Merge pull request #98 from sorairolake/feature/archives

Add bzip3 and LZ4 support

Author: bojand

README.md

@@ -157,6 +157,7 @@ assert_eq!(kind.extension(), "foo");
 - **rar** - `application/vnd.rar`
 - **gz** - `application/gzip`
 - **bz2** - `application/x-bzip2`
+- **bz3** - `application/vnd.bzip3`
 - **7z** - `application/x-7z-compressed`
 - **xz** - `application/x-xz`
 - **pdf** - `application/pdf`
@@ -175,6 +176,7 @@ assert_eq!(kind.extension(), "foo");
 - **rpm** - `application/x-rpm`
 - **dcm** - `application/dicom`
 - **zst** - `application/zstd`
+- **lz4** - `application/x-lz4`
 - **msi** - `application/x-ole-storage`
 - **cpio** - `application/x-cpio`
 

src/map.rs

@@ -462,6 +462,12 @@ matcher_map!(
         "bz2",
         matchers::archive::is_bz2
     ),
+    (
+        MatcherType::Archive,
+        "application/vnd.bzip3",
+        "bz3",
+        matchers::archive::is_bz3
+    ),
     (
         MatcherType::Archive,
         "application/x-7z-compressed",
@@ -570,6 +576,12 @@ matcher_map!(
         "zst",
         matchers::archive::is_zst
     ),
+    (
+        MatcherType::Archive,
+        "application/x-lz4",
+        "lz4",
+        matchers::archive::is_lz4
+    ),
     (
         MatcherType::Archive,
         "application/x-ole-storage",

src/matchers/archive.rs

@@ -52,11 +52,21 @@ pub fn is_gz(buf: &[u8]) -> bool {
     buf.len() > 2 && buf[0] == 0x1F && buf[1] == 0x8B && buf[2] == 0x8
 }
 
-/// Returns whether a buffer is a bzip archive.
+/// Returns whether a buffer is a bzip2 archive.
 pub fn is_bz2(buf: &[u8]) -> bool {
     buf.len() > 2 && buf[0] == 0x42 && buf[1] == 0x5A && buf[2] == 0x68
 }
 
+/// Returns whether a buffer is a bzip3 archive.
+pub fn is_bz3(buf: &[u8]) -> bool {
+    buf.len() > 4
+        && buf[0] == b'B'
+        && buf[1] == b'Z'
+        && buf[2] == b'3'
+        && buf[3] == b'v'
+        && buf[4] == b'1'
+}
+
 /// Returns whether a buffer is a 7z archive.
 pub fn is_7z(buf: &[u8]) -> bool {
     buf.len() > 5
@@ -239,6 +249,41 @@ pub fn is_zst(buf: &[u8]) -> bool {
     is_zst(next_frame)
 }
 
+/// Returns whether a buffer is a LZ4 archive.
+// LZ4 compressed data is made of one or more frames.
+// There are two frame formats defined by LZ4: LZ4 Frame format and Skippable frames.
+// See more details from https://github.com/lz4/lz4/blob/v1.9.4/doc/lz4_Frame_format.md
+pub fn is_lz4(buf: &[u8]) -> bool {
+    if buf.len() > 3 && buf[0] == 0x04 && buf[1] == 0x22 && buf[2] == 0x4D && buf[3] == 0x18 {
+        return true;
+    }
+
+    if buf.len() < 8 {
+        return false;
+    }
+
+    let magic = u32::from_le_bytes(buf[0..4].try_into().unwrap());
+    let Ok(magic) = usize::try_from(magic) else {
+        return false;
+    };
+
+    if magic & ZSTD_SKIP_MASK != ZSTD_SKIP_START {
+        return false;
+    }
+
+    let data_len = u32::from_le_bytes(buf[4..8].try_into().unwrap());
+    let Ok(data_len) = usize::try_from(data_len) else {
+        return false;
+    };
+
+    if buf.len() < 8 + data_len {
+        return false;
+    }
+
+    let next_frame = &buf[8 + data_len..];
+    is_lz4(next_frame)
+}
+
 /// Returns whether a buffer is a MSI Windows Installer archive.
 pub fn is_msi(buf: &[u8]) -> bool {
     buf.len() > 7

testdata/sample.tar.bz3

@@ -1,5 +1,12 @@
 mod common;
 
+test_format!(
+    Archive,
+    "application/vnd.bzip3",
+    "bz3",
+    bz3,
+    "sample.tar.bz3"
+);
 test_format!(
     Archive,
     "application/vnd.sqlite3",
@@ -9,6 +16,7 @@ test_format!(
 );
 
 test_format!(Archive, "application/zstd", "zst", zst, "sample.tar.zst");
+test_format!(Archive, "application/x-lz4", "lz4", lz4, "sample.tar.lz4");
 test_format!(Archive, "application/x-cpio", "cpio", cpio, "sample.cpio");
 test_format!(
     Archive,