Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -3,50 +3,49 @@ from docling.document_converter import DocumentConverter
|
|
3 |
from docling.datamodel.pipeline_options import PdfPipelineOptions, TesseractCliOcrOptions
|
4 |
import spaces
|
5 |
|
|
|
6 |
@spaces.GPU
|
7 |
def convert_document(file, output_format):
|
8 |
pdf_opts = PdfPipelineOptions(
|
9 |
do_ocr=True,
|
10 |
-
ocr_options=TesseractCliOcrOptions(lang=["eng"]),
|
11 |
)
|
12 |
-
|
13 |
converter = DocumentConverter(pipeline_options=pdf_opts)
|
14 |
result = converter.convert(file.name)
|
15 |
|
16 |
-
#
|
17 |
-
available_attributes = dir(result.document)
|
18 |
-
|
19 |
-
# Choose the output format
|
20 |
if output_format == "Markdown":
|
21 |
converted_text = result.document.export_to_markdown()
|
22 |
elif output_format == "JSON":
|
23 |
-
|
|
|
|
|
24 |
else:
|
25 |
-
converted_text = "Unsupported format"
|
26 |
|
27 |
-
#
|
28 |
-
metadata = {
|
29 |
-
"Available Attributes": available_attributes
|
30 |
-
}
|
31 |
|
32 |
return converted_text, metadata
|
33 |
|
34 |
-
|
35 |
with gr.Blocks() as app:
|
36 |
-
gr.Markdown("# Document Converter with Docling")
|
37 |
-
gr.Markdown("Upload a
|
|
|
|
|
|
|
|
|
38 |
|
39 |
-
|
40 |
-
format_input = gr.Radio(["Markdown", "JSON"], label="Choose Output Format")
|
41 |
-
output_text = gr.Textbox(label="Converted Document")
|
42 |
output_metadata = gr.JSON(label="Metadata")
|
43 |
|
44 |
-
# Define the process button and bind it to the function
|
45 |
convert_button = gr.Button("Convert")
|
46 |
convert_button.click(
|
47 |
-
convert_document,
|
48 |
inputs=[file_input, format_input],
|
49 |
outputs=[output_text, output_metadata]
|
50 |
)
|
51 |
|
52 |
-
app.launch(debug=True)
|
|
|
3 |
from docling.datamodel.pipeline_options import PdfPipelineOptions, TesseractCliOcrOptions
|
4 |
import spaces
|
5 |
|
6 |
+
# GPU decorator not really required for Docling OCR, but kept if you want
|
7 |
@spaces.GPU
|
8 |
def convert_document(file, output_format):
|
9 |
pdf_opts = PdfPipelineOptions(
|
10 |
do_ocr=True,
|
11 |
+
ocr_options=TesseractCliOcrOptions(lang=["eng"]),
|
12 |
)
|
13 |
+
|
14 |
converter = DocumentConverter(pipeline_options=pdf_opts)
|
15 |
result = converter.convert(file.name)
|
16 |
|
17 |
+
# Choose output format safely
|
|
|
|
|
|
|
18 |
if output_format == "Markdown":
|
19 |
converted_text = result.document.export_to_markdown()
|
20 |
elif output_format == "JSON":
|
21 |
+
# JSON needs to be dumped into a string for the Textbox
|
22 |
+
import json
|
23 |
+
converted_text = json.dumps(result.document.export_to_json(), indent=2)
|
24 |
else:
|
25 |
+
converted_text = "⚠️ Unsupported format"
|
26 |
|
27 |
+
# Metadata always JSON-friendly
|
28 |
+
metadata = {"Available Attributes": dir(result.document)}
|
|
|
|
|
29 |
|
30 |
return converted_text, metadata
|
31 |
|
32 |
+
|
33 |
with gr.Blocks() as app:
|
34 |
+
gr.Markdown("# 📄 Document Converter with Docling OCR")
|
35 |
+
gr.Markdown("Upload a PDF, choose the output format, and get the converted text + metadata.")
|
36 |
+
|
37 |
+
with gr.Row():
|
38 |
+
file_input = gr.File(label="Upload PDF", file_types=[".pdf"])
|
39 |
+
format_input = gr.Radio(["Markdown", "JSON"], label="Choose Output Format")
|
40 |
|
41 |
+
output_text = gr.Textbox(label="Converted Document", lines=20)
|
|
|
|
|
42 |
output_metadata = gr.JSON(label="Metadata")
|
43 |
|
|
|
44 |
convert_button = gr.Button("Convert")
|
45 |
convert_button.click(
|
46 |
+
fn=convert_document,
|
47 |
inputs=[file_input, format_input],
|
48 |
outputs=[output_text, output_metadata]
|
49 |
)
|
50 |
|
51 |
+
app.launch(debug=True)
|