refactor(tvix/store): move upload_chunk out of blobwriter r/5948

This is useful not only in blobwriter contexts.

Change-Id: I4c584b5264ff7b4bb3b1a9671affc39e18bf4ccf
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8245
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>

1 files changed, 30 insertions, 0 deletions

diff --git a/tvix/store/src/chunkservice/util.rs b/tvix/store/src/chunkservice/util.rs
new file mode 100644
index 0000000000..cf644fa51a
--- /dev/null
+++ b/tvix/store/src/chunkservice/util.rs
@@ -0,0 +1,30 @@
+use tracing::{debug, instrument};
+
+use crate::Error;
+
+use super::ChunkService;
+
+// upload a chunk to the chunk service, and return its digest (or an error) when done.
+#[instrument(skip_all, err)]
+pub fn upload_chunk<CS: ChunkService>(
+    chunk_service: &CS,
+    chunk_data: Vec<u8>,
+) -> Result<Vec<u8>, Error> {
+    let mut hasher = blake3::Hasher::new();
+    // TODO: benchmark this number and factor it out
+    if chunk_data.len() >= 128 * 1024 {
+        hasher.update_rayon(&chunk_data);
+    } else {
+        hasher.update(&chunk_data);
+    }
+    let digest = hasher.finalize();
+
+    if chunk_service.has(digest.as_bytes())? {
+        debug!("already has chunk, skipping");
+    }
+    let digest_resp = chunk_service.put(chunk_data)?;
+
+    assert_eq!(digest_resp, digest.as_bytes());
+
+    Ok(digest.as_bytes().to_vec())
+}