Hi! I recently found this after I started working on a map analyzer toy project, and thought it'd be neat if I could use this as a library to generate the map from a given seed. Unfortunately, right now all the code is in the binary module, so it can't really be used as a library.

This series of changes restructures the code a bit, mainly splitting it into a core library module and a thin CLI wrapper. I've also restructured a few of the top-level functions into primarily returning data, and split out the print statements into separate functions. This enabled me to do stuff like this, which uses this project as a WebAssembly library to generate the map.

Would you like this?

This includes everything in #2 and #3, and also adds a few questionable "improvements" that are definitely down to subjective taste, and some (962dd49a25f27d61be2fed1ce27a16ab6d8ee1fe, 564e686be4bbb808db14b1f4e4d2dd26c7185e4b) significantly change how parts of the program works. Again, you're welcome to cherry-pick whatever you want from here and ignore whatever you don't like.

Thank you for accept my pull request!

By the way, map generation by this tool sometimes seem to be wrong. for example,

sts_map_oracle --seed 673465884448

(673465884448 is AGCK93XT)

Act 2


14     R  R  R  R
(----snip----)
2      ?  M  ?     M  M
         \| \|   /      \
1         M  M  ?        M
        /  /      \    /
0      M  M        M  M

It should be below. (My StS version is 12-22-2020, and don't use MOD)

Act 2


14     R  R  R  R
(----snip----)
2      ?  M  ?     M  M
         \| \|   /      \
1         M  ?  ?        M
        /  /      \    /
0      M  M        M  M

I compared logic of room count in StS and sts_map_oracle.

◆ Slay the Spire decompiled

/*  630 */     int count = 0;
/*  631 */     for (ArrayList<MapRoomNode> a : map) {
/*  632 */       for (MapRoomNode n : a) {
/*  633 */         if (!n.hasEdges() || 
/*  634 */           n.y == map.size() - 2) {
/*      */           continue;
/*      */         }
/*  637 */         count++;
/*      */       } 
/*      */     } 
/*      */ 
/*      */ 
/*      */     
/*  643 */     generateRoomTypes(roomList, count);

◆ current sts_map_oracle logic

let count = map
    .iter()
    .flat_map(|row| row.iter())
    .filter(|n| !n.edges.is_empty() && n.y as usize != map.len() - 1)
    .count();
let room_list = generate_room_type(&room_chances, count);

But, filter condition of sts_map_oracle should be below

 .filter(|n| ((!n.edges.is_empty() || (n.y as usize == map.len()-1 && !n.parents.is_empty())) && n.y as usize != map.len()-2))

Can I send Pull Request?

This includes everything in #2, and also adds a bunch of refactoring. I personally think these refactorings all seem worthwhile, but in the end it's a matter of taste, so you're welcome to cherry-pick whatever you like and ignore whatever you don't like.

This first adds a couple of regression tests, and then changes a few math operations from unchecked to wrapping math. This eliminates the need for the unchecked_math feature, and thus the need to compile with nightly Rust, and still passes the tests. :slightly_smiling_face: