OXYGENHUB
# 3007d8f011fa oxygenflow / crates / flow-core / src / canonical.rs
4.3 KiB · Rust · last changed Job artifacts: output declarations, extraction and upload 1 month ago source blame history raw
1 
//! The canonical job graph (spec §3.2) and object hashes. Pipelines,
2 
//! jobs and receipts are encoded with Oxygen's deterministic CBOR profile
3 
//! (Oxygen spec §5.1: unsigned integer map keys in order) and hashed the
4 
//! way Oxygen hashes objects: `BLAKE3("oxy1 " || type || " " || len || 0x00 || body)`.
5
6 
use crate::pipeline::{Job, Pipeline, Sandbox};
7 
use oxy_core::cbor::{encode, Value};
8
9 
/// A 32-byte object reference, shown as hex.
10 
pub type Hash = [u8; 32];
11
12 
/// Oxygen's object hash construction for a Flow object type.
13 
pub fn object_hash(ty: &str, body: &[u8]) -> Hash {
14 
    let mut h = blake3::Hasher::new();
15 
    h.update(format!("oxy1 {ty} {}\0", body.len()).as_bytes());
16 
    h.update(body);
17 
    *h.finalize().as_bytes()
18 
}
19
20 
fn text(s: &str) -> Value {
21 
    Value::Text(s.to_string())
22 
}
23
24 
fn texts(v: &[String]) -> Value {
25 
    Value::Array(v.iter().map(|s| text(s)).collect())
26 
}
27
28 
fn hex32(s: &str) -> Value {
29 
    Value::Bytes(hex::decode(s).unwrap_or_default())
30 
}
31
32 
pub fn sandbox_value(s: &Sandbox) -> Value {
33 
    let mut fields = vec![
34 
        (0, text(&s.name)),
35 
        (1, text(s.runtime.as_str())),
36 
        (2, hex32(&s.rootfs)),
37 
        (3, Value::Uint(s.memory_mib as u64)),
38 
        (4, Value::Uint(s.cpus as u64)),
39 
        (5, text(s.network.as_str())),
40 
    ];
41 
    // Environment as sorted [name, value] pairs; absent when empty, so
42 
    // sandboxes without one hash as before.
43 
    if !s.env.is_empty() {
44 
        fields.push((6, Value::Array(s.env.iter().map(|(k, v)| Value::Array(vec![text(k), text(v)])).collect())));
45 
    }
46 
    Value::map(fields)
47 
}
48
49 
pub fn job_value(j: &Job) -> Value {
50 
    let mut fields = vec![
51 
        (0, text(&j.name)),
52 
        (1, text(&j.sandbox)),
53 
        (2, texts(&j.scope)),
54 
        (3, texts(&j.depends_on)),
55 
        // Environment as sorted [name, value] pairs: the profile has integer map keys only.
56 
        (4, Value::Array(j.env.iter().map(|(k, v)| Value::Array(vec![text(k), text(v)])).collect())),
57 
        (5, texts(&j.exec)),
58 
    ];
59 
    // Mounts as [source, target, size MiB] sorted by target; absent when
60 
    // there are none, so jobs without mounts hash as before.
61 
    if !j.mounts.is_empty() {
62 
        fields.push((
63 
            6,
64 
            Value::Array(
65 
                j.mounts
66 
                    .iter()
67 
                    .map(|m| Value::Array(vec![text(m.source.as_str()), text(&m.target), Value::Uint(m.size_mib as u64)]))
68 
                    .collect(),
69 
            ),
70 
        ));
71 
    }
72 
    // Artifacts as [name, path] sorted by name; absent when there are none.
73 
    if !j.outputs.is_empty() {
74 
        fields.push((7, Value::Array(j.outputs.iter().map(|o| Value::Array(vec![text(&o.name), text(&o.path)])).collect())));
75 
    }
76 
    Value::map(fields)
77 
}
78
79 
impl Pipeline {
80 
    /// The canonical CBOR job graph: sandboxes and jobs in name order.
81 
    pub fn canonical(&self) -> Vec<u8> {
82 
        encode(&Value::map([
83 
            (0, Value::Uint(self.schema_version as u64)),
84 
            (1, Value::Array(self.sandboxes.values().map(sandbox_value).collect())),
85 
            (2, Value::Array(self.jobs.values().map(job_value).collect())),
86 
        ]))
87 
    }
88
89 
    /// `pipeline_hash` of receipts and provenance.
90 
    pub fn hash(&self) -> Hash {
91 
        object_hash("pipeline", &self.canonical())
92 
    }
93 
}
94
95 
impl Job {
96 
    /// `Hash(J)` of spec §6.2: the job with its sandbox, so a change to
97 
    /// either is a different job.
98 
    pub fn hash(&self, sandbox: &Sandbox) -> Hash {
99 
        object_hash("job", &encode(&Value::Array(vec![job_value(self), sandbox_value(sandbox)])))
100 
    }
101 
}
102
103 
/// BLAKE3 tree hash of the files a scope selects (spec §6.2): sorted
104 
/// `(path, content hash)` pairs. An empty scope selects every file.
105 
pub fn scope_tree_hash<'a>(scope: &[String], files: impl IntoIterator<Item = (&'a str, Hash)>) -> Hash {
106 
    let set = {
107 
        let mut b = globset::GlobSetBuilder::new();
108 
        for g in scope {
109 
            if let Ok(g) = globset::Glob::new(g) {
110 
                b.add(g);
111 
            }
112 
        }
113 
        b.build().unwrap_or_else(|_| globset::GlobSet::empty())
114 
    };
115 
    let mut picked: Vec<(&str, Hash)> = files.into_iter().filter(|(p, _)| scope.is_empty() || set.is_match(p)).collect();
116 
    picked.sort();
117 
    let body = encode(&Value::Array(picked.iter().map(|(p, h)| Value::Array(vec![text(p), Value::Bytes(h.to_vec())])).collect()));
118 
    object_hash("scope", &body)
119 
}