File size: 8,222 Bytes
ed20377
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
from PIL import Image
import base64
from io import BytesIO
import json
import os
import requests
from typing import Optional
from huggingface_hub import InferenceClient
from transformers import AutoProcessor
from smolagents import Tool
import uuid
import mimetypes
from dotenv import load_dotenv

load_dotenv(override=True)

idefics_processor = AutoProcessor.from_pretrained("HuggingFaceM4/idefics2-8b-chatty")

def process_images_and_text(image_path, query, client):
    messages = [
        {
            "role": "user", "content": [
                {"type": "image"},
                {"type": "text", "text": query},
            ]
        },
    ]

    prompt_with_template = idefics_processor.apply_chat_template(messages, add_generation_prompt=True)

    # load images from local directory

    # encode images to strings which can be sent to the endpoint
    def encode_local_image(image_path):
        # load image
        image = Image.open(image_path).convert('RGB')

        # Convert the image to a base64 string
        buffer = BytesIO()
        image.save(buffer, format="JPEG")  # Use the appropriate format (e.g., JPEG, PNG)
        base64_image = base64.b64encode(buffer.getvalue()).decode('utf-8')

        # add string formatting required by the endpoint
        image_string = f"data:image/jpeg;base64,{base64_image}"

        return image_string
    

    image_string = encode_local_image(image_path)
    prompt_with_images = prompt_with_template.replace("<image>", "![]({}) ").format(image_string)


    payload = {
        "inputs": prompt_with_images,
        "parameters": {
            "return_full_text": False,
            "max_new_tokens": 200,
        }
    }

    return json.loads(client.post(json=payload).decode())[0]

# Function to encode the image
def encode_image(image_path):
    if image_path.startswith("http"):
        user_agent = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36 Edg/119.0.0.0"
        request_kwargs = {
            "headers": {"User-Agent": user_agent},
            "stream": True,
        }

        # Send a HTTP request to the URL
        response = requests.get(image_path, **request_kwargs)
        response.raise_for_status()
        content_type = response.headers.get("content-type", "")

        extension = mimetypes.guess_extension(content_type)
        if extension is None:
            extension = ".download"
    
        fname = str(uuid.uuid4()) + extension
        download_path = os.path.abspath(os.path.join("downloads", fname))

        with open(download_path, "wb") as fh:
            for chunk in response.iter_content(chunk_size=512):
                fh.write(chunk)

        image_path = download_path

    with open(image_path, "rb") as image_file:
        return base64.b64encode(image_file.read()).decode('utf-8')

headers = {
    "Content-Type": "application/json",
    "Authorization": f"Bearer {os.getenv('OPENAI_API_KEY')}"
}


def resize_image(image_path):
    img = Image.open(image_path)
    width, height = img.size
    img = img.resize((int(width / 2), int(height / 2)))
    new_image_path = f"resized_{image_path}"
    img.save(new_image_path)
    return new_image_path


class VisualQATool(Tool):
    name = "visualizer"
    description = "A tool that can answer questions about attached images."
    inputs = {
        "question": {
            "description": "the question to answer",
            "type": "string",
            "nullable": True,
        },
        "image_path": {
            "description": "The path to the image on which to answer the question",
            "type": "string",
        },
    }
    output_type = "string"

    client = InferenceClient("HuggingFaceM4/idefics2-8b-chatty")

    def forward(self, image_path: str, question: Optional[str] = None) -> str:
        add_note = False
        if not question:
            add_note = True
            question = "Please write a detailed caption for this image."
        try:
            output = process_images_and_text(image_path, question, self.client)
        except Exception as e:
            print(e)
            if "Payload Too Large" in str(e):
                new_image_path = resize_image(image_path)
                output = process_images_and_text(new_image_path, question, self.client)

        if add_note:
            output = f"You did not provide a particular question, so here is a detailed caption for the image: {output}"

        return output

# ////////////////////////////////////////////////////////////////////////
# import base64
# import json
# import os
# import uuid
# import mimetypes
# from io import BytesIO
# from typing import Optional
# from PIL import Image
# from dotenv import load_dotenv
# import requests
# from smolagents import Tool
# from huggingface_hub import InferenceClient

# load_dotenv()

# # === UTILS ===

# def encode_local_image(image_path):
#     image = Image.open(image_path).convert("RGB")
#     buffer = BytesIO()
#     image.save(buffer, format="JPEG")
#     base64_image = base64.b64encode(buffer.getvalue()).decode("utf-8")
#     return f"data:image/jpeg;base64,{base64_image}"

# def encode_image(image_path):
#     if image_path.startswith("http"):
#         user_agent = (
#             "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 "
#             "(KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36"
#         )
#         response = requests.get(image_path, headers={"User-Agent": user_agent}, stream=True)
#         response.raise_for_status()

#         ext = mimetypes.guess_extension(response.headers.get("content-type", ""))
#         fname = str(uuid.uuid4()) + (ext or ".jpg")
#         os.makedirs("downloads", exist_ok=True)
#         local_path = os.path.join("downloads", fname)

#         with open(local_path, "wb") as f:
#             for chunk in response.iter_content(chunk_size=1024):
#                 f.write(chunk)

#         image_path = local_path

#     with open(image_path, "rb") as img:
#         return base64.b64encode(img.read()).decode("utf-8")

# def resize_image(image_path):
#     img = Image.open(image_path)
#     width, height = img.size
#     img = img.resize((int(width / 2), int(height / 2)))
#     new_path = f"resized_{os.path.basename(image_path)}"
#     img.save(new_path)
#     return new_path

# # === IDEFICS2 Tool ===

# class VisualQATool(Tool):
#     name = "visualizer"
#     description = "A tool that can answer questions about attached images using IDEFICS2."
#     inputs = {
#         "question": {
#             "description": "The question to answer",
#             "type": "string",
#             "nullable": True,
#         },
#         "image_path": {
#             "description": "Path to the image (local or downloaded)",
#             "type": "string",
#         },
#     }
#     output_type = "string"

#     client = InferenceClient("HuggingFaceM4/idefics2-8b-chatty")

#     def forward(self, image_path: str, question: Optional[str] = None) -> str:
#         add_note = False
#         if not question:
#             add_note = True
#             question = "Please write a detailed caption for this image."

#         image_string = encode_local_image(image_path)
#         prompt = f"![]({image_string})\n\n{question}"

#         payload = {
#             "inputs": prompt,
#             "parameters": {
#                 "return_full_text": False,
#                 "max_new_tokens": 200,
#             },
#         }

#         try:
#             result = json.loads(self.client.post(json=payload).decode())[0]
#         except Exception as e:
#             if "Payload Too Large" in str(e):
#                 resized = resize_image(image_path)
#                 image_string = encode_local_image(resized)
#                 prompt = f"![]({image_string})\n\n{question}"
#                 payload["inputs"] = prompt
#                 result = json.loads(self.client.post(json=payload).decode())[0]
#             else:
#                 raise e

#         return (
#             f"You did not provide a particular question, so here is a detailed caption for the image: {result}"
#             if add_note else result
#         )