hdlConvertor
hdlConvertor copied to clipboard
Nic30
Missing source file information in parsed data
When parsing multiple files in the same context, the "Position" should include the source file context as well.
I guess https://github.com/Nic30/hdlConvertor/issues/88 and https://github.com/Nic30/hdlConvertor/issues/89 are the same. If not, you should provide example, otherwise we won't understand each other.
What you complain is that the original position of the token is lost, right ?
The parser is a 2 step processing. First preprocessing is executed, during this step token are correctly annotate by filename, line / pos. But the result of preprocessing is send to the SV parser as a std::string without those information.
In an old and closed ticket https://github.com/Nic30/hdlConvertor/issues/46#issuecomment-503313301 I proposed that preprocessor produce line directive that allow to keep track of file and line. So the SV Parser is able to correct filename and line number in the processed token.
@hs-apotell , True. The filename is required to resolve actual position in code. But there is another problem. The preprocessor takes the code from the files and generates new string out of them and parser itself works with this string. The solution is to put line directives in to preprocessed string.
`line line_number "file_name" level
And then use it in parser. It should be easy, but I did not have a time to implement it yet. I can not promisse but I will look in to it in 2 weeks as I am currently in ultimately bad situation.
I guess #88 and #89 are the same.
I consider them different. #88 is a bug (parsed values aren't propagated correctly) whereas #89 is a missing feature (nothing in the current exposed API holds the source filename).
I am planning on using this for intellisense (Ctrl+Click) functionality in my tool and both #88 and #89 are crucial to get it working correctly.