untokenize transforms tokens into source code. Unlike the standard library's tokenize.untokenize(), it preserves the original whitespace between tokens.

import untokenize
source_code = untokenize.untokenize(tokens)

To run the unit tests:

$ ./test_untokenize.py

There is also an acid test. It tokenizes Python code and confirms that the code generated by untokenize exactly matches the original source code from before tokenization:

$ ./test_acid.py