craft-fonts-build: reject faces under a non-UTF-8 path instead of embedding a lossy include_bytes! path

crates/fonts-build/src/embed.rs

@@ -181,8 +181,9 @@ impl Embed {
                 self.check_coverage(faces, dir, cargo);
             }
         }
-        let (table, faces) = match selected {
-            Ok(faces) => (table(&faces), faces.into_iter().map(|face| face.entry).collect()),
+        let tabled = selected.and_then(|faces| Ok((table(&faces)?, faces.into_iter().map(|face| face.entry).collect())));
+        let (table, faces) = match tabled {
+            Ok(tabled) => tabled,
             Err(e) if required => {
                 write(&out, EMPTY_TABLE)?;
                 return Err(e);
@@ -400,9 +401,12 @@ pub(crate) struct Env {
     pub out_dir: Option<OsString>,
 }
 
-fn table(faces: &[Face]) -> String {
+/// The generated source; `include_bytes!` takes a string, so a face under a path that isn't UTF-8
+/// is an error rather than a lossy path the consumer's build can't find.
+fn table(faces: &[Face]) -> Result<String, Error> {
     let mut src = String::from("// Generated by craft-fonts-build. Do not edit.\n&[\n");
     for Face { entry, path, index, .. } in faces {
+        let file = path.to_str().ok_or_else(|| Error::NonUtf8Path { path: path.clone() })?;
         let scripts: Vec<String> = entry.scripts.iter().map(|s| format!("{s:?}")).collect();
         let _ = writeln!(
             src,
@@ -411,11 +415,11 @@ fn table(faces: &[Face]) -> String {
             entry.style,
             scripts.join(", "),
             index,
-            path.display().to_string(),
+            file,
         );
     }
     src.push_str("]\n");
-    src
+    Ok(src)
 }
 
 fn write(path: &Path, contents: &str) -> Result<(), Error> {
@@ -446,6 +450,8 @@ pub enum Error {
     Subset { path: PathBuf, reason: String },
     /// [`Embed::documents`] names families the manifest lacks; `known` lists the manifest's.
     UnknownFamilies { manifest: PathBuf, families: Vec<String>, known: Vec<String> },
+    /// A face to embed has a path that isn't UTF-8, which `include_bytes!` can't name.
+    NonUtf8Path { path: PathBuf },
 }
 
 impl fmt::Display for Error {
@@ -477,6 +483,7 @@ impl fmt::Display for Error {
                 families.join(", "),
                 known.join(", ")
             ),
+            Self::NonUtf8Path { path } => write!(f, "cannot embed {}: include_bytes! needs a UTF-8 path", path.display()),
         }
     }
 }

crates/fonts-build/src/tests.rs

@@ -115,6 +115,20 @@ fn required_fails_without_a_usable_checkout() {
     assert!(matches!(run(Embed::new(), Some(&scratch.0), Some("1")).result, Err(Error::NoFaces { .. })));
 }
 
+#[cfg(unix)]
+#[test]
+fn rejects_a_checkout_under_a_non_utf8_path() {
+    use std::os::unix::ffi::OsStrExt;
+    let scratch = Scratch::new("non-utf8");
+    let dir = scratch.0.join(std::ffi::OsStr::from_bytes(b"craft-fonts-\xff"));
+    std::fs::create_dir_all(dir.join("fonts")).unwrap();
+    std::fs::copy(fixture().join("fonts/sans-arabic/SansArabic.ttf"), dir.join("fonts/a.ttf")).unwrap();
+    std::fs::write(dir.join("fonts/manifest.txt"), "A | Regular | fonts/a.ttf | Arab | OFL-1.1 | OFL.txt | 00 | src\n").unwrap();
+    let r = run(Embed::new(), Some(&dir), Some("1"));
+    assert!(matches!(r.result, Err(Error::NonUtf8Path { .. })), "{:?}", r.result);
+    assert_eq!(r.table.trim(), "&[]");
+}
+
 #[test]
 fn embeds_the_selected_scripts() {
     let r = run(Embed::new(), Some(&fixture()), Some("1"));