Skip to main content

kevy_embedded/
ops_view.rs

1//! Embedded views (server parity minus VIA/FIELDS hydration:
2//! in-process callers dereference and read fields directly).
3//!
4//! Mirrors the embedded index architecture: per-shard view states in
5//! `Inner` (maintained inside `commit_write` right after index
6//! maintenance, under the same shard lock), a store-level registry,
7//! synchronous builds, typed API (`Tree` passed directly — no text
8//! grammar in-process).
9
10use crate::{KevyError, KevyResult};
11use std::io;
12use std::sync::RwLock;
13
14use kevy_index::{
15    IndexValue, MaterializedSet, Tree, ViewCatalog, ViewMode, ViewSpec, eval_tree, key_in_tree,
16};
17
18use crate::ops_index::ShardSegs;
19use crate::store::{Store, lock_write};
20
21/// Store-level registry.
22#[derive(Default)]
23pub(crate) struct ViewReg {
24    pub(crate) catalog: RwLock<(u64, ViewCatalog)>,
25}
26
27/// One shard's view states (inside `Inner`, guarded by the shard lock).
28#[derive(Default)]
29pub(crate) struct ShardViews {
30    pub(crate) version: u64,
31    pub(crate) views: Vec<ViewState>,
32    /// `reserved_bytes` generation cache — see
33    /// `ShardSegs::stats_dirty`; same contract, view half. Tier-only,
34    /// like its twin.
35    #[cfg(all(feature = "tier", not(target_arch = "wasm32")))]
36    pub(crate) stats_dirty: bool,
37    #[cfg(all(feature = "tier", not(target_arch = "wasm32")))]
38    pub(crate) reserved_cache: u64,
39}
40
41pub(crate) struct ViewState {
42    spec: ViewSpec,
43    mat: Option<MaterializedSet>,
44    needs_rebuild: bool,
45}
46
47impl ShardViews {
48    /// Invalidate the cache — no-op without the tier backend; see
49    /// `ShardSegs::mark_stats_dirty`.
50    #[inline]
51    pub(crate) fn mark_stats_dirty(&mut self) {
52        #[cfg(all(feature = "tier", not(target_arch = "wasm32")))]
53        {
54            self.stats_dirty = true;
55        }
56    }
57
58    /// Σ approximate heap bytes of the materialized view sets — the
59    /// view half of the tier's `reserved_bytes` feed.
60    /// Virtual views hold no set and contribute nothing.
61    #[cfg(all(feature = "tier", not(target_arch = "wasm32")))]
62    pub(crate) fn reserved_bytes(&mut self) -> u64 {
63        if !self.stats_dirty {
64            return self.reserved_cache;
65        }
66        let sum = self
67            .views
68            .iter()
69            .map(|v| v.mat.as_ref().map_or(0, MaterializedSet::approx_bytes))
70            .sum();
71        self.reserved_cache = sum;
72        self.stats_dirty = false;
73        sum
74    }
75}
76
77/// One page of view members plus the resume cursor.
78pub type ViewPage = (Vec<(Vec<u8>, IndexValue)>, Option<(IndexValue, Vec<u8>)>);
79
80#[cfg(feature = "persist")]
81const SIDECAR: &str = "view-catalog.meta";
82
83impl Store {
84    /// Declare a view (typed tree; `via` is not supported embedded —
85    /// read fields in-process). Builds synchronously.
86    pub fn view_create(
87        &self,
88        name: &[u8],
89        tree: Tree,
90        order_by: &[u8],
91        desc: bool,
92        mode: ViewMode,
93    ) -> KevyResult<()> {
94        self.check_view_refs(&tree, order_by)?;
95        let spec = ViewSpec {
96            name: name.to_vec(),
97            tree,
98            order_by: order_by.to_vec(),
99            desc,
100            mode,
101            via: None,
102        };
103        {
104            let mut g =
105                self.views.catalog.write().unwrap_or_else(std::sync::PoisonError::into_inner);
106            let (ver, cat) = &mut *g;
107            cat.create(spec).map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))?;
108            *ver += 1;
109        }
110        self.persist_view_sidecar();
111        for shard in self.shards.iter() {
112            let mut g = lock_write(shard);
113            let inner = &mut *g;
114            crate::ops_index::sync_segs(&self.indexes, &mut inner.idx_segs, &mut inner.store);
115            sync_views(&self.views, &mut inner.view_segs, &inner.idx_segs);
116        }
117        Ok(())
118    }
119
120    /// Every index a view references (its leaves + ORDER BY) must
121    /// already be declared.
122    fn check_view_refs(&self, tree: &Tree, order_by: &[u8]) -> KevyResult<()> {
123        let mut names: Vec<Vec<u8>> = vec![order_by.to_vec()];
124        tree.each_leaf(&mut |l| names.push(l.index.clone()));
125        let g = self.indexes.catalog.read().unwrap_or_else(std::sync::PoisonError::into_inner);
126        for n in &names {
127            if g.1.get(n).is_none() {
128                return Err(KevyError::InvalidInput("view references unknown index".into()));
129            }
130        }
131        Ok(())
132    }
133
134    /// Drop a view; `false` if absent.
135    pub fn view_drop(&self, name: &[u8]) -> bool {
136        let hit = {
137            let mut g =
138                self.views.catalog.write().unwrap_or_else(std::sync::PoisonError::into_inner);
139            let (ver, cat) = &mut *g;
140            let hit = cat.drop_view(name);
141            if hit {
142                *ver += 1;
143            }
144            hit
145        };
146        if hit {
147            self.persist_view_sidecar();
148        }
149        hit
150    }
151
152    /// Ordered page across shards (`after` resumes exclusively; DESC
153    /// views page from the large end).
154    pub fn view_query(
155        &self,
156        name: &[u8],
157        after: Option<&(IndexValue, Vec<u8>)>,
158        limit: usize,
159    ) -> KevyResult<ViewPage> {
160        let limit = limit.clamp(1, 100_000);
161        let mut desc = false;
162        let mut all: Vec<(IndexValue, Vec<u8>)> = Vec::new();
163        let mut found = false;
164        for shard in self.shards.iter() {
165            let mut g = lock_write(shard);
166            let inner = &mut *g;
167            crate::ops_index::sync_segs(&self.indexes, &mut inner.idx_segs, &mut inner.store);
168            sync_views(&self.views, &mut inner.view_segs, &inner.idx_segs);
169            let Some(i) = inner.view_segs.views.iter().position(|v| v.spec.name == name) else {
170                continue;
171            };
172            found = true;
173            if inner.view_segs.views[i].needs_rebuild {
174                inner.view_segs.mark_stats_dirty();
175                rebuild(&mut inner.view_segs.views[i], &inner.idx_segs);
176            }
177            let vs = &inner.view_segs.views[i];
178            desc = vs.spec.desc;
179            match &vs.mat {
180                Some(m) => all.extend(m.page(after, limit, vs.spec.desc)),
181                None => stream_virtual(&vs.spec, &inner.idx_segs, after, limit, &mut all),
182            }
183        }
184        if !found {
185            return Err(KevyError::NotFound("no such view".into()));
186        }
187        all.sort();
188        if desc {
189            all.reverse();
190        }
191        all.truncate(limit);
192        let next = if all.len() == limit { all.last().cloned() } else { None };
193        Ok((all.into_iter().map(|(v, k)| (k, v)).collect(), next))
194    }
195
196    /// Declared views (name, mode, leaves).
197    pub fn view_list(&self) -> Vec<(Vec<u8>, ViewMode, usize)> {
198        let g = self.views.catalog.read().unwrap_or_else(std::sync::PoisonError::into_inner);
199        g.1.iter().map(|s| (s.name.clone(), s.mode, s.tree.leaves())).collect()
200    }
201
202    /// Summed member count across shards.
203    pub fn view_count(&self, name: &[u8]) -> KevyResult<u64> {
204        Ok(self.view_query(name, None, 100_000)?.0.len() as u64)
205    }
206
207    #[cfg(feature = "persist")]
208    fn persist_view_sidecar(&self) {
209        let Some(dir) = &self.config.data_dir else { return };
210        let g = self.views.catalog.read().unwrap_or_else(std::sync::PoisonError::into_inner);
211        let tmp = dir.join("view-catalog.meta.tmp");
212        if std::fs::write(&tmp, g.1.to_sidecar()).is_ok() {
213            let _ = std::fs::rename(&tmp, dir.join(SIDECAR));
214        }
215    }
216
217    /// Without `persist` there is no data dir — no sidecar to write
218    /// or load; both halves are no-ops.
219    #[cfg(not(feature = "persist"))]
220    fn persist_view_sidecar(&self) {}
221
222    #[cfg(not(feature = "persist"))]
223    pub(crate) fn view_boot(&self) {}
224
225    /// Boot half — load the persisted view catalog.
226    #[cfg(feature = "persist")]
227    pub(crate) fn view_boot(&self) {
228        let Some(dir) = &self.config.data_dir else { return };
229        if let Ok(text) = std::fs::read_to_string(dir.join(SIDECAR))
230            && let Some(cat) = ViewCatalog::from_sidecar(&text)
231            && !cat.is_empty()
232        {
233            let mut g =
234                self.views.catalog.write().unwrap_or_else(std::sync::PoisonError::into_inner);
235            *g = (g.0 + 1, cat);
236        }
237    }
238}
239
240fn resolver<'a>(segs: &'a ShardSegs) -> impl Fn(&[u8]) -> Option<&'a kevy_index::Segment> {
241    move |name: &[u8]| segs.segs.iter().find(|(s, _)| s.name == name).map(|(_, seg)| seg)
242}
243
244fn eval_shard(spec: &ViewSpec, segs: &ShardSegs) -> Vec<(IndexValue, Vec<u8>)> {
245    let r = resolver(segs);
246    let members = eval_tree(&spec.tree, &&r);
247    members
248        .into_iter()
249        .filter_map(|k| r(&spec.order_by).and_then(|s| s.verify_entry(&k)).map(|v| (v.clone(), k)))
250        .collect()
251}
252
253/// Virtual-mode page: order-driven streaming over the ORDER BY index,
254/// probing the tree per key (same clamp rationale as the server
255/// pager).
256fn stream_virtual(
257    spec: &ViewSpec,
258    segs: &ShardSegs,
259    after: Option<&(IndexValue, Vec<u8>)>,
260    limit: usize,
261    all: &mut Vec<(IndexValue, Vec<u8>)>,
262) {
263    let r = resolver(segs);
264    if let Some(order_seg) = r(&spec.order_by) {
265        let cursor = after.map(|(v, k)| kevy_index::Cursor { value: v.clone(), key: k.clone() });
266        let mut got = 0usize;
267        for (v, k) in order_seg.scan(cursor.as_ref(), spec.desc) {
268            if key_in_tree(&spec.tree, k, &&r) {
269                all.push((v.clone(), k.to_vec()));
270                got += 1;
271                if got == limit {
272                    break;
273                }
274            }
275        }
276    }
277}
278
279fn rebuild(vs: &mut ViewState, segs: &ShardSegs) {
280    let spec = vs.spec.clone();
281    let Some(mat) = &mut vs.mat else {
282        vs.needs_rebuild = false;
283        return;
284    };
285    mat.clear();
286    let mut rows = eval_shard(&spec, segs);
287    rows.sort();
288    if let ViewMode::Materialized { top_k } = spec.mode
289        && top_k > 0
290    {
291        if spec.desc {
292            rows.reverse();
293        }
294        rows.truncate((top_k + top_k / 4) as usize);
295    }
296    for (v, k) in rows {
297        mat.apply(&k, true, Some(v));
298    }
299    vs.needs_rebuild = false;
300}
301
302/// Reconcile with the catalog (under the shard lock).
303pub(crate) fn sync_views(reg: &ViewReg, sv: &mut ShardViews, segs: &ShardSegs) {
304    let g = reg.catalog.read().unwrap_or_else(std::sync::PoisonError::into_inner);
305    let (ver, cat) = &*g;
306    if sv.version == *ver {
307        return;
308    }
309    sv.mark_stats_dirty();
310    let mut next = Vec::new();
311    for spec in cat.iter() {
312        match sv.views.iter().position(|v| v.spec == *spec) {
313            Some(i) => next.push(sv.views.swap_remove(i)),
314            None => {
315                let mat = match spec.mode {
316                    ViewMode::Virtual => None,
317                    ViewMode::Materialized { top_k } => {
318                        Some(MaterializedSet::new(top_k, spec.desc))
319                    }
320                };
321                let mut vs = ViewState { spec: spec.clone(), needs_rebuild: mat.is_some(), mat };
322                if vs.needs_rebuild {
323                    rebuild(&mut vs, segs);
324                }
325                next.push(vs);
326            }
327        }
328    }
329    sv.views = next;
330    sv.version = *ver;
331}
332
333/// Write hook — call AFTER `ops_index::on_commit` (same shard lock).
334pub(crate) fn on_commit(reg: &ViewReg, sv: &mut ShardViews, segs: &ShardSegs, parts: &[&[u8]]) {
335    {
336        let g = reg.catalog.read().unwrap_or_else(std::sync::PoisonError::into_inner);
337        if g.1.is_empty() {
338            return;
339        }
340    }
341    sync_views(reg, sv, segs);
342    let verb = parts.first().copied().unwrap_or(b"");
343    if verb.eq_ignore_ascii_case(b"FLUSHALL") || verb.eq_ignore_ascii_case(b"FLUSHDB") {
344        for vs in &mut sv.views {
345            if let Some(m) = &mut vs.mat {
346                m.clear();
347            }
348        }
349        sv.mark_stats_dirty();
350        return;
351    }
352    // Same exact written-key walk as the index hook.
353    let mut touched = false;
354    let views = &mut sv.views;
355    crate::ops_index::each_written_key_pub(verb, parts, |key| {
356        for vs in &mut *views {
357            let Some(mat) = &mut vs.mat else { continue };
358            touched = true;
359            let r = resolver(segs);
360            let member = key_in_tree(&vs.spec.tree, key, &&r);
361            let order = r(&vs.spec.order_by).and_then(|s| s.verify_entry(key)).cloned();
362            if mat.apply(key, member, order) {
363                vs.needs_rebuild = true;
364            }
365        }
366    });
367    if touched {
368        sv.mark_stats_dirty();
369    }
370}