MDS file parsing stops early

[thokkat]

In OpenGothic the size of p.animations is smaller for some mds files than if using a pre-phoenix build. E.g. for HumanS.mds only the first 279 of 864 entries are loaded resulting in stuck npc behavior.

This is only a issue for G1, in G2 there's no difference.

Animation::Animation(phoenix::model_script &p, std::string_view name, const bool ignoreErrChunks) {
  ref = std::move(p.aliases);

  for(auto& ani : p.animations) {

https://github.com/Try/OpenGothic/blob/fc1c1343028fb92dbb7b5c1f7ed67b4294b800b3/game/graphics/mesh/animation.cpp#L33

[lmichaelis]

Hey @thokkat, thanks for the report! I did find something while pre-merge testing that is probably this exact issue though I wasn't able to find the cause. I'll look into it :)

zdump output before fix: HUMANS.MDS.txt

[lmichaelis]

Preliminary fix is now available. Changes made are detailed in the commit message :)

[thokkat]

Got this error message while parsing Lurker.mds

error: while parsing mds::dsl_integer
     │
 336 │ ⇨⇨⇨⇨*eventTag⇨("DEF_HIT_LIMB"⇨⇨"BIP01 L PONYTAIL1")
     │                ^ peek failure
error: while parsing mds::dsl_integer
     │
 336 │ ⇨⇨⇨⇨*eventTag⇨("DEF_HIT_LIMB"⇨⇨"BIP01 L PONYTAIL1")
     │                ^ expected digit.decimal

loading error: failed parsing resource of type syntax error

[lmichaelis]

I've checked all vanilla MDS files manually to make sure they are fully parsed. Trough that I've found two more quirks which are now fixed. This is just a temporary solution while I'm implementing a parser that does not use lexy.

There are just too many quirks and programming errors in the source files to keep the current parser around for long. It's already horrible to work with.

[thokkat]

Works, thanks for the quick fix.