Merge PR #509: Validate the full WAVE_FORMAT_EXTENSIBLE fmt chunk (#465)

crates/audio-io/src/wav.rs

@@ -50,13 +50,25 @@ fn parse_fmt(c: &[u8]) -> Result<Fmt> {
     fmt.valid_bits = fmt.bits;
     if fmt.tag == WAVE_FORMAT_EXTENSIBLE {
         // cbSize(2) validBits(2) channelMask(4) subFormat GUID(16)
-        if let (Some(valid), Some(sub)) = (le_u16(c, 18), le_u16(c, 24)) {
-            if valid != 0 {
-                fmt.valid_bits = valid;
-            }
-            fmt.tag = sub;
-        } else {
+        let (Some(cb_size), Some(valid), Some(sub)) = (le_u16(c, 16), le_u16(c, 18), le_u16(c, 24)) else {
             return Err(AudioError::Malformed("WAVE_FORMAT_EXTENSIBLE fmt chunk too short".into()));
+        };
+        let Some(tail) = c.get(26..40) else {
+            return Err(AudioError::Malformed("WAVE_FORMAT_EXTENSIBLE fmt chunk too short".into()));
+        };
+        if cb_size < 22 {
+            return Err(AudioError::Malformed("WAVE_FORMAT_EXTENSIBLE cbSize is below 22".into()));
+        }
+        if valid > fmt.bits {
+            return Err(AudioError::Malformed("WAVE_FORMAT_EXTENSIBLE valid bits exceed the container size".into()));
+        }
+        if valid != 0 {
+            fmt.valid_bits = valid;
+        }
+        // A subtype outside the KSDATAFORMAT_SUBTYPE_* family is not PCM/float, whatever its first two bytes say:
+        // keep the extensible tag so the file goes to the foreign decoder.
+        if tail == GUID_TAIL {
+            fmt.tag = sub;
         }
     }
     Ok(fmt)

crates/audio-io/tests/it/malformed.rs

@@ -511,3 +511,37 @@ proptest! {
         prop_assert_eq!(flac, wav);
     }
 }
+
+/// An extensible `fmt ` chunk: cbSize, valid bits, mask, then the 16-byte subtype GUID.
+fn extensible_fmt(cb_size: u16, valid_bits: u16, guid: &[u8]) -> Vec<u8> {
+    let mut f = fmt(0xFFFE, 1, 44_100, 16);
+    f.extend_from_slice(&cb_size.to_le_bytes());
+    f.extend_from_slice(&valid_bits.to_le_bytes());
+    f.extend_from_slice(&4u32.to_le_bytes());
+    f.extend_from_slice(guid);
+    f
+}
+
+const PCM_GUID: [u8; 16] = [1, 0, 0, 0, 0, 0, 0x10, 0, 0x80, 0, 0, 0xAA, 0, 0x38, 0x9B, 0x71];
+
+#[test]
+fn extensible_fmt_is_checked_in_full() {
+    let data = vec![0u8; 8];
+    let good = riff(&[(b"fmt ", extensible_fmt(22, 16, &PCM_GUID)), (b"data", data.clone())]);
+    assert_eq!(probe(&good, None).unwrap().sample_format, SampleFormat::Int16);
+    // Subtype GUID cut short after the format code.
+    let cut = riff(&[(b"fmt ", extensible_fmt(22, 16, &PCM_GUID[..4])), (b"data", data.clone())]);
+    assert!(matches!(probe(&cut, None), Err(AudioError::Malformed(_))));
+    assert!(matches!(decode(&cut, None), Err(AudioError::Malformed(_))));
+    // cbSize too small for the extension.
+    let small_cb = riff(&[(b"fmt ", extensible_fmt(0, 16, &PCM_GUID)), (b"data", data.clone())]);
+    assert!(matches!(probe(&small_cb, None), Err(AudioError::Malformed(_))));
+    // More valid bits than the 16-bit container holds.
+    let wide = riff(&[(b"fmt ", extensible_fmt(22, 24, &PCM_GUID)), (b"data", data.clone())]);
+    assert!(matches!(probe(&wide, None), Err(AudioError::Malformed(_))));
+    // PCM-like prefix but a foreign GUID tail is not read as native PCM.
+    let mut foreign = PCM_GUID;
+    foreign[15] ^= 0xFF;
+    let foreign = riff(&[(b"fmt ", extensible_fmt(22, 16, &foreign)), (b"data", data)]);
+    assert!(!matches!(probe(&foreign, None), Ok(i) if i.sample_format == SampleFormat::Int16));
+}