src/backup/data_blob.rs: new file format for binary blobs

diff --git a/src/backup.rs b/src/backup.rs
index 6cec36fe410348280b958c8b16fa12ce42369fe8..6e781e2a41013c08e91c353533a7ed47a6b2bba8 100644 (file)
--- a/src/backup.rs
+++ b/src/backup.rs
@@ -121,6 +121,18 @@ pub static COMPRESSED_CHUNK_MAGIC_1_0: [u8; 8] = [191, 237, 46, 195, 108, 17, 22
 // openssl::sha::sha256(b"Proxmox Backup zstd compressed encrypted chunk v1.0")[0..8]
 pub static ENCR_COMPR_CHUNK_MAGIC_1_0: [u8; 8] = [9, 40, 53, 200, 37, 150, 90, 196];
 
+// openssl::sha::sha256(b"Proxmox Backup uncompressed blob v1.0")[0..8]
+pub static UNCOMPRESSED_BLOB_MAGIC_1_0: [u8; 8] = [66, 171, 56, 7, 190, 131, 112, 161];
+
+//openssl::sha::sha256(b"Proxmox Backup zstd compressed blob v1.0")[0..8]
+pub static COMPRESSED_BLOB_MAGIC_1_0: [u8; 8] = [49, 185, 88, 66, 111, 182, 163, 127];
+
+// openssl::sha::sha256(b"Proxmox Backup encrypted blob v1.0")[0..8]
+pub static ENCRYPTED_BLOB_MAGIC_1_0: [u8; 8] = [123, 103, 133, 190, 34, 45, 76, 240];
+
+// openssl::sha::sha256(b"Proxmox Backup zstd compressed encrypted blob v1.0")[0..8]
+pub static ENCR_COMPR_BLOB_MAGIC_1_0: [u8; 8] = [230, 89, 27, 191, 11, 191, 216, 11];
+
 // openssl::sha::sha256(b"Proxmox Backup fixed sized chunk index v1.0")[0..8]
 pub static FIXED_SIZED_CHUNK_INDEX_1_0: [u8; 8] = [47, 127, 65, 237, 145, 253, 15, 205];
 
@@ -136,6 +148,9 @@ pub use key_derivation::*;
 mod data_chunk;
 pub use data_chunk::*;
 
+mod data_blob;
+pub use data_blob::*;
+
 mod chunk_stream;
 pub use chunk_stream::*;
 

diff --git a/src/backup/data_blob.rs b/src/backup/data_blob.rs
new file mode 100644 (file)
index 0000000..9a279b1
--- /dev/null
+++ b/src/backup/data_blob.rs
@@ -0,0 +1,107 @@
+use failure::*;
+use std::convert::TryInto;
+use std::io::Write;
+
+use super::*;
+
+/// Data blob binary storage format
+///
+/// Data blobs store arbitrary binary data (< 16MB), and can be
+/// compressed and encrypted. A simply binary format is used to store
+/// them on disk or transfer them over the network. Please use index
+/// files to store large data files (".fidx" of ".didx").
+///
+/// The format start with a 8 byte magic number to identify the type.
+/// Encrypted blobs contain a 16 byte IV, followed by a 18 byte AD
+/// tag, followed by the encrypted data (MAGIC || IV || TAG ||
+/// EncryptedData).
+///
+/// Unencrypted blobs simply contain the (compressed) data.
+///
+/// This is basically the same format we use for ``DataChunk``, but
+/// with other magic numbers so that we can distinguish them.
+pub struct DataBlob {
+    raw_data: Vec<u8>, // tagged, compressed, encryped data
+}
+
+impl DataBlob {
+
+    /// accessor to raw_data field
+    pub fn raw_data(&self) -> &[u8]  {
+        &self.raw_data
+    }
+
+    /// accessor to chunk type (magic number)
+    pub fn magic(&self) -> &[u8; 8] {
+        self.raw_data[0..8].try_into().unwrap()
+    }
+
+    pub fn encode(
+        data: &[u8],
+        config: Option<&CryptConfig>,
+        compress: bool,
+    ) -> Result<Self, Error> {
+
+        if data.len() > 16*1024*1024 {
+            bail!("data blob too large ({} bytes).", data.len());
+        }
+
+        if let Some(config) = config {
+
+            let enc_data = config.encode_chunk(
+                data,
+                compress,
+                &ENCRYPTED_BLOB_MAGIC_1_0,
+                &ENCR_COMPR_BLOB_MAGIC_1_0,
+            )?;
+            return Ok(DataBlob { raw_data: enc_data });
+        } else {
+
+            if compress {
+                let mut comp_data = Vec::with_capacity(data.len() + 8);
+
+                comp_data.write_all(&COMPRESSED_BLOB_MAGIC_1_0)?;
+                zstd::stream::copy_encode(data, &mut comp_data, 1)?;
+
+                if comp_data.len() < (data.len() + 8) {
+                    return Ok(DataBlob { raw_data: comp_data });
+                }
+            }
+
+            let mut raw_data = Vec::with_capacity(data.len() + 8);
+
+            raw_data.write_all(&UNCOMPRESSED_BLOB_MAGIC_1_0)?;
+            raw_data.extend_from_slice(data);
+
+            return Ok(DataBlob { raw_data });
+        }
+    }
+
+    /// Decode blob data
+    pub fn decode(self, config: Option<&CryptConfig>) -> Result<Vec<u8>, Error> {
+
+        let magic = self.magic();
+
+        if magic == &UNCOMPRESSED_BLOB_MAGIC_1_0 {
+             return Ok(self.raw_data);
+        } else if magic == &COMPRESSED_BLOB_MAGIC_1_0 {
+
+            let data = zstd::block::decompress(&self.raw_data[8..], 16*1024*1024)?;
+            return Ok(data);
+
+        } else if magic == &ENCR_COMPR_BLOB_MAGIC_1_0 || magic == &ENCRYPTED_BLOB_MAGIC_1_0 {
+            if let Some(config) = config  {
+                let data = if magic == &ENCR_COMPR_BLOB_MAGIC_1_0 {
+                    config.decode_compressed_chunk(&self.raw_data)?
+                } else {
+                    config.decode_uncompressed_chunk(&self.raw_data)?
+                };
+                return Ok(data);
+            } else {
+                bail!("unable to decrypt blob - missing CryptConfig");
+            }
+        } else {
+            bail!("Invalid blob magic number.");
+        }
+    }
+}