtered_intents
if not os.path.exists('models'):
os.makedirs('models')
async def train_and_save_model():
global lemmatizer, r
while True:
words, classes, documents = [], [], []
intents = load_and_filter_data()
user_questions = r.lrange('user_questions', 0, -1)
for question in user_questions:
question = question.decode('utf-8')
processed_words = nltk.word_tokenize(question)
documents.append((processed_words, "user_question"))
words.extend(processed_words)
for intent in intents['intents']:
for pattern in intent['patterns']:
processed_words = nltk.word_tokenize(pattern)
documents.append((processed_words, intent['tag']))
words.extend(processed_words)
if intent['tag'] not in classes:
classes.append(intent['tag'])
words = sorted(set(words))
classes = sorted(set(classes))
training = []
output_empty = [0] * len(classes)
for doc in tqdm(documents, desc="Preparando datos"):
bag = [0] * len(words)
pattern_words = [lemmatizer.lemmatize(word.lower()) for word in doc[0]]
for w in words:
bag[words.index(w)] = 1 if w in pattern_words else 0
output_row = list(output_empty)
output_row[classes.index(doc[1])] = 1
training.append([bag, output_row])
if not training:
await asyncio.sleep(60)
continue
train_x = np.array([row[0] for row in training])
train_y = np.array([row[1] for row in training])
if train_x.shape[1] != len(words):
logger.error(f"Error: Las dimensiones de entrada no coinciden. Esperado: {len(words)}, encontrado: {train_x.shape[1]}")
await asyncio.sleep(60)
continue
if r.exists('chatbot_model'):
with tempfile.NamedTemporaryFile(delete=False, suffix='.h5') as temp_file:
temp_file.write(r.get('chatbot_model'))
temp_file_name = temp_file.name
model = load_model(temp_file_name)
os.remove(temp_file.name)
else:
input_layer = Input(shape=(len(words),))
layer1 = Dense(128, activation='relu')(input_layer)
layer2 = Dropout(0.5)(layer1)
layer3 = Dense(64, activation='relu')(layer2)
layer4 = Dropout(0.5)(layer3)
output_layer = Dense(len(classes), activation='softmax')(layer4)
model = Sequential(layers=[input_layer, layer1, layer2, layer3, layer4, output_layer])
sgd = SGD(learning_rate=0.01, momentum=0.9, nesterov=True)
model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])
print("Entrenando el modelo...")
model.fit(train_x, train_y, epochs=1, batch_size=len(train_x), verbose=0)
r.set('words', pickle.dumps(words))
r.set('classes', pickle.dumps(classes))
with tempfile.NamedTemporaryFile(delete=False, suffix='.h5') as temp_file:
save_model(model, temp_file.name)
with open(temp_file.name, 'rb') as f:
r.set('chatbot_model', f.read())
os.remove(temp_file.name)
async def handle_new_message(message: str):
r.rpush('user_questions', message)
await train_and_save_model()
class ChatMessage(BaseModel):
message: str
@app.post("/chat")
async def chat(message: ChatMessage):
words = pickle.loads(r.get('words'))
classes = pickle.loads(r.get('classes'))
with tempfile.NamedTemporaryFile(delete=False) as temp_file:
temp_file.write(r.get('chatbot_model'))
temp_file_name = temp_file.name
model = load_model(temp_file_name)
sentence_words = nltk.word_tokenize(message.message)
bag = [0] * len(words)
for s in sentence_words:
for i, w in enumerate(words):
if w == s:
bag[i] = 1
p = model.predict(np.array([bag]))[0]
ERROR_THRESHOLD = 0.25
results = [[i, p] for i, p in enumerate(p) if p > ERROR_THRESHOLD]
results.sort(key=lambda x: x[1], reverse=True)
return_list = []
for i, p in results:
return_list.append({"intent": classes[i], "probability": str(p)})
await handle_new_message(message.message)
return return_list
@app.post("/tag")
async def tag_question(question: str, tag: str):
r.set(f"tag:{question}", tag)
return {"message": "Etiqueta guardada"}
html_code = """
<!DOCTYPE html>
<html>
<head>
<title>Chatbot</title>
<style>
body {
font-family: sans-serif;
background-color: #f4f4f4;
margin: 0;
padding: 0;
display: flex;
justify-content: center;
align-items: center;
min-height: 100vh;
}
#container {
background-color: #fff;
border-radius: 5px;
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.1);
padding: 30px;
width: 80%;
max-width: 600px;
}
h1 {
text-align: center;
margin-bottom: 20px;
color: #333;
}
#chatbox {
height: 300px;
overflow-y: auto;
padding: 10px;
border: 1px solid #ccc;
border-radius: 5px;
margin-bottom: 10px;
}
#chatbox p {
margin: 5px 0;
}
#user_input {
width: 100%;
padding: 10px;
border: 1px solid #ccc;
border-radius: 5px;
margin-bottom: 10px;
box-sizing: border-box;
}
button {
background-color: #4CAF50;
color: white;
padding: 10px 20px;
border: none;
border-radius: 5px;
cursor: pointer;
}
</style>
</head>
<body>
<div id="container">
<h1>Chatbot</h1>
<div id="chatbox"></div>
<input type="text" id="user_input" placeholder="Escribe tu mensaje...">
<button onclick="sendMessage()">Enviar</button>
</div>
<script>
function sendMessage() {
let userInput = document.getElementById('user_input').value;
document.getElementById('user_input').value = '';
document.getElementById('chatbox').innerHTML += '<p><b>Tú:</b> ' + userInput + '</p>';
fetch('/chat', {
method: 'POST',
headers: {
'Content-Type': 'application/json'
},
body: JSON.stringify({ message: userInput })
})
.then(response => response.json())
.then(data => {
data.forEach(response => {
document.getElementById('chatbox').innerHTML += '<p><b>Bot:</b> ' + response.intent + ' (' + response.probability + ')</p>';
});
document.getElementById('chatbox').scrollTop = document.getElementById('chatbox').scrollHeight;
});
}
</script>
</body>
</html>
"""
@app.get("/", response_class=HTMLResponse)
async def read_root():
return html_code
if __name__ == "__main__":
nltk.download('punkt')
nltk.download('wordnet')
nltk.download('omw-1.4')
nltk.download('averaged_perceptron_tagger')
nltk.download('punkt_tab')
uvicorn.run(app, host="0.0.0.0", port=8000)
|