Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
|
@@ -2,6 +2,7 @@ import base64
|
|
| 2 |
import io
|
| 3 |
import os
|
| 4 |
import zipfile
|
|
|
|
| 5 |
from threading import Thread
|
| 6 |
|
| 7 |
import dash
|
|
@@ -10,6 +11,10 @@ from dash import dcc, html, Input, Output, State, callback, MATCH, ALL
|
|
| 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
|
|
@@ -50,6 +55,7 @@ app.layout = dbc.Container([
|
|
| 50 |
dbc.Progress(id='progress-bar', className="my-3"),
|
| 51 |
dbc.Button("Download ZIP", id='download-button', color="success", className="mt-3", disabled=True),
|
| 52 |
dcc.Download(id="download-zip"),
|
|
|
|
| 53 |
], fluid=True)
|
| 54 |
|
| 55 |
@callback(
|
|
@@ -60,6 +66,7 @@ app.layout = dbc.Container([
|
|
| 60 |
)
|
| 61 |
def update_output(contents, filename):
|
| 62 |
if contents is not None:
|
|
|
|
| 63 |
return f"Selected file: {filename}", False
|
| 64 |
return "No file selected", True
|
| 65 |
|
|
@@ -75,6 +82,7 @@ def manage_ranges(add_clicks, remove_clicks, existing_ranges):
|
|
| 75 |
triggered_id = ctx.triggered[0]['prop_id'].split('.')[0]
|
| 76 |
|
| 77 |
if triggered_id == 'add-range':
|
|
|
|
| 78 |
new_index = len(existing_ranges)
|
| 79 |
new_range = dbc.Row([
|
| 80 |
dbc.Col(dbc.Input(id={'type': 'range-input', 'index': new_index}, type='text', placeholder='Enter page range (e.g., 1-3)'), width=10),
|
|
@@ -83,6 +91,7 @@ def manage_ranges(add_clicks, remove_clicks, existing_ranges):
|
|
| 83 |
existing_ranges.append(new_range)
|
| 84 |
elif 'remove-range' in triggered_id:
|
| 85 |
remove_index = json.loads(triggered_id)['index']
|
|
|
|
| 86 |
existing_ranges = [range for range in existing_ranges if json.loads(range['props']['children'][1]['props']['children']['props']['id'])['index'] != remove_index]
|
| 87 |
|
| 88 |
return existing_ranges
|
|
@@ -90,15 +99,18 @@ def manage_ranges(add_clicks, remove_clicks, existing_ranges):
|
|
| 90 |
def process_pdf(contents, filename, ranges):
|
| 91 |
global generated_file, progress
|
| 92 |
progress = 0
|
|
|
|
| 93 |
|
| 94 |
try:
|
| 95 |
content_type, content_string = contents.split(',')
|
| 96 |
decoded = base64.b64decode(content_string)
|
| 97 |
pdf = PdfReader(io.BytesIO(decoded))
|
|
|
|
| 98 |
|
| 99 |
writers = []
|
| 100 |
for range_str in ranges:
|
| 101 |
start, end = map(int, range_str.split('-'))
|
|
|
|
| 102 |
writer = PdfWriter()
|
| 103 |
for i in range(start - 1, min(end, len(pdf.pages))):
|
| 104 |
writer.add_page(pdf.pages[i])
|
|
@@ -108,6 +120,7 @@ def process_pdf(contents, filename, ranges):
|
|
| 108 |
with zipfile.ZipFile(zip_buffer, 'w', zipfile.ZIP_DEFLATED) as zf:
|
| 109 |
for i, writer in enumerate(writers):
|
| 110 |
progress = (i + 1) / len(writers) * 100
|
|
|
|
| 111 |
pdf_buffer = io.BytesIO()
|
| 112 |
writer.write(pdf_buffer)
|
| 113 |
pdf_buffer.seek(0)
|
|
@@ -116,13 +129,15 @@ def process_pdf(contents, filename, ranges):
|
|
| 116 |
zip_buffer.seek(0)
|
| 117 |
generated_file = zip_buffer.getvalue()
|
| 118 |
progress = 100
|
|
|
|
| 119 |
except Exception as e:
|
| 120 |
-
|
| 121 |
progress = -1
|
| 122 |
|
| 123 |
@callback(
|
| 124 |
Output('progress-bar', 'value'),
|
| 125 |
Output('download-button', 'disabled'),
|
|
|
|
| 126 |
Input('split-button', 'n_clicks'),
|
| 127 |
State('upload-pdf', 'contents'),
|
| 128 |
State('upload-pdf', 'filename'),
|
|
@@ -131,28 +146,35 @@ def process_pdf(contents, filename, ranges):
|
|
| 131 |
)
|
| 132 |
def split_pdf(n_clicks, contents, filename, ranges):
|
| 133 |
if not contents or not ranges:
|
|
|
|
| 134 |
raise PreventUpdate
|
| 135 |
|
|
|
|
| 136 |
ranges = [r for r in ranges if r] # Filter out empty ranges
|
|
|
|
|
|
|
| 137 |
thread = Thread(target=process_pdf, args=(contents, filename, ranges))
|
| 138 |
thread.start()
|
| 139 |
|
| 140 |
-
return 0, True
|
| 141 |
|
| 142 |
@callback(
|
| 143 |
Output('progress-bar', 'value', allow_duplicate=True),
|
| 144 |
Output('download-button', 'disabled', allow_duplicate=True),
|
|
|
|
| 145 |
Input('progress-bar', 'value'),
|
| 146 |
prevent_initial_call=True
|
| 147 |
)
|
| 148 |
def update_progress(value):
|
| 149 |
global progress
|
| 150 |
if progress == 100:
|
| 151 |
-
|
|
|
|
| 152 |
elif progress == -1:
|
| 153 |
-
|
|
|
|
| 154 |
else:
|
| 155 |
-
return progress, True
|
| 156 |
|
| 157 |
@callback(
|
| 158 |
Output("download-zip", "data"),
|
|
@@ -162,10 +184,12 @@ def update_progress(value):
|
|
| 162 |
def download_zip(n_clicks):
|
| 163 |
global generated_file
|
| 164 |
if generated_file is not None:
|
|
|
|
| 165 |
return dcc.send_bytes(generated_file, "split_pdfs.zip")
|
|
|
|
| 166 |
raise PreventUpdate
|
| 167 |
|
| 168 |
if __name__ == '__main__':
|
| 169 |
-
|
| 170 |
app.run(debug=True, host='0.0.0.0', port=7860)
|
| 171 |
-
|
|
|
|
| 2 |
import io
|
| 3 |
import os
|
| 4 |
import zipfile
|
| 5 |
+
import logging
|
| 6 |
from threading import Thread
|
| 7 |
|
| 8 |
import dash
|
|
|
|
| 11 |
from dash.exceptions import PreventUpdate
|
| 12 |
from PyPDF2 import PdfReader, PdfWriter
|
| 13 |
|
| 14 |
+
# Set up logging
|
| 15 |
+
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
|
| 16 |
+
logger = logging.getLogger(__name__)
|
| 17 |
+
|
| 18 |
app = dash.Dash(__name__, external_stylesheets=[dbc.themes.BOOTSTRAP])
|
| 19 |
|
| 20 |
# Global variables
|
|
|
|
| 55 |
dbc.Progress(id='progress-bar', className="my-3"),
|
| 56 |
dbc.Button("Download ZIP", id='download-button', color="success", className="mt-3", disabled=True),
|
| 57 |
dcc.Download(id="download-zip"),
|
| 58 |
+
html.Div(id='log-output', style={'whiteSpace': 'pre-line'}), # Add this line for log output
|
| 59 |
], fluid=True)
|
| 60 |
|
| 61 |
@callback(
|
|
|
|
| 66 |
)
|
| 67 |
def update_output(contents, filename):
|
| 68 |
if contents is not None:
|
| 69 |
+
logger.info(f"PDF file uploaded: {filename}")
|
| 70 |
return f"Selected file: {filename}", False
|
| 71 |
return "No file selected", True
|
| 72 |
|
|
|
|
| 82 |
triggered_id = ctx.triggered[0]['prop_id'].split('.')[0]
|
| 83 |
|
| 84 |
if triggered_id == 'add-range':
|
| 85 |
+
logger.info("Adding new range input")
|
| 86 |
new_index = len(existing_ranges)
|
| 87 |
new_range = dbc.Row([
|
| 88 |
dbc.Col(dbc.Input(id={'type': 'range-input', 'index': new_index}, type='text', placeholder='Enter page range (e.g., 1-3)'), width=10),
|
|
|
|
| 91 |
existing_ranges.append(new_range)
|
| 92 |
elif 'remove-range' in triggered_id:
|
| 93 |
remove_index = json.loads(triggered_id)['index']
|
| 94 |
+
logger.info(f"Removing range input at index {remove_index}")
|
| 95 |
existing_ranges = [range for range in existing_ranges if json.loads(range['props']['children'][1]['props']['children']['props']['id'])['index'] != remove_index]
|
| 96 |
|
| 97 |
return existing_ranges
|
|
|
|
| 99 |
def process_pdf(contents, filename, ranges):
|
| 100 |
global generated_file, progress
|
| 101 |
progress = 0
|
| 102 |
+
logger.info(f"Starting PDF processing for file: {filename}")
|
| 103 |
|
| 104 |
try:
|
| 105 |
content_type, content_string = contents.split(',')
|
| 106 |
decoded = base64.b64decode(content_string)
|
| 107 |
pdf = PdfReader(io.BytesIO(decoded))
|
| 108 |
+
logger.info(f"PDF loaded successfully. Total pages: {len(pdf.pages)}")
|
| 109 |
|
| 110 |
writers = []
|
| 111 |
for range_str in ranges:
|
| 112 |
start, end = map(int, range_str.split('-'))
|
| 113 |
+
logger.info(f"Processing range: {start}-{end}")
|
| 114 |
writer = PdfWriter()
|
| 115 |
for i in range(start - 1, min(end, len(pdf.pages))):
|
| 116 |
writer.add_page(pdf.pages[i])
|
|
|
|
| 120 |
with zipfile.ZipFile(zip_buffer, 'w', zipfile.ZIP_DEFLATED) as zf:
|
| 121 |
for i, writer in enumerate(writers):
|
| 122 |
progress = (i + 1) / len(writers) * 100
|
| 123 |
+
logger.info(f"Creating split PDF {i+1}/{len(writers)}")
|
| 124 |
pdf_buffer = io.BytesIO()
|
| 125 |
writer.write(pdf_buffer)
|
| 126 |
pdf_buffer.seek(0)
|
|
|
|
| 129 |
zip_buffer.seek(0)
|
| 130 |
generated_file = zip_buffer.getvalue()
|
| 131 |
progress = 100
|
| 132 |
+
logger.info("PDF processing completed successfully")
|
| 133 |
except Exception as e:
|
| 134 |
+
logger.error(f"Error processing PDF: {str(e)}", exc_info=True)
|
| 135 |
progress = -1
|
| 136 |
|
| 137 |
@callback(
|
| 138 |
Output('progress-bar', 'value'),
|
| 139 |
Output('download-button', 'disabled'),
|
| 140 |
+
Output('log-output', 'children'),
|
| 141 |
Input('split-button', 'n_clicks'),
|
| 142 |
State('upload-pdf', 'contents'),
|
| 143 |
State('upload-pdf', 'filename'),
|
|
|
|
| 146 |
)
|
| 147 |
def split_pdf(n_clicks, contents, filename, ranges):
|
| 148 |
if not contents or not ranges:
|
| 149 |
+
logger.warning("Split PDF clicked but no content or ranges provided")
|
| 150 |
raise PreventUpdate
|
| 151 |
|
| 152 |
+
logger.info("Split PDF button clicked")
|
| 153 |
ranges = [r for r in ranges if r] # Filter out empty ranges
|
| 154 |
+
logger.info(f"Processing {len(ranges)} ranges")
|
| 155 |
+
|
| 156 |
thread = Thread(target=process_pdf, args=(contents, filename, ranges))
|
| 157 |
thread.start()
|
| 158 |
|
| 159 |
+
return 0, True, "PDF splitting process started. Check console for detailed logs."
|
| 160 |
|
| 161 |
@callback(
|
| 162 |
Output('progress-bar', 'value', allow_duplicate=True),
|
| 163 |
Output('download-button', 'disabled', allow_duplicate=True),
|
| 164 |
+
Output('log-output', 'children', allow_duplicate=True),
|
| 165 |
Input('progress-bar', 'value'),
|
| 166 |
prevent_initial_call=True
|
| 167 |
)
|
| 168 |
def update_progress(value):
|
| 169 |
global progress
|
| 170 |
if progress == 100:
|
| 171 |
+
logger.info("PDF splitting completed")
|
| 172 |
+
return 100, False, "PDF splitting completed. Click 'Download ZIP' to get your files."
|
| 173 |
elif progress == -1:
|
| 174 |
+
logger.error("PDF splitting failed")
|
| 175 |
+
return 0, True, "Error occurred during PDF splitting. Check console for details."
|
| 176 |
else:
|
| 177 |
+
return progress, True, f"Processing... {progress:.0f}% complete"
|
| 178 |
|
| 179 |
@callback(
|
| 180 |
Output("download-zip", "data"),
|
|
|
|
| 184 |
def download_zip(n_clicks):
|
| 185 |
global generated_file
|
| 186 |
if generated_file is not None:
|
| 187 |
+
logger.info("Initiating ZIP file download")
|
| 188 |
return dcc.send_bytes(generated_file, "split_pdfs.zip")
|
| 189 |
+
logger.warning("Download attempted but no file generated")
|
| 190 |
raise PreventUpdate
|
| 191 |
|
| 192 |
if __name__ == '__main__':
|
| 193 |
+
logger.info("Starting the Dash application...")
|
| 194 |
app.run(debug=True, host='0.0.0.0', port=7860)
|
| 195 |
+
logger.info("Dash application has finished running.")
|