Spaces:
Sleeping
Sleeping
Commit ·
e44ba24
1
Parent(s): 8be0822
phase 6: curate Space requirements + fix src import (T11)
Browse filesReplace the uv-export requirements.txt (it pinned gradio/huggingface_hub, which
Spaces preinstall and manage, and pulled the full eval/batch dep set via -e .)
with a demo-only list: docling, google-genai, pydantic, pydantic-settings. Put
src/ on sys.path in app.py so doc_agent imports on the Space without installing
the package. Add short_description to the README front-matter.
Per the huggingface-spaces skill: never pin gradio/huggingface_hub in a Space.
- README.md +1 -0
- app.py +10 -3
- requirements.txt +12 -556
README.md
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
---
|
| 2 |
title: Document Extraction Agent
|
|
|
|
| 3 |
colorFrom: indigo
|
| 4 |
colorTo: blue
|
| 5 |
sdk: gradio
|
|
|
|
| 1 |
---
|
| 2 |
title: Document Extraction Agent
|
| 3 |
+
short_description: Validate and auto-accept or review invoice/receipt fields
|
| 4 |
colorFrom: indigo
|
| 5 |
colorTo: blue
|
| 6 |
sdk: gradio
|
app.py
CHANGED
|
@@ -1,13 +1,20 @@
|
|
| 1 |
"""Hugging Face Space entry point.
|
| 2 |
|
| 3 |
A thin launcher for the Gradio demo. All UI and pipeline logic lives in
|
| 4 |
-
``doc_agent.web.app``; this file only
|
| 5 |
-
|
|
|
|
| 6 |
(GEMINI_API_KEY, EXTRACTION_BACKEND, IMAGE_STRATEGY, GEMINI_MODEL) are read from
|
| 7 |
the environment -- set them as Space repository secrets, never in a file.
|
| 8 |
"""
|
| 9 |
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
demo = build_demo()
|
| 13 |
|
|
|
|
| 1 |
"""Hugging Face Space entry point.
|
| 2 |
|
| 3 |
A thin launcher for the Gradio demo. All UI and pipeline logic lives in
|
| 4 |
+
``doc_agent.web.app``; this file only makes the ``src/`` package importable on
|
| 5 |
+
the Space (which runs this file from the repo root without pip-installing the
|
| 6 |
+
package) and then builds and launches the demo. Configuration and secrets
|
| 7 |
(GEMINI_API_KEY, EXTRACTION_BACKEND, IMAGE_STRATEGY, GEMINI_MODEL) are read from
|
| 8 |
the environment -- set them as Space repository secrets, never in a file.
|
| 9 |
"""
|
| 10 |
|
| 11 |
+
import sys
|
| 12 |
+
from pathlib import Path
|
| 13 |
+
|
| 14 |
+
# src-layout: make `doc_agent` importable without installing the package.
|
| 15 |
+
sys.path.insert(0, str(Path(__file__).parent / "src"))
|
| 16 |
+
|
| 17 |
+
from doc_agent.web.app import build_demo # noqa: E402 (import after path setup)
|
| 18 |
|
| 19 |
demo = build_demo()
|
| 20 |
|
requirements.txt
CHANGED
|
@@ -1,556 +1,12 @@
|
|
| 1 |
-
#
|
| 2 |
-
#
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
# via aiohttp
|
| 14 |
-
annotated-doc==0.0.4
|
| 15 |
-
# via
|
| 16 |
-
# fastapi
|
| 17 |
-
# typer
|
| 18 |
-
annotated-types==0.7.0
|
| 19 |
-
# via pydantic
|
| 20 |
-
antlr4-python3-runtime==4.9.3
|
| 21 |
-
# via omegaconf
|
| 22 |
-
anyio==4.14.1
|
| 23 |
-
# via
|
| 24 |
-
# google-genai
|
| 25 |
-
# gradio
|
| 26 |
-
# httpx
|
| 27 |
-
# starlette
|
| 28 |
-
attrs==26.1.0
|
| 29 |
-
# via
|
| 30 |
-
# aiohttp
|
| 31 |
-
# jsonlines
|
| 32 |
-
# jsonschema
|
| 33 |
-
# referencing
|
| 34 |
-
audioop-lts==0.2.2 ; python_full_version >= '3.13'
|
| 35 |
-
# via gradio
|
| 36 |
-
beautifulsoup4==4.15.0
|
| 37 |
-
# via docling-slim
|
| 38 |
-
brotli==1.2.0
|
| 39 |
-
# via gradio
|
| 40 |
-
certifi==2026.6.17
|
| 41 |
-
# via
|
| 42 |
-
# docling-slim
|
| 43 |
-
# httpcore
|
| 44 |
-
# httpx
|
| 45 |
-
# requests
|
| 46 |
-
cffi==2.0.0 ; platform_python_implementation != 'PyPy'
|
| 47 |
-
# via cryptography
|
| 48 |
-
charset-normalizer==3.4.7
|
| 49 |
-
# via requests
|
| 50 |
-
click==8.4.2
|
| 51 |
-
# via
|
| 52 |
-
# huggingface-hub
|
| 53 |
-
# typer
|
| 54 |
-
# uvicorn
|
| 55 |
-
colorama==0.4.6 ; sys_platform == 'win32'
|
| 56 |
-
# via
|
| 57 |
-
# click
|
| 58 |
-
# colorlog
|
| 59 |
-
# tqdm
|
| 60 |
-
colorlog==6.10.1
|
| 61 |
-
# via rapidocr
|
| 62 |
-
cryptography==49.0.0
|
| 63 |
-
# via google-auth
|
| 64 |
-
cuda-bindings==13.3.1 ; sys_platform == 'linux'
|
| 65 |
-
# via torch
|
| 66 |
-
cuda-pathfinder==1.5.6 ; sys_platform == 'linux'
|
| 67 |
-
# via cuda-bindings
|
| 68 |
-
cuda-toolkit==13.0.2 ; sys_platform == 'linux'
|
| 69 |
-
# via torch
|
| 70 |
-
datasets==5.0.0
|
| 71 |
-
# via doc-agent
|
| 72 |
-
defusedxml==0.7.1
|
| 73 |
-
# via
|
| 74 |
-
# docling-core
|
| 75 |
-
# docling-slim
|
| 76 |
-
dill==0.4.1
|
| 77 |
-
# via
|
| 78 |
-
# datasets
|
| 79 |
-
# multiprocess
|
| 80 |
-
distro==1.9.0
|
| 81 |
-
# via google-genai
|
| 82 |
-
doclang==0.7.0
|
| 83 |
-
# via docling-core
|
| 84 |
-
docling==2.107.0
|
| 85 |
-
# via doc-agent
|
| 86 |
-
docling-core==2.85.0
|
| 87 |
-
# via
|
| 88 |
-
# docling-ibm-models
|
| 89 |
-
# docling-parse
|
| 90 |
-
# docling-slim
|
| 91 |
-
docling-ibm-models==3.13.3
|
| 92 |
-
# via docling-slim
|
| 93 |
-
docling-parse==7.2.0
|
| 94 |
-
# via docling-slim
|
| 95 |
-
docling-slim==2.107.0
|
| 96 |
-
# via docling
|
| 97 |
-
et-xmlfile==2.0.0
|
| 98 |
-
# via openpyxl
|
| 99 |
-
faker==40.25.0
|
| 100 |
-
# via polyfactory
|
| 101 |
-
fastapi==0.138.2
|
| 102 |
-
# via gradio
|
| 103 |
-
filelock==3.29.4
|
| 104 |
-
# via
|
| 105 |
-
# datasets
|
| 106 |
-
# huggingface-hub
|
| 107 |
-
# torch
|
| 108 |
-
filetype==1.2.0
|
| 109 |
-
# via docling-slim
|
| 110 |
-
frozenlist==1.8.0
|
| 111 |
-
# via
|
| 112 |
-
# aiohttp
|
| 113 |
-
# aiosignal
|
| 114 |
-
fsspec==2026.4.0
|
| 115 |
-
# via
|
| 116 |
-
# datasets
|
| 117 |
-
# gradio-client
|
| 118 |
-
# huggingface-hub
|
| 119 |
-
# torch
|
| 120 |
-
google-auth==2.55.1
|
| 121 |
-
# via google-genai
|
| 122 |
-
google-genai==2.10.0
|
| 123 |
-
# via doc-agent
|
| 124 |
-
gradio==6.19.0
|
| 125 |
-
# via doc-agent
|
| 126 |
-
gradio-client==2.5.0
|
| 127 |
-
# via
|
| 128 |
-
# gradio
|
| 129 |
-
# hf-gradio
|
| 130 |
-
groovy==0.1.2
|
| 131 |
-
# via gradio
|
| 132 |
-
h11==0.16.0
|
| 133 |
-
# via
|
| 134 |
-
# httpcore
|
| 135 |
-
# uvicorn
|
| 136 |
-
hf-gradio==0.4.1
|
| 137 |
-
# via gradio
|
| 138 |
-
hf-xet==1.5.1 ; platform_machine == 'AMD64' or platform_machine == 'aarch64' or platform_machine == 'amd64' or platform_machine == 'arm64' or platform_machine == 'x86_64'
|
| 139 |
-
# via huggingface-hub
|
| 140 |
-
httpcore==1.0.9
|
| 141 |
-
# via httpx
|
| 142 |
-
httpx==0.28.1
|
| 143 |
-
# via
|
| 144 |
-
# datasets
|
| 145 |
-
# docling-slim
|
| 146 |
-
# google-genai
|
| 147 |
-
# gradio
|
| 148 |
-
# gradio-client
|
| 149 |
-
# huggingface-hub
|
| 150 |
-
# safehttpx
|
| 151 |
-
huggingface-hub==1.21.0
|
| 152 |
-
# via
|
| 153 |
-
# accelerate
|
| 154 |
-
# datasets
|
| 155 |
-
# docling-ibm-models
|
| 156 |
-
# docling-slim
|
| 157 |
-
# gradio
|
| 158 |
-
# gradio-client
|
| 159 |
-
# tokenizers
|
| 160 |
-
# transformers
|
| 161 |
-
idna==3.18
|
| 162 |
-
# via
|
| 163 |
-
# anyio
|
| 164 |
-
# httpx
|
| 165 |
-
# requests
|
| 166 |
-
# yarl
|
| 167 |
-
jinja2==3.1.6
|
| 168 |
-
# via
|
| 169 |
-
# gradio
|
| 170 |
-
# torch
|
| 171 |
-
jsonlines==4.0.0
|
| 172 |
-
# via docling-ibm-models
|
| 173 |
-
jsonref==1.1.0
|
| 174 |
-
# via docling-core
|
| 175 |
-
jsonschema==4.26.0
|
| 176 |
-
# via docling-core
|
| 177 |
-
jsonschema-specifications==2025.9.1
|
| 178 |
-
# via jsonschema
|
| 179 |
-
latex2mathml==3.81.0
|
| 180 |
-
# via docling-core
|
| 181 |
-
lxml==6.1.1
|
| 182 |
-
# via
|
| 183 |
-
# doclang
|
| 184 |
-
# docling-slim
|
| 185 |
-
# python-docx
|
| 186 |
-
# python-pptx
|
| 187 |
-
mail-parser==4.4.0
|
| 188 |
-
# via docling-slim
|
| 189 |
-
markdown-it-py==4.2.0
|
| 190 |
-
# via rich
|
| 191 |
-
marko==2.2.3
|
| 192 |
-
# via docling-slim
|
| 193 |
-
markupsafe==3.0.3
|
| 194 |
-
# via
|
| 195 |
-
# gradio
|
| 196 |
-
# jinja2
|
| 197 |
-
mdurl==0.1.2
|
| 198 |
-
# via markdown-it-py
|
| 199 |
-
mpire==2.10.2
|
| 200 |
-
# via semchunk
|
| 201 |
-
mpmath==1.3.0
|
| 202 |
-
# via sympy
|
| 203 |
-
multidict==6.7.1
|
| 204 |
-
# via
|
| 205 |
-
# aiohttp
|
| 206 |
-
# yarl
|
| 207 |
-
multiprocess==0.70.19
|
| 208 |
-
# via
|
| 209 |
-
# datasets
|
| 210 |
-
# mpire
|
| 211 |
-
networkx==3.6.1
|
| 212 |
-
# via torch
|
| 213 |
-
numpy==2.4.6 ; python_full_version < '3.12'
|
| 214 |
-
# via
|
| 215 |
-
# accelerate
|
| 216 |
-
# datasets
|
| 217 |
-
# docling-ibm-models
|
| 218 |
-
# docling-slim
|
| 219 |
-
# gradio
|
| 220 |
-
# opencv-python
|
| 221 |
-
# pandas
|
| 222 |
-
# rapidocr
|
| 223 |
-
# safetensors
|
| 224 |
-
# scipy
|
| 225 |
-
# shapely
|
| 226 |
-
# torchvision
|
| 227 |
-
# transformers
|
| 228 |
-
numpy==2.5.0 ; python_full_version >= '3.12'
|
| 229 |
-
# via
|
| 230 |
-
# accelerate
|
| 231 |
-
# datasets
|
| 232 |
-
# docling-ibm-models
|
| 233 |
-
# docling-slim
|
| 234 |
-
# gradio
|
| 235 |
-
# opencv-python
|
| 236 |
-
# pandas
|
| 237 |
-
# rapidocr
|
| 238 |
-
# safetensors
|
| 239 |
-
# scipy
|
| 240 |
-
# shapely
|
| 241 |
-
# torchvision
|
| 242 |
-
# transformers
|
| 243 |
-
nvidia-cublas==13.1.1.3 ; sys_platform == 'linux'
|
| 244 |
-
# via
|
| 245 |
-
# nvidia-cudnn-cu13
|
| 246 |
-
# nvidia-cusolver
|
| 247 |
-
# torch
|
| 248 |
-
nvidia-cuda-cupti==13.0.85 ; sys_platform == 'linux'
|
| 249 |
-
# via cuda-toolkit
|
| 250 |
-
nvidia-cuda-nvrtc==13.0.88 ; sys_platform == 'linux'
|
| 251 |
-
# via
|
| 252 |
-
# cuda-toolkit
|
| 253 |
-
# nvidia-cublas
|
| 254 |
-
nvidia-cuda-runtime==13.0.96 ; sys_platform == 'linux'
|
| 255 |
-
# via cuda-toolkit
|
| 256 |
-
nvidia-cudnn-cu13==9.20.0.48 ; sys_platform == 'linux'
|
| 257 |
-
# via torch
|
| 258 |
-
nvidia-cufft==12.0.0.61 ; sys_platform == 'linux'
|
| 259 |
-
# via cuda-toolkit
|
| 260 |
-
nvidia-cufile==1.15.1.6 ; sys_platform == 'linux'
|
| 261 |
-
# via cuda-toolkit
|
| 262 |
-
nvidia-curand==10.4.0.35 ; sys_platform == 'linux'
|
| 263 |
-
# via cuda-toolkit
|
| 264 |
-
nvidia-cusolver==12.0.4.66 ; sys_platform == 'linux'
|
| 265 |
-
# via cuda-toolkit
|
| 266 |
-
nvidia-cusparse==12.6.3.3 ; sys_platform == 'linux'
|
| 267 |
-
# via
|
| 268 |
-
# cuda-toolkit
|
| 269 |
-
# nvidia-cusolver
|
| 270 |
-
nvidia-cusparselt-cu13==0.8.1 ; sys_platform == 'linux'
|
| 271 |
-
# via torch
|
| 272 |
-
nvidia-nccl-cu13==2.29.7 ; sys_platform == 'linux'
|
| 273 |
-
# via torch
|
| 274 |
-
nvidia-nvjitlink==13.0.88 ; sys_platform == 'linux'
|
| 275 |
-
# via
|
| 276 |
-
# cuda-toolkit
|
| 277 |
-
# nvidia-cufft
|
| 278 |
-
# nvidia-cusolver
|
| 279 |
-
# nvidia-cusparse
|
| 280 |
-
nvidia-nvshmem-cu13==3.4.5 ; sys_platform == 'linux'
|
| 281 |
-
# via torch
|
| 282 |
-
nvidia-nvtx==13.0.85 ; sys_platform == 'linux'
|
| 283 |
-
# via cuda-toolkit
|
| 284 |
-
omegaconf==2.3.1
|
| 285 |
-
# via rapidocr
|
| 286 |
-
opencv-python==4.13.0.92
|
| 287 |
-
# via rapidocr
|
| 288 |
-
openpyxl==3.1.5
|
| 289 |
-
# via docling-slim
|
| 290 |
-
orjson==3.11.9
|
| 291 |
-
# via gradio
|
| 292 |
-
packaging==26.2
|
| 293 |
-
# via
|
| 294 |
-
# accelerate
|
| 295 |
-
# datasets
|
| 296 |
-
# gradio
|
| 297 |
-
# gradio-client
|
| 298 |
-
# huggingface-hub
|
| 299 |
-
# transformers
|
| 300 |
-
pandas==3.0.3
|
| 301 |
-
# via
|
| 302 |
-
# datasets
|
| 303 |
-
# docling-core
|
| 304 |
-
# gradio
|
| 305 |
-
pillow==12.2.0
|
| 306 |
-
# via
|
| 307 |
-
# datasets
|
| 308 |
-
# doc-agent
|
| 309 |
-
# docling-core
|
| 310 |
-
# docling-ibm-models
|
| 311 |
-
# docling-parse
|
| 312 |
-
# docling-slim
|
| 313 |
-
# gradio
|
| 314 |
-
# python-pptx
|
| 315 |
-
# rapidocr
|
| 316 |
-
# torchvision
|
| 317 |
-
pluggy==1.6.0
|
| 318 |
-
# via docling-slim
|
| 319 |
-
polyfactory==3.3.0
|
| 320 |
-
# via docling-slim
|
| 321 |
-
propcache==0.5.2
|
| 322 |
-
# via
|
| 323 |
-
# aiohttp
|
| 324 |
-
# yarl
|
| 325 |
-
psutil==7.2.2
|
| 326 |
-
# via accelerate
|
| 327 |
-
pyarrow==24.0.0
|
| 328 |
-
# via datasets
|
| 329 |
-
pyasn1==0.6.3
|
| 330 |
-
# via pyasn1-modules
|
| 331 |
-
pyasn1-modules==0.4.2
|
| 332 |
-
# via google-auth
|
| 333 |
-
pyclipper==1.4.0
|
| 334 |
-
# via rapidocr
|
| 335 |
-
pycparser==3.0 ; implementation_name != 'PyPy' and platform_python_implementation != 'PyPy'
|
| 336 |
-
# via cffi
|
| 337 |
-
pydantic==2.13.4
|
| 338 |
-
# via
|
| 339 |
-
# doc-agent
|
| 340 |
-
# docling-core
|
| 341 |
-
# docling-ibm-models
|
| 342 |
-
# docling-parse
|
| 343 |
-
# docling-slim
|
| 344 |
-
# fastapi
|
| 345 |
-
# google-genai
|
| 346 |
-
# gradio
|
| 347 |
-
# pydantic-settings
|
| 348 |
-
pydantic-core==2.46.4
|
| 349 |
-
# via pydantic
|
| 350 |
-
pydantic-settings==2.14.2
|
| 351 |
-
# via
|
| 352 |
-
# doc-agent
|
| 353 |
-
# docling-core
|
| 354 |
-
# docling-slim
|
| 355 |
-
pydub==0.25.1
|
| 356 |
-
# via gradio
|
| 357 |
-
pygments==2.20.0
|
| 358 |
-
# via
|
| 359 |
-
# mpire
|
| 360 |
-
# rich
|
| 361 |
-
pylatexenc==2.10
|
| 362 |
-
# via docling-slim
|
| 363 |
-
pypdfium2==5.11.0
|
| 364 |
-
# via docling-slim
|
| 365 |
-
python-dateutil==2.9.0.post0
|
| 366 |
-
# via pandas
|
| 367 |
-
python-docx==1.2.0
|
| 368 |
-
# via docling-slim
|
| 369 |
-
python-dotenv==1.2.2
|
| 370 |
-
# via
|
| 371 |
-
# docling-slim
|
| 372 |
-
# pydantic-settings
|
| 373 |
-
python-multipart==0.0.32
|
| 374 |
-
# via gradio
|
| 375 |
-
python-pptx==1.0.2
|
| 376 |
-
# via docling-slim
|
| 377 |
-
pytz==2026.2
|
| 378 |
-
# via gradio
|
| 379 |
-
pywin32==312 ; sys_platform == 'win32'
|
| 380 |
-
# via
|
| 381 |
-
# docling-parse
|
| 382 |
-
# mpire
|
| 383 |
-
pyyaml==6.0.3
|
| 384 |
-
# via
|
| 385 |
-
# accelerate
|
| 386 |
-
# datasets
|
| 387 |
-
# docling-core
|
| 388 |
-
# gradio
|
| 389 |
-
# huggingface-hub
|
| 390 |
-
# omegaconf
|
| 391 |
-
# rapidocr
|
| 392 |
-
# transformers
|
| 393 |
-
rapidocr==3.9.0
|
| 394 |
-
# via docling-slim
|
| 395 |
-
referencing==0.37.0
|
| 396 |
-
# via
|
| 397 |
-
# jsonschema
|
| 398 |
-
# jsonschema-specifications
|
| 399 |
-
regex==2026.6.28
|
| 400 |
-
# via transformers
|
| 401 |
-
requests==2.34.2
|
| 402 |
-
# via
|
| 403 |
-
# datasets
|
| 404 |
-
# docling-slim
|
| 405 |
-
# google-auth
|
| 406 |
-
# google-genai
|
| 407 |
-
# rapidocr
|
| 408 |
-
rich==15.0.0
|
| 409 |
-
# via
|
| 410 |
-
# docling-slim
|
| 411 |
-
# typer
|
| 412 |
-
rpds-py==2026.6.3
|
| 413 |
-
# via
|
| 414 |
-
# jsonschema
|
| 415 |
-
# referencing
|
| 416 |
-
rtree==1.4.1
|
| 417 |
-
# via
|
| 418 |
-
# docling-ibm-models
|
| 419 |
-
# docling-slim
|
| 420 |
-
safehttpx==0.1.7
|
| 421 |
-
# via gradio
|
| 422 |
-
safetensors==0.8.0
|
| 423 |
-
# via
|
| 424 |
-
# accelerate
|
| 425 |
-
# docling-ibm-models
|
| 426 |
-
# transformers
|
| 427 |
-
saxonche==13.0.0
|
| 428 |
-
# via doclang
|
| 429 |
-
scipy==1.17.1 ; python_full_version < '3.12'
|
| 430 |
-
# via docling-slim
|
| 431 |
-
scipy==1.18.0 ; python_full_version >= '3.12'
|
| 432 |
-
# via docling-slim
|
| 433 |
-
semantic-version==2.10.0
|
| 434 |
-
# via gradio
|
| 435 |
-
semchunk==3.2.5
|
| 436 |
-
# via docling-core
|
| 437 |
-
setuptools==81.0.0
|
| 438 |
-
# via torch
|
| 439 |
-
shapely==2.1.2
|
| 440 |
-
# via rapidocr
|
| 441 |
-
shellingham==1.5.4
|
| 442 |
-
# via typer
|
| 443 |
-
six==1.17.0
|
| 444 |
-
# via
|
| 445 |
-
# python-dateutil
|
| 446 |
-
# rapidocr
|
| 447 |
-
sniffio==1.3.1
|
| 448 |
-
# via google-genai
|
| 449 |
-
soupsieve==2.8.4
|
| 450 |
-
# via beautifulsoup4
|
| 451 |
-
starlette==1.3.1
|
| 452 |
-
# via
|
| 453 |
-
# fastapi
|
| 454 |
-
# gradio
|
| 455 |
-
sympy==1.14.0
|
| 456 |
-
# via torch
|
| 457 |
-
tabulate==0.10.0
|
| 458 |
-
# via docling-core
|
| 459 |
-
tenacity==9.1.4
|
| 460 |
-
# via google-genai
|
| 461 |
-
tokenizers==0.22.2
|
| 462 |
-
# via transformers
|
| 463 |
-
tomlkit==0.14.0
|
| 464 |
-
# via gradio
|
| 465 |
-
torch==2.12.1
|
| 466 |
-
# via
|
| 467 |
-
# accelerate
|
| 468 |
-
# docling-ibm-models
|
| 469 |
-
# docling-slim
|
| 470 |
-
# safetensors
|
| 471 |
-
# torchvision
|
| 472 |
-
torchvision==0.27.1
|
| 473 |
-
# via
|
| 474 |
-
# docling-ibm-models
|
| 475 |
-
# docling-slim
|
| 476 |
-
tqdm==4.68.3
|
| 477 |
-
# via
|
| 478 |
-
# datasets
|
| 479 |
-
# docling-ibm-models
|
| 480 |
-
# docling-slim
|
| 481 |
-
# huggingface-hub
|
| 482 |
-
# mpire
|
| 483 |
-
# rapidocr
|
| 484 |
-
# semchunk
|
| 485 |
-
# transformers
|
| 486 |
-
transformers==5.8.1
|
| 487 |
-
# via
|
| 488 |
-
# docling-core
|
| 489 |
-
# docling-ibm-models
|
| 490 |
-
tree-sitter==0.26.0
|
| 491 |
-
# via docling-core
|
| 492 |
-
tree-sitter-c==0.24.2
|
| 493 |
-
# via docling-core
|
| 494 |
-
tree-sitter-javascript==0.25.0
|
| 495 |
-
# via docling-core
|
| 496 |
-
tree-sitter-python==0.25.0
|
| 497 |
-
# via docling-core
|
| 498 |
-
tree-sitter-typescript==0.23.2
|
| 499 |
-
# via docling-core
|
| 500 |
-
triton==3.7.1 ; sys_platform == 'linux'
|
| 501 |
-
# via torch
|
| 502 |
-
typer==0.24.2
|
| 503 |
-
# via
|
| 504 |
-
# doclang
|
| 505 |
-
# docling-core
|
| 506 |
-
# docling-slim
|
| 507 |
-
# gradio
|
| 508 |
-
# hf-gradio
|
| 509 |
-
# huggingface-hub
|
| 510 |
-
# transformers
|
| 511 |
-
typing-extensions==4.15.0
|
| 512 |
-
# via
|
| 513 |
-
# aiohttp
|
| 514 |
-
# aiosignal
|
| 515 |
-
# anyio
|
| 516 |
-
# beautifulsoup4
|
| 517 |
-
# docling-core
|
| 518 |
-
# fastapi
|
| 519 |
-
# google-genai
|
| 520 |
-
# gradio
|
| 521 |
-
# gradio-client
|
| 522 |
-
# huggingface-hub
|
| 523 |
-
# polyfactory
|
| 524 |
-
# pydantic
|
| 525 |
-
# pydantic-core
|
| 526 |
-
# python-docx
|
| 527 |
-
# python-pptx
|
| 528 |
-
# referencing
|
| 529 |
-
# starlette
|
| 530 |
-
# torch
|
| 531 |
-
# typing-inspection
|
| 532 |
-
typing-inspection==0.4.2
|
| 533 |
-
# via
|
| 534 |
-
# fastapi
|
| 535 |
-
# pydantic
|
| 536 |
-
# pydantic-settings
|
| 537 |
-
tzdata==2026.2 ; sys_platform == 'emscripten' or sys_platform == 'win32'
|
| 538 |
-
# via
|
| 539 |
-
# faker
|
| 540 |
-
# pandas
|
| 541 |
-
urllib3==2.7.0
|
| 542 |
-
# via requests
|
| 543 |
-
uvicorn==0.49.0
|
| 544 |
-
# via gradio
|
| 545 |
-
watchdog==6.0.0
|
| 546 |
-
# via doc-agent
|
| 547 |
-
websockets==16.0
|
| 548 |
-
# via
|
| 549 |
-
# docling-slim
|
| 550 |
-
# google-genai
|
| 551 |
-
xlsxwriter==3.2.9
|
| 552 |
-
# via python-pptx
|
| 553 |
-
xxhash==3.8.0
|
| 554 |
-
# via datasets
|
| 555 |
-
yarl==1.24.2
|
| 556 |
-
# via aiohttp
|
|
|
|
| 1 |
+
# Hugging Face Space -- web-demo dependencies only.
|
| 2 |
+
#
|
| 3 |
+
# Deliberately curated, not `uv export`: Spaces preinstall and manage `gradio`
|
| 4 |
+
# and `huggingface_hub`, so listing/pinning them causes resolution failures.
|
| 5 |
+
# `datasets` (eval-only), `pillow` (eval-only), and `watchdog` (batch watcher)
|
| 6 |
+
# are not imported by the demo, so they are omitted to keep the build light.
|
| 7 |
+
# `docling` pulls torch for the native-PDF path; the image path uses Gemini
|
| 8 |
+
# vision and needs only google-genai.
|
| 9 |
+
docling>=2.107.0
|
| 10 |
+
google-genai>=2.10.0
|
| 11 |
+
pydantic>=2.13.4
|
| 12 |
+
pydantic-settings>=2.14.2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|