@nullvalue which OCR did you use? I just did some trials with tesseract, which is not specfically designed for the recognition of source code text, but gives relatively good results with some training. One problem still is accurate whitespace reproduction (for indentation), the other problem is...