ackwell / ironworks Goto Github PK

Currently, most configuration is hardcoded in inappropriate locations. Look into building config file(s) for storing and allowing user configuration. Ideally, environment variables would also be read.

As of last look, the crates config and figment look like close contenders for this functionality.

Encountered when trying to extract every file in the game (lmao).

use ironworks::{ffxiv, sqpack::SqPack, Ironworks};
use std::path::Path;

fn main() {
    let game_path = Path::new("G:/Steam/steamapps/common/FINAL FANTASY XIV Online");
    let ironworks = Ironworks::new().resource(SqPack::new(ffxiv::FsResource::at(game_path)));
    let file = ironworks
        .file::<Vec<u8>>("chara/monster/m0023/obj/body/b0003/texture/v01_m0023b0003_s.tex");

    println!("{:#?}", file);
}

thread 'main' panicked at 'assertion failed: `(left == right)`
  left: `16000`,
 right: `6`: Block info and header decompressed size differs.', C:\Users\Julian\.cargo\registry\src\github.com-1ecc6299db9ec823\ironworks-0.3.1\src\sqpack\file.rs:55:5

Some scalar fields have additional semantics, i.e. a numeric column can be an icon, which implies a path within the game data to an icon texture, despite no explicit table relation existing for this lookup.

This will require changes to the schema to include a (presumably optional) metadata field on scalars.

Of further note, relationships are technically a scalar in the source sheet, so there may be room to redefine relationships as metadata on a scalar.

Schema provider currently clones StC on first boot and never updates it. Need to add capability to schema module to update the repo, and a regular interval in bm to call it.

thread 'main' panicked at 'index out of bounds: the len is 0 but the index is 0', C:\Users\Julian\.cargo\git\checkouts\ironworks-836e4c5d0f0f4516\ed0af56\ironworks\src\sqpack\block\stream.rs:62:25

Search will need per-column explicit value filtering, i.e. numeric range checks and so on. Given search indices are decoupled from sheet schemas, and hence do not resolve any references, this will be a bit fiddly.

If the column filters are built to be sufficiently structured, it will likely be possible to trace a set of filters to the leaf fields, and then walk back up with a search at each reference to find matches.

Likely lots of optimisation opportunities on that.

Search field specifier can probably share some concept or implementation with reference targeting in schemas.

Using version 0.3.0 ironworks fails to compile if the ffxiv feature is enabled without excel.

cargo 1.75.0-nightly (794d0a825 2023-10-03), rustc 1.75.0-nightly (cae0791da 2023-10-05). After a cargo clean && cargo build:

error[E0422]: cannot find struct, variant or union type `LineColumn` in crate `proc_macro`
   --> C:\Users\Julian\.cargo\registry\src\index.crates.io-6f17d22bba15001f\proc-macro2-1.0.53\src\wrapper.rs:475:33
    |
475 |                 let proc_macro::LineColumn { line, column } = s.start();
    |                                 ^^^^^^^^^^ not found in `proc_macro`
    |
help: consider importing this struct through its public re-export
    |
1   + use crate::LineColumn;
    |
help: if you import `LineColumn`, refer to it directly
    |
475 -                 let proc_macro::LineColumn { line, column } = s.start();
475 +                 let LineColumn { line, column } = s.start();
    |

error[E0422]: cannot find struct, variant or union type `LineColumn` in crate `proc_macro`
   --> C:\Users\Julian\.cargo\registry\src\index.crates.io-6f17d22bba15001f\proc-macro2-1.0.53\src\wrapper.rs:489:33
    |
489 |                 let proc_macro::LineColumn { line, column } = s.end();
    |                                 ^^^^^^^^^^ not found in `proc_macro`
    |
help: consider importing this struct through its public re-export
    |
1   + use crate::LineColumn;
    |
help: if you import `LineColumn`, refer to it directly
    |
489 -                 let proc_macro::LineColumn { line, column } = s.end();
489 +                 let LineColumn { line, column } = s.end();
    |

error[E0635]: unknown feature `proc_macro_span_shrink`
  --> C:\Users\Julian\.cargo\registry\src\index.crates.io-6f17d22bba15001f\proc-macro2-1.0.53\src\lib.rs:92:30
   |
92 |     feature(proc_macro_span, proc_macro_span_shrink)
   |

Running cargo update proc-macro2 leads to even more spooky errors, too long to post.

see https://github.com/jansol/texpresso

Sheet endpoints would benefit from a query argument(s) to control the columns that will be returned in the response. While the query arguments are necessarily strings, they should be parsed into some form of structured representation that can be "walked" alongside the schema.

Open questions:

• Schema package has some need for column specifiers on references, is this something relevant to that?
  It likely isn't relevant - reference specifiers are naturally single-column targets, while column filtering must target 0-to-many. Search filter specifiers will probably be more appropriate for this.