Consolidate required CI gate (#115)

tools/ci-matrix/src/provenance.rs

@@ -917,7 +917,7 @@ fn parse_lock_string(value: &str) -> Result<String> {
 
 fn package_notices(root: &Path, explicit: Option<&str>) -> Result<Vec<(String, String)>> {
     let mut candidates = Vec::new();
-    collect_notice_files(root, 0, &mut candidates)?;
+    collect_notice_files(root, &mut candidates)?;
     if let Some(explicit) = explicit {
         let path = root.join(explicit);
         if path.is_file() {
@@ -946,10 +946,10 @@ fn package_notices(root: &Path, explicit: Option<&str>) -> Result<Vec<(String, S
     Ok(notices)
 }
 
-fn collect_notice_files(directory: &Path, depth: u8, output: &mut Vec<PathBuf>) -> Result<()> {
-    if depth > 2 {
-        return Ok(());
-    }
+fn collect_notice_files(directory: &Path, output: &mut Vec<PathBuf>) -> Result<()> {
+    // Cargo build scripts may populate nested source directories (rust-skia
+    // creates `skia/` in its registry checkout). Only package-root notices and
+    // Cargo's explicit `license_file` are immutable package metadata.
     let mut entries = fs::read_dir(directory)?.collect::<std::io::Result<Vec<_>>>()?;
     entries.sort_by_key(std::fs::DirEntry::file_name);
     for entry in entries {
@@ -969,8 +969,6 @@ fn collect_notice_files(directory: &Path, depth: u8, output: &mut Vec<PathBuf>)
             .any(|prefix| name.starts_with(prefix))
         {
             output.push(path);
-        } else if kind.is_dir() && depth < 2 {
-            collect_notice_files(&path, depth + 1, output)?;
         }
     }
     Ok(())
@@ -1095,6 +1093,10 @@ fn source_paths(root: &Path) -> Result<Vec<String>> {
                 .map_err(|_| MatrixError::new("source archive contains a non-UTF-8 path"))
         })
         .collect::<Result<Vec<_>>>()?;
+    // `git ls-files --cached` retains index entries for working-tree deletions
+    // until the consolidation commit is created. A source distribution always
+    // describes files that actually exist, including during a pre-commit audit.
+    paths.retain(|path| root.join(path).is_file());
     for path in &paths {
         validate_relative(path)?;
     }
@@ -1126,8 +1128,18 @@ fn compare_report(root: &Path, relative: &str, expected: &[u8]) -> Result<()> {
         MatrixError::new(format!("generated report {}: {error}", path.display()))
     })?;
     if actual != expected {
+        if let Some(directory) = std::env::var_os("METACRATE_PROVENANCE_DIAGNOSTICS_DIR") {
+            let directory = PathBuf::from(directory);
+            fs::create_dir_all(&directory)?;
+            let name = path
+                .file_name()
+                .ok_or_else(|| MatrixError::new("generated report path has no file name"))?;
+            fs::write(directory.join(name), expected)?;
+        }
         return Err(MatrixError::new(format!(
-            "{relative} is stale; run `cargo run --locked -p metacrate-ci-matrix -- provenance-report`"
+            "{relative} is stale (checked-in sha256 {}, generated sha256 {}); run `cargo run --locked -p metacrate-ci-matrix -- provenance-report`",
+            sha256(&actual),
+            sha256(expected),
         )));
     }
     Ok(())
@@ -1253,6 +1265,27 @@ version = "0.0.1"
         );
     }
 
+    #[test]
+    fn package_notice_scan_ignores_build_generated_subtrees() {
+        let root =
+            std::env::temp_dir().join(format!("metacrate-package-notices-{}", std::process::id()));
+        let _ = fs::remove_dir_all(&root);
+        fs::create_dir_all(root.join("generated")).unwrap();
+        fs::write(root.join("LICENSE"), "package license\n").unwrap();
+        fs::write(root.join("generated/LICENSE"), "generated license\n").unwrap();
+        fs::write(root.join("generated/NOTICE.explicit"), "explicit notice\n").unwrap();
+
+        let notices = package_notices(&root, Some("generated/NOTICE.explicit")).unwrap();
+        assert_eq!(
+            notices
+                .iter()
+                .map(|(path, _)| path.as_str())
+                .collect::<Vec<_>>(),
+            ["LICENSE", "generated/NOTICE.explicit"]
+        );
+        fs::remove_dir_all(root).unwrap();
+    }
+
     #[test]
     fn unsafe_paths_are_rejected() {
         for path in ["", "../escape", "a/../b", "/absolute"] {