Spaces:
Sleeping
Sleeping
add preview
Browse files- .gradio/flagged/dataset1.csv +2 -0
- app.py +7 -12
.gradio/flagged/dataset1.csv
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
Input Type,Document or Image URL,Extracted Plain Text,Preview,timestamp
|
2 |
+
URL,,,[],2025-04-23 12:25:00.449275
|
app.py
CHANGED
@@ -4,6 +4,7 @@ from io import BytesIO
|
|
4 |
import gradio as gr
|
5 |
from mistralai import Mistral, TextChunk, ImageURLChunk
|
6 |
from PIL import Image
|
|
|
7 |
|
8 |
|
9 |
def process_msg(client, target):
|
@@ -15,14 +16,6 @@ def process_msg(client, target):
|
|
15 |
)
|
16 |
processed_output = image_response.pages[0].markdown
|
17 |
|
18 |
-
images=[]
|
19 |
-
base64_str = image_response.pages[0].images[0].image_base64
|
20 |
-
if "," in base64_str:
|
21 |
-
base64_str = base64_str.split(",")[1]
|
22 |
-
img_bytes = base64.b64decode(base64_str)
|
23 |
-
img_pil = Image.open(BytesIO(img_bytes))
|
24 |
-
images.append(img_pil)
|
25 |
-
|
26 |
messages = [
|
27 |
{
|
28 |
"role": "user",
|
@@ -39,7 +32,7 @@ def process_msg(client, target):
|
|
39 |
]
|
40 |
}
|
41 |
]
|
42 |
-
return messages
|
43 |
|
44 |
def chat_response(client, model, messages):
|
45 |
response = client.chat.complete(
|
@@ -73,8 +66,10 @@ def do_ocr(input_type, url=None):
|
|
73 |
#ocr = PaddleOCR(use_angle_cls=True, lang="en",use_gpu=False)
|
74 |
#ocr_response = ocr.ocr(document_source[document_source["type"]], cls=True)[0]
|
75 |
|
76 |
-
message
|
77 |
response = chat_response(client, model, message)
|
|
|
|
|
78 |
return response.choices[0].message.content, images
|
79 |
|
80 |
custom_css = """
|
@@ -114,7 +109,7 @@ with gr.Blocks(
|
|
114 |
img_example = gr.Button("Image", elem_classes="example-button")
|
115 |
with gr.Column(scale=2):
|
116 |
cleaned_output = gr.Textbox(label="Extracted Plain Text", lines=10, show_copy_button=True)
|
117 |
-
image_output = gr.Gallery(label="
|
118 |
|
119 |
def update_visibility(choice):
|
120 |
return gr.update(visible=(choice == "URL")), gr.update(visible=(choice == "Upload file"))
|
@@ -132,7 +127,7 @@ with gr.Blocks(
|
|
132 |
submit_btn.click(
|
133 |
fn=do_ocr,
|
134 |
inputs=[input_type, url_input],
|
135 |
-
outputs=[cleaned_output,
|
136 |
)
|
137 |
|
138 |
if __name__ == "__main__":
|
|
|
4 |
import gradio as gr
|
5 |
from mistralai import Mistral, TextChunk, ImageURLChunk
|
6 |
from PIL import Image
|
7 |
+
import requests
|
8 |
|
9 |
|
10 |
def process_msg(client, target):
|
|
|
16 |
)
|
17 |
processed_output = image_response.pages[0].markdown
|
18 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
messages = [
|
20 |
{
|
21 |
"role": "user",
|
|
|
32 |
]
|
33 |
}
|
34 |
]
|
35 |
+
return messages
|
36 |
|
37 |
def chat_response(client, model, messages):
|
38 |
response = client.chat.complete(
|
|
|
66 |
#ocr = PaddleOCR(use_angle_cls=True, lang="en",use_gpu=False)
|
67 |
#ocr_response = ocr.ocr(document_source[document_source["type"]], cls=True)[0]
|
68 |
|
69 |
+
message = process_msg(client, document_source[document_source["type"]])
|
70 |
response = chat_response(client, model, message)
|
71 |
+
images=[]
|
72 |
+
images.append(Image.open(BytesIO(requests.get(url).content)))
|
73 |
return response.choices[0].message.content, images
|
74 |
|
75 |
custom_css = """
|
|
|
109 |
img_example = gr.Button("Image", elem_classes="example-button")
|
110 |
with gr.Column(scale=2):
|
111 |
cleaned_output = gr.Textbox(label="Extracted Plain Text", lines=10, show_copy_button=True)
|
112 |
+
image_output = gr.Gallery(label="Preview", height="contain")
|
113 |
|
114 |
def update_visibility(choice):
|
115 |
return gr.update(visible=(choice == "URL")), gr.update(visible=(choice == "Upload file"))
|
|
|
127 |
submit_btn.click(
|
128 |
fn=do_ocr,
|
129 |
inputs=[input_type, url_input],
|
130 |
+
outputs=[cleaned_output, image_output]
|
131 |
)
|
132 |
|
133 |
if __name__ == "__main__":
|