app.py

import gradio as gr
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer, StoppingCriteria
import gradio as gr
import spaces
import torch
import numpy as np
import torch
import torchvision.transforms as T
from PIL import Image
from torchvision.transforms.functional import InterpolationMode
from transformers import AutoModel, AutoTokenizer
from PIL import Image, ExifTags
from threading import Thread
import re
import time 
from PIL import Image
import torch
import spaces
import subprocess
import os

subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)

torch.set_default_device('cuda')


IMAGENET_MEAN = (0.485, 0.456, 0.406)
IMAGENET_STD = (0.229, 0.224, 0.225)

def build_transform(input_size):
    MEAN, STD = IMAGENET_MEAN, IMAGENET_STD
    transform = T.Compose([
        T.Lambda(lambda img: img.convert('RGB') if img.mode != 'RGB' else img),
        T.Resize((input_size, input_size), interpolation=InterpolationMode.BICUBIC),
        T.ToTensor(),
        T.Normalize(mean=MEAN, std=STD)
    ])
    return transform

def find_closest_aspect_ratio(aspect_ratio, target_ratios, width, height, image_size):
    best_ratio_diff = float('inf')
    best_ratio = (1, 1)
    area = width * height
    for ratio in target_ratios:
        target_aspect_ratio = ratio[0] / ratio[1]
        ratio_diff = abs(aspect_ratio - target_aspect_ratio)
        if ratio_diff < best_ratio_diff:
            best_ratio_diff = ratio_diff
            best_ratio = ratio
        elif ratio_diff == best_ratio_diff:
            if area > 0.5 * image_size * image_size * ratio[0] * ratio[1]:
                best_ratio = ratio
    return best_ratio

def dynamic_preprocess(image, min_num=1, max_num=12, image_size=448, use_thumbnail=False):
    orig_width, orig_height = image.size
    aspect_ratio = orig_width / orig_height

    # calculate the existing image aspect ratio
    target_ratios = set(
        (i, j) for n in range(min_num, max_num + 1) for i in range(1, n + 1) for j in range(1, n + 1) if
        i * j <= max_num and i * j >= min_num)
    target_ratios = sorted(target_ratios, key=lambda x: x[0] * x[1])

    # find the closest aspect ratio to the target
    target_aspect_ratio = find_closest_aspect_ratio(
        aspect_ratio, target_ratios, orig_width, orig_height, image_size)

    # calculate the target width and height
    target_width = image_size * target_aspect_ratio[0]
    target_height = image_size * target_aspect_ratio[1]
    blocks = target_aspect_ratio[0] * target_aspect_ratio[1]

    # resize the image
    resized_img = image.resize((target_width, target_height))
    processed_images = []
    for i in range(blocks):
        box = (
            (i % (target_width // image_size)) * image_size,
            (i // (target_width // image_size)) * image_size,
            ((i % (target_width // image_size)) + 1) * image_size,
            ((i // (target_width // image_size)) + 1) * image_size
        )
        # split the image
        split_img = resized_img.crop(box)
        processed_images.append(split_img)
    assert len(processed_images) == blocks
    if use_thumbnail and len(processed_images) != 1:
        thumbnail_img = image.resize((image_size, image_size))
        processed_images.append(thumbnail_img)
    return processed_images

def correct_image_orientation(image_path):
    # Mở ảnh
    image = Image.open(image_path)

    # Kiểm tra dữ liệu Exif (nếu có)
    try:
        exif = image._getexif()
        if exif is not None:
            for tag, value in exif.items():
                if ExifTags.TAGS.get(tag) == "Orientation":
                    # Sửa hướng dựa trên Orientation
                    if value == 3:
                        image = image.rotate(180, expand=True)
                    elif value == 6:
                        image = image.rotate(-90, expand=True)
                    elif value == 8:
                        image = image.rotate(90, expand=True)
                    break
    except Exception as e:
        print("Không thể xử lý Exif:", e)

    return image
    
def load_image(image_file, input_size=448, max_num=12):
    image = correct_image_orientation(image_file).convert('RGB')
    width, height = image.size
    image = image.resize((width * 2, height * 2), Image.LANCZOS)
    print("Image size: ", image.size)
    transform = build_transform(input_size=input_size)
    images = dynamic_preprocess(image, image_size=input_size, use_thumbnail=True, max_num=max_num)
    pixel_values = [transform(image) for image in images]
    pixel_values = torch.stack(pixel_values)
    return pixel_values

def extract_conclusion(text):
    match = re.search(r"<CONCLUSION>(.*?)</CONCLUSION>", text, re.DOTALL)
    return match.group(1).strip() if match else ""

def extract_think(text):
    text = re.sub(r"<.*?>", "", text.split("<CONCLUSION>")[0])  # Loại bỏ tất cả các tag <...>
    conclusion_part = extract_conclusion(text)
    return text.replace(conclusion_part, "").strip()

def wrap_text(text, max_words=20):
    lines = text.split('\n')  # Cắt theo dòng trước
    wrapped_lines = []
    
    for line in lines:
        words = line.split()
        if len(words) > max_words:
            wrapped_lines.extend([' '.join(words[i:i+max_words]) for i in range(0, len(words), max_words)])
        else:
            wrapped_lines.append(line)
    
    return '\n'.join(wrapped_lines)
    
model = AutoModel.from_pretrained(
    "5CD-AI/Vintern-3B-R-beta",
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
    trust_remote_code=True,
    use_flash_attn=True,
).eval().cuda()

tokenizer = AutoTokenizer.from_pretrained("5CD-AI/Vintern-3B-R-beta", trust_remote_code=True, use_fast=False)

global_think_mode =False
think_prompt = """Bạn là người rất cẩn thận và đa nghi, vui lòng trả lời câu hỏi dưới đây bằng tiếng Việt. Khi suy luận bạn thường liệt kê ra các bằng chứng để chỉ ra các đáp án khả thi, suy luận và giải thích tại sao lại lựa chọn và loại bỏ trước khi đưa ra câu trả lời cuối cùng.

Câu hỏi:
{question_input}

Hãy trả lời rất dài theo định dạng sau:
<SUMMARY>...</SUMMARY>

<CAPTION>...</CAPTION>

<INFORMATION_EXTRACT>...</INFORMATION_EXTRACT>

<EXTERNAL_KNOWLEDGE_EXPANSION>...</EXTERNAL_KNOWLEDGE_EXPANSION>

<FIND_CANDIDATES_REASONING>...</FIND_CANDIDATES_REASONING>

<TOP3_CANDIDATES>...</TOP3_CANDIDATES>

<REASONING_PLAN>...</REASONING_PLAN>

<REASONING>...</REASONING>

<COUNTER_ARGUMENTS>...</COUNTER_ARGUMENTS>

<VALIDATION_REASONING>...</VALIDATION_REASONING>

<CONCLUSION>...</CONCLUSION>
"""


@spaces.GPU(duration=120)
def chat(message, history):
    global global_think_mode
    print("------------------------> RUN with global_think_mode: ",global_think_mode)
    print("history",history)
    print("message",message)

    if len(history) != 0 and len(message["files"]) != 0:
        return """Chúng tôi hiện chỉ hổ trợ 1 ảnh ở đầu ngữ cảnh! Vui lòng tạo mới cuộc trò chuyện.
We currently only support one image at the start of the context! Please start a new conversation."""
    
    if len(history) == 0 and len(message["files"]) != 0:
        if "path" in message["files"][0]:
            test_image = message["files"][0]["path"]
        else:
            test_image = message["files"][0]
        pixel_values = load_image(test_image, max_num=6).to(torch.bfloat16).cuda()
    elif len(history) == 0 and len(message["files"]) == 0:
        pixel_values = None     
    elif history[0][0][0] is not None and os.path.isfile(history[0][0][0]):
        test_image = history[0][0][0]
        pixel_values = load_image(test_image, max_num=6).to(torch.bfloat16).cuda()
    else:
        pixel_values = None 
            
    if not global_think_mode:
        generation_config = dict(max_new_tokens= 700, do_sample=False, num_beams = 3, repetition_penalty=2.5)
    
        if len(history) == 0:
            if pixel_values is not None:
                question = '<image>\n'+message["text"]
            else:
                question = message["text"]
            response, conv_history = model.chat(tokenizer, pixel_values, question, generation_config, history=None, return_history=True)
        else:
            conv_history = []
            if history[0][0][0] is not None and os.path.isfile(history[0][0][0]):
                start_index = 1
            else:
                start_index = 0
            
            for i, chat_pair in enumerate(history[start_index:]):
                if i == 0 and start_index == 1:
                     conv_history.append(tuple(['<image>\n'+chat_pair[0],chat_pair[1]]))
                else:
                    conv_history.append(tuple(chat_pair))
    
                
            print("conv_history",conv_history)
            question = message["text"]
            response, conv_history = model.chat(tokenizer, pixel_values, question, generation_config, history=conv_history, return_history=True)
            
        print(f'User: {question}\nAssistant: {response}')
    
        # return response
        buffer = ""
        for new_text in response:
          buffer += new_text
          generated_text_without_prompt = buffer[:]
          time.sleep(0.02)
          yield generated_text_without_prompt
    else:
        ####################################################### thinking #######################################################
        generation_config = dict(max_new_tokens= 2000, do_sample=True, num_beams = 2, repetition_penalty=2.5, temperature=0.5)
    
        if len(history) == 0:
            if pixel_values is not None:
                question = '<image>\n'+ think_prompt.format(question_input=message["text"])
            else:
                question = think_prompt.format(question_input=message["text"])
            response, conv_history = model.chat(tokenizer, pixel_values, question, generation_config, history=None, return_history=True)
        else:
            conv_history = []
            if history[0][0][0] is not None and os.path.isfile(history[0][0][0]):
                start_index = 1
            else:
                start_index = 0
            
            for i, chat_pair in enumerate(history[start_index:]):
                if i == 0 and start_index == 1:
                     conv_history.append(tuple(['<image>\n'+chat_pair[0],chat_pair[1]]))
                else:
                    conv_history.append(tuple(chat_pair))
    
                
            print("conv_history",conv_history)
            question = message["text"]
            response, conv_history = model.chat(tokenizer, pixel_values, question, generation_config, history=conv_history, return_history=True)
            
        print(f'User: {question}\nAssistant: {response}')
        think_part = wrap_text(extract_think(response))
        conclusion_part = extract_conclusion(response)
        if conclusion_part == "":
            conclusion_part = think_part
            
        buffer = ""
        thinking = think_part

        accumulated_text = "💡 **Thinking process:**\n\n"
        accumulated_text += "<pre><code>\n"

        temp_text = ""
        for char in thinking:
            temp_text += char
            yield accumulated_text + temp_text + "\n</code></pre>\n"
            time.sleep(0.01)

        accumulated_text += temp_text + "\n</code></pre>\n"

        # Yield phần kết luận
        accumulated_text += "🎯 **Conclusion:**\n\n"
        
        temp_text = ""
        for char in conclusion_part:
            temp_text += char
            yield accumulated_text + temp_text 
            time.sleep(0.02)
        
        accumulated_text += temp_text 


CSS ="""
#component-10 {
  height: 70dvh !important;
  transform-origin: top; /* Đảm bảo rằng phần tử mở rộng từ trên xuống */
  border-style: solid;
  overflow: hidden;
  flex-grow: 1;
  min-width: min(160px, 100%);
  border-width: var(--block-border-width);
}

#component-12 {
    height: 50dvh !important;
    border-style: solid;
    overflow: auto;
    flex-grow: 1;
    min-width: min(160px, 100%);
    border-width: var(--block-border-width);
}

#component-15 {
    border-style: solid;
    overflow: hidden;
    flex-grow: 7;
    min-width: min(160px, 100%); 
    border-width: var(--block-border-width);
    height: 20dvh !important;
}

#think-button{
    width: 40% !important;
}

/* Đảm bảo ảnh bên trong nút hiển thị đúng cách cho các nút có aria-label chỉ định */
button.svelte-1lcyrx4[aria-label="user's message: a file of type image/jpeg, "] img.svelte-1pijsyv {
  width: 100%;
  object-fit: contain;
  height: 100%;
  border-radius: 13px; /* Thêm bo góc cho ảnh */
  max-width: 50vw;     /* Giới hạn chiều rộng ảnh */
}
/* Đặt chiều cao cho nút và cho phép chọn văn bản chỉ cho các nút có aria-label chỉ định */
button.svelte-1lcyrx4[aria-label="user's message: a file of type image/jpeg, "] {
  user-select: text;
  text-align: left;
  height: 300px;
}
/* Thêm bo góc và giới hạn chiều rộng cho ảnh không thuộc avatar container */
.message-wrap.svelte-1lcyrx4 > div.svelte-1lcyrx4 .svelte-1lcyrx4:not(.avatar-container) img {
  border-radius: 13px;
  max-width: 50vw;
}
.message-wrap.svelte-1lcyrx4 .message.svelte-1lcyrx4 img {
    margin: var(--size-2);
    max-height: 500px;
}
.image-preview-close-button {
  position: relative; /* Nếu cần định vị trí */
  width: 5%; /* Chiều rộng nút */
  height: 5%; /* Chiều cao nút */
  display: flex;
  justify-content: center;
  align-items: center;
  padding: 0; /* Để tránh ảnh hưởng từ padding mặc định */
  border: none; /* Tùy chọn để loại bỏ đường viền */
  background: none; /* Tùy chọn để loại bỏ nền */
}

.example-image-container.svelte-9pi8y1 {
    width: calc(var(--size-8) * 5);
    height: calc(var(--size-8) * 5);
    border-radius: var(--radius-lg);
    overflow: hidden;
    position: relative;
    margin-bottom: var(--spacing-lg);
}
"""

js = """
function forceLightTheme() {
    const url = new URL(window.location);

    // Cập nhật __theme thành light nếu giá trị không đúng
    if (url.searchParams.get('__theme') !== 'light') {
        url.searchParams.set('__theme', 'light');
        // Thay đổi URL mà không tải lại trang nếu cần
        window.history.replaceState({}, '', url.href);
    }

    // Đảm bảo document luôn áp dụng theme light
    document.documentElement.setAttribute('data-theme', 'light');
}
"""

def toggle_think_mode(current_state):
    global global_think_mode
    new_state = not current_state
    global_think_mode = not global_think_mode
    print("global_think_mode: ",global_think_mode,"="*20)
    button_label = "🧠DeepThink💡1minute⏳" if global_think_mode else "🧠Think"
    return new_state, button_label

def reset_think_mode():
    return False, "🧠Think"  # Trả về trạng thái mặc định
    
demo = gr.Blocks(css=CSS,js=js, theme='NoCrypt/miku')
# demo = gr.Blocks( theme='NoCrypt/miku')


with demo:
    think_mode = gr.State(False)  # Lưu trạng thái Think Mode
    
    chat_demo_interface = gr.ChatInterface(
        fn=chat,
        description="""**Vintern-3B-R-beta** This Gradio demo is not complete yet; I am still working on it. :) """,
        examples=[
            [{"text": "Trích xuất các thông tin từ ảnh trả về markdown.", "files":["./demo_1.jpg"]}, False,False],
            [{"text": "Liệt kê toàn bộ văn bản.", "files":["./demo_2.jpg"]}, False,False],
            [{"text": "Trích xuất thông tin kiện hàng trong ảnh và trả về dạng JSON.", "files":["./demo_4.jpg"]}, False,False]
        ],
        # additional_inputs=[think_mode],
        title="❄️Vintern-3B-R-beta❄️",
        multimodal=True,
        css=CSS,
        js=js,
        theme='NoCrypt/miku'
    )
    
    think_button = gr.Button("🧠Think", elem_id="think-button", variant="secondary") 
    
    # Khi nhấn nút, trạng thái think_mode thay đổi + đổi nhãn nút
    think_button.click(toggle_think_mode, inputs=[think_mode], outputs=[think_mode, think_button])
    # Reset nút Think sau khi chat hoàn tất
    # chat_demo_interface.submit(reset_think_mode, inputs=[], outputs=[think_mode, think_button])

demo.queue().launch()