Update tokenizer.py

[BBC-Esq]

Added New Custom Exception Class

• TokenizationError: Created at the top to handle tokenization-specific errors, allowing for precise error differentiation.

Enhanced decode() Method

• Added try-except block with:
  • Input validation: Ensures non-empty token sequences and positive integer tokens.
  • Error handling: Raises TokenizationError with context-specific messages ("No valid text tokens" and "Invalid token values") and preserves original exceptions using from e.

Enhanced decode_with_timestamps() Method

• Added try-except block with:
  • Input validation: Verifies non-empty token sequences and valid token types/values.
  • Output validation: Ensures non-empty decoded results.
  • Error handling: Raises TokenizationError with relevant messages ("Empty token sequence," "Invalid token values," "No valid output") and maintains original exceptions with from e.

Consistent Error Propagation

• Unified error handling: TokenizationError wraps all errors, preserving original exceptions and context about the failing operation across methods.

[MahmoudAshraf97]

Hi and thanks for the contribution, unfortunately the diff is unreadable, can you try following the formatting steps in the contributor guideline?

isort .
black .
flake8 .

make sure all three pass without warnings or errors

[BBC-Esq]

Sure, I'll give it a shot...although I am limited since this is not my profession. lol