Create func_ai.py

func_ai.py

@@ -0,0 +1,58 @@
+import requests
+import torch
+from googletrans import Translator
+from transformers import pipeline
+import time
+import os 
+VECTOR_API_URL = os.getenv('API_URL')
+
+translator = Translator()
+
+
+sentiment_model = pipeline(
+    'sentiment-analysis',
+    model='cardiffnlp/twitter-xlm-roberta-base-sentiment',
+    tokenizer='cardiffnlp/twitter-xlm-roberta-base-sentiment',
+    device=0 if torch.cuda.is_available() else -1
+)
+
+
+
+classifier = pipeline(
+    "zero-shot-classification",
+    model="valhalla/distilbart-mnli-12-6",
+    device=0 if torch.cuda.is_available() else -1
+)
+
+
+def classify_comment(text):
+    translated_text = translator.translate(text, dest='en').text
+    result = classifier(translated_text, ["interrogative", "non-interrogative"], clean_up_tokenization_spaces=True)
+    top_class = result['labels'][0]
+    return top_class
+
+
+def retrieve_from_vdb(query):
+    print(f"Отправка запроса к FastAPI сервису: {query}")
+    response = requests.get(f"{VECTOR_API_URL}/search/", json={"query": query})
+    if response.status_code == 200:
+        results = response.json().get("results", [])
+        print(f"Получено {len(results)} результатов.")
+        return results
+    else:
+        print(f"Ошибка при поиске: {response.text}")
+        return []
+
+
+def analyze_sentiment(comments):
+    print("Начинаем анализ настроений.")
+    results = []
+    for i in range(0, len(comments), 50):
+        batch = comments[i:i + 50]
+        print(f"Анализируем батч с {i} по {i + len(batch)} комментарий.")
+        batch_results = sentiment_model(batch)
+        results.extend(batch_results)
+        time.sleep(1)  # Задержка для предотвращения перегрузки
+    print("Анализ настроений завершен.")
+    return results
+