# # SPDX-FileCopyrightText: Hadad # SPDX-License-Identifier: CC-BY-NC-SA-4.0 # import gradio as gr import requests import json import os import random import time import pytesseract import pdfplumber import docx import pandas as pd import pptx import fitz import io import uuid from pathlib import Path from PIL import Image from pptx import Presentation os.system("apt-get update -q -y && apt-get install -q -y tesseract-ocr tesseract-ocr-eng tesseract-ocr-ind libleptonica-dev libtesseract-dev") LINUX_SERVER_HOSTS = [host for host in json.loads(os.getenv("LINUX_SERVER_HOST", "[]")) if host] LINUX_SERVER_PROVIDER_KEYS = [key for key in json.loads(os.getenv("LINUX_SERVER_PROVIDER_KEY", "[]")) if key] AI_TYPES = {f"AI_TYPE_{i}": os.getenv(f"AI_TYPE_{i}") for i in range(1, 7)} RESPONSES = {f"RESPONSE_{i}": os.getenv(f"RESPONSE_{i}") for i in range(1, 10)} MODEL_MAPPING = json.loads(os.getenv("MODEL_MAPPING", "{}")) MODEL_CONFIG = json.loads(os.getenv("MODEL_CONFIG", "{}")) MODEL_CHOICES = list(MODEL_MAPPING.values()) DEFAULT_CONFIG = json.loads(os.getenv("DEFAULT_CONFIG", "{}")) META_TAGS = os.getenv("META_TAGS") ALLOWED_EXTENSIONS = json.loads(os.getenv("ALLOWED_EXTENSIONS")) class SessionWithID(requests.Session): def __init__(self): super().__init__() self.headers.update({"Connection": "keep-alive"}) self.session_id = str(uuid.uuid4()) def create_session(): return SessionWithID() def get_model_key(display_name): return next((k for k, v in MODEL_MAPPING.items() if v == display_name), MODEL_CHOICES[0]) def extract_file_content(file_path): ext = Path(file_path).suffix.lower() content = "" try: if ext == ".pdf": with pdfplumber.open(file_path) as pdf: for page in pdf.pages: text = page.extract_text() if text: content += text + "\n" tables = page.extract_tables() if tables: for table in tables: table_str = "\n".join([", ".join(row) for row in table if row]) content += "\n" + table_str + "\n" elif ext in [".doc", ".docx"]: doc = docx.Document(file_path) for para in doc.paragraphs: content += para.text + "\n" elif ext in [".xlsx", ".xls"]: df = pd.read_excel(file_path) content += df.to_csv(index=False) elif ext in [".ppt", ".pptx"]: prs = Presentation(file_path) for slide in prs.slides: for shape in slide.shapes: if hasattr(shape, "text") and shape.text: content += shape.text + "\n" elif ext in [".png", ".jpg", ".jpeg", ".tiff", ".bmp", ".gif", ".webp"]: try: pytesseract.pytesseract.tesseract_cmd = "/usr/bin/tesseract" image = Image.open(file_path) text = pytesseract.image_to_string(image) content += text + "\n" except Exception as e: content += f"{e}\n" else: content = Path(file_path).read_text(encoding="utf-8") except Exception as e: content = f"{file_path}: {e}" return content.strip() def chat_with_model(history, user_input, selected_model_display, sess): if not LINUX_SERVER_PROVIDER_KEYS or not LINUX_SERVER_HOSTS: return RESPONSES["RESPONSE_3"] if not hasattr(sess, "session_id"): sess.session_id = str(uuid.uuid4()) selected_model = get_model_key(selected_model_display) model_config = MODEL_CONFIG.get(selected_model, DEFAULT_CONFIG) messages = [{"role": "user", "content": user} for user, _ in history] messages += [{"role": "assistant", "content": assistant} for _, assistant in history if assistant] messages.append({"role": "user", "content": user_input}) data = {"model": selected_model, "messages": messages, **model_config, "session_id": sess.session_id} random.shuffle(LINUX_SERVER_PROVIDER_KEYS) random.shuffle(LINUX_SERVER_HOSTS) for api_key in LINUX_SERVER_PROVIDER_KEYS: for host in LINUX_SERVER_HOSTS: try: response = sess.post(host, json=data, headers={"Authorization": f"Bearer {api_key}"}, timeout=5) if response.status_code < 400: ai_text = response.json().get("choices", [{}])[0].get("message", {}).get("content", RESPONSES["RESPONSE_2"]) return ai_text except requests.exceptions.RequestException: continue return RESPONSES["RESPONSE_3"] def respond(multi_input, history, selected_model_display, sess): message = {"text": multi_input.get("text", "").strip(), "files": multi_input.get("files", [])} if not message["text"] and not message["files"]: yield history, gr.MultimodalTextbox(value=None, interactive=True), sess return combined_input = "" for file_item in message["files"]: if isinstance(file_item, dict) and "name" in file_item: file_path = file_item["name"] else: file_path = file_item file_content = extract_file_content(file_path) combined_input += f"{Path(file_path).name}\n\n{file_content}\n\n" if message["text"]: combined_input += message["text"] history.append([combined_input, ""]) ai_response = chat_with_model(history, combined_input, selected_model_display, sess) history[-1][1] = "" for character in ai_response: history[-1][1] += character time.sleep(0.0009) yield history, gr.MultimodalTextbox(value=None, interactive=True), sess def change_model(new_model_display): return [], create_session(), new_model_display with gr.Blocks(fill_height=True, fill_width=True, title=AI_TYPES["AI_TYPE_4"], head=META_TAGS) as jarvis: user_history = gr.State([]) user_session = gr.State(create_session()) selected_model = gr.State(MODEL_CHOICES[0]) chatbot = gr.Chatbot(label=AI_TYPES["AI_TYPE_1"], show_copy_button=True, scale=1, elem_id=AI_TYPES["AI_TYPE_2"]) model_dropdown = gr.Dropdown(show_label=False, choices=MODEL_CHOICES, value=MODEL_CHOICES[0]) with gr.Row(): msg = gr.MultimodalTextbox(show_label=False, placeholder=RESPONSES["RESPONSE_5"], interactive=True, file_count="single", file_types=ALLOWED_EXTENSIONS) model_dropdown.change(fn=change_model, inputs=[model_dropdown], outputs=[user_history, user_session, selected_model]) msg.submit(fn=respond, inputs=[msg, user_history, selected_model, user_session], outputs=[chatbot, msg, user_session]) jarvis.launch(show_api=False, max_file_size="1mb")