Update rag_pipeline.py

rag_pipeline.py

@@ -5,16 +5,16 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 class RAGPipeline:
     def __init__(self, logger):
         self.logger = logger
-        self.logger("[RAG] Initializing tokenizer and model...")
+        self.logger("[RAG] جاري تحميل النموذج والمحول...")
         self.tokenizer = AutoTokenizer.from_pretrained("aubmindlab/aragpt2-mega", trust_remote_code=True)
         self.generator = AutoModelForCausalLM.from_pretrained("aubmindlab/aragpt2-mega", trust_remote_code=True)
         self.chunk_embeddings = []
         self.index = []
-        self.logger("[RAG] Initialization done.")
+        self.logger("[RAG] تم التحميل بنجاح.")
 
     def build_index(self, chunks):
         start = time.time()
-        self.logger(f"[RAG] Building index for {len(chunks)} chunks...")
+        self.logger(f"[RAG] بناء الفهرس لـ {len(chunks)} مقاطع...")
         self.chunk_embeddings = []
         self.index = []
 
@@ -23,28 +23,26 @@ class RAGPipeline:
             self.chunk_embeddings.append(embedding)
             self.index.append(chunk)
             if (i+1) % 10 == 0 or (i+1) == len(chunks):
-                self.logger(f"[RAG] Processed {i+1}/{len(chunks)} chunks.")
+                self.logger(f"[RAG] تم معالجة {i+1}/{len(chunks)} مقاطع.")
 
         self.chunk_embeddings = np.array(self.chunk_embeddings)
         dim = self.chunk_embeddings.shape[1] if len(self.chunk_embeddings) > 0 else 0
         elapsed = time.time() - start
-        self.logger(f"[RAG] Index built with dimension {dim} in {elapsed:.2f}s.")
-        return "Index built successfully."
+        self.logger(f"[RAG] تم بناء الفهرس بأبعاد {dim} في {elapsed:.2f} ثانية.")
+        return "تم بناء الفهرس بنجاح."
 
     def _dummy_embedding(self, text):
         return np.random.rand(768)
 
     def generate_answer(self, question, top_k=3):
         start = time.time()
-        self.logger(f"[RAG] Generating answer for question:\n{question}")
+        self.logger(f"[RAG] توليد إجابة للسؤال: {question}")
 
         if len(self.index) == 0:
-            self.logger("[RAG] Warning: index is empty, please build index first.")
+            self.logger("[RAG] تحذير: الفهرس فارغ، الرجاء بناء الفهرس أولاً.")
             return "لم يتم بناء الفهرس بعد.", []
 
-        # بحث مبسط لأقرب النصوص (dummy - عشوائي)
         passages = self.index[:top_k]
-
         prompt = question + "\n\nالمراجع:\n" + "\n".join(passages)
 
         inputs = self.tokenizer(prompt, return_tensors="pt")
@@ -52,5 +50,5 @@ class RAGPipeline:
         response = self.tokenizer.decode(output[0], skip_special_tokens=True)
 
         elapsed = time.time() - start
-        self.logger(f"[RAG] Answer generated in {elapsed:.2f}s.")
+        self.logger(f"[RAG] تم توليد الإجابة في {elapsed:.2f} ثانية.")
         return response, passages