Visibility incorrectly parsed as wind speed, visibility not included in output about metar-taf-parser HOT 2 CLOSED

Alright, I think I have identified the cause. Two things I have noted:

• 00000KT is technically not correctly parsed. The regex only matches the first 4 digits, which means that VIDP 270200Z 00000MPS is incorrectly parsed as 0 knots wind (I assume KT is the default unit)
  
  
  
  This is technically not related to the original issue, so I opened #80 to track it separately!

• The same regex also matches visibility stubs if the visibility is less than 100 meters. This match occurs later than the wind match, so it looks like the result overrides the real wind value resulting in the originally mentioned issue. I assume that this also means that the visibility parser never runs for this stub, which explains why there is no visibility reported in the output.
  
  
  

from metar-taf-parser.

Ok, so I dug a bit deeper. The culprit seems to be the double zero-digit matcher in the beginning of the regex.

It seems to be there in order to satisfy this test case, which includes an invalid METAR string according to the specifications I can find. As usual, most are not too explicit on the matter but this is one example:

Wind speeds (ff) are two digits (or three digits if required), in knots.
https://meteocentre.com/doc/metar.html

So I guess this is a case of choosing which case to support? If we change the regex by adding an additional required zero:

(VRB|000|[0-3]\d{2})(\d{2})G?(\d{2,3})?(KT|MPS|KM\/H)?

The result is much better, solving several issues:

• Wind unit is correctly registered
• Wind speed is correctly registered (not that it matters, it's all zeros anyway)
• Visibility is no longer incorrectly matched

The downside is that the test case will start to fail. Is the METAR included in that test case a common occurrence? I mean, is it common for real-world METAR reports to report winds calm with only four digits?

from metar-taf-parser.