Don't allow invalid Source Locations

[BraisGabin]

Fix #7024

A line or column value less than 1 is not valid. Detekt didn't check this but sarif those so we were generating invalid sarif.

Other way to solve the issue is to move this code to the sarif serializer but I think that it's more correct to fix it in our own code and don't allow to point to a source location that doesn't exist.

[codecov]

Codecov Report

Attention: Patch coverage is 77.77778% with 2 lines in your changes are missing coverage. Please review.

Project coverage is 83.79%. Comparing base (9cb4976) to head (8f4457e).
Report is 17 commits behind head on main.

Files	Patch %	Lines
...kotlin/io/gitlab/arturbosch/detekt/api/Location.kt	50.00%	0 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff            @@
##               main    #7030   +/-   ##
=========================================
  Coverage     83.78%   83.79%           
- Complexity     3929     3938    +9     
=========================================
  Files           578      578           
  Lines         12099    12123   +24     
  Branches       2505     2514    +9     
=========================================
+ Hits          10137    10158   +21     
+ Misses          710      709    -1     
- Partials       1252     1256    +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[3flex]

Won't coercing values mask underlying problems rule or core logic? I don't think we should do that here.

[BraisGabin]

What do you think about the new implementation? If a file is empty then DiagnosticUtils.getLineAndColumnInPsiFile doesn't provide a correct value. You can see it here:

    @NotNull
    public static LineAndColumn offsetToLineAndColumn(@Nullable Document document, int offset) {
        if (document != null && document.getTextLength() != 0) {
            int lineNumber = document.getLineNumber(offset);
            int lineStartOffset = document.getLineStartOffset(lineNumber);
            int column = offset - lineStartOffset;
            int lineEndOffset = document.getLineEndOffset(lineNumber);
            CharSequence lineContent = document.getCharsSequence().subSequence(lineStartOffset, lineEndOffset);
            return new LineAndColumn(lineNumber + 1, column + 1, lineContent.toString());
        } else {
            return new LineAndColumn(-1, offset, (String)null);
        }
    }

So in that case I return null. And then the default value that we have for the cases where we don't get the line/column returns a valid line/column.

I don't think I should fix this at the rule level because any rule that reports an empty file would crash.