Merge pull request 'fix: skip NTFS DOS (8.3) short-name entries when extracting' (#7) from jujuforce/fsdecrypt:fix/ntfs-dos-short-name-extraction into trunk

src/vhd.rs

@@ -7,7 +7,10 @@ use std::{
 
 use anyhow::{anyhow, Result};
 use indicatif::{ProgressBar, ProgressStyle};
-use ntfs::{structured_values::NtfsStandardInformation, Ntfs, NtfsAttributeType, NtfsTime};
+use ntfs::{
+    structured_values::{NtfsFileNamespace, NtfsStandardInformation},
+    Ntfs, NtfsAttributeType, NtfsTime,
+};
 
 // ---------------------------------------------------------------------------
 // Constants
@@ -563,6 +566,17 @@ fn is_ntfs_system_entry(name: &str) -> bool {
     name.starts_with('$') || name == "." || name == ".." || name == "System Volume Information"
 }
 
+/// Whether a directory-index entry should be skipped when extracting.
+///
+/// Besides NTFS system metadata, this skips DOS (8.3) short-name aliases: a file
+/// that has a separate short name appears in the index *twice* — once with its
+/// Win32 long name and once with the `Dos` short name. Without this, every such
+/// entry is extracted a second time under its mangled `NAME~1.EXT` name (and
+/// short-named directories get their whole subtree duplicated).
+fn skip_index_entry(namespace: NtfsFileNamespace, name: &str) -> bool {
+    namespace == NtfsFileNamespace::Dos || is_ntfs_system_entry(name)
+}
+
 fn ntfs_time_to_system_time(t: NtfsTime) -> SystemTime {
     let nanos = (t.nt_timestamp() - WINDOWS_EPOCH_OFFSET) * 100;
     SystemTime::UNIX_EPOCH + Duration::from_nanos(nanos)
@@ -602,7 +616,7 @@ fn extract_ntfs_dir<T: Read + Seek>(
         let entry = entry?;
         let key = entry.key().ok_or_else(|| anyhow!("missing key"))??;
         let name = key.name().to_string_lossy();
-        if is_ntfs_system_entry(&name) {
+        if skip_index_entry(key.namespace(), &name) {
             continue;
         }
 
@@ -649,7 +663,7 @@ fn calculate_ntfs_size<T: Read + Seek>(
     while let Some(entry) = iter.next(fs) {
         let entry = entry?;
         let key = entry.key().ok_or_else(|| anyhow!("missing key"))??;
-        if is_ntfs_system_entry(&key.name().to_string_lossy().as_ref()) {
+        if skip_index_entry(key.namespace(), key.name().to_string_lossy().as_ref()) {
             continue;
         }
         let file = entry.to_file(ntfs, fs)?;
@@ -748,3 +762,32 @@ fn read_be_u32(buf: &[u8], offset: usize) -> u32 {
 fn read_be_u64(buf: &[u8], offset: usize) -> u64 {
     u64::from_be_bytes(buf[offset..offset + 8].try_into().unwrap())
 }
+
+#[cfg(test)]
+mod tests {
+    use super::skip_index_entry;
+    use ntfs::structured_values::NtfsFileNamespace;
+
+    #[test]
+    fn skips_dos_short_name_aliases() {
+        // 8.3 aliases duplicate a Win32 entry and must not be extracted again.
+        assert!(skip_index_entry(NtfsFileNamespace::Dos, "OXGETH~1.EXE"));
+        assert!(skip_index_entry(NtfsFileNamespace::Dos, "PROGRA~1"));
+    }
+
+    #[test]
+    fn keeps_long_and_native_names() {
+        assert!(!skip_index_entry(NtfsFileNamespace::Win32, "oxGetHwInfo.exe"));
+        // A name that is its own short name (no separate Dos entry) is kept.
+        assert!(!skip_index_entry(NtfsFileNamespace::Win32AndDos, "game.bat"));
+        assert!(!skip_index_entry(NtfsFileNamespace::Posix, "readme"));
+    }
+
+    #[test]
+    fn skips_system_entries_regardless_of_namespace() {
+        assert!(skip_index_entry(NtfsFileNamespace::Win32, "$MFT"));
+        assert!(skip_index_entry(NtfsFileNamespace::Win32, "System Volume Information"));
+        assert!(skip_index_entry(NtfsFileNamespace::Win32, "."));
+        assert!(skip_index_entry(NtfsFileNamespace::Win32, ".."));
+    }
+}