feat: add basic compressing support for squashfs

src/archive/squashfs.rs

@@ -1,17 +1,23 @@
 use std::{
-    fs,
-    io::{self, BufWriter, Write},
-    path::Path,
+    env, fs,
+    io::{self, BufWriter, Seek, Write},
+    path::{Path, PathBuf},
+    time::SystemTime,
 };
 
-use backhand::{FilesystemReader, InnerNode, SquashfsFileReader};
-use filetime_creation::{set_file_handle_times, set_file_mtime, FileTime};
+use backhand::{
+    compression::Compressor, FilesystemCompressor, FilesystemReader, FilesystemWriter, InnerNode, NodeHeader,
+    SquashfsFileReader,
+};
+use filetime_creation::{set_file_handle_times, FileTime};
+use same_file::Handle;
 
 use crate::{
+    error::FinalError,
     list::FileInArchive,
     utils::{
         logger::{info, warning},
-        Bytes,
+        Bytes, FileVisibilityPolicy,
     },
 };
 
@@ -131,3 +137,205 @@ pub fn unpack_archive(archive: FilesystemReader<'_>, output_folder: &Path, quiet
 
     Ok(unpacked_files)
 }
+
+pub fn build_archive_from_paths<W>(
+    input_filenames: &[PathBuf],
+    output_path: &Path,
+    mut writer: W,
+    file_visibility_policy: FileVisibilityPolicy,
+    quiet: bool,
+    follow_symlinks: bool,
+) -> crate::Result<W>
+where
+    W: Write + Seek,
+{
+    let root_dir = match input_filenames {
+        [path] if path.is_dir() => path,
+        _ => {
+            let error = FinalError::with_title("Cannot build squashfs")
+                .detail("Squashfs requires a single directory input for root directory")
+                .detail(if input_filenames.len() != 1 {
+                    "Multiple paths are provided".into()
+                } else {
+                    format!("Not a directory: {:?}", input_filenames[0])
+                });
+            return Err(error.into());
+        }
+    };
+
+    let output_handle = Handle::from_path(output_path);
+
+    let mut fs_writer = FilesystemWriter::default();
+    // Set the default compression to Gzip with default level, matching mksquashfs's default.
+    // The default choice of `backhand` is Xz which is not enabled by us.
+    // TODO: We do not support customization argument for archive formats.
+    fs_writer.set_compressor(FilesystemCompressor::new(Compressor::Gzip, None).expect("gzip is supported"));
+
+    // cd *into* the source directory, using it as the archive root.
+    let previous_cwd = env::current_dir()?;
+    env::set_current_dir(root_dir)?;
+
+    for entry in file_visibility_policy.build_walker(".") {
+        let entry = entry?;
+        let path = entry.path();
+
+        if let Ok(handle) = &output_handle {
+            if matches!(Handle::from_path(path), Ok(x) if &x == handle) {
+                warning(format!(
+                    "Cannot compress `{}` into itself, skipping",
+                    output_path.display()
+                ));
+            }
+        }
+
+        if !quiet {
+            // `fs_writer.push_*` only maintains metadata. We do not want to give
+            // users a false information that we are compressing during the
+            // traversal. So this is not "compressing".
+            // File reading, compression and writing are done in
+            // `fs_writer.write` below, after the hierarchy tree gets finalized.
+            info(format!("Found {path:?}"));
+        }
+
+        let metadata = entry.metadata()?;
+        let file_type = metadata.file_type();
+
+        let mut header = NodeHeader::default();
+        header.mtime = match metadata.modified() {
+            // Not available.
+            Err(_) => 0,
+            Ok(mtime) => {
+                let mtime = mtime
+                    .duration_since(SystemTime::UNIX_EPOCH)
+                    .ok()
+                    .and_then(|dur| u32::try_from(dur.as_secs()).ok());
+                if mtime.is_none() {
+                    warning(format!(
+                        "Modification time of {path:?} exceeds the representable range (1970-01-01 ~ 2106-02-07) \
+                        of squashfs. Recorded as 1970-01-01."
+                    ));
+                }
+                mtime.unwrap_or(0)
+            }
+        };
+
+        #[cfg(not(unix))]
+        {
+            header.permissions = 0o777;
+        }
+
+        #[cfg(unix)]
+        {
+            use std::os::unix::fs::{MetadataExt, PermissionsExt};
+
+            // Only permission bits, not file type bits.
+            header.permissions = metadata.permissions().mode() as u16 & !(libc::S_IFMT as u16);
+            header.uid = metadata.uid();
+            header.gid = metadata.gid();
+        }
+
+        // Root directory is special cased.
+        if path == Path::new(".") {
+            fs_writer.set_root_mode(header.permissions);
+            fs_writer.set_root_uid(header.uid);
+            fs_writer.set_root_gid(header.gid);
+            continue;
+        }
+
+        if file_type.is_dir() {
+            fs_writer.push_dir(path, header)?;
+        } else if !follow_symlinks && file_type.is_symlink() {
+            let target = fs::read_link(path)?;
+            fs_writer.push_symlink(target, path, header)?;
+        } else if maybe_push_unix_special_inode(&mut fs_writer, path, &metadata, header)? {
+            // Already handled.
+        } else {
+            // Fallback case: read as a regular file.
+            // See comments of `LazyFile` for why not `File::open` here.
+            let reader = LazyFile::Path(path.to_path_buf());
+            fs_writer.push_file(reader, path, header)?;
+        }
+    }
+
+    if !quiet {
+        info(format!("Compressing data"));
+    }
+
+    // Finalize the superblock and write data. This should be done before
+    // resetting current directory, because `LazyFile`s store relative paths.
+    fs_writer.write(&mut writer)?;
+
+    env::set_current_dir(previous_cwd)?;
+
+    Ok(writer)
+}
+
+#[cfg(not(unix))]
+fn maybe_push_unix_special_inode(
+    _writer: &mut FilesystemWriter,
+    _path: &Path,
+    _metadata: &fs::Metadata,
+    _header: NodeHeader,
+) -> io::Result<bool> {
+    Ok(false)
+}
+
+#[cfg(unix)]
+fn maybe_push_unix_special_inode(
+    writer: &mut FilesystemWriter,
+    path: &Path,
+    metadata: &fs::Metadata,
+    header: NodeHeader,
+) -> io::Result<bool> {
+    use std::os::unix::fs::{FileTypeExt, MetadataExt};
+
+    let file_type = metadata.file_type();
+    if file_type.is_fifo() {
+        writer.push_fifo(path, header)?;
+    } else if file_type.is_socket() {
+        writer.push_socket(path, header)?;
+    } else if file_type.is_block_device() {
+        let dev = metadata.rdev() as u32;
+        writer.push_block_device(dev, path, header)?;
+    } else if file_type.is_char_device() {
+        let dev = metadata.rdev() as u32;
+        writer.push_char_device(dev, path, header)?;
+    } else {
+        return Ok(false);
+    }
+    Ok(true)
+}
+
+/// Delay file opening until the first read and close it as soon as the EOF is encountered.
+///
+/// Due to design of `backhand`, we need to store all `impl Read` into the
+/// builder during traversal and write out the squashfs later. But we cannot
+/// open and store all file handles during traversal or it will exhaust file
+/// descriptors on *NIX if there are thousands of files (a pretty low limit!).
+///
+/// Upstream discussion: https://github.com/wcampbell0x2a/backhand/discussions/614
+enum LazyFile {
+    Path(PathBuf),
+    Opened(fs::File),
+    Closed,
+}
+
+impl io::Read for LazyFile {
+    fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
+        match self {
+            LazyFile::Path(path) => {
+                let file = fs::File::open(path)?;
+                *self = Self::Opened(file);
+                self.read(buf)
+            }
+            LazyFile::Opened(file) => {
+                let cnt = file.read(buf)?;
+                if buf.len() != 0 && cnt == 0 {
+                    *self = Self::Closed;
+                }
+                Ok(cnt)
+            }
+            LazyFile::Closed => Ok(0),
+        }
+    }
+}

src/commands/compress.rs

@@ -124,21 +124,22 @@ pub fn compress_files(
             )?;
             writer.flush()?;
         }
-        Squashfs => todo!(),
-        Zip => {
+        Zip | Squashfs => {
+            let is_zip = matches!(first_format, Zip);
+
             if !formats.is_empty() {
                 // Locking necessary to guarantee that warning and question
                 // messages stay adjacent
                 let _locks = lock_and_flush_output_stdio();
 
-                warn_user_about_loading_in_memory(".zip");
+                warn_user_about_loading_in_memory(if is_zip { ".zip" } else { ".sqfs" });
                 if !user_wants_to_continue(output_path, question_policy, QuestionAction::Compression)? {
                     return Ok(false);
                 }
             }
 
             let mut vec_buffer = Cursor::new(vec![]);
-
+            if is_zip {
                 archive::zip::build_archive_from_paths(
                     &files,
                     output_path,
@@ -147,6 +148,17 @@ pub fn compress_files(
                     quiet,
                     follow_symlinks,
                 )?;
+            } else {
+                archive::squashfs::build_archive_from_paths(
+                    &files,
+                    output_path,
+                    &mut vec_buffer,
+                    file_visibility_policy,
+                    quiet,
+                    follow_symlinks,
+                )?;
+            }
+
             vec_buffer.rewind()?;
             io::copy(&mut vec_buffer, &mut writer)?;
         }