re: read the defaulted weapon stats out of live guest memory

Canary backs the whole guest address space with one shared-memory file, so the
game's RAM is readable from the host while it runs -- no debugger, no emulator
patch. That turns the parked Route-B question (fields the IDXD omits because
they sit at their default) from "unreachable" into a table lookup.

gmem.py maps guest VAs into /dev/shm/xenia_memory_* via Xenia's fixed table and
searches only the allocated extents, so a full-RAM scan is ~0.2 s.

weapon_runtime.py finds the parsed objects by scanning for their vtable, then
*solves* the struct layout instead of guessing it: every (field, offset,
encoding) triple is scored against the disc records, and a binding is accepted
only with zero contradictions -- and marked confirmed only when >=10 records
agree on >=3 distinct values, because a field whose samples are all one number
matches any offset holding that constant.

Result: Weapon (0xc0) and Shell (0x200) mapped, 4393 values the disc does not
carry, for all 126 weapons at 5 % save progress. Cross-checks hold -- the
Arsenal DATA SHEET read 4 for wep_05/wep_60 Max. Lock Ons and the memory read
agrees; the in-flight HUD's 06000/00300 match LoadingCount; all 252 objects are
byte-identical across a screen change, so this is definition data, not state.

Two findings fell out: `IsCharging = Yes` switches LoadingCount and
TriggerShotCount to float32 (it partitions the 8 float-encoded records exactly),
and two .tbl entries declare Shell_TCAF_Ship_AAGun with conflicting Power -- the
runtime keeps the one that omits it.

Where the defaulted values *come from* (the IDXD's undecoded binary node region
vs code defaults) is not settled here; they vary per weapon, so they are not one
constructor constant.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

crates/sylpheed-formats/examples/idxd_tokens.rs

@@ -0,0 +1,102 @@
+//! RE diagnostic: emit every `weapon\Weapon_*.tbl` in a pak as machine-readable
+//! records, split at the sub-object boundaries the schema declares.
+//!
+//! An IDXD `.tbl` for a weapon holds `count` sub-records — a `Weapon` and its
+//! `Shell` — flattened into one string pool. `sylpheed-cli pak dump` shows them
+//! merged, which is ambiguous (both declare `ID`/`Name`). The title's schema
+//! name pool gives the split point: the pool contains the literal type names
+//! (`Weapon`, `Shell`), and each sub-record's fields follow its type name.
+//!
+//! Here we split on a type-name token appearing as a *key* position, so each
+//! record is emitted with its own ID and field set:
+//!
+//! ```text
+//! REC <tbl-hash> <index> <TypeName> <ID>
+//! F   <key> <value>
+//! ```
+//!
+//! Run: cargo run --release -p sylpheed-formats --example idxd_tokens -- <pak> [types...]
+
+use sylpheed_formats::idxd::IdxdObject;
+use sylpheed_formats::pak::PakArchive;
+
+fn is_number(s: &str) -> bool {
+    let s = s.strip_suffix(['f', 'F']).unwrap_or(s);
+    let t = s.strip_prefix(['-', '+']).unwrap_or(s);
+    !t.is_empty() && t.chars().all(|c| c.is_ascii_digit() || c == '.')
+}
+
+/// `Yes`/`No`-style scalar literals are values, not field names, even though
+/// they are identifier-shaped.
+fn is_enum_value(s: &str) -> bool {
+    matches!(s, "Yes" | "No" | "YES" | "NO" | "On" | "Off" | "ON" | "OFF")
+}
+
+fn is_key_like(s: &str) -> bool {
+    !s.is_empty()
+        && s.chars().all(|c| c.is_ascii_alphanumeric() || c == '_')
+        && s.chars().next().is_some_and(|c| c.is_ascii_alphabetic() || c == '_')
+        && !is_number(s)
+        && !is_enum_value(s)
+}
+
+fn main() {
+    let mut args = std::env::args().skip(1);
+    let pak_path = args.next().expect("usage: idxd_tokens <pak> [TypeName...]");
+    let types: Vec<String> = {
+        let v: Vec<String> = args.collect();
+        if v.is_empty() {
+            vec!["Weapon".into(), "Shell".into()]
+        } else {
+            v
+        }
+    };
+
+    let pak = PakArchive::open(&pak_path).expect("open pak");
+    for entry in pak.entries() {
+        let Ok(bytes) = pak.read(entry) else { continue };
+        let Ok(obj) = IdxdObject::parse(&bytes) else { continue };
+        let toks = obj.tokens();
+        if !toks.iter().any(|t| t.starts_with("Weapon_")) {
+            continue;
+        }
+
+        if std::env::var("SYLPH_RAW").is_ok() {
+            println!("RAW {:08x}", entry.name_hash);
+            for t in toks {
+                println!("T {t}");
+            }
+        }
+
+        // Sub-record boundaries: a bare type-name token. The first one is the
+        // schema's own declaration (`EnumWeapon`-style header lives in title
+        // code, not here), so we take every occurrence in order.
+        let mut bounds: Vec<(usize, &str)> = Vec::new();
+        for (i, t) in toks.iter().enumerate() {
+            // The pool-start heuristic can glue one stray binary byte onto the
+            // first token ("YWeapon"), so match a short suffix, not equality.
+            if let Some(ty) = types
+                .iter()
+                .find(|ty| t == *ty || (t.ends_with(ty.as_str()) && t.len() <= ty.len() + 2))
+            {
+                bounds.push((i, ty.as_str()));
+            }
+        }
+        for (n, &(start, ty)) in bounds.iter().enumerate() {
+            let end = bounds.get(n + 1).map(|b| b.0).unwrap_or(toks.len());
+            let slice = &toks[start..end];
+            // The record's own ID value is the token straight after its type
+            // name (`Shell` -> `Shell_DSaber_P_wep_01_Beam`). The `ID`/`Name`
+            // *keys* are interned once in the pool, so only the first record
+            // shows them -- position-of-key lookup would miss the rest.
+            let id = slice.get(1).map(|s| s.as_str()).unwrap_or("?");
+            println!("REC {:08x} {n} {ty} {id}", entry.name_hash);
+            for i in 1..slice.len() {
+                let (k, v) = (slice[i].as_str(), slice[i - 1].as_str());
+                if is_key_like(k) && (!is_key_like(v) || is_number(v)) {
+                    println!("F {k} {v}");
+                }
+            }
+        }
+    }
+}