[Test] Add best effort test to validate game ID rules

[Douile]

An attempt to implement the rules specified in #108 as a programmatic test.

[Douile]

The games that currently fail are either wrong (based on current rules as I understand):

• left 4 dead
• seven days to die
• quake 3: arena

Or ambiguous:

• minecraft legacy...
  • how the protocol version should be encoded into ID is not specified

---

Further work (that I won't include in this PR):

• Validate the game definition matches the ID used for its module
  • This isn't needed if we generate them both from the same macro

[CosminPerRam]

Listed games are indeed wrongly identified. Rule #6 is formed a bit strange, will reword it.

how the protocol version should be encoded into ID is not specified

Well said, will also modify. I will make a PR with these fixes and I will link it here.

[Douile]

I'm not sure on darkest hour there isn't a rule that specifies it should be how we currently have it but the ID we expect is not great.

[CosminPerRam]

Darkest Hour: Europe '44-'45 should be dhe4445:

2. More than 2 words
3. [...] If a number is not in the first position, its entire numeric digits will be used instead of the acronym of that number's digits, so for '44-'45 it will be 4445 as I understand it.

Sorry for not coming back to this, I said that I'll try to modify it myself but I ended up not having the right time to put into it.

[Douile]

Ah okay that makes sense. but it's going to be hard to implement I think. Yeah I don't blame you this is a bit spaghetti I should probably refactor.

[Douile]

After a further look at the messy code the reason this is ambiguous is because some games use - in-between words and we count them as different words (Counter-Strike: Global Offensive). Therefore the numbers at the end of Darkest Hour get split into two words.

I guess I could add a band-aid solution of if there are two numbers at the end of the word list combine them, this is a "best effort" after all. The nuance comes down to human interpretation and the code should be changed to match that.

[Douile]

I'm not sure why CI isn't running here.

[CosminPerRam]

CI passes, good to merge?

[Douile]

Assuming you are okay with these assumptions made about IDs:

• Numbers containing "-" are split numbers e.g. "10-40" == "1040"
  • Numbers with trailing "-" will panic e.g. "10-fourty"
• If the last word in a name is a number then it is the year
• Identifying roman numerals is hard: Only parse numerals if they are not the first word
• If expected ID doesn't match try again for the part after a "-" (for protocols)
• Parts in (brackets) are optional and ignored in most cases

Also should I remove the node-gamedig test?

[CosminPerRam]

Sounds good to me!