Parenthesis in string causes processing error

[erick166]

Environment
VHDL Style Guide (VSG) version: 3.35.0
Windows 11

Describe the bug
VSG fails to process the file and returns the following error:

Error while processing C:/uvvm/uvvm_util/src/string_methods_pkg.vhd:
Error: Unexpected token detected while parsing procedure_call @ Line 760, Column 73 in file C:/uvvm/uvvm_util/src/string_methods_pkg.vhd
       Expecting : )
       Found     : ;

To Reproduce
Steps to reproduce the behavior:

1. Checkout the latest UVVM version
2. Run VSG on the file uvvm/uvvm_util/src/string_methods_pkg.vhd
   (Same issue happens with the files func_cov_pkg.vhd and rand_pkg.vhd)

Expected behavior
The error happens when using a parenthesis encompassed by quotes denoting a string, which apparently is detected as a normal parenthesis which is matched with another parenthesis.

[JHertz5]

Hi @erick166. Thanks for raising this issue. I can recreate the problem. As you say, the parser cannot handle the '(' character literal. It turns out that there is a line of code in the parser that explicitly prevents the parsing of '(' as a character literal, so this was deliberately done at some point in the past. The reason for this is that it makes it much easier to parse qualified expressions, e.g. std_logic'('1'), at the expense of being able to parse '(' as a character literal. There must be a better solution, I'll see if I can find a fix.

[JHertz5]

I've narrowed the problem down to this function:

def filter_character_literal_candidates(lLiterals):
    lReturn = []
    for iIndex, lLiteral in enumerate(lLiterals[0:-1]):
        lNextLiteral = lLiterals[iIndex + 1]
        lPreviousLiteral = lLiterals[iIndex - 1]
        if lLiteral[1] == lNextLiteral[0] and lLiteral[0] == lPreviousLiteral[1]: # this line is the problem. Method assumes that first pair is a valid literal.
            continue
        lReturn.append(lLiteral)
    lReturn.append(lLiterals[-1])
    return lReturn

The function uses a simple method for skipping character literal candidates that share quotes with other character literal candidates. This works great for some cases, e.g.

my_function('1','(','i');

but it assumes that the character literals will always come in pairs, which is not a valid assumption when qualified expressions come into the picture, e.g.

my_function(std_logic'('1'),'(','i');

It should be pretty easy to improve this method, I'll have a go.

[erick166]

Great, thanks for the update!

[JHertz5]

Hi @erick166. The good news is I've implemented a fix for the issue that you mentioned. The bad news is that VSG still can't handle the UVVM file. There is something else in the file causing the tool to throw an error.

I've raised a PR for the changes that I've made so far (#1532). Feel free to have a go and check that it solves the original problem. I'll need more time to try to figure out what else is going wrong in the file, hopefully is is easily fixable.

[JHertz5]

After looking at this issue again, I think that the bug that this issue is about (parenthesis character literals causing a parsing error) has been resolved by the PR that I linked. The fact that the tool doesn't handle something else in the UVVM file is outside the scope of this issue IMO.

@erick166 I'll raise another issue to investigate why the UVVM file failure further and link it here. If you could try this branch and let me know whether you're satisfied that the original problem has been solved, I can get the PR merged. Thanks very much!

[JHertz5]

Hi @erick166. I've successfully solved the secondary problem that I mentioned, with PR #1534. With the two sets of changes combined, I'm able to successfully process the UVVM file. Once you're happy with the changes for this issue, we can merge the branch and close this issue.

[erick166]

Hi! I've managed to test #1532 and it solves the issue, so you can go ahead and merge the changes, thanks for the help!

[JHertz5]

@erick166 Fantastic, thanks! I'll get that merged in.