Skip to main content

omgbase_surface/
planner.rs

1//! The tier-3 pushdown planner: reduces the row set a query scans by
2//! translating its pushable top-level `where` conjuncts into ONE SQL
3//! statement against the store, then handing the produced rows (plus the
4//! untranslatable residual) back to the in-memory engine to finish. Port of
5//! `packages/core/src/oqx-js/planner.ts` over the [`oqx`] seam
6//! ([`QueryPlanner`], [`Plan`], [`oqx::partition_pushable`],
7//! [`oqx::residual_query`], [`oqx::ROWS_ROOT`]).
8//!
9//! Correctness is guaranteed by the residual fallback — anything
10//! [`crate::translate`] declines stays in-memory — and verified by the
11//! differential gate (planned == in-memory over every corpus-backed query
12//! case and the conformance list). Deliberately conservative: only simple
13//! top-level target scans (no `follow`, no `from E` re-projection) whose
14//! source is a bare `docs|blocks|nodes|links` or `<target>`, with at
15//! least one pushable scalar conjunct, are planned; everything else declines
16//! to a full in-memory run.
17//!
18//! One shape difference from the reference, forced by ownership: the
19//! reference attaches a store context to its plan (`makeStoreContext(…,
20//! rowsRoot)`), but a [`crate::StoreContext`] borrows the store's connection
21//! and `Plan::context` must be `'static`, so the plan carries no context and
22//! the runner ([`mod@crate::query`]) builds the rows-root context itself
23//! ([`crate::StoreContext::with_rows_root`]) — the same six lines
24//! `oqx::PlannedEngine::run` would execute.
25
26use oqx::Consumer;
27use oqx::ast::{Expr, Follow, FollowDestination, OpNode, Query, SelectItem, Subquery, Where};
28use oqx::{Plan, QueryPlanner, Value, partition_pushable, residual_query};
29use rusqlite::Connection;
30use rusqlite::types::Value as SqlValue;
31
32use crate::context::{REVISION_COLS, REVISION_FROM, Target, fetch_rows, tag_rows};
33use crate::translate::{TranslateCtx, translate_predicate};
34
35/// The SQL aliases of the scanned row (`self`) and its owning doc (`doc`).
36fn aliases(t: Target) -> (&'static str, &'static str) {
37    match t {
38        Target::Docs => ("d", "d"),
39        Target::Blocks => ("b", "d"),
40        Target::Nodes => ("n", "d"),
41        Target::Links => ("e", "d"),
42        Target::Revisions => ("r", "d"),
43    }
44}
45
46/// The scan's `FROM`. The `revisions` arms here and below (4.7) are the
47/// context's own scan spelled as clauses, so a caller keyed by target
48/// (`names::present`, a probe) can speak of the kind; the planner itself
49/// never plans a revisions scan ([`compile`]).
50pub(crate) fn from_clause(t: Target) -> &'static str {
51    match t {
52        Target::Docs => "docs d",
53        Target::Blocks => "blocks b JOIN docs d ON d.doc_id = b.doc_id",
54        Target::Nodes => "nodes n JOIN docs d ON d.doc_id = n.doc_id",
55        Target::Links => "edges e JOIN docs d ON d.doc_id = e.src_doc",
56        Target::Revisions => REVISION_FROM,
57    }
58}
59
60/// Row columns + the owning-doc path as `__path` (matches the context's
61/// roots so produced rows are indistinguishable from a full scan's).
62pub(crate) fn columns(t: Target) -> &'static str {
63    match t {
64        Target::Docs => "d.*",
65        Target::Blocks => "b.*, d.path AS __path",
66        Target::Nodes => "n.*, d.path AS __path",
67        Target::Links => "e.*, d.path AS __path",
68        Target::Revisions => REVISION_COLS,
69    }
70}
71
72/// The root order (`spec/surface` §1.1).
73pub(crate) fn order_clause(t: Target) -> &'static str {
74    match t {
75        Target::Docs => "d.path, d.doc_id",
76        Target::Blocks => "d.path, b.block_id",
77        Target::Nodes => "d.path, n.node_id",
78        Target::Links => "d.path, e.edge_id",
79        Target::Revisions => "r.path, r.rev_id",
80    }
81}
82
83/// The liveness guards of the root scan, with the repo id bound. The history
84/// has no liveness: every revision of every document of the repository,
85/// tombstoned ones included (§1.1, 4.7) — the repository guard through the
86/// owning `docs` row, since `revisions` carries no `repo_id`.
87pub(crate) fn guards(t: Target) -> &'static str {
88    match t {
89        Target::Docs => "d.repo_id = ? AND d.deleted_commit IS NULL",
90        Target::Blocks => "b.repo_id = ? AND b.deleted_commit IS NULL AND d.deleted_commit IS NULL",
91        Target::Nodes => "n.repo_id = ? AND d.deleted_commit IS NULL",
92        Target::Links => "e.repo_id = ? AND e.to_commit IS NULL AND d.deleted_commit IS NULL",
93        Target::Revisions => "d.repo_id = ?",
94    }
95}
96
97/// The joined targets driven FROM the document: the same rows and columns as
98/// [`from_clause`], with the loop order fixed by `CROSS JOIN` so a predicate on
99/// the document (`d.path = ?`) is the outer search and the target's rows are
100/// reached through their `doc_id` / `src_doc` index (the store indexes'
101/// `path` probes). [`guards_by_doc`] goes with it; `Docs` is `from_clause`.
102pub(crate) fn from_by_doc(t: Target) -> &'static str {
103    match t {
104        Target::Docs | Target::Revisions => from_clause(t),
105        Target::Blocks => "docs d CROSS JOIN blocks b ON b.doc_id = d.doc_id",
106        Target::Nodes => "docs d CROSS JOIN nodes n ON n.doc_id = d.doc_id",
107        Target::Links => "docs d CROSS JOIN edges e ON e.src_doc = d.doc_id",
108    }
109}
110
111/// [`guards`] for a [`from_by_doc`] statement: the same tests, with the
112/// document's repo first and the target's repo term behind SQLite's unary `+`
113/// so it stays a filter and never selects a `(repo_id, …)` index over the
114/// `doc_id` one. Two repo-id binds (`Docs`: one, as `guards`).
115pub(crate) fn guards_by_doc(t: Target) -> &'static str {
116    match t {
117        Target::Docs | Target::Revisions => guards(t),
118        Target::Blocks => {
119            "d.repo_id = ? AND +b.repo_id = ? AND b.deleted_commit IS NULL AND d.deleted_commit IS NULL"
120        }
121        Target::Nodes => "d.repo_id = ? AND +n.repo_id = ? AND d.deleted_commit IS NULL",
122        Target::Links => {
123            "d.repo_id = ? AND +e.repo_id = ? AND e.to_commit IS NULL AND d.deleted_commit IS NULL"
124        }
125    }
126}
127
128/// The root collection a query scans, if it is a bare
129/// `docs|blocks|nodes|links|revisions` source at the root scope (else `None` —
130/// a nested `^docs` is a block's receiver, never the query's source). The
131/// runner asks too: a bare root scan's rows are store rows by construction.
132/// Five-wide (the reference's `ROOTS` test); [`compile`] plans four — a
133/// `revisions` scan (4.7) always runs in memory.
134pub(crate) fn root_target(source: &Expr) -> Option<Target> {
135    match source {
136        Expr::Ident { name, .. } => Target::parse(name),
137        _ => None,
138    }
139}
140
141/// Whether `name` is a relation of `target`'s rows that is spelled like a
142/// target (§1.2: `docs.nodes`, `docs.blocks`, `docs.revisions` (4.7),
143/// `blocks.nodes`, `section.blocks`).
144fn is_target_relation_of(target: Target, name: &str) -> bool {
145    matches!(
146        (target, name),
147        (Target::Docs, "nodes" | "blocks" | "revisions")
148            | (Target::Blocks, "nodes")
149            | (Target::Nodes, "blocks")
150    )
151}
152
153/// The row a reach-through head reaches (`doc.x`, `block.x`, `section.x`,
154/// and a link's `dst.x`).
155fn reach_through(head: &str) -> Option<Target> {
156    match head {
157        "doc" | "dst" => Some(Target::Docs),
158        "block" => Some(Target::Blocks),
159        "section" => Some(Target::Nodes),
160        _ => None,
161    }
162}
163
164// ---- the residual-error decline (surface 1.1 patch, §1) ---------------------------------
165
166/// Whether a residual `where` could raise an OQX eval error the pushed
167/// conjuncts might hide by emptying the scan. In memory every conjunct of the
168/// original `&&` is evaluated for the first row and a throwing one aborts the
169/// run; planned, a pushed conjunct that matches no row means the residual
170/// never runs and the error vanishes. So a residual containing any function
171/// or method call (an unknown function, a bad regex, a wrong arity…), a
172/// nested block with the `single` consumer (more than one row raises), a
173/// `^`-escaped name (no enclosing scope at the top level) sends the whole
174/// query to the in-memory engine unplanned (before this batch of 3.0 a bare
175/// reserved docs basename did too; the attributes are bare now and never
176/// raise). Only comparisons, logical
177/// operators, `in`, `!`, literals, bindings and plain reads keep the push.
178///
179/// The walk is generic over the whole `oqx` AST — every `Where` node, every
180/// nested block (its `from`, `where`, `select`, `order by`, `follow`,
181/// `limit` / `offset`) and every `Expr`. A `^name:` lift in a nested select
182/// counts as a `^`-escaped name. Same decisions as the reference's
183/// `residualMayRaise`.
184fn residual_may_raise(w: &Where, target: Target) -> bool {
185    where_may_raise(w, target, true)
186}
187
188fn where_may_raise(w: &Where, target: Target, root: bool) -> bool {
189    match w {
190        Where::And { parts, .. } | Where::Or { parts, .. } => {
191            parts.iter().any(|p| where_may_raise(p, target, root))
192        }
193        Where::Not { expr, .. } => where_may_raise(expr, target, root),
194        Where::Scalar { expr, .. } => expr_may_raise(expr, target, root),
195        Where::Op(op) => op_may_raise(op, target, root),
196    }
197}
198
199fn op_may_raise(op: &OpNode, target: Target, root: bool) -> bool {
200    op.op == Consumer::Single
201        || expr_may_raise(&op.receiver, target, root)
202        || subquery_may_raise(&op.sub, target)
203}
204
205fn subquery_may_raise(sub: &Subquery, target: Target) -> bool {
206    let inner = |e: &Expr| expr_may_raise(e, target, false);
207    sub.from.iter().any(inner)
208        || sub
209            .r#where
210            .as_ref()
211            .is_some_and(|w| where_may_raise(w, target, false))
212        || sub.select.iter().any(|item| match item {
213            SelectItem::Field { expr, lift, .. } => *lift > 0 || inner(expr),
214            SelectItem::Collect { op, .. } => op_may_raise(op, target, false),
215        })
216        || sub.order_by.iter().flatten().any(|o| inner(&o.expr))
217        || sub
218            .follow
219            .as_ref()
220            .is_some_and(|f| follow_may_raise(f, target))
221        || sub.frontier.as_ref().is_some_and(inner)
222        // A flattened block (oqx 0.25) evaluates each `flatten` expression per row.
223        || sub.flatten.iter().any(|f| inner(&f.expr))
224        // A grouped block (oqx 0.24) evaluates its keys, `having` and its
225        // aggregates (`sum` over a string raises) per group.
226        || sub
227            .group_by
228            .as_ref()
229            .is_some_and(|g| g.keys.iter().any(|k| inner(&k.expr)))
230        || sub
231            .having
232            .as_ref()
233            .is_some_and(|h| where_may_raise(h, target, false))
234        || sub.limit.as_ref().is_some_and(inner)
235        || sub.offset.as_ref().is_some_and(inner)
236}
237
238// A destination is a relation read in the frontier row's scope, or a destination
239// block (whose receiver, body, `frontier` and `depth` are its own — oqx 0.20).
240fn follow_may_raise(f: &Follow, target: Target) -> bool {
241    let inner = |e: &Expr| expr_may_raise(e, target, false);
242    f.destinations.iter().any(|d| match d {
243        FollowDestination::Relation(e) => inner(e),
244        FollowDestination::Block(op) => op_may_raise(op, target, false),
245    })
246}
247
248fn expr_may_raise(e: &Expr, target: Target, root: bool) -> bool {
249    let again = |e: &Expr| expr_may_raise(e, target, root);
250    match e {
251        Expr::Lit { .. } | Expr::Binding { .. } => false,
252        // A bare target name read as a property raises (the root-row rule,
253        // `crate::query`) unless it is a relation of the rows at hand: known at
254        // the root scope, unknown below.
255        Expr::Ident { name, .. } if Target::parse(name).is_some() => {
256            !root || !is_target_relation_of(target, name)
257        }
258        // A bare attribute never raises (3.0: there is no loud-error set).
259        Expr::Ident { .. } => false,
260        Expr::Outer { .. } | Expr::Call { .. } => true,
261        // `<head>.<target>` raises unless the head is a reach-through whose row
262        // has that relation.
263        Expr::Member { recv, name, .. } => {
264            (Target::parse(name).is_some()
265                && !matches!(&**recv, Expr::Ident { name: head, .. }
266                        if reach_through(head).is_some_and(|t| is_target_relation_of(t, name))))
267                || again(recv)
268        }
269        Expr::Unary { expr, .. } => again(expr),
270        // `x!` raises when absent; a value-position directive may `single`-fail
271        // or read anything its block reads.
272        Expr::Required { .. } => true,
273        Expr::Op(op) => op_may_raise(op, target, root),
274        // A regex operator (oqx 0.22) is a `matches` call: an invalid pattern
275        // raises, exactly as the call would.
276        Expr::Binary {
277            op, left, right, ..
278        } => op.is_regex() || again(left) || again(right),
279        Expr::Logical { left, right, .. } | Expr::In { left, right, .. } => {
280            again(left) || again(right)
281        }
282        Expr::Range { lo, hi, .. } => {
283            lo.as_deref().is_some_and(again) || hi.as_deref().is_some_and(again)
284        }
285        // A tuple (oqx 0.20) is never pushed down; it raises iff an element does.
286        Expr::Tuple { items, .. } => items.iter().any(again),
287        // A template literal (oqx 0.21) is never pushed down; interpolating a
288        // list or a record is an eval error only the data decides, so with any
289        // interpolation it may raise.
290        Expr::Template { parts, .. } => oqx::walk::template_exprs(parts).next().is_some(),
291        // An aggregate (oqx 0.24) raises on a non-number under `sum`/`avg`,
292        // which only the data decides; a grouped query is never planned anyway
293        // (`compile`).
294        Expr::Aggregate { .. } => true,
295    }
296}
297
298/// A compiled plan before execution: the statement, its params and the
299/// residual query. Pure — what the planner would run, for inspection.
300#[derive(Clone, Debug, PartialEq)]
301pub struct Compiled {
302    pub target: Target,
303    pub sql: String,
304    pub params: Vec<SqlValue>,
305    pub residual: Query,
306}
307
308/// Compile `query` against `repo_id`, or `None` when the shape declines: a
309/// `follow`, a `from E` re-projection, a `group by` (oqx 0.24) or a `flatten`
310/// (oqx 0.25) — a grouped or flattened query is declined whole: its `where`
311/// conjuncts could be pushed (the fan-out and the grouping happen after them,
312/// so the residual would flatten and group the produced rows correctly), but
313/// a pushed conjunct that empties the scan would hide an eval error a
314/// `flatten` expression, an aggregate or `having` raises in memory, so decline
315/// (c) extends to the post-where clauses; SQL `GROUP BY` / `json_each`
316/// pushdown is a later round — a source that is not a root scan, or no
317/// pushable conjunct at all (the engine then does everything).
318#[must_use]
319pub fn compile(query: &Query, params: &[Value], repo_id: &str) -> Option<Compiled> {
320    if query.follow.is_some()
321        || !query.from.is_empty()
322        || query.group_by.is_some()
323        || !query.flatten.is_empty()
324    {
325        return None;
326    }
327    let target = root_target(&query.source)?;
328    // The history (4.7) is never planned: no store index, no pushdown — the
329    // scan runs in memory, as the reference's four-wide `rootTarget` decides.
330    if target == Target::Revisions {
331        return None;
332    }
333    let (self_alias, doc_alias) = aliases(target);
334    let ctx = TranslateCtx {
335        target,
336        self_alias,
337        doc_alias,
338        params,
339    };
340    let (pushed, residual) = partition_pushable(query.r#where.as_ref(), |e| {
341        translate_predicate(e, &ctx).is_some()
342    });
343    if pushed.is_empty() {
344        return None;
345    }
346    // Decline (c): a residual that could raise must not be hidden behind an
347    // emptied scan — the whole query runs in memory.
348    if residual
349        .as_ref()
350        .is_some_and(|w| residual_may_raise(w, target))
351    {
352        return None;
353    }
354    let mut where_sql = guards(target).to_owned();
355    let mut sql_params = vec![SqlValue::Text(repo_id.to_owned())];
356    for e in &pushed {
357        let frag = translate_predicate(e, &ctx).expect("accepted by partition_pushable");
358        where_sql.push_str(" AND (");
359        where_sql.push_str(&frag.sql);
360        where_sql.push(')');
361        sql_params.extend(frag.params);
362    }
363    let sql = format!(
364        "SELECT {} FROM {} WHERE {where_sql} ORDER BY {}",
365        columns(target),
366        from_clause(target),
367        order_clause(target)
368    );
369    Some(Compiled {
370        target,
371        sql,
372        params: sql_params,
373        residual: residual_query(query, residual),
374    })
375}
376
377/// The SQLite planner over one repo of the store.
378pub struct SqlitePlanner<'a> {
379    conn: &'a Connection,
380    repo_id: String,
381}
382
383impl<'a> SqlitePlanner<'a> {
384    #[must_use]
385    pub fn new(conn: &'a Connection, repo_id: &str) -> Self {
386        Self {
387            conn,
388            repo_id: repo_id.to_owned(),
389        }
390    }
391
392    /// Plan `query`: `Ok(None)` when the shape declines, `Ok(Some(plan))`
393    /// with the produced rows (tagged like the context's root rows) and the
394    /// residual, `Err` when the statement itself failed (a store error, not
395    /// a decline — the runner reports it rather than silently rescanning).
396    pub fn try_plan(&self, query: &Query, params: &[Value]) -> rusqlite::Result<Option<Plan>> {
397        let Some(compiled) = compile(query, params, &self.repo_id) else {
398            return Ok(None);
399        };
400        let rows = fetch_rows(self.conn, &compiled.sql, &compiled.params)?;
401        Ok(Some(Plan::new(
402            tag_rows(rows, compiled.target),
403            compiled.residual,
404        )))
405    }
406}
407
408impl QueryPlanner for SqlitePlanner<'_> {
409    /// The seam's shape: a failed statement declines (the engine falls back
410    /// to a full scan). The runner uses [`Self::try_plan`] to surface it.
411    fn plan(&self, query: &Query, params: &[Value]) -> Option<Plan> {
412        self.try_plan(query, params).ok().flatten()
413    }
414}
415
416#[cfg(test)]
417mod tests {
418    use super::*;
419    use oqx::ROWS_ROOT;
420    use oqx::ast::Where;
421
422    fn parse(src: &str) -> Query {
423        oqx::parse_string(src).expect("parses")
424    }
425
426    fn text(s: &str) -> SqlValue {
427        SqlValue::Text(s.to_owned())
428    }
429
430    #[test]
431    fn a_pushable_scan_compiles_to_one_statement_in_root_order() {
432        let c =
433            compile(&parse("from docs where path == \"index.md\""), &[], "r_1").expect("planned");
434        assert_eq!(c.target, Target::Docs);
435        assert_eq!(
436            c.sql,
437            "SELECT d.* FROM docs d WHERE d.repo_id = ? AND d.deleted_commit IS NULL AND ((('/' || d.path) IS ?)) ORDER BY d.path, d.doc_id"
438        );
439        assert_eq!(c.params, vec![text("r_1"), text("index.md")]);
440        assert_eq!(
441            c.residual.source,
442            Expr::Ident {
443                name: ROWS_ROOT.to_owned(),
444                span: oqx::Span::EMPTY,
445            }
446        );
447        assert_eq!(c.residual.r#where, None);
448    }
449
450    #[test]
451    fn every_target_has_its_join_columns_guards_and_order() {
452        let b = compile(
453            &parse("from blocks where path.startsWith(\"lab/\")"),
454            &[],
455            "r",
456        )
457        .unwrap();
458        assert_eq!(
459            b.sql,
460            "SELECT b.*, d.path AS __path FROM blocks b JOIN docs d ON d.doc_id = b.doc_id \
461             WHERE b.repo_id = ? AND b.deleted_commit IS NULL AND d.deleted_commit IS NULL \
462             AND ((substr(('/' || d.path), 1, length(?)) = ?)) ORDER BY d.path, b.block_id"
463        );
464        assert_eq!(b.params, vec![text("r"), text("lab/"), text("lab/")]);
465        let n = compile(
466            &parse("nodes count { where kind == \"md:task\" }"),
467            &[],
468            "r",
469        )
470        .unwrap();
471        assert_eq!(n.target, Target::Nodes);
472        assert!(n.sql.starts_with(
473            "SELECT n.*, d.path AS __path FROM nodes n JOIN docs d ON d.doc_id = n.doc_id WHERE n.repo_id = ? AND d.deleted_commit IS NULL AND ((n.kind IS ?))"
474        ));
475        assert!(n.sql.ends_with("ORDER BY d.path, n.node_id"));
476        let e = compile(
477            &parse("from links where predicate == \"references\""),
478            &[],
479            "r",
480        )
481        .unwrap();
482        assert!(e.sql.starts_with(
483            "SELECT e.*, d.path AS __path FROM edges e JOIN docs d ON d.doc_id = e.src_doc WHERE e.repo_id = ? AND e.to_commit IS NULL AND d.deleted_commit IS NULL AND ((e.predicate IS ?))"
484        ));
485        assert!(e.sql.ends_with("ORDER BY d.path, e.edge_id"));
486    }
487
488    #[test]
489    fn mixed_conjunctions_push_the_translatable_parts_and_keep_the_rest() {
490        let q = parse(
491            "from docs where path.startsWith(\"processes/\") && nodes exists { where kind == \"md:task\" } && layer == \"canon\"",
492        );
493        let c = compile(&q, &[], "r").expect("planned");
494        assert!(
495            c.sql
496                .contains("(substr(('/' || d.path), 1, length(?)) = ?)"),
497            "{}",
498            c.sql
499        );
500        assert!(c.sql.contains("p.key = 'layer'"), "{}", c.sql);
501        // Guards first, then the fragments' params in statement order.
502        assert_eq!(
503            c.params,
504            vec![
505                text("r"),
506                text("processes/"),
507                text("processes/"),
508                text("canon")
509            ]
510        );
511        // One conjunct left → it stands alone, not wrapped in an `and`.
512        assert!(
513            matches!(c.residual.r#where, Some(Where::Op(_))),
514            "{:?}",
515            c.residual.r#where
516        );
517        assert!(c.residual.from.is_empty());
518        assert_eq!(c.residual.select, q.select);
519        assert_eq!(c.residual.consumer, q.consumer);
520    }
521
522    #[test]
523    fn declined_shapes_return_none() {
524        // nothing pushable → let the engine do it all
525        assert!(compile(&parse("from docs"), &[], "r").is_none());
526        assert!(compile(&parse("from docs where era in 800..1680"), &[], "r").is_none());
527        assert!(compile(&parse("from docs where !verified"), &[], "r").is_none());
528        assert!(
529            compile(
530                &parse("from docs where path == \"a\" || path == \"b\""),
531                &[],
532                "r"
533            )
534            .is_none()
535        );
536        assert!(
537            compile(
538                &parse("from docs where nodes exists { where kind == \"md:task\" }"),
539                &[],
540                "r"
541            )
542            .is_none()
543        );
544        // follow
545        assert!(
546            compile(
547                &parse("from docs where path == \"a.md\" follow distinct out_links collect { from dst }"),
548                &[],
549                "r"
550            )
551            .is_none()
552        );
553        // flatten (oqx 0.25): declined whole, like `group by`
554        assert!(
555            compile(
556                &parse("select path, t from docs where layer == \"canon\" flatten tags as t"),
557                &[],
558                "r"
559            )
560            .is_none()
561        );
562        assert!(
563            compile(
564                &parse("from docs where layer == \"canon\" flatten blocks { where type == \"task\" } as b"),
565                &[],
566                "r"
567            )
568            .is_none()
569        );
570        // follow with several destinations: relations and a destination block
571        // decline exactly like a single relation (0.14)
572        assert!(
573            compile(
574                &parse("from docs where path == \"a.md\" follow out_links collect { from dst }, in_links collect { from doc }"),
575                &[],
576                "r"
577            )
578            .is_none()
579        );
580        assert!(
581            compile(
582                &parse(
583                    "from docs where path == \"a.md\" follow ^docs collect { where after.contains(^path) }"
584                ),
585                &[],
586                "r"
587            )
588            .is_none()
589        );
590        // a body-level `frontier` (oqx 0.20) plans like the `follow` it needs
591        assert!(
592            compile(
593                &parse("from docs where path == \"a.md\" frontier type == \"hub\" follow distinct out_links collect { from dst frontier type == \"practitioner\" }"),
594                &[],
595                "r"
596            )
597            .is_none()
598        );
599        // a grouped query is declined whole (oqx 0.24), pushable `where` or not
600        assert!(
601            compile(
602                &parse("select layer, n: count() from docs where path == \"a.md\" group by layer"),
603                &[],
604                "r"
605            )
606            .is_none()
607        );
608        // not a root scan
609        assert!(compile(&parse("from things where path == \"a.md\""), &[], "r").is_none());
610        assert!(compile(&parse("from nope where path == \"a.md\""), &[], "r").is_none());
611        assert!(compile(&parse("from docs.nodes where kind == \"x\""), &[], "r").is_none());
612        // a `from E` re-projection (an AST-level shape)
613        let mut q = parse("from docs where path == \"a.md\"");
614        q.from.push(Expr::Ident {
615            name: "nodes".to_owned(),
616            span: oqx::Span::EMPTY,
617        });
618        assert!(compile(&q, &[], "r").is_none());
619    }
620
621    // -- decline (c): a residual that could raise sends the whole query in memory --
622
623    #[test]
624    fn a_residual_that_could_raise_declines_the_whole_query() {
625        let declined = |src: &str| {
626            assert!(
627                compile(&parse(src), &[], "r").is_none(),
628                "should decline: {src}"
629            );
630        };
631        let planned = |src: &str| {
632            assert!(
633                compile(&parse(src), &[], "r").is_some(),
634                "should plan: {src}"
635            );
636        };
637        // a bare attribute is never a raise (3.0: no loud-error set) — `path`
638        // pushes, `raw`/`updated_at`/`tags` are plain residuals
639        planned("from docs where path == \"x\" && path == \"nope.md\"");
640        planned("from docs where path == \"nope.md\" && !raw");
641        planned("from docs where path == \"nope.md\" && updated_at");
642        planned("from docs where path == \"nope.md\" && tags");
643        // `doc.<attribute>` on any target, at any depth, is a plain read too
644        planned("from blocks where path == \"x\" && doc.path == \"y\"");
645        planned("from blocks where path == \"x\" && nodes exists { where doc.path == \"y\" }");
646        // a function or method call
647        declined("from docs where path.matches(\"[\") && path == \"nope.md\"");
648        declined("from docs where nope(\"x\") && path == \"nope.md\"");
649        declined("from docs where path == \"x\" && size(tags) > 1");
650        declined("from docs where path == \"x\" && nodes exists { where name.lower() == \"a\" }");
651        declined(
652            "from docs where path == \"x\" && nodes count { where kind == \"a\" order by size(name) } > 1",
653        );
654        // a `^`-escaped name: an outer reference or a lift
655        declined("from docs where path == \"x\" && ^slug == \"y\"");
656        declined("from docs where path == \"x\" && nodes exists { where name == ^title }");
657        declined("from docs where path == \"x\" && nodes collect { ^first_task: name }");
658        // a nested block with the `single` consumer
659        declined(
660            "from docs where path == \"x\" && blocks exists { select t: nodes single { where kind == \"md:task\" } }",
661        );
662        // the pushed side raising is impossible; only the residual matters
663        planned("from docs where path.startsWith(\"lab/\") && path == \"x\"");
664        // plain reads, comparisons, `in`, `!`, literals and bindings keep the push
665        planned("from docs where path == \"x\" && era in 800..1680");
666        planned("from docs where path == \"x\" && !verified");
667        planned("from docs where path == \"x\" && (layer == \"a\" || layer == \"b\")");
668        planned("from docs where path == \"x\" && verified == true");
669        planned("from docs where path == \"x\" && nodes exists { where kind == \"md:task\" }");
670        planned(
671            "from docs where path == \"x\" && nodes count { where kind == \"md:task\" limit 5 } > 1",
672        );
673        // on blocks a bare `path` is an attribute, and inside a block the rows
674        // are another scope — an ordinary read, not the guard
675        planned("from blocks where path == \"x\" && !path");
676        planned("from docs where path == \"x\" && nodes exists { where path == \"y\" }");
677        planned("from docs where path == \"x\" && frontmatter.path == \"y\"");
678        // a key read through a source handle is pushed (3.0), scoped to its
679        // source; a WHOLE handle reference and `entries(…)` stay residual (and
680        // the function call declines the query)
681        let fm = compile(
682            &parse("from docs where frontmatter.layer == \"canon\""),
683            &[],
684            "r",
685        )
686        .expect("planned");
687        assert!(fm.sql.contains("p.source = 'frontmatter'"), "{}", fm.sql);
688        let data =
689            compile(&parse("from docs where data.layer == \"canon\""), &[], "r").expect("planned");
690        assert!(!data.sql.contains("p.source"), "{}", data.sql);
691        planned("from blocks where doc.inline.element == \"fire\"");
692        planned("from docs where path == \"x\" && frontmatter == null");
693        declined("from docs where path == \"x\" && entries(frontmatter) exists { }");
694    }
695
696    #[test]
697    fn a_top_level_not_or_or_is_the_whole_residual_and_declines() {
698        // `!(…)` and `||` at the top are `Where::Not` / `Where::Or`, never scalar
699        // parts, so nothing is pushed even when their leaves would be.
700        assert!(compile(&parse("from docs where !(path == \"a\")"), &[], "r").is_none());
701        assert!(
702            compile(
703                &parse("from docs where (path == \"a\" || path == \"b\") && layer == \"canon\""),
704                &[],
705                "r"
706            )
707            .is_some()
708        );
709    }
710}