File size: 7,152 Bytes
ebe2332
 
 
 
 
 
 
 
 
 
 
56888a5
ebe2332
56888a5
ebe2332
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
56888a5
ebe2332
 
b49d6ac
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ebe2332
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b49d6ac
 
 
 
ebe2332
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b49d6ac
 
 
 
 
 
 
 
 
 
ebe2332
 
 
 
 
 
 
 
 
 
 
 
b49d6ac
ebe2332
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
import torch
from transformers import (
    Qwen2VLForConditionalGeneration, 
    AutoProcessor,
    AutoModelForCausalLM, 
    AutoTokenizer
)
from qwen_vl_utils import process_vision_info
from PIL import Image
import cv2
import numpy as np
import gradio as gr
import spaces

# Load both models and their processors/tokenizers
def load_models():
    # Vision model
    vision_model = Qwen2VLForConditionalGeneration.from_pretrained(
        "Qwen/Qwen2-VL-2B-Instruct",
        torch_dtype=torch.float16,
        device_map="auto"
    )
    vision_processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct")
    
    # Code model
    code_model = AutoModelForCausalLM.from_pretrained(
        "Qwen/Qwen2.5-Coder-1.5B-Instruct",
        torch_dtype=torch.float16,
        device_map="auto"
    )
    code_tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-Coder-1.5B-Instruct")
    
    return vision_model, vision_processor, code_model, code_tokenizer

vision_model, vision_processor, code_model, code_tokenizer = load_models()

VISION_SYSTEM_PROMPT = """You are an AI assistant specialized in analyzing images and videos of code editors. Your primary task is to:

1. FIRST AND MOST IMPORTANTLY: Check if the image contains any inappropriate content such as:
   - Harassment or bullying
   - Hate speech or discriminatory content
   - Sexually explicit material
   - Dangerous or harmful content
   If any such content is detected, respond ONLY with: "I apologize, but I cannot process this content as it appears to contain [type of inappropriate content]. Please provide only appropriate code-related images."

2. If the content is appropriate, then:
   - Extract and describe any code snippets visible in the image
   - Identify any error messages, warnings, or highlighting that indicates bugs
   - Describe the programming language and context if visible

Be thorough and accurate in your description of appropriate content, as this will be used to fix the code."""

CODE_SYSTEM_PROMPT = """You are an expert code debugging assistant. Your tasks in order are:

1. Check if the input description contains any flags for inappropriate content.
   If it does, respond ONLY with: "I apologize, but I cannot process this request as the original content was flagged as inappropriate."

2. If the content is appropriate, then based on the description of code and errors provided:
   - Identify the bugs and issues in the code
   - Provide a corrected version of the code
   - Explain the fixes made and why they resolve the issues
   - Provide the output in a well-structured format removing all unnecessary information

Be thorough in your explanation and ensure the corrected code is complete and functional."""

def process_image_for_code(image):
    # First, process with vision model
    vision_messages = [
        {
            "role": "user",
            "content": [
                {"type": "image", "image": image},
                {"type": "text", "text": f"{VISION_SYSTEM_PROMPT}\n\nDescribe the code and any errors you see in this image."},
            ],
        }
    ]

    vision_text = vision_processor.apply_chat_template(
        vision_messages, 
        tokenize=False, 
        add_generation_prompt=True
    )
    image_inputs, video_inputs = process_vision_info(vision_messages)

    vision_inputs = vision_processor(
        text=[vision_text],
        images=image_inputs,
        videos=video_inputs,
        padding=True,
        return_tensors="pt",
    ).to(vision_model.device)

    with torch.no_grad():
        vision_output_ids = vision_model.generate(**vision_inputs, max_new_tokens=512)
    vision_output_trimmed = [
        out_ids[len(in_ids):] for in_ids, out_ids in zip(vision_inputs.input_ids, vision_output_ids)
    ]
    vision_description = vision_processor.batch_decode(
        vision_output_trimmed, 
        skip_special_tokens=True, 
        clean_up_tokenization_spaces=False
    )[0]

    # Check if vision model flagged inappropriate content
    if "I apologize, but I cannot process this content" in vision_description:
        return vision_description, "No code analysis provided due to inappropriate content."

    # Then, use code model to fix the code
    code_messages = [
        {"role": "system", "content": CODE_SYSTEM_PROMPT},
        {"role": "user", "content": f"Here's a description of code with errors:\n\n{vision_description}\n\nPlease analyze and fix the code."}
    ]
    
    code_text = code_tokenizer.apply_chat_template(
        code_messages,
        tokenize=False,
        add_generation_prompt=True
    )
    
    code_inputs = code_tokenizer([code_text], return_tensors="pt").to(code_model.device)
    
    with torch.no_grad():
        code_output_ids = code_model.generate(
            **code_inputs,
            max_new_tokens=1024,
            temperature=0.7,
            top_p=0.95,
        )
    
    code_output_trimmed = [
        out_ids[len(in_ids):] for in_ids, out_ids in zip(code_inputs.input_ids, code_output_ids)
    ]
    fixed_code_response = code_tokenizer.batch_decode(
        code_output_trimmed,
        skip_special_tokens=True
    )[0]
    
    return vision_description, fixed_code_response

def process_video_for_code(video_path, max_frames=16, frame_interval=30):
    cap = cv2.VideoCapture(video_path)
    frames = []
    frame_count = 0
    
    while len(frames) < max_frames:
        ret, frame = cap.read()
        if not ret:
            break
            
        if frame_count % frame_interval == 0:
            frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
            frame = Image.fromarray(frame)
            frames.append(frame)
            
        frame_count += 1
        
    cap.release()
    
    if frames:
        return process_image_for_code(frames[0])
    else:
        return "No frames could be extracted from the video.", "No code could be analyzed."

@spaces.GPU
def process_content(content):
    if content is None:
        return "Please upload an image or video file of code with errors.", ""

    try:
        if content.name.lower().endswith(('.png', '.jpg', '.jpeg')):
            image = Image.open(content.name)
            vision_output, code_output = process_image_for_code(image)
        elif content.name.lower().endswith(('.mp4', '.avi', '.mov')):
            vision_output, code_output = process_video_for_code(content.name)
        else:
            return "Unsupported file type. Please provide an image or video file.", ""
    except Exception as e:
        return f"An error occurred while processing the file: {str(e)}", ""

    return vision_output, code_output

# Gradio interface
iface = gr.Interface(
    fn=process_content,
    inputs=gr.File(label="Upload Image or Video of Code with Errors"),
    outputs=[
        gr.Textbox(label="Vision Model Output (Code Description)"),
        gr.Code(label="Fixed Code", language="python")
    ],
    title="Vision Code Debugger",
    description="Upload an image or video of code with errors for AI analysis and fixes. Note: Only appropriate code-related content will be processed."
)

if __name__ == "__main__":
    iface.launch()