File size: 11,427 Bytes
6293d43 acb6ed9 6895902 e298683 105e133 3340b1c 50587b1 6b03e9a 40c33c8 acb6ed9 3340b1c 50587b1 879e7eb a2a58e7 e298683 a5501b7 d388d30 aa18bae 3340b1c 6debdfd 9ff5f83 3340b1c 6c9f8ba acb6ed9 1e9c143 3340b1c bc0f75e 879e7eb acb6ed9 3340b1c acb6ed9 3340b1c acb6ed9 bc0f75e 9a485d3 bc0f75e 35c1a65 fbfe5d9 08860c0 a5501b7 6293d43 a674ac1 6293d43 50587b1 bc0f75e fbfe5d9 50587b1 6b03e9a 6293d43 a201f43 6293d43 ed21e94 a5501b7 bc0f75e 3340b1c 9a485d3 6b03e9a 40c33c8 e298683 ed21e94 3340b1c 35c1a65 9a485d3 5910af2 6b03e9a 3340b1c 9a485d3 3340b1c 9a485d3 3340b1c 9a485d3 73dbfa8 50587b1 8910a27 9a485d3 6b03e9a a5501b7 9fa3a58 d388d30 6293d43 1d11933 aa18bae 6293d43 9a485d3 6b03e9a a5501b7 9a485d3 6b03e9a 5910af2 9a485d3 fbfe5d9 bda1f51 6293d43 bc0f75e e298683 879e7eb bc0f75e 879e7eb bc0f75e a5501b7 9fa3a58 d388d30 bc0f75e 879e7eb bc0f75e 8910a27 bc0f75e 73dbfa8 bc0f75e 73dbfa8 bc0f75e 879e7eb bc0f75e d388d30 9f4a9e1 bc0f75e 6895902 bc0f75e 3bac2a0 bc0f75e 3bac2a0 bc0f75e a5501b7 3340b1c 6293d43 e298683 40c33c8 879e7eb |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 |
import uvicorn
import nltk
nltk.download('punkt')
nltk.download('wordnet')
nltk.download('omw-1.4')
nltk.download('averaged_perceptron_tagger')
nltk.download('punkt_tab')
from nltk.stem import WordNetLemmatizer
from nltk.corpus import wordnet
from tqdm import tqdm
from tqdm.keras import TqdmCallback
import json
import pickle
import random
import asyncio
import concurrent.futures
import multiprocessing
import io
import os
import tempfile
import numpy as np
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
import redis
import os
from dotenv import load_dotenv
from fastapi import FastAPI
from fastapi.responses import HTMLResponse
from pydantic import BaseModel
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)
# Load existing data from files if they exist and upload to Redis
def load_data_to_redis():
files_to_load = {
'intents.json': 'intents',
'classes.pkl': 'classes',
'words.pkl': 'words',
'chatbot_model.h5': 'chatbot_model'
}
for file_name, redis_key in files_to_load.items():
if os.path.exists(file_name) and not r.exists(redis_key):
print(f"Loading {file_name} to Redis...")
if file_name.endswith('.json'):
with open(file_name) as f:
data = json.load(f)
r.set(redis_key, json.dumps(data))
elif file_name.endswith('.h5'):
with open(file_name, 'rb') as f:
r.set(redis_key, f.read())
else:
with open(file_name, 'rb') as f:
r.set(redis_key, pickle.dumps(pickle.load(f)))
# Ensure directories exist
if not os.path.exists('models'):
os.makedirs('models')
def initialize_redis():
global r
try:
r.ping()
print("Redis connection successful.")
load_data_to_redis()
except redis.exceptions.ConnectionError:
print("Error connecting to Redis. Exiting.")
exit(1)
async def train_and_save_model():
global lemmatizer, r
while True:
words = []
classes = []
documents = []
ignore_words = ['?', '!']
intents = json.loads(r.get('intents'))
print("Loading user questions from Redis...")
if not r.exists('user_questions_loaded'):
user_questions = r.lrange('user_questions', 0, -1)
for question in user_questions:
question = question.decode('utf-8')
try:
existing_tag = r.get(f"tag:{question}").decode('utf-8')
documents.append((nltk.word_tokenize(question), existing_tag))
if existing_tag not in classes:
classes.append(existing_tag)
except AttributeError:
documents.append((nltk.word_tokenize(question), "unknown"))
if "unknown" not in classes:
classes.append("unknown")
r.set('user_questions_loaded', 1)
print("Processing intents from Redis...")
for intent in intents['intents']:
for pattern in intent['patterns']:
w = nltk.word_tokenize(pattern)
words.extend(w)
documents.append((w, intent['tag']))
if intent['tag'] not in classes:
classes.append(intent['tag'])
print(f"Generating synonyms for intent '{intent['tag']}'...")
with multiprocessing.Pool() as pool:
results = []
for _ in tqdm(range(100000), desc="Generating synonyms", leave=False):
if not intent['patterns']:
break
results.append(pool.apply_async(generate_synonym_pattern, (intent['patterns'],)))
for result in results:
new_pattern = result.get()
if new_pattern:
intent['patterns'].append(new_pattern)
words = [lemmatizer.lemmatize(w.lower()) for w in words if w not in ignore_words]
words = sorted(list(set(words)))
classes = sorted(list(set(classes)))
print("Creating training data...")
training = []
output_empty = [0] * len(classes)
for doc in documents:
bag = []
pattern_words = doc[0]
pattern_words = [lemmatizer.lemmatize(word.lower()) for word in pattern_words]
for w in words:
bag.append(1) if w in pattern_words else bag.append(0)
output_row = list(output_empty)
output_row[classes.index(doc[1])] = 1
training.append([bag, output_row])
if not training:
print("No training data yet. Waiting...")
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])
print("Loading or creating model...")
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(train_x[0]),))
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("Training the model...")
model.fit(train_x, train_y, epochs=1, batch_size=len(train_x), verbose=0, callbacks=[TqdmCallback(verbose=2)])
print("Saving data to Redis...")
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)
print("Data and model saved. Re-training...")
def generate_synonym_pattern(patterns):
new_pattern = []
for word in random.choice(patterns).split():
synonyms = wordnet.synsets(word)
if synonyms:
synonym = random.choice(synonyms[0].lemmas()).name()
new_pattern.append(synonym)
else:
new_pattern.append(word)
return " ".join(new_pattern)
def start_training_loop():
loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop)
loop.run_until_complete(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 io.BytesIO(r.get('chatbot_model')) as f:
with tempfile.NamedTemporaryFile(delete=False, suffix='.h5') as temp_file:
temp_file.write(f.read())
temp_file_name = temp_file.name
model = load_model(temp_file_name)
os.remove(temp_file_name)
sentence_words = nltk.word_tokenize(message.message)
sentence_words = [lemmatizer.lemmatize(word.lower()) for word in sentence_words]
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)})
r.rpush('user_questions', message.message)
asyncio.create_task(train_and_save_model())
return return_list
@app.post("/tag")
async def tag_question(question: str, tag: str):
r.set(f"tag:{question}", tag)
return {"message": "Tag saved"}
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="Type your message...">
<button onclick="sendMessage()">Send</button>
</div>
<script>
function sendMessage() {
let userInput = document.getElementById('user_input').value;
document.getElementById('user_input').value = '';
fetch('/chat', {
method: 'POST',
headers: {'Content-Type': 'application/json'},
body: JSON.stringify({"message": userInput})
})
.then(response => response.json())
.then(data => {
let chatbox = document.getElementById('chatbox');
chatbox.innerHTML += '<p><b>You:</b> ' + userInput + '</p>';
data.forEach(item => {
chatbox.innerHTML += '<p><b>Bot:</b> ' + item.intent + ' (Probability: ' + item.probability + ')</p>';
});
});
}
</script>
</body>
</html>
"""
@app.get("/", response_class=HTMLResponse)
async def root():
return html_code
if __name__ == "__main__":
initialize_redis()
training_process = multiprocessing.Process(target=start_training_loop)
training_process.start()
uvicorn.run(app, host="0.0.0.0", port=7860) |