| 1 | //! Uploaded attachments: content-addressed files stored outside git. |
| 2 | //! |
| 3 | //! The bytes live on disk under `storage::attachment_path` (never in a git |
| 4 | //! repository); this module owns the database rows that index them per repo |
| 5 | //! and the content hashing used as their address. See [`crate::models::Attachment`]. |
| 6 | |
| 7 | use sha2::{ |
| 8 | Digest, |
| 9 | Sha256, |
| 10 | }; |
| 11 | |
| 12 | use crate::{ |
| 13 | error::Result, |
| 14 | models::Attachment, |
| 15 | }; |
| 16 | |
| 17 | /// Lowercase hex SHA-256 of `bytes` — the content address used as both the |
| 18 | /// dedup key and the on-disk filename. |
| 19 | pub fn content_hash(bytes: &[u8]) -> String { |
| 20 | Sha256::digest(bytes) |
| 21 | .iter() |
| 22 | .map(|b| format!("{b:02x}")) |
| 23 | .collect() |
| 24 | } |
| 25 | |
| 26 | /// The attachment row for `(repo_id, hash)`, if one exists. |
| 27 | pub async fn find(db: &toasty::Db, repo_id: i64, hash: &str) -> Result<Option<Attachment>> { |
| 28 | let mut conn = db.clone(); |
| 29 | let row = Attachment::filter(Attachment::fields().repo_id().eq(repo_id)) |
| 30 | .filter(Attachment::fields().hash().eq(hash)) |
| 31 | .first() |
| 32 | .exec(&mut conn) |
| 33 | .await?; |
| 34 | Ok(row) |
| 35 | } |
| 36 | |
| 37 | /// Record an attachment for a repo, deduping on content: if `hash` is already |
| 38 | /// recorded for `repo_id` the existing row is returned and no new row is made. |
| 39 | /// The caller writes the bytes to [`crate::storage::attachment_path`] itself. |
| 40 | pub async fn add( |
| 41 | db: &toasty::Db, |
| 42 | repo_id: i64, |
| 43 | hash: &str, |
| 44 | content_type: &str, |
| 45 | size: i64, |
| 46 | uploader_id: i64, |
| 47 | ) -> Result<Attachment> { |
| 48 | if let Some(existing) = find(db, repo_id, hash).await? { |
| 49 | return Ok(existing); |
| 50 | } |
| 51 | let mut conn = db.clone(); |
| 52 | let row = toasty::create!(Attachment { |
| 53 | repo_id: repo_id, |
| 54 | hash: hash, |
| 55 | content_type: content_type, |
| 56 | size: size, |
| 57 | uploader_id: uploader_id, |
| 58 | created_at: crate::now(), |
| 59 | }) |
| 60 | .exec(&mut conn) |
| 61 | .await?; |
| 62 | Ok(row) |
| 63 | } |
| 64 | |
| 65 | #[cfg(test)] |
| 66 | mod tests { |
| 67 | use super::*; |
| 68 | |
| 69 | #[test] |
| 70 | fn content_hash_is_stable_lowercase_hex_sha256() { |
| 71 | // Known SHA-256 of "hello". |
| 72 | assert_eq!( |
| 73 | content_hash(b"hello"), |
| 74 | "2cf24dba5fb0a30e26e83b2ac5b9e29e1b161e5c1fa7425e73043362938b9824" |
| 75 | ); |
| 76 | assert_eq!(content_hash(b"a"), content_hash(b"a")); |
| 77 | assert_ne!(content_hash(b"a"), content_hash(b"b")); |
| 78 | } |
| 79 | } |