import gradio as gr import os from rag_pipeline import RAGPipeline from PyPDF2 import PdfReader import docx logs = [] def logger(message): logs.append(message) if len(logs) > 100: logs.pop(0) rag = RAGPipeline(logger=logger) def extract_text_from_file(file): text = "" if file.name.endswith(".pdf"): reader = PdfReader(file.name) for page in reader.pages: text += page.extract_text() + "\n" elif file.name.endswith(".docx"): doc = docx.Document(file.name) for para in doc.paragraphs: text += para.text + "\n" elif file.name.endswith(".txt"): with open(file.name, "r", encoding="utf-8") as f: text = f.read() else: logger(f"[RAG] تنسيق الملف غير مدعوم: {file.name}") return text def process_documents(files): all_text = "" for file in files: all_text += extract_text_from_file(file) + "\n" chunks = [t.strip() for t in all_text.split("\n") if t.strip()] msg = rag.build_index(chunks) return msg, "\n".join(logs) def answer_question(question): answer, sources = rag.generate_answer(question) return answer, "\n".join(sources), "\n".join(logs) with gr.Blocks() as demo: gr.Markdown("## 🧠 نظام RAG باللغة العربية") with gr.Tab("📄 رفع المستندات وبناء الفهرس"): file_input = gr.File(label="اختر ملفات المستندات", file_types=[".pdf", ".docx", ".txt"], file_count="multiple") build_btn = gr.Button("🔨 بناء الفهرس") build_status = gr.Textbox(label="حالة بناء الفهرس", lines=2) logs_box = gr.Textbox(label="📜 سجلات التشغيل", lines=10) build_btn.click(fn=process_documents, inputs=file_input, outputs=[build_status, logs_box]) with gr.Tab("❓ طرح الأسئلة"): question_input = gr.Textbox(label="اكتب سؤالك هنا", lines=2) ask_btn = gr.Button("📩 إرسال السؤال") answer_output = gr.Textbox(label="الإجابة", lines=5) sources_output = gr.Textbox(label="المراجع المستخدمة", lines=5) logs_box2 = gr.Textbox(label="📜 سجلات التشغيل", lines=10) ask_btn.click(fn=answer_question, inputs=question_input, outputs=[answer_output, sources_output, logs_box2]) demo.launch()