skytemple / explorerscript Goto Github PK

See https://canary.discord.com/channels/710190644152369162/712343594568253512/809199677085974546

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #reverse-engineering.

Source code that's causing issues: https://canary.discord.com/channels/710190644152369162/787260430728691712/808283984952819753
Example call: https://cdn.discordapp.com/attachments/787260430728691712/808284196354523166/unknown.png

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #scripting-help.

Originally reported at https://discord.com/channels/710190644152369162/1002325768682090636/1002330772272857239

The following code triggers a KeyError after compiling the script, deleting the original ExplorerScript source and trying to decompile it again:

def 0 {
    call @something;
    end;
    §something;
    return;
}

The call opcode can be replaced with a jump, both produce the same result.

The exception is raised at

Originally reported at https://discord.com/channels/710190644152369162/724746586072023182/1002499328184692807

Seems like SkyTemple allows specifying out-of-bounds IDs to WaitExecute commands (in particular, an actor ID where an object ID is expected and vice-versa). The script compiles and works, probably because actors and objects are read the same way by the game. However, trying to reopen a script with out-of-bounds IDs in SkyTemple causes an error.

It would be a good idea to add some validation to the parameters of these opcodes (and also to other opcodes that might take entity IDs that are supposed to be in a certain range, if there's any) to prevent this.

Long scripts cause stack overflows.
See SkyTemple/skytemple-ssb-debugger#4.

Example:

ValueError: After a simple opcode there must be exactly 0 or 1 immediate opcode.
     args = ('After a simple opcode there must be exactly 0 or 1 immediate opcode.',)
     with_traceback = <built-in method with_traceback of ValueError object> 

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #skytemple-tester.

Currently not allowed in the syntax, and probably not properly implemented

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #scripting-help.

Support for strings that go over one line using actual line breaks. Automatic trimming of spaces?

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #skytemple.

The MacroResolutionOrderVisitor visitor for ExplorerScript is completely broken. It fails with every slightly more complex script.

Rewrite it and implement the projects first unit tests to prove the new version is working.

Allow multiple non-branching statements in with-Blocks for easier readability (generate multiple lives/object/performer opcodes, one for each instruction

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #rom-hacking-support.

Currently the compiler only actually supports base 10 integers.

No content

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #beta-tester.

put it in quotes

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #rom-hacking-support.

See https://canary.discord.com/channels/710190644152369162/787260430728691712/787376443050295328

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #scripting-help.

See https://cdn.discordapp.com/attachments/712343594568253512/862306128435740692/SetAnimation2.txt - this is gonna be a huge thing to support, should probably impelement something generic for bitflag params. Will also require files and ssb-debugger changes

Beep, boop, I'm a bot! This issue was created by @Parakoopa in #reverse-engineering.