fix(xtask): derive scorecard corpus counts from pinned manifests

xtask/src/scorecard.rs

@@ -5,6 +5,7 @@
 //!   numbers against the budgets). A fresh `cargo xtask perf` result in `target/` is not used:
 //!   publish it with `cargo xtask perf --update-baseline`.
 //! - `crates/io/tests/corpus.rs`: the corpus floors (`Source { .. }` constants).
+//! - `xtask/*-corpus.sha256`: file counts for the pinned corpora.
 //! - `scorecard/*.toml`: per-area checklists (id, target, status, issue, note).
 //! - the prefs audit: `Preferences` fields that no code outside `prefs.rs` / `prefs_ui.rs` reads.
 //! - counted from the tree: never-crash attribute coverage, `docs/parity-checklist.md`'s live count.
@@ -474,7 +475,15 @@ pub fn gather(root: &Path) -> Result<Inputs, String> {
         Ok(t) => Some(serde_json::from_str(&t).map_err(|e| format!("{}: {e}", perf::BASELINE))?),
         Err(_) => None,
     };
-    let corpora = parse_corpus(&std::fs::read_to_string(root.join(CORPUS_RS)).map_err(|e| format!("read {CORPUS_RS}: {e}"))?);
+    let mut corpora = parse_corpus(&std::fs::read_to_string(root.join(CORPUS_RS)).map_err(|e| format!("read {CORPUS_RS}: {e}"))?);
+    for corpus in &mut corpora {
+        let Some(pinned) = crate::corpus_pins::ALL.iter().find(|p| corpus.dir == format!("corpus/{}", p.dest)) else { continue };
+        let manifest = std::fs::read_to_string(root.join(pinned.manifest)).map_err(|e| format!("read {}: {e}", pinned.manifest))?;
+        corpus.files = Some(
+            u32::try_from(manifest.lines().filter(|line| !line.trim().is_empty()).count())
+                .map_err(|_| format!("{}: corpus file count exceeds u32", pinned.manifest))?,
+        );
+    }
     Ok(Inputs {
         budgets,
         baseline,
@@ -522,7 +531,7 @@ pub fn render(inp: &Inputs) -> String {
 
     let mut s = String::from(
         "# PhotoCraft scorecard\n\nGenerated by `cargo xtask scorecard` from `perf/budgets.toml`, `perf/baseline.json`, \
-`crates/io/tests/corpus.rs`, `scorecard/*.toml` and the source tree. Do not edit by hand: change the sources and \
+`crates/io/tests/corpus.rs`, `xtask/*-corpus.sha256`, `scorecard/*.toml` and the source tree. Do not edit by hand: change the sources and \
 regenerate (CI runs `cargo xtask scorecard --check`). Checklist statuses are verified against the code; numbers \
 here supersede estimates elsewhere.\n\n## Summary\n\n| Area | Done | Partial | Missing | Key numbers |\n|---|---:|---:|---:|---|\n",
     );
@@ -791,6 +800,30 @@ const PHOTOSHOP: Source = Source { label: "photoshop oracles", dir: "corpus/phot
         assert!(parse_corpus("const X: u32 = 1;").is_empty());
     }
 
+    #[test]
+    fn scorecard_uses_pinned_corpus_file_counts() {
+        let root = crate::root();
+        let inp = gather(&root).unwrap();
+        let rendered = render(&inp);
+        for corpus in &inp.corpora {
+            let pinned = crate::corpus_pins::ALL.iter().find(|p| corpus.dir == format!("corpus/{}", p.dest)).unwrap();
+            let manifest = std::fs::read_to_string(root.join(pinned.manifest)).unwrap();
+            let files = manifest.lines().filter(|line| !line.trim().is_empty()).count();
+            assert!(
+                rendered.contains(&format!(
+                    "{}: oracle ≥ {}/{files}, round trip ≥ {}/{files}",
+                    corpus.label,
+                    corpus.pass_floor.unwrap(),
+                    corpus.roundtrip_floor.unwrap()
+                )),
+                "{} must use the {files} files in {} instead of a count from a doc comment",
+                corpus.label,
+                pinned.manifest
+            );
+            assert!(rendered.contains(&format!("| {} | `{}` | {files} |", corpus.label, corpus.dir)));
+        }
+    }
+
     #[test]
     fn checklists_validate() {
         let ok = "area = \"tools\"\ntitle = \"Tools\"\norder = 3\n[[item]]\nid = \"T-1\"\ntarget = \"Pencil\"\nstatus = \"partial\"\nissue = 213\n";