Spaces:
Running
Running
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from pdfitdown.pdfconversion import convert_to_pdf, convert_markdown_to_pdf
|
2 |
+
import warnings
|
3 |
+
from typing import List
|
4 |
+
import gradio as gr
|
5 |
+
|
6 |
+
class FileNotConvertedWarning(Warning):
|
7 |
+
"""The file was not in one of the specified formats for conversion to PDF,thus it was not converted"""
|
8 |
+
|
9 |
+
def to_pdf(files: List[str]) -> List[str]:
|
10 |
+
"""
|
11 |
+
Converts various file formats to PDF.
|
12 |
+
|
13 |
+
Args:
|
14 |
+
files: List of file paths to convert. Supports .docx, .pdf, .html, .pptx,
|
15 |
+
.csv, .xml, and .md files.
|
16 |
+
|
17 |
+
Returns:
|
18 |
+
List of paths to converted PDF files. For files already in PDF format,
|
19 |
+
returns original path.
|
20 |
+
|
21 |
+
Raises:
|
22 |
+
FileNotConvertedWarning: When file format is not supported.
|
23 |
+
"""
|
24 |
+
pdfs = []
|
25 |
+
for f in files:
|
26 |
+
if f.endswith(".docx"):
|
27 |
+
newfile = f.replace(".docx", ".pdf")
|
28 |
+
file_to_add = convert_to_pdf(f, newfile, newfile.split(".")[0])
|
29 |
+
pdfs.append(file_to_add)
|
30 |
+
elif f.endswith(".pdf"):
|
31 |
+
pdfs.append(f)
|
32 |
+
elif f.endswith(".html"):
|
33 |
+
newfile = f.replace(".html", ".pdf")
|
34 |
+
file_to_add = convert_to_pdf(f, newfile, newfile.split(".")[0])
|
35 |
+
pdfs.append(file_to_add)
|
36 |
+
elif f.endswith(".pptx"):
|
37 |
+
newfile = f.replace(".pptx", ".pdf")
|
38 |
+
file_to_add = convert_to_pdf(f, newfile, newfile.split(".")[0])
|
39 |
+
pdfs.append(file_to_add)
|
40 |
+
elif f.endswith(".csv"):
|
41 |
+
newfile = f.replace(".csv", ".pdf")
|
42 |
+
file_to_add = convert_to_pdf(f, newfile, newfile.split(".")[0])
|
43 |
+
pdfs.append(file_to_add)
|
44 |
+
elif f.endswith(".xml"):
|
45 |
+
newfile = f.replace(".xml", ".pdf")
|
46 |
+
file_to_add = convert_to_pdf(f, newfile, newfile.split(".")[0])
|
47 |
+
pdfs.append(file_to_add)
|
48 |
+
elif f.endswith(".md"):
|
49 |
+
newfile = f.replace(".md", ".pdf")
|
50 |
+
file_to_add = convert_markdown_to_pdf(f, newfile, newfile.split(".")[0])
|
51 |
+
pdfs.append(file_to_add)
|
52 |
+
else:
|
53 |
+
warnings.warn(f"File {f} was not converted to PDF because its file format is not included in those that can be converted", FileNotConvertedWarning)
|
54 |
+
continue
|
55 |
+
return pdfs
|
56 |
+
|
57 |
+
def convert(file: str) -> str:
|
58 |
+
files = [file]
|
59 |
+
pdfs = to_pdf(files)
|
60 |
+
return pdfs[0]
|
61 |
+
|
62 |
+
iface = gr.Interface(
|
63 |
+
fn=convert,
|
64 |
+
inputs=gr.File(label="Upload your file"),
|
65 |
+
outputs=gr.File(label="Converted PDF"),
|
66 |
+
title="File to PDF Converter",
|
67 |
+
description="Upload a file in .docx, .pdf, .html, .pptx, .csv, .xml, or .md format, and get it converted to PDF."
|
68 |
+
)
|
69 |
+
iface.launch()
|