Correct reading of length

app.py

@@ -94,9 +94,7 @@ def extract_text_from_pptx(pptx_data, clean=True):
     return text, len(text)
 
 def read_document(file, clean=True):
-    """Reads content from various document formats."""
     file_path = file.name
-    # No file extension used
 
     with open(file_path, "rb") as f:
         file_content = f.read()
@@ -109,7 +107,6 @@ def read_document(file, clean=True):
         mime = kind.mime
 
     if mime == "application/pdf":
-        # PDF Handling (unchanged)
         try:
             pdf_reader = PyPDF2.PdfReader(io.BytesIO(file_content))
             content = ''
@@ -117,11 +114,10 @@ def read_document(file, clean=True):
                 content += pdf_reader.pages[page].extract_text()
             if clean:
                 content = clean_text(content)
-            return content, len(content)
+            return content, len(repr(content))
         except Exception as e:
             return f"Error reading PDF: {e}", 0
     elif mime == "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet":
-        # XLSX Handling (unchanged)
         try:
             wb = load_workbook(io.BytesIO(file_content))
             content = ''
@@ -132,7 +128,7 @@ def read_document(file, clean=True):
                             content += str(cell.value) + ' '
             if clean:
                 content = clean_text(content)
-            return content, len(content)
+            return content, len(repr(content))
         except Exception as e:
             return f"Error reading XLSX: {e}", 0
     elif mime == "text/plain":
@@ -140,7 +136,7 @@ def read_document(file, clean=True):
             content = file_content.decode('utf-8')
             if clean:
                 content = clean_text(content)
-            return content, len(content)
+            return content, len(repr(content))
         except Exception as e:
             return f"Error reading TXT file: {e}", 0
     elif mime == "text/csv":
@@ -148,7 +144,7 @@ def read_document(file, clean=True):
             content = file_content.decode('utf-8')
             if clean:
                 content = clean_text(content)
-            return content, len(content)
+            return content, len(repr(content))
         except Exception as e:
             return f"Error reading CSV file: {e}", 0
     elif mime == "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
@@ -167,11 +163,12 @@ def read_document(file, clean=True):
             content = file_content.decode('utf-8')
             if clean:
                 content = clean_text(content)
-            return content, len(content)
+            return content, len(repr(content))
         except Exception as e:
             return f"Error reading file: {e}", 0
 
 
+
 # --- Chat Functions ---
 
 def generate_mistral_response(message):