fix: route archive inspection through byte-safe boundaries

crates/fparkan-corpus/src/lib.rs

@@ -35,6 +35,8 @@ use std::collections::{BTreeMap, BTreeSet};
 use std::fmt;
 use std::fs;
 use std::io::Write;
+#[cfg(unix)]
+use std::os::unix::ffi::OsStrExt;
 use std::path::{Path, PathBuf};
 use std::sync::Arc;
 
@@ -69,6 +71,8 @@ pub struct DiscoverOptions {
 pub struct ManifestEntry {
     /// Normalized relative path.
     pub path: String,
+    /// Byte-exact relative host path used for reopening corpus files.
+    pub host_rel_path: PathBuf,
     /// File size in bytes.
     pub size: u64,
     /// SHA-256 content fingerprint.
@@ -188,7 +192,7 @@ pub fn discover(root: &Path, options: DiscoverOptions) -> Result<CorpusManifest,
     }
     let mut files = Vec::new();
     walk(root, root, options, &mut files)?;
-    files.sort_by(|a, b| a.path.cmp(&b.path));
+    files.sort_by(|a, b| a.host_rel_path.cmp(&b.host_rel_path));
 
     let kind = classify(root, &files);
     let casefold_collisions = detect_casefold_collisions(&files);
@@ -243,17 +247,22 @@ fn walk(
         let rel = path
             .strip_prefix(root)
             .map_err(|_| CorpusError::InvalidPath(path.display().to_string()))?;
-        let rel_text = rel
+        #[cfg(unix)]
+        let rel_bytes = rel.as_os_str().as_bytes();
+        #[cfg(not(unix))]
+        let rel_bytes = rel
             .to_str()
-            .ok_or_else(|| CorpusError::InvalidPath(path.display().to_string()))?;
+            .ok_or_else(|| CorpusError::InvalidPath(path.display().to_string()))?
-        let normalized = normalize_relative(rel_text.as_bytes(), PathPolicy::HostCompatible)
+            .as_bytes();
-            .map_err(|_| CorpusError::InvalidPath(rel_text.to_string()))?;
+        let normalized = normalize_relative(rel_bytes, PathPolicy::HostCompatible)
+            .map_err(|_| CorpusError::InvalidPath(path.display().to_string()))?;
         let bytes = fs::read(&path).map_err(|source| CorpusError::Io {
             path: path.clone(),
             source,
         })?;
         out.push(ManifestEntry {
-            path: normalized.as_str().to_string(),
+            path: normalized.display_lossy().to_string(),
+            host_rel_path: rel.to_path_buf(),
             size: metadata.len(),
             hash: sha256(&bytes),
         });
@@ -285,7 +294,7 @@ fn detect_casefold_collisions(files: &[ManifestEntry]) -> Vec<Vec<String>> {
     let mut grouped: BTreeMap<Vec<u8>, BTreeSet<String>> = BTreeMap::new();
     for file in files {
         grouped
-            .entry(ascii_lookup_key(file.path.as_bytes()).0)
+            .entry(ascii_lookup_key(path_identity_bytes(&file.host_rel_path)).0)
             .or_default()
             .insert(file.path.clone());
     }
@@ -353,7 +362,7 @@ fn inspect_report_file(
 ) -> CorpusFileRecord {
     let lower = entry.path.to_ascii_lowercase();
     let mut variant = inspect_path_metrics(&lower, metrics);
-    let path = root.join(&entry.path);
+    let path = root.join(&entry.host_rel_path);
     let bytes = match fs::read(&path) {
         Ok(bytes) => bytes,
         Err(source) => {
@@ -439,6 +448,17 @@ fn inspect_report_file(
     }
 }
 
+fn path_identity_bytes(path: &Path) -> &[u8] {
+    #[cfg(unix)]
+    {
+        path.as_os_str().as_bytes()
+    }
+    #[cfg(not(unix))]
+    {
+        path.to_str().unwrap_or_default().as_bytes()
+    }
+}
+
 fn inspect_path_metrics(lower: &str, metrics: &mut BTreeMap<String, u64>) -> String {
     let mut variant = "file";
     if lower.ends_with("data.tma") {
@@ -767,11 +787,7 @@ mod tests {
     fn report_json_contains_metrics_and_hashes_not_paths_or_payloads() {
         let manifest = CorpusManifest {
             kind: CorpusKind::Part1,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry("secret/payload.bin", 4, sha256(b"DATA"))],
-                path: "secret/payload.bin".to_string(),
-                size: 4,
-                hash: sha256(b"DATA"),
-            }],
             casefold_collisions: Vec::new(),
         };
         let report = report(Path::new("."), &manifest).expect("report");
@@ -791,11 +807,7 @@ mod tests {
         let root = temp_dir("report-missing");
         let manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry("missing.lib", 1, sha256(b"missing"))],
-                path: "missing.lib".to_string(),
-                size: 1,
-                hash: sha256(b"missing"),
-            }],
             casefold_collisions: Vec::new(),
         };
 
@@ -814,11 +826,7 @@ mod tests {
         fs::write(root.join("bad.lib"), b"NRes").expect("bad nres");
         let manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry("bad.lib", 4, sha256(b"NRes"))],
-                path: "bad.lib".to_string(),
-                size: 4,
-                hash: sha256(b"NRes"),
-            }],
             casefold_collisions: Vec::new(),
         };
 
@@ -857,11 +865,11 @@ mod tests {
         fs::write(root.join("archive.lib"), &archive).expect("archive");
         let manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry(
-                path: "archive.lib".to_string(),
+                "archive.lib",
-                size: u64::try_from(archive.len()).expect("archive size"),
+                u64::try_from(archive.len()).expect("archive size"),
-                hash: sha256(&archive),
+                sha256(&archive),
-            }],
+            )],
             casefold_collisions: Vec::new(),
         };
 
@@ -886,11 +894,7 @@ mod tests {
         fs::write(root.join("WORLD/MAP/land.map"), build_nres(&[])).expect("land map");
         let manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry("WORLD/MAP/land.map", 16, sha256(b"land.map"))],
-                path: "WORLD/MAP/land.map".to_string(),
-                size: 16,
-                hash: sha256(b"land.map"),
-            }],
             casefold_collisions: Vec::new(),
         };
 
@@ -909,11 +913,7 @@ mod tests {
         fs::write(root.join("WORLD/MAP/land.msh"), build_nres(&[])).expect("land msh");
         let manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry("WORLD/MAP/land.msh", 16, sha256(b"land.msh"))],
-                path: "WORLD/MAP/land.msh".to_string(),
-                size: 16,
-                hash: sha256(b"land.msh"),
-            }],
             casefold_collisions: Vec::new(),
         };
 
@@ -932,11 +932,11 @@ mod tests {
         fs::write(root.join("MISSIONS/test/data.tma"), b"malformed tma").expect("tma");
         let manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry(
-                path: "MISSIONS/test/data.tma".to_string(),
+                "MISSIONS/test/data.tma",
-                size: 12,
+                12,
-                hash: sha256(b"malformed tma"),
+                sha256(b"malformed tma"),
-            }],
+            )],
             casefold_collisions: Vec::new(),
         };
 
@@ -955,11 +955,7 @@ mod tests {
         fs::write(root.join("units/unit.dat"), vec![0u8; 120]).expect("unit");
         let manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry("units/unit.dat", 120, sha256(&[0u8; 120]))],
-                path: "units/unit.dat".to_string(),
-                size: 120,
-                hash: sha256(&[0u8; 120]),
-            }],
             casefold_collisions: Vec::new(),
         };
 
@@ -977,11 +973,7 @@ mod tests {
         fs::write(root.join("patch.nl"), b"NL malformed").expect("rsli");
         let manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry("patch.nl", 12, sha256(b"NL malformed"))],
-                path: "patch.nl".to_string(),
-                size: 12,
-                hash: sha256(b"NL malformed"),
-            }],
             casefold_collisions: Vec::new(),
         };
 
@@ -1052,16 +1044,8 @@ mod tests {
         let manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
             files: vec![
-                ManifestEntry {
+                manifest_entry("Textures/Foo.TEX", 1, sha256(b"first")),
-                    path: "Textures/Foo.TEX".to_string(),
+                manifest_entry("textures/foo.tex", 1, sha256(b"second")),
-                    size: 1,
-                    hash: sha256(b"first"),
-                },
-                ManifestEntry {
-                    path: "textures/foo.tex".to_string(),
-                    size: 1,
-                    hash: sha256(b"second"),
-                },
             ],
             casefold_collisions: Vec::new(),
         };
@@ -1081,11 +1065,7 @@ mod tests {
     fn fingerprint_changes() {
         let mut manifest = CorpusManifest {
             kind: CorpusKind::Unknown,
-            files: vec![ManifestEntry {
+            files: vec![manifest_entry("a", 1, sha256(b"before"))],
-                path: "a".to_string(),
-                size: 1,
-                hash: sha256(b"before"),
-            }],
             casefold_collisions: Vec::new(),
         };
         let a = fingerprint(&manifest);
@@ -1118,6 +1098,29 @@ mod tests {
         let _ = fs::remove_file(tmp);
     }
 
+    #[cfg(unix)]
+    #[test]
+    fn discover_supports_non_utf8_host_paths() {
+        use std::ffi::OsString;
+        use std::os::unix::ffi::OsStringExt;
+
+        let root = temp_dir("non-utf8");
+        let file_name = OsString::from_vec(vec![0xFF, b'.', b'b', b'i', b'n']);
+        let file_path = root.join(&file_name);
+        if let Err(err) = fs::write(&file_path, b"raw") {
+            assert_eq!(err.kind(), std::io::ErrorKind::PermissionDenied);
+            let _ = fs::remove_dir_all(root);
+            return;
+        }
+
+        let manifest = discover(&root, DiscoverOptions::default()).expect("manifest");
+
+        assert_eq!(manifest.files.len(), 1);
+        assert_eq!(manifest.files[0].path, "\u{FFFD}.bin");
+        assert_eq!(manifest.files[0].host_rel_path, PathBuf::from(&file_name));
+        let _ = fs::remove_dir_all(root);
+    }
+
     struct TestNresEntry<'a> {
         name: &'a str,
         type_id: u32,
@@ -1164,6 +1167,15 @@ mod tests {
         out
     }
 
+    fn manifest_entry(path: &str, size: u64, hash: Sha256Digest) -> ManifestEntry {
+        ManifestEntry {
+            path: path.to_string(),
+            host_rel_path: PathBuf::from(path),
+            size,
+            hash,
+        }
+    }
+
     fn push_u32(out: &mut Vec<u8>, value: u32) {
         out.extend_from_slice(&value.to_le_bytes());
     }

crates/fparkan-inspection/Cargo.toml

@@ -6,8 +6,10 @@ license.workspace = true
 repository.workspace = true
 
 [dependencies]
+fparkan-diagnostics = { path = "../fparkan-diagnostics" }
 fparkan-msh = { path = "../fparkan-msh" }
 fparkan-nres = { path = "../fparkan-nres" }
+fparkan-path = { path = "../fparkan-path" }
 fparkan-rsli = { path = "../fparkan-rsli" }
 fparkan-resource = { path = "../fparkan-resource" }
 fparkan-terrain-format = { path = "../fparkan-terrain-format" }

crates/fparkan-inspection/src/lib.rs

@@ -20,14 +20,20 @@
 )]
 //! Shared inspection helpers for format-backed tooling.
 
-use fparkan_msh::{decode_msh, validate_msh};
+use fparkan_diagnostics::{
+    diagnostic, render_human, Diagnostic, DiagnosticCode, DiagnosticContext, Phase,
+};
+use fparkan_msh::{decode_msh, validate_msh, ModelAsset};
 use fparkan_nres::{decode as decode_nres, NresDocument, ReadProfile};
+use fparkan_path::{normalize_relative, PathPolicy};
 use fparkan_resource::{archive_path, resource_name, CachedResourceRepository, ResourceRepository};
 use fparkan_rsli::decode as decode_rsli;
 use fparkan_terrain_format::{decode_land_map, decode_land_msh};
 use fparkan_texm::decode_texm;
-use fparkan_vfs::DirectoryVfs;
+use fparkan_vfs::{DirectoryVfs, Vfs};
 use std::fs;
+#[cfg(unix)]
+use std::os::unix::ffi::OsStrExt;
 use std::path::Path;
 use std::sync::Arc;
 
@@ -131,7 +137,70 @@ pub enum LandFileKind {
 ///
 /// Returns a string error when the archive cannot be read or decoded.
 pub fn inspect_archive_file(path: &Path, sample_limit: usize) -> Result<ArchiveInspection, String> {
-    let bytes = fs::read(path).map_err(|err| format!("{}: {err}", path.display()))?;
+    inspect_archive_file_diagnostic(path, sample_limit).map_err(|diagnostic| render_human(&diagnostic))
+}
+
+/// Inspects a format archive and returns a structured diagnostic on failure.
+///
+/// # Errors
+///
+/// Returns a [`Diagnostic`] when the archive cannot be read or decoded.
+pub fn inspect_archive_file_diagnostic(
+    path: &Path,
+    sample_limit: usize,
+) -> Result<ArchiveInspection, Diagnostic> {
+    let parent = path.parent().unwrap_or_else(|| Path::new("."));
+    let file_name = path.file_name().ok_or_else(|| {
+        diagnostic(
+            DiagnosticCode("S1.VFS.PATH"),
+            format!("{}: archive path has no file name", path.display()),
+        )
+        .with_context(DiagnosticContext {
+            phase: Some(Phase::Read),
+            path: Some(path.display().to_string()),
+            ..DiagnosticContext::default()
+        })
+    })?;
+    #[cfg(unix)]
+    let raw_name = file_name.as_bytes();
+    #[cfg(not(unix))]
+    let raw_name = file_name
+        .to_str()
+        .ok_or_else(|| {
+            diagnostic(
+                DiagnosticCode("S1.VFS.PATH"),
+                format!("{}: archive file name is not valid text", path.display()),
+            )
+            .with_context(DiagnosticContext {
+                phase: Some(Phase::Read),
+                path: Some(path.display().to_string()),
+                ..DiagnosticContext::default()
+            })
+        })?
+        .as_bytes();
+    let normalized = normalize_relative(raw_name, PathPolicy::HostCompatible).map_err(|err| {
+        diagnostic(
+            DiagnosticCode("S1.VFS.PATH"),
+            format!("{}: {err}", path.display()),
+        )
+        .with_context(DiagnosticContext {
+            phase: Some(Phase::Read),
+            path: Some(path.display().to_string()),
+            ..DiagnosticContext::default()
+        })
+    })?;
+    let vfs = DirectoryVfs::new(parent);
+    let bytes = vfs.read(&normalized).map_err(|err| {
+        diagnostic(
+            DiagnosticCode("S1.VFS.READ"),
+            format!("{}: {err}", path.display()),
+        )
+        .with_context(DiagnosticContext {
+            phase: Some(Phase::Read),
+            path: Some(path.display().to_string()),
+            ..DiagnosticContext::default()
+        })
+    })?;
     inspect_archive_bytes(&bytes, sample_limit, Some(path))
 }
 
@@ -140,13 +209,13 @@ fn inspect_archive_bytes(
     bytes: &[u8],
     sample_limit: usize,
     source: Option<&Path>,
-) -> Result<ArchiveInspection, String> {
+) -> Result<ArchiveInspection, Diagnostic> {
     if bytes.starts_with(b"NRes") {
         let document = decode_nres(
             Arc::from(bytes.to_vec().into_boxed_slice()),
             ReadProfile::Compatible,
         )
-        .map_err(|err| err.to_string())?;
+        .map_err(|err| archive_parse_diagnostic("S1.NRES.DECODE", source, err.to_string()))?;
         let mut sample = Vec::new();
         for entry in document.entries().iter().take(sample_limit) {
             sample.push(NresEntrySummary {
@@ -165,15 +234,16 @@ fn inspect_archive_bytes(
             Arc::from(bytes.to_vec().into_boxed_slice()),
             fparkan_rsli::ReadProfile::Compatible,
         )
-        .map_err(|err| err.to_string())?;
+        .map_err(|err| archive_parse_diagnostic("S1.RSLI.DECODE", source, err.to_string()))?;
         Ok(ArchiveInspection::Rsli {
             entries: document.entries().len(),
         })
     } else {
-        match source {
+        Err(archive_parse_diagnostic(
-            Some(path) => Err(format!("{}: unsupported archive magic", path.display())),
+            "S1.RESOURCE.UNSUPPORTED_ARCHIVE",
-            None => Err("unsupported archive magic".to_string()),
+            source,
-        }
+            "unsupported archive magic".to_string(),
+        ))
     }
 }
 
@@ -202,6 +272,22 @@ pub fn inspect_model_from_root(
     })
 }
 
+/// Loads and validates a model resource through repository-backed lookup.
+///
+/// # Errors
+///
+/// Returns a string error when the resource cannot be resolved or parsed as a
+/// valid model payload.
+pub fn load_model_from_root(
+    root: &Path,
+    archive: &str,
+    resource: &str,
+) -> Result<ModelAsset, String> {
+    let document = load_model_document_from_root(root, archive, resource)?;
+    let msh = decode_msh(&document).map_err(|err| err.to_string())?;
+    validate_msh(&msh).map_err(|err| err.to_string())
+}
+
 /// Inspects a texture through repository-backed resource lookup.
 ///
 /// # Errors
@@ -288,6 +374,27 @@ fn read_resource_bytes(root: &Path, archive: &str, name: &str) -> Result<Arc<[u8
     Ok(Arc::from(bytes.into_owned()))
 }
 
+fn load_model_document_from_root(
+    root: &Path,
+    archive: &str,
+    resource: &str,
+) -> Result<NresDocument, String> {
+    let bytes = read_resource_bytes(root, archive, resource)?;
+    decode_nres(bytes, ReadProfile::Compatible).map_err(|err| err.to_string())
+}
+
+fn archive_parse_diagnostic(
+    code: &'static str,
+    source: Option<&Path>,
+    message: String,
+) -> Diagnostic {
+    diagnostic(DiagnosticCode(code), message).with_context(DiagnosticContext {
+        phase: Some(Phase::Parse),
+        path: source.map(|path| path.display().to_string()),
+        ..DiagnosticContext::default()
+    })
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -306,6 +413,23 @@ mod tests {
         assert!(error.contains("entry table out of bounds"));
     }
 
+    #[test]
+    fn archive_diagnostic_preserves_source_path() {
+        let dir = temp_dir("inspect-diagnostic");
+        let path = dir.join("broken.nres");
+        fs::write(&path, b"NRes").expect("broken nres");
+
+        let diagnostic =
+            inspect_archive_file_diagnostic(&path, 0).expect_err("diagnostic failure");
+
+        assert_eq!(diagnostic.code.0, "S1.NRES.DECODE");
+        let expected_path = path.display().to_string();
+        assert_eq!(
+            diagnostic.context.path.as_deref(),
+            Some(expected_path.as_str())
+        );
+    }
+
     #[test]
     fn nres_entry_summary_fields_are_readable() {
         let dir = temp_dir("inspect-nres");