Spaces:
Running
Running
Update app2.py
Browse files
app2.py
CHANGED
@@ -4,23 +4,19 @@ import re
|
|
4 |
import time
|
5 |
import logging
|
6 |
import mimetypes
|
7 |
-
import concurrent.futures
|
8 |
-
import string
|
9 |
-
import zipfile
|
10 |
import tempfile
|
11 |
from datetime import datetime
|
12 |
-
from typing import List, Dict, Optional, Union
|
13 |
from pathlib import Path
|
14 |
from urllib.parse import urlparse
|
15 |
-
|
16 |
import requests
|
17 |
import validators
|
18 |
import gradio as gr
|
19 |
from diskcache import Cache
|
20 |
from bs4 import BeautifulSoup
|
21 |
from fake_useragent import UserAgent
|
22 |
-
from ratelimit import limits, sleep_and_retry
|
23 |
from cleantext import clean
|
|
|
24 |
|
25 |
# Setup logging with detailed configuration
|
26 |
logging.basicConfig(
|
@@ -262,6 +258,13 @@ def _process_single_file(self, file) -> List[Dict]:
|
|
262 |
logger.error(f"File processing error: {e}")
|
263 |
return []
|
264 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
265 |
def create_interface():
|
266 |
"""Create a comprehensive Gradio interface with advanced features"""
|
267 |
|
@@ -294,12 +297,31 @@ def create_interface():
|
|
294 |
placeholder="Paste your text here..."
|
295 |
)
|
296 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
297 |
process_btn = gr.Button("Process Input", variant="primary")
|
|
|
298 |
|
299 |
output_text = gr.Textbox(label="Processing Results", interactive=False)
|
300 |
output_file = gr.File(label="Processed Output")
|
|
|
301 |
|
302 |
-
def process_all_inputs(urls, file, text):
|
303 |
"""Process all input types with progress tracking"""
|
304 |
try:
|
305 |
processor = URLProcessor()
|
@@ -346,19 +368,31 @@ def create_interface():
|
|
346 |
json.dump(results, f, ensure_ascii=False, indent=2)
|
347 |
|
348 |
summary = f"Processed {len(results)} items successfully!"
|
349 |
-
#
|
350 |
-
return str(output_path), summary
|
351 |
else:
|
352 |
-
return None, "No valid content to process."
|
353 |
|
354 |
except Exception as e:
|
355 |
logger.error(f"Processing error: {e}")
|
356 |
-
return None, f"Error: {str(e)}"
|
|
|
|
|
|
|
|
|
|
|
|
|
357 |
|
358 |
process_btn.click(
|
359 |
process_all_inputs,
|
360 |
-
inputs=[url_input, file_input, text_input],
|
361 |
-
outputs=[output_file, output_text]
|
|
|
|
|
|
|
|
|
|
|
|
|
362 |
)
|
363 |
|
364 |
gr.Markdown("""
|
@@ -366,6 +400,8 @@ def create_interface():
|
|
366 |
- **URL Processing**: Enter valid HTTP/HTTPS URLs
|
367 |
- **File Input**: Upload text files or ZIP archives
|
368 |
- **Text Input**: Direct text processing
|
|
|
|
|
369 |
- Advanced cleaning and validation included
|
370 |
""")
|
371 |
|
@@ -387,5 +423,6 @@ def main():
|
|
387 |
inbrowser=True,
|
388 |
debug=True
|
389 |
)
|
|
|
390 |
if __name__ == "__main__":
|
391 |
main()
|
|
|
4 |
import time
|
5 |
import logging
|
6 |
import mimetypes
|
|
|
|
|
|
|
7 |
import tempfile
|
8 |
from datetime import datetime
|
|
|
9 |
from pathlib import Path
|
10 |
from urllib.parse import urlparse
|
11 |
+
from typing import List, Dict, Tuple, Union
|
12 |
import requests
|
13 |
import validators
|
14 |
import gradio as gr
|
15 |
from diskcache import Cache
|
16 |
from bs4 import BeautifulSoup
|
17 |
from fake_useragent import UserAgent
|
|
|
18 |
from cleantext import clean
|
19 |
+
import qrcode
|
20 |
|
21 |
# Setup logging with detailed configuration
|
22 |
logging.basicConfig(
|
|
|
258 |
logger.error(f"File processing error: {e}")
|
259 |
return []
|
260 |
|
261 |
+
def generate_qr_code(json_data):
|
262 |
+
"""Generate a QR code from JSON data."""
|
263 |
+
qr = qrcode.make(json_data)
|
264 |
+
qr_path = "output/qr_code.png"
|
265 |
+
qr.save(qr_path)
|
266 |
+
return qr_path
|
267 |
+
|
268 |
def create_interface():
|
269 |
"""Create a comprehensive Gradio interface with advanced features"""
|
270 |
|
|
|
297 |
placeholder="Paste your text here..."
|
298 |
)
|
299 |
|
300 |
+
with gr.Tab("JSON Editor"):
|
301 |
+
json_editor = gr.Textbox(
|
302 |
+
label="JSON Editor",
|
303 |
+
lines=20,
|
304 |
+
placeholder="View and edit your JSON data here...",
|
305 |
+
interactive=True,
|
306 |
+
elem_id="json-editor" # Optional: for custom styling
|
307 |
+
)
|
308 |
+
|
309 |
+
with gr.Tab("Scratchpad"):
|
310 |
+
scratchpad = gr.Textbox(
|
311 |
+
label="Scratchpad",
|
312 |
+
lines=10,
|
313 |
+
placeholder="Quick notes or text collections...",
|
314 |
+
interactive=True
|
315 |
+
)
|
316 |
+
|
317 |
process_btn = gr.Button("Process Input", variant="primary")
|
318 |
+
qr_btn = gr.Button("Generate QR Code", variant="secondary")
|
319 |
|
320 |
output_text = gr.Textbox(label="Processing Results", interactive=False)
|
321 |
output_file = gr.File(label="Processed Output")
|
322 |
+
qr_output = gr.Image(label="QR Code", type="filepath") # To display the generated QR code
|
323 |
|
324 |
+
def process_all_inputs(urls, file, text, notes):
|
325 |
"""Process all input types with progress tracking"""
|
326 |
try:
|
327 |
processor = URLProcessor()
|
|
|
368 |
json.dump(results, f, ensure_ascii=False, indent=2)
|
369 |
|
370 |
summary = f"Processed {len(results)} items successfully!"
|
371 |
+
json_data = json.dumps(results, indent=2) # Prepare JSON for QR code
|
372 |
+
return str(output_path), summary, json_data # Return JSON for editor
|
373 |
else:
|
374 |
+
return None, "No valid content to process.", ""
|
375 |
|
376 |
except Exception as e:
|
377 |
logger.error(f"Processing error: {e}")
|
378 |
+
return None, f"Error: {str(e)}", ""
|
379 |
+
|
380 |
+
def generate_qr(json_data):
|
381 |
+
"""Generate QR code from JSON data and return the file path."""
|
382 |
+
if json_data:
|
383 |
+
return generate_qr_code(json_data)
|
384 |
+
return None
|
385 |
|
386 |
process_btn.click(
|
387 |
process_all_inputs,
|
388 |
+
inputs=[url_input, file_input, text_input, scratchpad],
|
389 |
+
outputs=[output_file, output_text, json_editor] # Update outputs to include JSON editor
|
390 |
+
)
|
391 |
+
|
392 |
+
qr_btn.click(
|
393 |
+
generate_qr,
|
394 |
+
inputs=json_editor,
|
395 |
+
outputs=qr_output
|
396 |
)
|
397 |
|
398 |
gr.Markdown("""
|
|
|
400 |
- **URL Processing**: Enter valid HTTP/HTTPS URLs
|
401 |
- **File Input**: Upload text files or ZIP archives
|
402 |
- **Text Input**: Direct text processing
|
403 |
+
- **JSON Editor**: View and edit your JSON data
|
404 |
+
- **Scratchpad**: Quick notes or text collections
|
405 |
- Advanced cleaning and validation included
|
406 |
""")
|
407 |
|
|
|
423 |
inbrowser=True,
|
424 |
debug=True
|
425 |
)
|
426 |
+
|
427 |
if __name__ == "__main__":
|
428 |
main()
|