harmont_cli/orchestrator/
archive.rs

1//! Per-run source archive store.
2//!
3//! On build start the orchestrator tar.gzs the user's working
4//! directory once (via [`crate::orchestrator::source::build_archive_bytes`])
5//! and registers the bytes under an opaque `ArchiveId`. Step-executor
6//! plugins receive that ID in their `ExecutorInput` and pull bytes
7//! via `hm_archive_read`. The host caches archives in memory keyed
8//! by ID for the duration of a single `orchestrator::run` invocation.
9
10use std::collections::HashMap;
11use std::sync::Mutex;
12
13use hm_plugin_protocol::ArchiveId;
14use uuid::Uuid;
15
16#[derive(Debug, Default)]
17pub struct ArchiveStore {
18    archives: Mutex<HashMap<ArchiveId, Vec<u8>>>,
19}
20
21impl ArchiveStore {
22    #[must_use]
23    pub fn new() -> Self {
24        Self::default()
25    }
26
27    /// Register a new archive. Returns the freshly-minted ID.
28    pub fn register(&self, bytes: Vec<u8>) -> ArchiveId {
29        let id = ArchiveId::from(Uuid::new_v4());
30        let _ = self.archives.lock().map(|mut m| m.insert(id, bytes));
31        id
32    }
33
34    /// Total size of the archive identified by `id`, or `0` if no
35    /// such archive is registered.
36    #[must_use]
37    pub fn total_size(&self, id: ArchiveId) -> u64 {
38        self.archives
39            .lock()
40            .ok()
41            .and_then(|m| m.get(&id).map(|b| b.len() as u64))
42            .unwrap_or(0)
43    }
44
45    /// Return a clone of the full archive bytes, or `None` if unknown.
46    #[must_use]
47    pub fn get_bytes(&self, id: ArchiveId) -> Option<Vec<u8>> {
48        self.archives.lock().ok()?.get(&id).cloned()
49    }
50
51    /// Read up to `max` bytes from offset `offset`. Returns empty
52    /// when offset is beyond end, or when the archive is unknown.
53    #[must_use]
54    pub fn read(&self, id: ArchiveId, offset: u64, max: u64) -> Vec<u8> {
55        // We must hold the lock guard across the read so the bytes
56        // slice we copy out is consistent. `let...else` here would
57        // require holding `bytes` across the early-return branch.
58        let Ok(g) = self.archives.lock() else {
59            return Vec::new();
60        };
61        let Some(bytes) = g.get(&id) else {
62            return Vec::new();
63        };
64        // Archive sizes fit in `usize` on 64-bit hosts (the only
65        // supported targets); local-mode archives are at most a few
66        // hundred MB. The cast cannot truncate in practice.
67        #[allow(
68            clippy::cast_possible_truncation,
69            reason = "archive sizes fit in usize on supported 64-bit hosts"
70        )]
71        let start = (offset as usize).min(bytes.len());
72        #[allow(
73            clippy::cast_possible_truncation,
74            reason = "archive sizes fit in usize on supported 64-bit hosts"
75        )]
76        let max_us = max as usize;
77        let end = start.saturating_add(max_us).min(bytes.len());
78        bytes[start..end].to_vec()
79    }
80}
81
82#[cfg(test)]
83#[allow(clippy::unwrap_used, clippy::expect_used, clippy::panic)]
84mod tests {
85    use super::*;
86
87    #[test]
88    fn register_then_read_round_trip() {
89        let s = ArchiveStore::new();
90        let id = s.register(b"hello world".to_vec());
91        assert_eq!(s.total_size(id), 11);
92        assert_eq!(s.read(id, 0, 5), b"hello");
93        assert_eq!(s.read(id, 6, 5), b"world");
94        assert_eq!(s.read(id, 100, 5), Vec::<u8>::new());
95    }
96
97    #[test]
98    fn unknown_id_returns_empty() {
99        let s = ArchiveStore::new();
100        let bogus = ArchiveId(Uuid::new_v4());
101        assert_eq!(s.total_size(bogus), 0);
102        assert_eq!(s.read(bogus, 0, 100), Vec::<u8>::new());
103    }
104}
harmont_cli/orchestrator/archive.rs

harmont_cli/orchestrator/
archive.rs
