Update app2.py

app2.py

@@ -17,7 +17,10 @@ from bs4 import BeautifulSoup
 from fake_useragent import UserAgent
 from cleantext import clean
 import qrcode
-import zipfile
+if sys.version_info >= (3, 6):
+    import zipfile
+else:
+    import zipfile36 as zipfile
 
 # Setup logging with detailed configuration
 logging.basicConfig(
@@ -42,13 +45,12 @@ class URLProcessor:
             'Connection': 'keep-alive',
             'Upgrade-Insecure-Requests': '1'
         })
-
+    
     def advanced_text_cleaning(self, text: str) -> str:
         """Robust text cleaning with version compatibility"""
         try:
             cleaned_text = clean(
                 text,
-                fix_unicode=True,
                 to_ascii=True,
                 lower=True,
                 no_line_breaks=True,
@@ -208,7 +210,7 @@ class FileProcessor:
             for extracted_file in os.listdir(extraction_directory):
                 extracted_file_path = os.path.join(extraction_directory, extracted_file)
                 process_file(extracted_file_path)
-    
+                
 def _process_single_file(self, file) -> List[Dict]:
     try:
         file_stat = os.stat(file.name)
@@ -247,13 +249,6 @@ def _process_single_file(self, file) -> List[Dict]:
 
 def generate_qr_code(json_data):
     """Generate a QR code from JSON data."""
-    # Limit the size of json_data to avoid exceeding QR code version limits
-    max_length = 2953  # Maximum length for version 40 (the highest version)
-    
-    if len(json_data) > max_length:
-        logger.warning("JSON data is too large for QR code generation. Truncating data.")
-        json_data = json_data[:max_length]  # Truncate the data
-
     qr = qrcode.make(json_data)
     qr_path = "output/qr_code.png"
     qr.save(qr_path)