Add native deterministic tar writer

1 files changed, 113 insertions, 0 deletions

diff --git a/src/archive.rs b/src/archive.rs
new file mode 100644
index 0000000..0076364
--- /dev/null
+++ b/src/archive.rs
@@ -0,0 +1,113 @@
+use std::fs::{self, File};
+use std::path::Path;
+
+use anyhow::{bail, Context, Result};
+use tar::{Builder, EntryType, Header};
+
+/// Writes a deterministic rootfs tar archive using numeric ownership metadata.
+///
+/// Extended ACLs and xattrs are intentionally not represented by this initial
+/// portable archive format. Callers that require them must use a future archive
+/// policy which can verify their availability before packaging.
+#[derive(Debug, Default, Clone, Copy)]
+pub struct NativeTarWriter;
+
+impl NativeTarWriter {
+    pub fn new() -> Self {
+        Self
+    }
+
+    pub fn write(&self, rootfs: impl AsRef<Path>, output: impl AsRef<Path>) -> Result<()> {
+        let rootfs = rootfs.as_ref();
+        if !rootfs.is_dir() {
+            bail!("rootfs is not a directory: {}", rootfs.display());
+        }
+
+        let output = output.as_ref();
+        let file = File::create(output).with_context(|| format!("create archive {}", output.display()))?;
+        let mut archive = Builder::new(file);
+        append_tree(&mut archive, rootfs, Path::new(""))?;
+        archive
+            .finish()
+            .with_context(|| format!("finish archive {}", output.display()))?;
+        Ok(())
+    }
+}
+
+fn append_tree(archive: &mut Builder<File>, rootfs: &Path, relative: &Path) -> Result<()> {
+    let directory = rootfs.join(relative);
+    let mut entries = fs::read_dir(&directory)
+        .with_context(|| format!("read rootfs directory {}", directory.display()))?
+        .collect::<std::result::Result<Vec<_>, _>>()?;
+    entries.sort_by_key(|entry| entry.file_name());
+
+    for entry in entries {
+        let name = entry.file_name();
+        if name == ".host" || name == ".fakedata" {
+            continue;
+        }
+        let path = entry.path();
+        let archive_path = relative.join(&name);
+        let metadata = fs::symlink_metadata(&path)
+            .with_context(|| format!("inspect rootfs entry {}", path.display()))?;
+        let file_type = metadata.file_type();
+
+        if file_type.is_dir() {
+            append_directory(archive, &archive_path, &metadata)?;
+            append_tree(archive, rootfs, &archive_path)?;
+        } else if file_type.is_file() {
+            append_file(archive, &path, &archive_path, &metadata)?;
+        } else if file_type.is_symlink() {
+            append_symlink(archive, &path, &archive_path, &metadata)?;
+        } else {
+            bail!("unsupported rootfs entry type: {}", path.display());
+        }
+    }
+    Ok(())
+}
+
+fn deterministic_header(metadata: &fs::Metadata, entry_type: EntryType, size: u64) -> Header {
+    use std::os::unix::fs::PermissionsExt;
+
+    let mut header = Header::new_gnu();
+    header.set_entry_type(entry_type);
+    header.set_mode(metadata.permissions().mode());
+    header.set_uid(0);
+    header.set_gid(0);
+    header.set_mtime(0);
+    header.set_size(size);
+    header
+}
+
+fn append_directory(archive: &mut Builder<File>, path: &Path, metadata: &fs::Metadata) -> Result<()> {
+    let mut header = deterministic_header(metadata, EntryType::Directory, 0);
+    archive
+        .append_data(&mut header, path, std::io::empty())
+        .with_context(|| format!("append directory {}", path.display()))
+}
+
+fn append_file(
+    archive: &mut Builder<File>,
+    source: &Path,
+    path: &Path,
+    metadata: &fs::Metadata,
+) -> Result<()> {
+    let mut header = deterministic_header(metadata, EntryType::Regular, metadata.len());
+    let input = File::open(source).with_context(|| format!("open rootfs file {}", source.display()))?;
+    archive
+        .append_data(&mut header, path, input)
+        .with_context(|| format!("append file {}", path.display()))
+}
+
+fn append_symlink(
+    archive: &mut Builder<File>,
+    source: &Path,
+    path: &Path,
+    metadata: &fs::Metadata,
+) -> Result<()> {
+    let target = fs::read_link(source).with_context(|| format!("read rootfs symlink {}", source.display()))?;
+    let mut header = deterministic_header(metadata, EntryType::Symlink, 0);
+    archive
+        .append_link(&mut header, path, target)
+        .with_context(|| format!("append symlink {}", path.display()))
+}