Spaces:
Runtime error
Runtime error
Delete app.py
Browse files
app.py
DELETED
|
@@ -1,117 +0,0 @@
|
|
| 1 |
-
import os, tempfile, traceback
|
| 2 |
-
import gradio as gr
|
| 3 |
-
import spaces
|
| 4 |
-
import requests
|
| 5 |
-
|
| 6 |
-
# Persist caches in /data so future runs are fast
|
| 7 |
-
os.environ.setdefault("HF_HOME", "/data/.cache/huggingface")
|
| 8 |
-
os.environ.setdefault("TOKENIZERS_PARALLELISM", "false")
|
| 9 |
-
|
| 10 |
-
from docling.datamodel.base_models import InputFormat
|
| 11 |
-
from docling.document_converter import DocumentConverter, PdfFormatOption
|
| 12 |
-
from docling.pipeline.vlm_pipeline import VlmPipeline
|
| 13 |
-
|
| 14 |
-
# (Optional) reduce thread pressure on ZeroGPU
|
| 15 |
-
try:
|
| 16 |
-
import torch
|
| 17 |
-
torch.set_num_threads(max(1, int(os.environ.get("OMP_NUM_THREADS", "2"))))
|
| 18 |
-
HAS_CUDA = torch.cuda.is_available()
|
| 19 |
-
except Exception:
|
| 20 |
-
HAS_CUDA = False
|
| 21 |
-
|
| 22 |
-
# Build converters once
|
| 23 |
-
std_converter = DocumentConverter(
|
| 24 |
-
format_options={InputFormat.PDF: PdfFormatOption()}
|
| 25 |
-
)
|
| 26 |
-
vlm_converter = DocumentConverter(
|
| 27 |
-
format_options={InputFormat.PDF: PdfFormatOption(pipeline_cls=VlmPipeline)}
|
| 28 |
-
)
|
| 29 |
-
|
| 30 |
-
def _success(md: str, html: str):
|
| 31 |
-
tmpdir = tempfile.gettempdir()
|
| 32 |
-
md_path = os.path.join(tmpdir, "output.md")
|
| 33 |
-
html_path = os.path.join(tmpdir, "output.html")
|
| 34 |
-
with open(md_path, "w", encoding="utf-8") as f: f.write(md)
|
| 35 |
-
with open(html_path, "w", encoding="utf-8") as f: f.write(html)
|
| 36 |
-
return md, md_path, html_path
|
| 37 |
-
|
| 38 |
-
def _fail(msg: str):
|
| 39 |
-
# Show the full error text in the Markdown preview
|
| 40 |
-
return f"**Conversion failed**:\n```\n{msg}\n```", None, None
|
| 41 |
-
|
| 42 |
-
def _convert_source(path: str, use_vlm: bool):
|
| 43 |
-
try:
|
| 44 |
-
converter = vlm_converter if use_vlm else std_converter
|
| 45 |
-
doc = converter.convert(source=path).document
|
| 46 |
-
md = doc.export_to_markdown()
|
| 47 |
-
html = doc.export_to_html()
|
| 48 |
-
return _success(md, html)
|
| 49 |
-
except Exception as e:
|
| 50 |
-
return _fail(f"{e}\n\n{traceback.format_exc()}")
|
| 51 |
-
|
| 52 |
-
@spaces.GPU(duration=600)
|
| 53 |
-
def run_convert_file(file, mode):
|
| 54 |
-
if file is None:
|
| 55 |
-
return _fail("No file provided.")
|
| 56 |
-
use_vlm = mode.startswith("VLM")
|
| 57 |
-
return _convert_source(file.name, use_vlm)
|
| 58 |
-
|
| 59 |
-
@spaces.GPU(duration=600)
|
| 60 |
-
def run_convert_url(url, mode):
|
| 61 |
-
if not url:
|
| 62 |
-
return _fail("No URL provided.")
|
| 63 |
-
# Download to a temp file so Docling always sees a local path
|
| 64 |
-
try:
|
| 65 |
-
r = requests.get(url, stream=True, timeout=60)
|
| 66 |
-
r.raise_for_status()
|
| 67 |
-
fd, tmp_path = tempfile.mkstemp(suffix=".pdf")
|
| 68 |
-
with os.fdopen(fd, "wb") as tmp:
|
| 69 |
-
for chunk in r.iter_content(chunk_size=1 << 20):
|
| 70 |
-
if chunk:
|
| 71 |
-
tmp.write(chunk)
|
| 72 |
-
except Exception as e:
|
| 73 |
-
return _fail(f"Failed to download URL: {e}")
|
| 74 |
-
use_vlm = mode.startswith("VLM")
|
| 75 |
-
out = _convert_source(tmp_path, use_vlm)
|
| 76 |
-
try:
|
| 77 |
-
os.remove(tmp_path)
|
| 78 |
-
except Exception:
|
| 79 |
-
pass
|
| 80 |
-
return out
|
| 81 |
-
|
| 82 |
-
subtitle = "Device: **CUDA (ZeroGPU)**" if HAS_CUDA else "Device: **CPU** (GPU warms on first call)"
|
| 83 |
-
|
| 84 |
-
with gr.Blocks(title="Granite-Docling 258M β PDF β Markdown/HTML") as demo:
|
| 85 |
-
gr.Markdown(
|
| 86 |
-
f"""# Granite-Docling 258M β PDF β Markdown / HTML
|
| 87 |
-
{subtitle}
|
| 88 |
-
|
| 89 |
-
**Modes**
|
| 90 |
-
- **Standard (faster)** β PDFs with a text layer
|
| 91 |
-
- **VLM (Granite β better for complex/scanned)** β scans / heavy tables / formulas
|
| 92 |
-
|
| 93 |
-
_First call may be slow while models download and ZeroGPU warms. Cache lives in `/data`._
|
| 94 |
-
"""
|
| 95 |
-
)
|
| 96 |
-
|
| 97 |
-
mode = gr.Radio(
|
| 98 |
-
["Standard (faster)", "VLM (Granite β better for complex/scanned)"],
|
| 99 |
-
value="Standard (faster)", label="Mode"
|
| 100 |
-
)
|
| 101 |
-
|
| 102 |
-
with gr.Tab("Upload PDF"):
|
| 103 |
-
fi = gr.File(file_types=[".pdf"], label="PDF")
|
| 104 |
-
out_md = gr.Markdown(label="Markdown Preview")
|
| 105 |
-
dl_md = gr.File(label="Download Markdown")
|
| 106 |
-
dl_html = gr.File(label="Download HTML")
|
| 107 |
-
gr.Button("Convert").click(run_convert_file, [fi, mode], [out_md, dl_md, dl_html])
|
| 108 |
-
|
| 109 |
-
with gr.Tab("Convert from URL"):
|
| 110 |
-
url = gr.Textbox(label="Public PDF URL", placeholder="https://.../file.pdf")
|
| 111 |
-
out2_md = gr.Markdown(label="Markdown Preview")
|
| 112 |
-
dl2_md = gr.File(label="Download Markdown")
|
| 113 |
-
dl2_html = gr.File(label="Download HTML")
|
| 114 |
-
gr.Button("Convert").click(run_convert_url, [url, mode], [out2_md, dl2_md, dl2_html])
|
| 115 |
-
|
| 116 |
-
# Explicit bind/queue so Spaces picks up the interface
|
| 117 |
-
demo.queue().launch(server_name="0.0.0.0", server_port=int(os.environ.get("PORT", 7860)))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|