added support for listing and decompressing .rar archives

.github/workflows/build-and-test.yml

@@ -75,6 +75,7 @@ jobs:
         run: |
           sudo apt-get update
           sudo apt-get install musl-tools
+          sudo ln -s "/usr/bin/g++" "/usr/bin/musl-g++"
 
       - name: Set up extra cargo flags
         if: matrix.no-zstd-thin

CHANGELOG.md

@@ -19,6 +19,8 @@ Categories Used:
 **Bullet points in chronological order by PR**
 
 ## [Unreleased](https://github.com/ouch-org/ouch/compare/0.4.2...HEAD)
+- Add support for listing and decompressing `.rar` archives
+- Fix mime type detection
 
 ### Bug Fixes
 

Cargo.lock

@@ -763,6 +763,7 @@ dependencies = [
  "tempfile",
  "test-strategy",
  "time",
+ "unrar",
  "xz2",
  "zip",
  "zstd",
@@ -1253,6 +1254,29 @@ version = "1.0.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "301abaae475aa91687eb82514b328ab47a211a533026cb25fc3e519b86adfc3c"
 
+[[package]]
+name = "unrar"
+version = "0.5.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "850dc99da3a3c12a9400a52a3c510057bf2b16c6c9a495492850d25ed9d757ab"
+dependencies = [
+ "bitflags 1.3.2",
+ "regex",
+ "unrar_sys",
+ "widestring",
+]
+
+[[package]]
+name = "unrar_sys"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "95e536227cb6a91f8e88adb0219912e004a107c9724a094f90baad9229281efa"
+dependencies = [
+ "cc",
+ "libc",
+ "winapi",
+]
+
 [[package]]
 name = "utf8parse"
 version = "0.2.1"
@@ -1350,6 +1374,12 @@ version = "0.2.87"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ca6ad05a4870b2bf5fe995117d3728437bd27d7cd5f06f13c17443ef369775a1"
 
+[[package]]
+name = "widestring"
+version = "1.0.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "653f141f39ec16bba3c5abe400a0c60da7468261cc2cbf36805022876bc721a8"
+
 [[package]]
 name = "winapi"
 version = "0.3.9"

src/archive/mod.rs

@@ -1,4 +1,5 @@
 //! Archive compression algorithms
 
+pub mod rar;
 pub mod tar;
 pub mod zip;

src/archive/rar.rs

@@ -0,0 +1,56 @@
+//! Contains RAR-specific building and unpacking functions
+
+use std::path::Path;
+
+use unrar::{self, Archive};
+
+use crate::{info, list::FileInArchive, warning};
+
+/// Unpacks the archive given by `archive_path` into the folder given by `output_folder`.
+/// Assumes that output_folder is empty
+pub fn unpack_archive(archive_path: &Path, output_folder: &Path, quiet: bool) -> crate::Result<usize> {
+    assert!(output_folder.read_dir().expect("dir exists").count() == 0);
+
+    let mut archive = Archive::new(archive_path).open_for_processing()?;
+    let mut unpacked = 0;
+
+    while let Some(header) = archive.read_header()? {
+        let entry = header.entry();
+        archive = if entry.is_file() {
+            if !quiet {
+                info!(
+                    inaccessible,
+                    "{} extracted. ({})",
+                    entry.filename.display(),
+                    entry.unpacked_size
+                );
+            }
+            unpacked += 1;
+            header.extract_with_base(output_folder)?
+        } else {
+            header.skip()?
+        };
+    }
+
+    Ok(unpacked)
+}
+
+/// List contents of `archive_path`, returning a vector of archive entries
+pub fn list_archive(archive_path: &Path) -> impl Iterator<Item = crate::Result<FileInArchive>> {
+    Archive::new(archive_path)
+        .open_for_listing()
+        .expect("cannot open archive")
+        .map(|item| {
+            let item = item?;
+            let is_dir = item.is_directory();
+            let path = item.filename;
+
+            Ok(FileInArchive { path, is_dir })
+        })
+}
+
+pub fn no_compression_notice() {
+    const MESSAGE: &str = "Creating '.rar' archives is not supported due to licensing restrictions";
+
+    warning!("{}", MESSAGE);
+}

src/commands/compress.rs

@@ -79,7 +79,7 @@ pub fn compress_files(
                 //     is `clamp`ed and therefore guaranteed to be valid
                 Box::new(zstd_encoder.unwrap().auto_finish())
             }
-            Tar | Zip => unreachable!(),
+            Tar | Zip | Rar => unreachable!(),
         };
         Ok(encoder)
     };
@@ -122,6 +122,10 @@ pub fn compress_files(
             vec_buffer.rewind()?;
             io::copy(&mut vec_buffer, &mut writer)?;
         }
+        Rar => {
+            archive::rar::no_compression_notice();
+            return Ok(false);
+        }
     }
 
     Ok(true)

src/commands/decompress.rs

@@ -86,7 +86,7 @@ pub fn decompress_file(
             Lzma => Box::new(xz2::read::XzDecoder::new(decoder)),
             Snappy => Box::new(snap::read::FrameDecoder::new(decoder)),
             Zstd => Box::new(zstd::stream::Decoder::new(decoder)?),
-            Tar | Zip => unreachable!(),
+            Tar | Zip | Rar => unreachable!(),
         };
         Ok(decoder)
     };
@@ -146,6 +146,24 @@ pub fn decompress_file(
                 return Ok(());
             }
         }
+        Rar => {
+            type UnpackResult = crate::Result<usize>;
+            let unpack_fn: Box<dyn FnOnce(&Path) -> UnpackResult> = if formats.len() > 1 {
+                let mut temp_file = tempfile::NamedTempFile::new()?;
+                io::copy(&mut reader, &mut temp_file)?;
+                Box::new(move |output_dir| crate::archive::rar::unpack_archive(temp_file.path(), output_dir, quiet))
+            } else {
+                Box::new(|output_dir| crate::archive::rar::unpack_archive(input_file_path, output_dir, quiet))
+            };
+
+            if let ControlFlow::Continue(files) =
+                smart_unpack(unpack_fn, output_dir, &output_file_path, question_policy)?
+            {
+                files
+            } else {
+                return Ok(());
+            }
+        }
     };
 
     // this is only printed once, so it doesn't result in much text. On the other hand,

src/commands/list.rs

@@ -52,7 +52,7 @@ pub fn list_archive_contents(
                 Lzma => Box::new(xz2::read::XzDecoder::new(decoder)),
                 Snappy => Box::new(snap::read::FrameDecoder::new(decoder)),
                 Zstd => Box::new(zstd::stream::Decoder::new(decoder)?),
-                Tar | Zip => unreachable!(),
+                Tar | Zip | Rar => unreachable!(),
             };
             Ok(decoder)
         };
@@ -78,6 +78,15 @@ pub fn list_archive_contents(
 
             Box::new(crate::archive::zip::list_archive(zip_archive))
         }
+        Rar => {
+            if formats.len() > 1 {
+                let mut temp_file = tempfile::NamedTempFile::new()?;
+                io::copy(&mut reader, &mut temp_file)?;
+                Box::new(crate::archive::rar::list_archive(temp_file.path()))
+            } else {
+                Box::new(crate::archive::rar::list_archive(archive_path))
+            }
+        }
         Gzip | Bzip | Lz4 | Lzma | Snappy | Zstd => {
             panic!("Not an archive! This should never happen, if it does, something is wrong with `CompressionFormat::is_archive()`. Please report this error!");
         }

src/error.rs

@@ -178,6 +178,14 @@ impl From<zip::result::ZipError> for Error {
     }
 }
 
+impl From<unrar::error::UnrarError> for Error {
+    fn from(err: unrar::error::UnrarError) -> Self {
+        Self::Custom {
+            reason: FinalError::with_title("Unexpected error in rar archive").detail(format!("{:?}", err.code)),
+        }
+    }
+}
+
 impl From<ignore::Error> for Error {
     fn from(err: ignore::Error) -> Self {
         Self::WalkdirError {

src/extension.rs

@@ -7,9 +7,9 @@ use bstr::ByteSlice;
 use self::CompressionFormat::*;
 use crate::{error::Error, warning};
 
-pub const SUPPORTED_EXTENSIONS: &[&str] = &["tar", "zip", "bz", "bz2", "gz", "lz4", "xz", "lzma", "sz", "zst"];
+pub const SUPPORTED_EXTENSIONS: &[&str] = &["tar", "zip", "bz", "bz2", "gz", "lz4", "xz", "lzma", "sz", "zst", "rar"];
 pub const SUPPORTED_ALIASES: &[&str] = &["tgz", "tbz", "tlz4", "txz", "tzlma", "tsz", "tzst"];
-pub const PRETTY_SUPPORTED_EXTENSIONS: &str = "tar, zip, bz, bz2, gz, lz4, xz, lzma, sz, zst";
+pub const PRETTY_SUPPORTED_EXTENSIONS: &str = "tar, zip, bz, bz2, gz, lz4, xz, lzma, sz, zst, rar";
 pub const PRETTY_SUPPORTED_ALIASES: &str = "tgz, tbz, tlz4, txz, tzlma, tsz, tzst";
 
 /// A wrapper around `CompressionFormat` that allows combinations like `tgz`
@@ -72,6 +72,8 @@ pub enum CompressionFormat {
     Zstd,
     /// .zip
     Zip,
+    /// .rar
+    Rar,
 }
 
 impl CompressionFormat {
@@ -79,7 +81,7 @@ impl CompressionFormat {
     fn is_archive_format(&self) -> bool {
         // Keep this match like that without a wildcard `_` so we don't forget to update it
         match self {
-            Tar | Zip => true,
+            Tar | Zip | Rar => true,
             Gzip => false,
             Bzip => false,
             Lz4 => false,
@@ -107,6 +109,7 @@ fn to_extension(ext: &[u8]) -> Option<Extension> {
             b"xz" | b"lzma" => &[Lzma],
             b"sz" => &[Snappy],
             b"zst" => &[Zstd],
+            b"rar" => &[Rar],
             _ => return None,
         },
         ext.to_str_lossy(),

src/utils/fs.rs

@@ -86,6 +86,11 @@ pub fn try_infer_extension(path: &Path) -> Option<Extension> {
     fn is_zst(buf: &[u8]) -> bool {
         buf.starts_with(&[0x28, 0xB5, 0x2F, 0xFD])
     }
+    fn is_rar(buf: &[u8]) -> bool {
+        buf.len() >= 7
+            && buf.starts_with(&[0x52, 0x61, 0x72, 0x21, 0x1A, 0x07])
+            && (buf[6] == 0x00 || (buf.len() >= 8 && buf[6..=7] == [0x01, 0x00]))
+    }
 
     let buf = {
         let mut buf = [0; 270];
@@ -117,6 +122,8 @@ pub fn try_infer_extension(path: &Path) -> Option<Extension> {
         Some(Extension::new(&[Snappy], "sz"))
     } else if is_zst(&buf) {
         Some(Extension::new(&[Zstd], "zst"))
+    } else if is_rar(&buf) {
+        Some(Extension::new(&[Rar], "rar"))
     } else {
         None
     }

tests/integration.rs

@@ -1,7 +1,10 @@
 #[macro_use]
 mod utils;
 
-use std::{iter::once, path::PathBuf};
+use std::{
+    iter::once,
+    path::{Path, PathBuf},
+};
 
 use fs_err as fs;
 use parse_display::Display;
@@ -144,3 +147,26 @@ fn multiple_files(
     ouch!("-A", "d", archive, "-d", after);
     assert_same_directory(before, after, !matches!(ext, DirectoryExtension::Zip));
 }
+
+// test .rar decompression
+fn test_unpack_rar_single(input: &Path) -> Result<(), Box<dyn std::error::Error>> {
+    let dir = tempdir()?;
+    let dirpath = dir.path();
+    let unpacked_path = &dirpath.join("testfile.txt");
+    ouch!("-A", "d", input, "-d", dirpath);
+    let content = fs::read_to_string(unpacked_path)?;
+    assert_eq!(content, "Testing 123\n");
+
+    Ok(())
+}
+
+#[test]
+fn unpack_rar() -> Result<(), Box<dyn std::error::Error>> {
+    let mut datadir = PathBuf::from(std::env::var("CARGO_MANIFEST_DIR")?);
+    datadir.push("tests/data");
+    ["testfile.rar3.rar.gz", "testfile.rar5.rar"]
+        .iter()
+        .try_for_each(|path| test_unpack_rar_single(&datadir.join(path)))?;
+
+    Ok(())
+}

tests/snapshots/ui__ui_test_err_decompress_missing_extension-2.snap

@@ -7,6 +7,6 @@ expression: "run_ouch(\"ouch decompress a b.unknown\", dir)"
  - Files with missing extensions: <FOLDER>/a
  - Decompression formats are detected automatically from file extension
 
-hint: Supported extensions are: tar, zip, bz, bz2, gz, lz4, xz, lzma, sz, zst
+hint: Supported extensions are: tar, zip, bz, bz2, gz, lz4, xz, lzma, sz, zst, rar
 hint: Supported aliases are: tgz, tbz, tlz4, txz, tzlma, tsz, tzst
 

tests/snapshots/ui__ui_test_err_decompress_missing_extension-3.snap

@@ -6,7 +6,7 @@ expression: "run_ouch(\"ouch decompress b.unknown\", dir)"
  - Files with unsupported extensions: <FOLDER>/b.unknown
  - Decompression formats are detected automatically from file extension
 
-hint: Supported extensions are: tar, zip, bz, bz2, gz, lz4, xz, lzma, sz, zst
+hint: Supported extensions are: tar, zip, bz, bz2, gz, lz4, xz, lzma, sz, zst, rar
 hint: Supported aliases are: tgz, tbz, tlz4, txz, tzlma, tsz, tzst
 hint: 
 hint: Alternatively, you can pass an extension to the '--format' flag:

tests/snapshots/ui__ui_test_err_decompress_missing_extension.snap

@@ -6,7 +6,7 @@ expression: "run_ouch(\"ouch decompress a\", dir)"
  - Files with missing extensions: <FOLDER>/a
  - Decompression formats are detected automatically from file extension
 
-hint: Supported extensions are: tar, zip, bz, bz2, gz, lz4, xz, lzma, sz, zst
+hint: Supported extensions are: tar, zip, bz, bz2, gz, lz4, xz, lzma, sz, zst, rar
 hint: Supported aliases are: tgz, tbz, tlz4, txz, tzlma, tsz, tzst
 hint: 
 hint: Alternatively, you can pass an extension to the '--format' flag:

tests/snapshots/ui__ui_test_ok_decompress.snap

@@ -2,7 +2,6 @@
 source: tests/ui.rs
 expression: "run_ouch(\"ouch decompress output.zst\", dir)"
 ---
-[INFO] Failed to confirm the format of `output` by sniffing the contents, file might be misnamed
 [INFO] Successfully decompressed archive in current directory.
 [INFO] Files unpacked: 1