- add doc strings

2026-08-07 17:00:28 +02:00
parent b950f4a39d
commit 5ad9a44ecd
16 changed files with 993 additions and 428 deletions
@@ -0,0 +1,322 @@
+use crate::axes::{Axis, Operation};
+use indexmap::{Equivalent, IndexMap};
+use ndarray::{ArrayD, SliceInfoElem};
+use std::any::Any;
+use std::fmt::{Debug, Formatter};
+use std::hash::{Hash, Hasher};
+use std::path::{Path, PathBuf};
+use std::sync::{Arc, Mutex, OnceLock};
+
+use crate::readers::Frame;
+
+/// maximum number of frames held in the cache
+const DEFAULT_FRAME_CACHE_SIZE: usize = 128;
+
+/// maximum number of materialized arrays held in the cache
+const DEFAULT_ARRAY_CACHE_SIZE: usize = 32;
+
+/// identity of the reader a frame was read from
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+pub(crate) struct ReaderKey {
+    pub(crate) name: String,
+    pub(crate) path: PathBuf,
+    pub(crate) series: usize,
+    pub(crate) position: usize,
+}
+
+/// borrowed view of [`ReaderKey`] for cache lookups without allocation.
+/// hashes byte-identically to [`ReaderKey`] (str/String and Path/PathBuf hash
+/// the same), so it only matches the same reader identity.
+#[derive(Debug, PartialEq, Eq)]
+pub(crate) struct ReaderKeyRef<'a> {
+    pub(crate) name: &'a str,
+    pub(crate) path: &'a Path,
+    pub(crate) series: usize,
+    pub(crate) position: usize,
+}
+
+impl Hash for ReaderKeyRef<'_> {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.name.hash(state);
+        self.path.hash(state);
+        self.series.hash(state);
+        self.position.hash(state);
+    }
+}
+
+impl Equivalent<ReaderKey> for ReaderKeyRef<'_> {
+    fn equivalent(&self, key: &ReaderKey) -> bool {
+        self.name == key.name
+            && self.path == key.path.as_path()
+            && self.series == key.series
+            && self.position == key.position
+    }
+}
+
+impl ReaderKeyRef<'_> {
+    pub(crate) fn to_owned(&self) -> ReaderKey {
+        ReaderKey {
+            name: self.name.to_string(),
+            path: self.path.to_path_buf(),
+            series: self.series,
+            position: self.position,
+        }
+    }
+}
+
+pub(crate) type FrameKey = (ReaderKey, usize, usize, usize);
+
+/// borrowed view of [`FrameKey`] for cache lookups without allocation.
+#[derive(Debug, PartialEq, Eq)]
+pub(crate) struct FrameKeyRef<'a> {
+    pub(crate) reader: ReaderKeyRef<'a>,
+    pub(crate) c: usize,
+    pub(crate) z: usize,
+    pub(crate) t: usize,
+}
+
+impl Hash for FrameKeyRef<'_> {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.reader.hash(state);
+        self.c.hash(state);
+        self.z.hash(state);
+        self.t.hash(state);
+    }
+}
+
+impl Equivalent<FrameKey> for FrameKeyRef<'_> {
+    fn equivalent(&self, key: &FrameKey) -> bool {
+        let (rk, c, z, t) = key;
+        self.reader.equivalent(rk) && self.c == *c && self.z == *z && self.t == *t
+    }
+}
+
+impl FrameKeyRef<'_> {
+    pub(crate) fn to_owned(&self) -> FrameKey {
+        (self.reader.to_owned(), self.c, self.z, self.t)
+    }
+}
+
+/// identity of a materialized array in the process-wide cache
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+pub(crate) struct ArrayKey {
+    pub(crate) reader: ReaderKey,
+    pub(crate) dtype: &'static str,
+    pub(crate) slice: Vec<SliceInfoElem>,
+    pub(crate) axes: Vec<Axis>,
+    pub(crate) operations: Vec<(Axis, Operation)>,
+}
+
+/// borrowed view of [`ArrayKey`] for cache lookups without allocation.
+#[derive(Debug, PartialEq, Eq)]
+pub(crate) struct ArrayKeyRef<'a> {
+    pub(crate) reader: ReaderKeyRef<'a>,
+    pub(crate) dtype: &'static str,
+    pub(crate) slice: &'a [SliceInfoElem],
+    pub(crate) axes: &'a [Axis],
+    pub(crate) operations: &'a IndexMap<Axis, Operation>,
+}
+
+impl Hash for ArrayKeyRef<'_> {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.reader.hash(state);
+        self.dtype.hash(state);
+        self.slice.hash(state);
+        self.axes.hash(state);
+        self.operations.len().hash(state);
+        for (ax, op) in self.operations.iter() {
+            ax.hash(state);
+            op.hash(state);
+        }
+    }
+}
+
+impl Equivalent<ArrayKey> for ArrayKeyRef<'_> {
+    fn equivalent(&self, key: &ArrayKey) -> bool {
+        self.reader.equivalent(&key.reader)
+            && self.dtype == key.dtype
+            && self.slice == key.slice
+            && self.axes == key.axes
+            && self.operations.len() == key.operations.len()
+            && self
+                .operations
+                .iter()
+                .zip(&key.operations)
+                .all(|((ax, op), (key_ax, key_op))| ax == key_ax && op == key_op)
+    }
+}
+
+impl ArrayKeyRef<'_> {
+    pub(crate) fn to_owned(&self) -> ArrayKey {
+        ArrayKey {
+            reader: self.reader.to_owned(),
+            dtype: self.dtype,
+            slice: self.slice.to_vec(),
+            axes: self.axes.to_vec(),
+            operations: self
+                .operations
+                .iter()
+                .map(|(ax, op)| (*ax, op.clone()))
+                .collect(),
+        }
+    }
+}
+
+/// process-wide LRU cache of frames, shared between all views and threads
+static GLOBAL_FRAME_CACHE: OnceLock<FrameCache> = OnceLock::new();
+
+/// thread-safe LRU cache of frames read from the underlying reader
+pub(crate) struct FrameCache {
+    inner: Mutex<FrameCacheInner>,
+}
+
+struct FrameCacheInner {
+    map: IndexMap<FrameKey, Arc<Frame>>,
+    capacity: usize,
+}
+
+impl Default for FrameCache {
+    fn default() -> Self {
+        Self::new(DEFAULT_FRAME_CACHE_SIZE)
+    }
+}
+
+impl FrameCache {
+    pub(crate) fn new(capacity: usize) -> Self {
+        Self {
+            inner: Mutex::new(FrameCacheInner {
+                map: IndexMap::with_capacity(capacity),
+                capacity,
+            }),
+        }
+    }
+
+    pub(crate) fn global() -> &'static FrameCache {
+        GLOBAL_FRAME_CACHE.get_or_init(FrameCache::default)
+    }
+
+    pub(crate) fn capacity(&self) -> usize {
+        self.inner.lock().unwrap().capacity
+    }
+
+    pub(crate) fn set_capacity(&self, capacity: usize) {
+        let mut inner = self.inner.lock().unwrap();
+        inner.capacity = capacity;
+        while inner.map.len() > inner.capacity {
+            inner.map.shift_remove_index(0);
+        }
+    }
+
+    pub(crate) fn get<Q>(&self, key: &Q) -> Option<Arc<Frame>>
+    where
+        Q: ?Sized + Hash + Equivalent<FrameKey>,
+    {
+        let mut inner = self.inner.lock().unwrap();
+        if let Some(idx) = inner.map.get_index_of(key) {
+            let (key, frame) = inner.map.shift_remove_index(idx).unwrap();
+            inner.map.insert(key, frame.clone());
+            Some(frame)
+        } else {
+            None
+        }
+    }
+
+    pub(crate) fn insert(&self, key: FrameKey, frame: Arc<Frame>) {
+        let mut inner = self.inner.lock().unwrap();
+        inner.map.insert(key, frame);
+        while inner.map.len() > inner.capacity {
+            inner.map.shift_remove_index(0);
+        }
+    }
+
+    #[cfg(test)]
+    pub(crate) fn len(&self) -> usize {
+        self.inner.lock().unwrap().map.len()
+    }
+}
+
+impl Debug for FrameCache {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("FrameCache").finish_non_exhaustive()
+    }
+}
+
+/// process-wide LRU cache of materialized arrays, shared between all views and threads
+static GLOBAL_ARRAY_CACHE: OnceLock<ArrayCache> = OnceLock::new();
+
+/// thread-safe LRU cache of materialized arrays produced by `as_array_dyn`
+pub(crate) struct ArrayCache {
+    inner: Mutex<ArrayCacheInner>,
+}
+
+struct ArrayCacheInner {
+    map: IndexMap<ArrayKey, Arc<dyn Any + Send + Sync>>,
+    capacity: usize,
+}
+
+impl Default for ArrayCache {
+    fn default() -> Self {
+        Self::new(DEFAULT_ARRAY_CACHE_SIZE)
+    }
+}
+
+impl ArrayCache {
+    pub(crate) fn new(capacity: usize) -> Self {
+        Self {
+            inner: Mutex::new(ArrayCacheInner {
+                map: IndexMap::with_capacity(capacity),
+                capacity,
+            }),
+        }
+    }
+
+    pub(crate) fn global() -> &'static ArrayCache {
+        GLOBAL_ARRAY_CACHE.get_or_init(ArrayCache::default)
+    }
+
+    pub(crate) fn capacity(&self) -> usize {
+        self.inner.lock().unwrap().capacity
+    }
+
+    pub(crate) fn set_capacity(&self, capacity: usize) {
+        let mut inner = self.inner.lock().unwrap();
+        inner.capacity = capacity;
+        while inner.map.len() > inner.capacity {
+            inner.map.shift_remove_index(0);
+        }
+    }
+
+    pub(crate) fn get<T, Q>(&self, key: &Q) -> Option<Arc<ArrayD<T>>>
+    where
+        Q: ?Sized + Hash + Equivalent<ArrayKey>,
+        T: Any + Send + Sync,
+    {
+        let mut inner = self.inner.lock().unwrap();
+        if let Some(idx) = inner.map.get_index_of(key) {
+            let (key, array) = inner.map.shift_remove_index(idx).unwrap();
+            inner.map.insert(key, array.clone());
+            array.downcast::<ArrayD<T>>().ok()
+        } else {
+            None
+        }
+    }
+
+    pub(crate) fn insert<T: Any + Send + Sync>(&self, key: ArrayKey, array: ArrayD<T>) {
+        let mut inner = self.inner.lock().unwrap();
+        inner.map.insert(key, Arc::new(array));
+        while inner.map.len() > inner.capacity {
+            inner.map.shift_remove_index(0);
+        }
+    }
+
+    #[cfg(test)]
+    pub(crate) fn len(&self) -> usize {
+        self.inner.lock().unwrap().map.len()
+    }
+}
+
+impl Debug for ArrayCache {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("ArrayCache").finish_non_exhaustive()
+    }
+}