add sanity check (decoded blob == batch bytes)

aggregator/src/eip4844.rs

@@ -54,3 +54,36 @@ pub fn get_blob_bytes(batch_bytes: &[u8]) -> Vec<u8> {
 
     blob_bytes
 }
+
+/// Given the blob's bytes, take into account the first byte, i.e. enable_encoding? and either spit
+/// out the raw bytes or zstd decode them.
+pub fn decode_blob(blob_bytes: &[u8]) -> std::io::Result<Vec<u8>> {
+    let enable_encoding = blob_bytes[0].eq(&1);
+
+    // If not encoded, spit out the rest of the bytes, as it is.
+    if !enable_encoding {
+        return Ok(blob_bytes[1..].to_vec());
+    }
+
+    // The bytes following the first byte represent the zstd-encoded bytes.
+    let mut encoded_bytes = blob_bytes[1..].to_vec();
+    let mut encoded_len = encoded_bytes.len();
+    let mut decoded_bytes = Vec::with_capacity(5 * 4096 * 32);
+    loop {
+        let mut decoder = zstd_encoder::zstd::stream::read::Decoder::new(encoded_bytes.as_slice())?;
+        decoder.include_magicbytes(false)?;
+        decoder.window_log_max(30)?;
+
+        decoded_bytes.clear();
+
+        if std::io::copy(&mut decoder, &mut decoded_bytes).is_ok() {
+            break;
+        }
+
+        // The error above means we need to truncate the suffix 0-byte.
+        encoded_len -= 1;
+        encoded_bytes.truncate(encoded_len);
+    }
+
+    Ok(decoded_bytes)
+}

prover/src/aggregator/prover.rs

@@ -1,6 +1,8 @@
 use std::{collections::BTreeMap, env, iter::repeat};
 
-use aggregator::{BatchHash, BatchHeader, ChunkInfo, MAX_AGG_SNARKS};
+use aggregator::{
+    eip4844::decode_blob, BatchData, BatchHash, BatchHeader, ChunkInfo, MAX_AGG_SNARKS,
+};
 use anyhow::{bail, Result};
 use eth_types::H256;
 use halo2_proofs::{halo2curves::bn256::Bn256, poly::kzg::commitment::ParamsKZG};
@@ -212,6 +214,16 @@ impl<'params> Prover<'params> {
         let batch_hash = batch_header.batch_hash();
         let batch_info: BatchHash<N_SNARKS> =
             BatchHash::construct(&chunk_hashes, batch_header, &batch.blob_bytes);
+        let batch_data: BatchData<N_SNARKS> = BatchData::from(&batch_info);
+
+        // sanity check:
+        // - conditionally decoded blob should match batch data.
+        let batch_bytes = batch_data.get_batch_data_bytes();
+        let decoded_blob_bytes = decode_blob(&batch.blob_bytes)?;
+        assert_eq!(
+            batch_bytes, decoded_blob_bytes,
+            "BatchProvingTask(sanity) mismatch batch bytes and decoded blob bytes",
+        );
 
         let layer3_snark = self.prover_impl.load_or_gen_agg_snark(
             name,