Struct DiskCache 

pub struct DiskCache { /* private fields */ }

Persistent disk cache for analysis results.

Stores serialized analysis outputs in a directory hierarchy, with per-shard advisory locking to prevent concurrent writes to the same entry. Supports atomic writes via NamedTempFile::persist and graceful degradation on I/O errors.

Implementations

impl DiskCache

pub fn drain_write_failures(&self) -> u64

Returns the number of write failures accumulated since the last call and resets the per-drain counter. The cumulative total_write_failures is never reset.

pub fn is_degraded(&self) -> bool

Returns true when cumulative write failures have reached DISK_CACHE_DEGRADED_THRESHOLD. Callers can use this to emit a degraded health signal without polling the counter.

pub fn cache_stats(&self) -> (u64, u64)

Returns cache statistics as (entry_count, total_size_bytes). Note: size_bytes is approximate; deleted files decrement entry_count but not size_bytes.

impl DiskCache

pub fn new(base: PathBuf, disabled: bool) -> Self

Creates the cache directory (mode 0700) and returns a new instance. If disabled is true, or if directory creation fails, all operations are no-ops.

pub fn entry_path(&self, tool: &str, key: &Hash) -> PathBuf

pub fn get<T: DeserializeOwned>(&self, tool: &str, key: &Hash) -> Option<T>

Retrieves a cached entry by key, decompressing and deserializing on success. Returns None if the entry does not exist, is corrupted, or deserialization fails.

pub fn put<T: Serialize>(&self, tool: &str, key: &Hash, value: &T)

Atomic write via NamedTempFile::persist (rename(2)). Silently drops all errors.

pub fn evict_stale(&self, retention_days: u64)

Removes files not accessed within retention_days. Best-effort; silently drops errors.