Spaces:
Running
on
Zero
Running
on
Zero
Liam Dyer
commited on
initial
Browse files- README.md +1 -1
- app.py +31 -0
- requirements.txt +1 -0
README.md
CHANGED
@@ -1,4 +1,4 @@
|
|
1 |
---
|
2 |
license: gpl-3.0
|
3 |
-
title:
|
4 |
---
|
|
|
1 |
---
|
2 |
license: gpl-3.0
|
3 |
+
title: PDF to Markdown
|
4 |
---
|
app.py
ADDED
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import argparse
|
2 |
+
import os
|
3 |
+
|
4 |
+
from marker.convert import convert_single_pdf
|
5 |
+
from marker.logger import configure_logging
|
6 |
+
from marker.models import load_all_models
|
7 |
+
|
8 |
+
import spaces
|
9 |
+
import gradio as gr
|
10 |
+
from marker.output import save_markdown
|
11 |
+
|
12 |
+
|
13 |
+
@spaces.GPU
|
14 |
+
def convert(file_path):
|
15 |
+
model_lst = load_all_models()
|
16 |
+
full_text, images, out_meta = convert_single_pdf(
|
17 |
+
file_path,
|
18 |
+
model_lst,
|
19 |
+
max_pages=None,
|
20 |
+
langs=None,
|
21 |
+
batch_multiplier=16,
|
22 |
+
)
|
23 |
+
|
24 |
+
return full_text
|
25 |
+
|
26 |
+
|
27 |
+
gr.Interface(
|
28 |
+
convert,
|
29 |
+
inputs=gr.File(label="PDF file", sources=["upload"], type="filepath"),
|
30 |
+
outputs=gr.Markdown(label="Markdown"),
|
31 |
+
).launch()
|
requirements.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
marker-pdf=0.2.5
|