haematite 0.4.1

Content-addressed, branchable, actor-native storage engine
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97

//! HBR1 record codec and ref-file naming for the branch ref store (§2.1,
//! §2.2). Split out of `refstore.rs` via `#[path]` so that file stays within
//! the branch module's 500-line cap; the record layout is documented on the
//! parent module.

use std::path::Path;

use super::{BranchRefError, BranchRefRecord, BranchShardRef};
use crate::branch::persist::{Reader, push_bytes, push_u64};

const REF_MAGIC: &[u8; 4] = b"HBR1";
pub(super) const REF_EXTENSION: &str = "ref";

/// `<hex(blake3(name)[0..16])>.ref` (§2.1): the file name is derived from the
/// name hash so arbitrary byte-string names are filesystem-safe; the full name
/// lives inside the record.
pub(super) fn ref_file_name(name: &str) -> String {
    const HEX: &[u8; 16] = b"0123456789abcdef";
    let digest = blake3::hash(name.as_bytes());
    let mut file_name = String::with_capacity(36);
    for byte in &digest.as_bytes()[..16] {
        file_name.push(char::from(HEX[usize::from(byte >> 4)]));
        file_name.push(char::from(HEX[usize::from(byte & 0x0f)]));
    }
    file_name.push('.');
    file_name.push_str(REF_EXTENSION);
    file_name
}

/// Rewraps a decode failure so the error names the offending file (§2.2).
pub(super) fn corrupt_in_file(path: &Path, error: &BranchRefError) -> BranchRefError {
    BranchRefError::Corrupt(format!("{}: {error}", path.display()))
}

pub(super) fn encode_record(record: &BranchRefRecord) -> Vec<u8> {
    let mut bytes = Vec::new();
    bytes.extend_from_slice(REF_MAGIC);
    push_bytes(&mut bytes, record.name.as_bytes());
    push_u64(&mut bytes, record.created);
    push_u64(&mut bytes, record.seq);
    push_u64(&mut bytes, record.timestamp);
    push_u64(&mut bytes, record.shards.len() as u64);
    for shard in &record.shards {
        push_u64(&mut bytes, shard.shard_id as u64);
        bytes.extend_from_slice(shard.fork_anchor.as_bytes());
        bytes.extend_from_slice(shard.head.as_bytes());
    }
    push_u64(&mut bytes, record.parents.len() as u64);
    for parent in &record.parents {
        bytes.extend_from_slice(parent.as_bytes());
    }
    bytes
}

pub(super) fn decode_record(bytes: &[u8]) -> Result<BranchRefRecord, BranchRefError> {
    let mut reader = Reader::new(bytes);
    reader.expect_magic(*REF_MAGIC)?;
    let name = String::from_utf8(reader.read_bytes()?)
        .map_err(|_error| BranchRefError::Corrupt("branch name is not valid UTF-8".into()))?;
    let created = reader.read_u64()?;
    let seq = reader.read_u64()?;
    let timestamp = reader.read_u64()?;
    let shard_count = reader.read_usize()?;
    let mut shards: Vec<BranchShardRef> = Vec::with_capacity(shard_count);
    for _ in 0..shard_count {
        let shard_id = reader.read_usize()?;
        if shards
            .last()
            .is_some_and(|previous| previous.shard_id >= shard_id)
        {
            return Err(BranchRefError::Corrupt(
                "shard ids are not strictly ascending".into(),
            ));
        }
        let fork_anchor = reader.read_hash()?;
        let head = reader.read_hash()?;
        shards.push(BranchShardRef {
            shard_id,
            fork_anchor,
            head,
        });
    }
    let parent_count = reader.read_usize()?;
    let mut parents = Vec::with_capacity(parent_count);
    for _ in 0..parent_count {
        parents.push(reader.read_hash()?);
    }
    reader.finish()?;
    Ok(BranchRefRecord {
        name,
        created,
        seq,
        timestamp,
        shards,
        parents,
    })
}