Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -35,6 +35,114 @@ from magic_pdf.data.data_reader_writer import FileBasedDataReader
|
|
| 35 |
from magic_pdf.libs.hash_utils import compute_sha256
|
| 36 |
from magic_pdf.tools.common import do_parse, prepare_env
|
| 37 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
def read_fn(path):
|
| 39 |
disk_rw = FileBasedDataReader(os.path.dirname(path))
|
| 40 |
return disk_rw.read(os.path.basename(path))
|
|
@@ -113,6 +221,18 @@ def to_markdown(file_path, end_pages, is_ocr, layout_mode, formula_enable, table
|
|
| 113 |
new_pdf_path = os.path.join(local_md_dir, file_name + "_layout.pdf")
|
| 114 |
return md_content, txt_content, archive_zip_path, new_pdf_path
|
| 115 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
latex_delimiters = [{"left": "$$", "right": "$$", "display": True},
|
| 117 |
{"left": '$', "right": '$', "display": False}]
|
| 118 |
|
|
@@ -152,36 +272,81 @@ other_lang = ['ch', 'en', 'korean', 'japan', 'chinese_cht', 'ta', 'te', 'ka']
|
|
| 152 |
all_lang = ['', 'auto']
|
| 153 |
all_lang.extend([*other_lang, *latin_lang, *arabic_lang, *cyrillic_lang, *devanagari_lang])
|
| 154 |
|
| 155 |
-
def to_pdf(file_path):
|
| 156 |
-
with pymupdf.open(file_path) as f:
|
| 157 |
-
if f.is_pdf:
|
| 158 |
-
return file_path
|
| 159 |
-
else:
|
| 160 |
-
pdf_bytes = f.convert_to_pdf()
|
| 161 |
-
unique_filename = f"{uuid.uuid4()}.pdf"
|
| 162 |
-
tmp_file_path = os.path.join(os.path.dirname(file_path), unique_filename)
|
| 163 |
-
with open(tmp_file_path, 'wb') as tmp_pdf_file:
|
| 164 |
-
tmp_pdf_file.write(pdf_bytes)
|
| 165 |
-
return tmp_file_path
|
| 166 |
-
|
| 167 |
if __name__ == "__main__":
|
| 168 |
-
with gr.Blocks(title="OCR FLEX") as demo:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 169 |
with gr.Row():
|
|
|
|
| 170 |
with gr.Column(variant='panel', scale=5):
|
| 171 |
-
file = gr.File(
|
| 172 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 173 |
with gr.Row():
|
| 174 |
-
layout_mode = gr.Dropdown(
|
| 175 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 176 |
with gr.Row():
|
| 177 |
-
formula_enable = gr.Checkbox(
|
| 178 |
-
|
| 179 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 180 |
with gr.Row():
|
| 181 |
-
change_bu = gr.Button(
|
| 182 |
-
|
| 183 |
-
|
| 184 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 185 |
example_root = os.path.join(os.path.dirname(__file__), "examples")
|
| 186 |
gr.Examples(
|
| 187 |
examples=[os.path.join(example_root, _) for _ in os.listdir(example_root) if
|
|
@@ -189,18 +354,60 @@ if __name__ == "__main__":
|
|
| 189 |
inputs=file
|
| 190 |
)
|
| 191 |
|
|
|
|
| 192 |
with gr.Column(variant='panel', scale=5):
|
| 193 |
-
output_file = gr.File(
|
| 194 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 195 |
with gr.Tab("๋งํฌ๋ค์ด ๋ ๋๋ง"):
|
| 196 |
-
md = gr.Markdown(
|
| 197 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 198 |
with gr.Tab("๋งํฌ๋ค์ด ํ
์คํธ"):
|
| 199 |
-
md_text = gr.TextArea(
|
| 200 |
-
|
| 201 |
-
|
| 202 |
-
|
| 203 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 204 |
clear_bu.add([file, md, pdf_show, md_text, output_file, is_ocr])
|
| 205 |
|
| 206 |
-
|
|
|
|
|
|
| 35 |
from magic_pdf.libs.hash_utils import compute_sha256
|
| 36 |
from magic_pdf.tools.common import do_parse, prepare_env
|
| 37 |
|
| 38 |
+
def create_css():
|
| 39 |
+
return """
|
| 40 |
+
/* ์ ์ฒด ์คํ์ผ */
|
| 41 |
+
.gradio-container {
|
| 42 |
+
background: linear-gradient(135deg, #EFF6FF 0%, #F5F3FF 100%);
|
| 43 |
+
max-width: 1200px !important;
|
| 44 |
+
margin: 0 auto !important;
|
| 45 |
+
padding: 2rem !important;
|
| 46 |
+
}
|
| 47 |
+
/* ์ ๋ชฉ ์คํ์ผ */
|
| 48 |
+
.title-area {
|
| 49 |
+
text-align: center;
|
| 50 |
+
margin-bottom: 2rem;
|
| 51 |
+
padding: 1rem;
|
| 52 |
+
background: white;
|
| 53 |
+
border-radius: 1rem;
|
| 54 |
+
box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
|
| 55 |
+
}
|
| 56 |
+
.title-area h1 {
|
| 57 |
+
background: linear-gradient(90deg, #2563EB 0%, #7C3AED 100%);
|
| 58 |
+
-webkit-background-clip: text;
|
| 59 |
+
-webkit-text-fill-color: transparent;
|
| 60 |
+
font-size: 2.5rem;
|
| 61 |
+
font-weight: bold;
|
| 62 |
+
margin-bottom: 0.5rem;
|
| 63 |
+
}
|
| 64 |
+
.title-area p {
|
| 65 |
+
color: #6B7280;
|
| 66 |
+
font-size: 1.1rem;
|
| 67 |
+
}
|
| 68 |
+
/* ์ปดํฌ๋ํธ ์คํ์ผ๋ง */
|
| 69 |
+
.gr-box, .gr-panel {
|
| 70 |
+
border: 2px solid #E0E7FF !important;
|
| 71 |
+
border-radius: 12px !important;
|
| 72 |
+
box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1) !important;
|
| 73 |
+
background: white !important;
|
| 74 |
+
}
|
| 75 |
+
/* ํ์ผ ์
๋ก๋ ์์ญ */
|
| 76 |
+
.file-upload {
|
| 77 |
+
border: 2px dashed #93C5FD !important;
|
| 78 |
+
border-radius: 8px !important;
|
| 79 |
+
padding: 2rem !important;
|
| 80 |
+
background: #F0F9FF !important;
|
| 81 |
+
transition: all 0.3s ease;
|
| 82 |
+
}
|
| 83 |
+
.file-upload:hover {
|
| 84 |
+
background: #E0F2FE !important;
|
| 85 |
+
border-color: #60A5FA !important;
|
| 86 |
+
}
|
| 87 |
+
/* ๋ฒํผ ์คํ์ผ๋ง */
|
| 88 |
+
.gr-button.primary-button {
|
| 89 |
+
background: linear-gradient(90deg, #2563EB 0%, #7C3AED 100%) !important;
|
| 90 |
+
color: white !important;
|
| 91 |
+
border: none !important;
|
| 92 |
+
border-radius: 8px !important;
|
| 93 |
+
padding: 0.75rem 1.5rem !important;
|
| 94 |
+
font-weight: bold !important;
|
| 95 |
+
transition: opacity 0.2s !important;
|
| 96 |
+
}
|
| 97 |
+
.gr-button.primary-button:hover {
|
| 98 |
+
opacity: 0.9 !important;
|
| 99 |
+
}
|
| 100 |
+
.gr-button.secondary-button {
|
| 101 |
+
background: white !important;
|
| 102 |
+
color: #4B5563 !important;
|
| 103 |
+
border: 1px solid #D1D5DB !important;
|
| 104 |
+
border-radius: 8px !important;
|
| 105 |
+
padding: 0.75rem 1.5rem !important;
|
| 106 |
+
}
|
| 107 |
+
.gr-button.secondary-button:hover {
|
| 108 |
+
background: #F9FAFB !important;
|
| 109 |
+
}
|
| 110 |
+
/* ์ฌ๋ผ์ด๋ ์คํ์ผ๋ง */
|
| 111 |
+
.gr-slider {
|
| 112 |
+
background: #E0E7FF !important;
|
| 113 |
+
}
|
| 114 |
+
.gr-slider .gr-slider-handle {
|
| 115 |
+
background: #4F46E5 !important;
|
| 116 |
+
}
|
| 117 |
+
/* ์ฒดํฌ๋ฐ์ค ์คํ์ผ๋ง */
|
| 118 |
+
.gr-checkbox {
|
| 119 |
+
border-color: #6366F1 !important;
|
| 120 |
+
}
|
| 121 |
+
.gr-checkbox:checked {
|
| 122 |
+
background-color: #4F46E5 !important;
|
| 123 |
+
}
|
| 124 |
+
/* ํญ ์คํ์ผ๋ง */
|
| 125 |
+
.gr-tabs {
|
| 126 |
+
border-bottom: 2px solid #E0E7FF !important;
|
| 127 |
+
}
|
| 128 |
+
.gr-tab-button {
|
| 129 |
+
color: #6B7280 !important;
|
| 130 |
+
padding: 0.75rem 1rem !important;
|
| 131 |
+
font-weight: 500 !important;
|
| 132 |
+
}
|
| 133 |
+
.gr-tab-button.selected {
|
| 134 |
+
color: #4F46E5 !important;
|
| 135 |
+
border-bottom: 2px solid #4F46E5 !important;
|
| 136 |
+
}
|
| 137 |
+
/* ๋งํฌ๋ค์ด ์ถ๋ ฅ ์์ญ */
|
| 138 |
+
.markdown-output {
|
| 139 |
+
background: white !important;
|
| 140 |
+
border-radius: 8px !important;
|
| 141 |
+
padding: 1rem !important;
|
| 142 |
+
box-shadow: inset 0 2px 4px rgba(0, 0, 0, 0.05) !important;
|
| 143 |
+
}
|
| 144 |
+
"""
|
| 145 |
+
|
| 146 |
def read_fn(path):
|
| 147 |
disk_rw = FileBasedDataReader(os.path.dirname(path))
|
| 148 |
return disk_rw.read(os.path.basename(path))
|
|
|
|
| 221 |
new_pdf_path = os.path.join(local_md_dir, file_name + "_layout.pdf")
|
| 222 |
return md_content, txt_content, archive_zip_path, new_pdf_path
|
| 223 |
|
| 224 |
+
def to_pdf(file_path):
|
| 225 |
+
with pymupdf.open(file_path) as f:
|
| 226 |
+
if f.is_pdf:
|
| 227 |
+
return file_path
|
| 228 |
+
else:
|
| 229 |
+
pdf_bytes = f.convert_to_pdf()
|
| 230 |
+
unique_filename = f"{uuid.uuid4()}.pdf"
|
| 231 |
+
tmp_file_path = os.path.join(os.path.dirname(file_path), unique_filename)
|
| 232 |
+
with open(tmp_file_path, 'wb') as tmp_pdf_file:
|
| 233 |
+
tmp_pdf_file.write(pdf_bytes)
|
| 234 |
+
return tmp_file_path
|
| 235 |
+
|
| 236 |
latex_delimiters = [{"left": "$$", "right": "$$", "display": True},
|
| 237 |
{"left": '$', "right": '$', "display": False}]
|
| 238 |
|
|
|
|
| 272 |
all_lang = ['', 'auto']
|
| 273 |
all_lang.extend([*other_lang, *latin_lang, *arabic_lang, *cyrillic_lang, *devanagari_lang])
|
| 274 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 275 |
if __name__ == "__main__":
|
| 276 |
+
with gr.Blocks(title="OCR FLEX", css=create_css()) as demo:
|
| 277 |
+
# ํ์ดํ ์์ญ
|
| 278 |
+
with gr.Row(elem_classes="title-area"):
|
| 279 |
+
gr.HTML("""
|
| 280 |
+
<h1>OCR FLEX</h1>
|
| 281 |
+
<p>PDF์ ์ด๋ฏธ์ง์์ ํ
์คํธ๋ฅผ ๋น ๋ฅด๊ณ ์ ํํ๊ฒ ์ถ์ถํ์ธ์</p>
|
| 282 |
+
""")
|
| 283 |
+
|
| 284 |
with gr.Row():
|
| 285 |
+
# ์ผ์ชฝ ํจ๋
|
| 286 |
with gr.Column(variant='panel', scale=5):
|
| 287 |
+
file = gr.File(
|
| 288 |
+
label="PDF ๋๋ ์ด๋ฏธ์ง ํ์ผ์ ์
๋ก๋ํ์ธ์",
|
| 289 |
+
file_types=[".pdf", ".png", ".jpeg", ".jpg"],
|
| 290 |
+
elem_classes="file-upload"
|
| 291 |
+
)
|
| 292 |
+
|
| 293 |
+
max_pages = gr.Slider(
|
| 294 |
+
1, 20, 10,
|
| 295 |
+
step=1,
|
| 296 |
+
label='์ต๋ ๋ณํ ํ์ด์ง ์',
|
| 297 |
+
elem_classes="custom-slider"
|
| 298 |
+
)
|
| 299 |
+
|
| 300 |
with gr.Row():
|
| 301 |
+
layout_mode = gr.Dropdown(
|
| 302 |
+
["layoutlmv3", "doclayout_yolo"],
|
| 303 |
+
label="๋ ์ด์์ ๋ชจ๋ธ",
|
| 304 |
+
value="doclayout_yolo",
|
| 305 |
+
elem_classes="custom-dropdown"
|
| 306 |
+
)
|
| 307 |
+
language = gr.Dropdown(
|
| 308 |
+
all_lang,
|
| 309 |
+
label="์ธ์ด",
|
| 310 |
+
value='auto',
|
| 311 |
+
elem_classes="custom-dropdown"
|
| 312 |
+
)
|
| 313 |
+
|
| 314 |
with gr.Row():
|
| 315 |
+
formula_enable = gr.Checkbox(
|
| 316 |
+
label="์์ ์ธ์ ํ์ฑํ",
|
| 317 |
+
value=True,
|
| 318 |
+
elem_classes="custom-checkbox"
|
| 319 |
+
)
|
| 320 |
+
is_ocr = gr.Checkbox(
|
| 321 |
+
label="OCR ๊ฐ์ ํ์ฑํ",
|
| 322 |
+
value=False,
|
| 323 |
+
elem_classes="custom-checkbox"
|
| 324 |
+
)
|
| 325 |
+
table_enable = gr.Checkbox(
|
| 326 |
+
label="ํ ์ธ์ ํ์ฑํ(ํ
์คํธ)",
|
| 327 |
+
value=True,
|
| 328 |
+
elem_classes="custom-checkbox"
|
| 329 |
+
)
|
| 330 |
+
|
| 331 |
with gr.Row():
|
| 332 |
+
change_bu = gr.Button(
|
| 333 |
+
"๋ณํ",
|
| 334 |
+
elem_classes="primary-button"
|
| 335 |
+
)
|
| 336 |
+
clear_bu = gr.ClearButton(
|
| 337 |
+
value="์ด๊ธฐํ",
|
| 338 |
+
elem_classes="secondary-button"
|
| 339 |
+
)
|
| 340 |
+
|
| 341 |
+
pdf_show = PDF(
|
| 342 |
+
label='PDF ๋ฏธ๋ฆฌ๋ณด๊ธฐ',
|
| 343 |
+
interactive=False,
|
| 344 |
+
visible=True,
|
| 345 |
+
height=800,
|
| 346 |
+
elem_classes="pdf-preview"
|
| 347 |
+
)
|
| 348 |
+
|
| 349 |
+
with gr.Accordion("์์ :", open=False):
|
| 350 |
example_root = os.path.join(os.path.dirname(__file__), "examples")
|
| 351 |
gr.Examples(
|
| 352 |
examples=[os.path.join(example_root, _) for _ in os.listdir(example_root) if
|
|
|
|
| 354 |
inputs=file
|
| 355 |
)
|
| 356 |
|
| 357 |
+
# ์ค๋ฅธ์ชฝ ํจ๋
|
| 358 |
with gr.Column(variant='panel', scale=5):
|
| 359 |
+
output_file = gr.File(
|
| 360 |
+
label="๋ณํ ๊ฒฐ๊ณผ",
|
| 361 |
+
interactive=False,
|
| 362 |
+
elem_classes="output-file"
|
| 363 |
+
)
|
| 364 |
+
|
| 365 |
+
with gr.Tabs() as tabs:
|
| 366 |
with gr.Tab("๋งํฌ๋ค์ด ๋ ๋๋ง"):
|
| 367 |
+
md = gr.Markdown(
|
| 368 |
+
label="๋งํฌ๋ค์ด ๋ ๋๋ง",
|
| 369 |
+
height=1100,
|
| 370 |
+
show_copy_button=True,
|
| 371 |
+
latex_delimiters=latex_delimiters,
|
| 372 |
+
line_breaks=True,
|
| 373 |
+
elem_classes="markdown-output"
|
| 374 |
+
)
|
| 375 |
+
|
| 376 |
with gr.Tab("๋งํฌ๋ค์ด ํ
์คํธ"):
|
| 377 |
+
md_text = gr.TextArea(
|
| 378 |
+
lines=45,
|
| 379 |
+
show_copy_button=True,
|
| 380 |
+
elem_classes="markdown-text"
|
| 381 |
+
)
|
| 382 |
+
|
| 383 |
+
# ์ด๋ฒคํธ ํธ๋ค๋ฌ
|
| 384 |
+
file.change(
|
| 385 |
+
fn=to_pdf,
|
| 386 |
+
inputs=file,
|
| 387 |
+
outputs=pdf_show
|
| 388 |
+
)
|
| 389 |
+
|
| 390 |
+
change_bu.click(
|
| 391 |
+
fn=to_markdown,
|
| 392 |
+
inputs=[
|
| 393 |
+
file,
|
| 394 |
+
max_pages,
|
| 395 |
+
is_ocr,
|
| 396 |
+
layout_mode,
|
| 397 |
+
formula_enable,
|
| 398 |
+
table_enable,
|
| 399 |
+
language
|
| 400 |
+
],
|
| 401 |
+
outputs=[
|
| 402 |
+
md,
|
| 403 |
+
md_text,
|
| 404 |
+
output_file,
|
| 405 |
+
pdf_show
|
| 406 |
+
],
|
| 407 |
+
api_name=False
|
| 408 |
+
)
|
| 409 |
+
|
| 410 |
clear_bu.add([file, md, pdf_show, md_text, output_file, is_ocr])
|
| 411 |
|
| 412 |
+
# ์ฑ ์คํ
|
| 413 |
+
demo.launch(ssr_mode=True)
|