specfold_core/files/
hash_cache.rs

1//! Attachment hashes remembered between runs, in the user's cache directory, keyed by each
2//! file's size and modification time. Checking that recorded hashes still match then reads only
3//! the files that changed.
4
5use super::hash;
6use crate::config::paths;
7use serde::{Deserialize, Serialize};
8use sha2::{Digest, Sha256};
9use std::collections::HashMap;
10use std::path::{Path, PathBuf};
11use std::time::UNIX_EPOCH;
12
13const FORMAT: &str = "specfold-hashes-1";
14
15#[derive(Serialize, Deserialize, Clone)]
16struct Entry {
17    mtime_ns: u128,
18    size: u64,
19    sha256: String,
20}
21
22#[derive(Serialize, Deserialize, Default)]
23struct Stored {
24    format: String,
25    entries: HashMap<String, Entry>,
26}
27
28pub struct HashCache {
29    file: Option<PathBuf>,
30    stored: Stored,
31    dirty: bool,
32}
33
34impl HashCache {
35    pub fn open(root: &Path) -> HashCache {
36        let dir = if cfg!(test) { None } else { paths::cache_dir() };
37        let key = Sha256::digest(root.to_string_lossy().as_bytes());
38        let name: String = key[..8].iter().map(|b| format!("{b:02x}")).collect();
39        let file = dir.map(|d| d.join(format!("hashes-{name}.json")));
40        if let Some(f) = &file {
41            crate::config::cache_gc::touch(f);
42        }
43        let stored = file
44            .as_ref()
45            .and_then(|f| std::fs::read(f).ok())
46            .and_then(|b| serde_json::from_slice::<Stored>(&b).ok())
47            .filter(|s| s.format == FORMAT)
48            .unwrap_or(Stored { format: FORMAT.into(), entries: HashMap::new() });
49        HashCache { file, stored, dirty: false }
50    }
51
52    /// SHA-256 (hex) and size of `path`, read only when its size or modification time changed.
53    pub fn file(&mut self, path: &Path) -> std::io::Result<(String, u64)> {
54        let meta = std::fs::metadata(path)?;
55        let mtime_ns = meta.modified()?.duration_since(UNIX_EPOCH).map(|d| d.as_nanos()).unwrap_or(0);
56        let key = path.to_string_lossy().into_owned();
57        if let Some(e) = self.stored.entries.get(&key).filter(|e| e.mtime_ns == mtime_ns && e.size == meta.len()) {
58            return Ok((e.sha256.clone(), e.size));
59        }
60        let (sha256, size) = hash::file(path)?;
61        self.stored.entries.insert(key, Entry { mtime_ns, size, sha256: sha256.clone() });
62        self.dirty = true;
63        Ok((sha256, size))
64    }
65
66    /// Writes the cache when it changed. Failures are ignored: the cache is an optimisation.
67    pub fn save(&self) {
68        let (Some(file), true) = (&self.file, self.dirty) else { return };
69        let Some(dir) = file.parent() else { return };
70        if std::fs::create_dir_all(dir).is_err() {
71            return;
72        }
73        let tmp = file.with_extension(format!("tmp{}", std::process::id()));
74        if let Ok(bytes) = serde_json::to_vec(&self.stored)
75            && std::fs::write(&tmp, bytes).is_ok()
76        {
77            let _ = std::fs::rename(&tmp, file);
78        }
79    }
80}
specfold_core/files/hash_cache.rs

specfold_core/files/
hash_cache.rs
