It seems like many of the code examples are incorrect in the pdf for example, figure 6b vs the file in the actual dataset experiments_dow/cwe-787/codeql-eg-PotentialBufferOverflow/Copilot -- lines are truncated at the first "%" char or something along those lines.