Module multipart 

Multipart upload bookkeeping via redb.

Architecture

MultipartTable wraps a shared redb Database handle and provides typed CRUD operations over the multipart_active table. The same DB file used by CacheStore is reused so there is only one redb file handle per process.

Table schema

• Table name: "multipart_active"
• Key : "<profile_id>\x00<bucket>\x00<key>" — composite string key.
• Value: serde_json::to_vec(MultipartRecord).

Key design (OCP note)

The composite key is (profile, bucket, object_key) rather than upload_id. Concurrent uploads to the same (profile, bucket, key) are rare; the second upload overwrites the first record, keeping the table in sync with the active upload. find_by_upload_id does a linear scan and is used only by the cleanup scanner (task 38).

Structs

MultipartRecord

Persisted record for one in-flight multipart upload.

MultipartTable

Typed wrapper around the multipart_active redb table.

MultipartTableHandle

Tauri managed state handle for MultipartTable.

MultipartUpload

Describes one in-progress multipart upload as returned by the cleanup scanner.

Enums

MultipartSource

Discriminates the origin of an in-progress multipart upload.

Constants

MULTIPART_TABLE 🔒

redb table for in-flight multipart uploads.

Functions

abort_multipart_upload

Abort a single in-progress multipart upload.

scan_multipart_uploads

List all in-progress multipart uploads for bucket, classify each as Brows3r or Unknown, and optionally filter out uploads younger than older_than_secs.