Spaces:
Paused
Paused
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,155 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import base64
|
| 2 |
+
import io
|
| 3 |
+
import os
|
| 4 |
+
import zipfile
|
| 5 |
+
from threading import Thread
|
| 6 |
+
|
| 7 |
+
import dash
|
| 8 |
+
import dash_bootstrap_components as dbc
|
| 9 |
+
from dash import dcc, html, Input, Output, State, callback
|
| 10 |
+
from dash.exceptions import PreventUpdate
|
| 11 |
+
from PyPDF2 import PdfReader, PdfWriter
|
| 12 |
+
|
| 13 |
+
app = dash.Dash(__name__, external_stylesheets=[dbc.themes.BOOTSTRAP])
|
| 14 |
+
|
| 15 |
+
# Global variables
|
| 16 |
+
generated_file = None
|
| 17 |
+
progress = 0
|
| 18 |
+
|
| 19 |
+
# Layout
|
| 20 |
+
app.layout = dbc.Container([
|
| 21 |
+
html.H1("PDF Splitter", className="my-4"),
|
| 22 |
+
dcc.Upload(
|
| 23 |
+
id='upload-pdf',
|
| 24 |
+
children=html.Div(['Drag and Drop or ', html.A('Select PDF')]),
|
| 25 |
+
style={
|
| 26 |
+
'width': '100%',
|
| 27 |
+
'height': '60px',
|
| 28 |
+
'lineHeight': '60px',
|
| 29 |
+
'borderWidth': '1px',
|
| 30 |
+
'borderStyle': 'dashed',
|
| 31 |
+
'borderRadius': '5px',
|
| 32 |
+
'textAlign': 'center',
|
| 33 |
+
'margin': '10px'
|
| 34 |
+
},
|
| 35 |
+
multiple=False
|
| 36 |
+
),
|
| 37 |
+
html.Div(id='pdf-name'),
|
| 38 |
+
dbc.Card([
|
| 39 |
+
dbc.CardBody([
|
| 40 |
+
dbc.Input(id='page-range', type='text', placeholder='Enter page range (e.g., 1-3)'),
|
| 41 |
+
dbc.Button("Add Range", id='add-range', color="secondary", className="mt-2"),
|
| 42 |
+
html.Div(id='ranges-list'),
|
| 43 |
+
])
|
| 44 |
+
], className="my-3"),
|
| 45 |
+
dbc.Button("Split PDF", id='split-button', color="primary", className="mt-3", disabled=True),
|
| 46 |
+
dcc.Progress(id='progress-bar', className="my-3"),
|
| 47 |
+
dbc.Button("Download ZIP", id='download-button', color="success", className="mt-3", disabled=True),
|
| 48 |
+
dcc.Download(id="download-zip"),
|
| 49 |
+
], fluid=True)
|
| 50 |
+
|
| 51 |
+
@callback(
|
| 52 |
+
Output('pdf-name', 'children'),
|
| 53 |
+
Output('split-button', 'disabled'),
|
| 54 |
+
Input('upload-pdf', 'contents'),
|
| 55 |
+
State('upload-pdf', 'filename')
|
| 56 |
+
)
|
| 57 |
+
def update_output(contents, filename):
|
| 58 |
+
if contents is not None:
|
| 59 |
+
return f"Selected file: {filename}", False
|
| 60 |
+
return "No file selected", True
|
| 61 |
+
|
| 62 |
+
@callback(
|
| 63 |
+
Output('ranges-list', 'children'),
|
| 64 |
+
Input('add-range', 'n_clicks'),
|
| 65 |
+
State('page-range', 'value'),
|
| 66 |
+
State('ranges-list', 'children')
|
| 67 |
+
)
|
| 68 |
+
def add_range(n_clicks, new_range, existing_ranges):
|
| 69 |
+
if n_clicks is None or not new_range:
|
| 70 |
+
return existing_ranges or []
|
| 71 |
+
new_item = html.Div(f"Range: {new_range}")
|
| 72 |
+
return existing_ranges + [new_item] if existing_ranges else [new_item]
|
| 73 |
+
|
| 74 |
+
def process_pdf(contents, filename, ranges):
|
| 75 |
+
global generated_file, progress
|
| 76 |
+
progress = 0
|
| 77 |
+
|
| 78 |
+
try:
|
| 79 |
+
content_type, content_string = contents.split(',')
|
| 80 |
+
decoded = base64.b64decode(content_string)
|
| 81 |
+
pdf = PdfReader(io.BytesIO(decoded))
|
| 82 |
+
|
| 83 |
+
writers = []
|
| 84 |
+
for range_str in ranges:
|
| 85 |
+
start, end = map(int, range_str.split('-'))
|
| 86 |
+
writer = PdfWriter()
|
| 87 |
+
for i in range(start - 1, min(end, len(pdf.pages))):
|
| 88 |
+
writer.add_page(pdf.pages[i])
|
| 89 |
+
writers.append(writer)
|
| 90 |
+
|
| 91 |
+
zip_buffer = io.BytesIO()
|
| 92 |
+
with zipfile.ZipFile(zip_buffer, 'w', zipfile.ZIP_DEFLATED) as zf:
|
| 93 |
+
for i, writer in enumerate(writers):
|
| 94 |
+
progress = (i + 1) / len(writers) * 100
|
| 95 |
+
pdf_buffer = io.BytesIO()
|
| 96 |
+
writer.write(pdf_buffer)
|
| 97 |
+
pdf_buffer.seek(0)
|
| 98 |
+
zf.writestr(f"split_{i+1}.pdf", pdf_buffer.getvalue())
|
| 99 |
+
|
| 100 |
+
zip_buffer.seek(0)
|
| 101 |
+
generated_file = zip_buffer.getvalue()
|
| 102 |
+
progress = 100
|
| 103 |
+
except Exception as e:
|
| 104 |
+
print(f"Error processing PDF: {str(e)}")
|
| 105 |
+
progress = -1
|
| 106 |
+
|
| 107 |
+
@callback(
|
| 108 |
+
Output('progress-bar', 'value'),
|
| 109 |
+
Output('download-button', 'disabled'),
|
| 110 |
+
Input('split-button', 'n_clicks'),
|
| 111 |
+
State('upload-pdf', 'contents'),
|
| 112 |
+
State('upload-pdf', 'filename'),
|
| 113 |
+
State('ranges-list', 'children'),
|
| 114 |
+
prevent_initial_call=True
|
| 115 |
+
)
|
| 116 |
+
def split_pdf(n_clicks, contents, filename, ranges):
|
| 117 |
+
if not contents or not ranges:
|
| 118 |
+
raise PreventUpdate
|
| 119 |
+
|
| 120 |
+
ranges = [r['props']['children'].split(': ')[1] for r in ranges]
|
| 121 |
+
thread = Thread(target=process_pdf, args=(contents, filename, ranges))
|
| 122 |
+
thread.start()
|
| 123 |
+
|
| 124 |
+
return 0, True
|
| 125 |
+
|
| 126 |
+
@callback(
|
| 127 |
+
Output('progress-bar', 'value', allow_duplicate=True),
|
| 128 |
+
Output('download-button', 'disabled', allow_duplicate=True),
|
| 129 |
+
Input('progress-bar', 'value'),
|
| 130 |
+
prevent_initial_call=True
|
| 131 |
+
)
|
| 132 |
+
def update_progress(value):
|
| 133 |
+
global progress
|
| 134 |
+
if progress == 100:
|
| 135 |
+
return 100, False
|
| 136 |
+
elif progress == -1:
|
| 137 |
+
return 0, True
|
| 138 |
+
else:
|
| 139 |
+
return progress, True
|
| 140 |
+
|
| 141 |
+
@callback(
|
| 142 |
+
Output("download-zip", "data"),
|
| 143 |
+
Input("download-button", "n_clicks"),
|
| 144 |
+
prevent_initial_call=True
|
| 145 |
+
)
|
| 146 |
+
def download_zip(n_clicks):
|
| 147 |
+
global generated_file
|
| 148 |
+
if generated_file is not None:
|
| 149 |
+
return dcc.send_bytes(generated_file, "split_pdfs.zip")
|
| 150 |
+
raise PreventUpdate
|
| 151 |
+
|
| 152 |
+
if __name__ == '__main__':
|
| 153 |
+
print("Starting the Dash application...")
|
| 154 |
+
app.run(debug=True, host='0.0.0.0', port=7860)
|
| 155 |
+
print("Dash application has finished running.")
|