1use 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
35fn 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
46pub(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
60pub(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
72pub(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
83pub(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
97pub(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
111pub(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
128pub(crate) fn root_target(source: &Expr) -> Option<Target> {
135 match source {
136 Expr::Ident { name, .. } => Target::parse(name),
137 _ => None,
138 }
139}
140
141fn 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
153fn 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
164fn 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 || sub
225 .group_by
226 .as_ref()
227 .is_some_and(|g| g.keys.iter().any(|k| inner(&k.expr)))
228 || sub
229 .having
230 .as_ref()
231 .is_some_and(|h| where_may_raise(h, target, false))
232 || sub.limit.as_ref().is_some_and(inner)
233 || sub.offset.as_ref().is_some_and(inner)
234}
235
236fn follow_may_raise(f: &Follow, target: Target) -> bool {
239 let inner = |e: &Expr| expr_may_raise(e, target, false);
240 f.destinations.iter().any(|d| match d {
241 FollowDestination::Relation(e) => inner(e),
242 FollowDestination::Block(op) => op_may_raise(op, target, false),
243 })
244}
245
246fn expr_may_raise(e: &Expr, target: Target, root: bool) -> bool {
247 let again = |e: &Expr| expr_may_raise(e, target, root);
248 match e {
249 Expr::Lit { .. } | Expr::Binding { .. } => false,
250 Expr::Ident { name, .. } if Target::parse(name).is_some() => {
254 !root || !is_target_relation_of(target, name)
255 }
256 Expr::Ident { .. } => false,
258 Expr::Outer { .. } | Expr::Call { .. } => true,
259 Expr::Member { recv, name, .. } => {
262 (Target::parse(name).is_some()
263 && !matches!(&**recv, Expr::Ident { name: head, .. }
264 if reach_through(head).is_some_and(|t| is_target_relation_of(t, name))))
265 || again(recv)
266 }
267 Expr::Unary { expr, .. } => again(expr),
268 Expr::Required { .. } => true,
271 Expr::Op(op) => op_may_raise(op, target, root),
272 Expr::Binary {
275 op, left, right, ..
276 } => op.is_regex() || again(left) || again(right),
277 Expr::Logical { left, right, .. } | Expr::In { left, right, .. } => {
278 again(left) || again(right)
279 }
280 Expr::Range { lo, hi, .. } => {
281 lo.as_deref().is_some_and(again) || hi.as_deref().is_some_and(again)
282 }
283 Expr::Tuple { items, .. } => items.iter().any(again),
285 Expr::Template { parts, .. } => oqx::walk::template_exprs(parts).next().is_some(),
289 Expr::Aggregate { .. } => true,
293 }
294}
295
296#[derive(Clone, Debug, PartialEq)]
299pub struct Compiled {
300 pub target: Target,
301 pub sql: String,
302 pub params: Vec<SqlValue>,
303 pub residual: Query,
304}
305
306#[must_use]
315pub fn compile(query: &Query, params: &[Value], repo_id: &str) -> Option<Compiled> {
316 if query.follow.is_some() || !query.from.is_empty() || query.group_by.is_some() {
317 return None;
318 }
319 let target = root_target(&query.source)?;
320 if target == Target::Revisions {
323 return None;
324 }
325 let (self_alias, doc_alias) = aliases(target);
326 let ctx = TranslateCtx {
327 target,
328 self_alias,
329 doc_alias,
330 params,
331 };
332 let (pushed, residual) = partition_pushable(query.r#where.as_ref(), |e| {
333 translate_predicate(e, &ctx).is_some()
334 });
335 if pushed.is_empty() {
336 return None;
337 }
338 if residual
341 .as_ref()
342 .is_some_and(|w| residual_may_raise(w, target))
343 {
344 return None;
345 }
346 let mut where_sql = guards(target).to_owned();
347 let mut sql_params = vec![SqlValue::Text(repo_id.to_owned())];
348 for e in &pushed {
349 let frag = translate_predicate(e, &ctx).expect("accepted by partition_pushable");
350 where_sql.push_str(" AND (");
351 where_sql.push_str(&frag.sql);
352 where_sql.push(')');
353 sql_params.extend(frag.params);
354 }
355 let sql = format!(
356 "SELECT {} FROM {} WHERE {where_sql} ORDER BY {}",
357 columns(target),
358 from_clause(target),
359 order_clause(target)
360 );
361 Some(Compiled {
362 target,
363 sql,
364 params: sql_params,
365 residual: residual_query(query, residual),
366 })
367}
368
369pub struct SqlitePlanner<'a> {
371 conn: &'a Connection,
372 repo_id: String,
373}
374
375impl<'a> SqlitePlanner<'a> {
376 #[must_use]
377 pub fn new(conn: &'a Connection, repo_id: &str) -> Self {
378 Self {
379 conn,
380 repo_id: repo_id.to_owned(),
381 }
382 }
383
384 pub fn try_plan(&self, query: &Query, params: &[Value]) -> rusqlite::Result<Option<Plan>> {
389 let Some(compiled) = compile(query, params, &self.repo_id) else {
390 return Ok(None);
391 };
392 let rows = fetch_rows(self.conn, &compiled.sql, &compiled.params)?;
393 Ok(Some(Plan::new(
394 tag_rows(rows, compiled.target),
395 compiled.residual,
396 )))
397 }
398}
399
400impl QueryPlanner for SqlitePlanner<'_> {
401 fn plan(&self, query: &Query, params: &[Value]) -> Option<Plan> {
404 self.try_plan(query, params).ok().flatten()
405 }
406}
407
408#[cfg(test)]
409mod tests {
410 use super::*;
411 use oqx::ROWS_ROOT;
412 use oqx::ast::Where;
413
414 fn parse(src: &str) -> Query {
415 oqx::parse_string(src).expect("parses")
416 }
417
418 fn text(s: &str) -> SqlValue {
419 SqlValue::Text(s.to_owned())
420 }
421
422 #[test]
423 fn a_pushable_scan_compiles_to_one_statement_in_root_order() {
424 let c =
425 compile(&parse("from docs where path == \"index.md\""), &[], "r_1").expect("planned");
426 assert_eq!(c.target, Target::Docs);
427 assert_eq!(
428 c.sql,
429 "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"
430 );
431 assert_eq!(c.params, vec![text("r_1"), text("index.md")]);
432 assert_eq!(
433 c.residual.source,
434 Expr::Ident {
435 name: ROWS_ROOT.to_owned(),
436 span: oqx::Span::EMPTY,
437 }
438 );
439 assert_eq!(c.residual.r#where, None);
440 }
441
442 #[test]
443 fn every_target_has_its_join_columns_guards_and_order() {
444 let b = compile(
445 &parse("from blocks where path.startsWith(\"lab/\")"),
446 &[],
447 "r",
448 )
449 .unwrap();
450 assert_eq!(
451 b.sql,
452 "SELECT b.*, d.path AS __path FROM blocks b JOIN docs d ON d.doc_id = b.doc_id \
453 WHERE b.repo_id = ? AND b.deleted_commit IS NULL AND d.deleted_commit IS NULL \
454 AND ((substr(('/' || d.path), 1, length(?)) = ?)) ORDER BY d.path, b.block_id"
455 );
456 assert_eq!(b.params, vec![text("r"), text("lab/"), text("lab/")]);
457 let n = compile(
458 &parse("nodes count { where kind == \"md:task\" }"),
459 &[],
460 "r",
461 )
462 .unwrap();
463 assert_eq!(n.target, Target::Nodes);
464 assert!(n.sql.starts_with(
465 "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 ?))"
466 ));
467 assert!(n.sql.ends_with("ORDER BY d.path, n.node_id"));
468 let e = compile(
469 &parse("from links where predicate == \"references\""),
470 &[],
471 "r",
472 )
473 .unwrap();
474 assert!(e.sql.starts_with(
475 "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 ?))"
476 ));
477 assert!(e.sql.ends_with("ORDER BY d.path, e.edge_id"));
478 }
479
480 #[test]
481 fn mixed_conjunctions_push_the_translatable_parts_and_keep_the_rest() {
482 let q = parse(
483 "from docs where path.startsWith(\"processes/\") && nodes exists { where kind == \"md:task\" } && layer == \"canon\"",
484 );
485 let c = compile(&q, &[], "r").expect("planned");
486 assert!(
487 c.sql
488 .contains("(substr(('/' || d.path), 1, length(?)) = ?)"),
489 "{}",
490 c.sql
491 );
492 assert!(c.sql.contains("p.key = 'layer'"), "{}", c.sql);
493 assert_eq!(
495 c.params,
496 vec![
497 text("r"),
498 text("processes/"),
499 text("processes/"),
500 text("canon")
501 ]
502 );
503 assert!(
505 matches!(c.residual.r#where, Some(Where::Op(_))),
506 "{:?}",
507 c.residual.r#where
508 );
509 assert!(c.residual.from.is_empty());
510 assert_eq!(c.residual.select, q.select);
511 assert_eq!(c.residual.consumer, q.consumer);
512 }
513
514 #[test]
515 fn declined_shapes_return_none() {
516 assert!(compile(&parse("from docs"), &[], "r").is_none());
518 assert!(compile(&parse("from docs where era in 800..1680"), &[], "r").is_none());
519 assert!(compile(&parse("from docs where !verified"), &[], "r").is_none());
520 assert!(
521 compile(
522 &parse("from docs where path == \"a\" || path == \"b\""),
523 &[],
524 "r"
525 )
526 .is_none()
527 );
528 assert!(
529 compile(
530 &parse("from docs where nodes exists { where kind == \"md:task\" }"),
531 &[],
532 "r"
533 )
534 .is_none()
535 );
536 assert!(
538 compile(
539 &parse("from docs where path == \"a.md\" follow distinct out_links collect { from dst }"),
540 &[],
541 "r"
542 )
543 .is_none()
544 );
545 assert!(
548 compile(
549 &parse("from docs where path == \"a.md\" follow out_links collect { from dst }, in_links collect { from doc }"),
550 &[],
551 "r"
552 )
553 .is_none()
554 );
555 assert!(
556 compile(
557 &parse(
558 "from docs where path == \"a.md\" follow ^docs collect { where after.contains(^path) }"
559 ),
560 &[],
561 "r"
562 )
563 .is_none()
564 );
565 assert!(
567 compile(
568 &parse("from docs where path == \"a.md\" frontier type == \"hub\" follow distinct out_links collect { from dst frontier type == \"practitioner\" }"),
569 &[],
570 "r"
571 )
572 .is_none()
573 );
574 assert!(
576 compile(
577 &parse("select layer, n: count() from docs where path == \"a.md\" group by layer"),
578 &[],
579 "r"
580 )
581 .is_none()
582 );
583 assert!(compile(&parse("from things where path == \"a.md\""), &[], "r").is_none());
585 assert!(compile(&parse("from nope where path == \"a.md\""), &[], "r").is_none());
586 assert!(compile(&parse("from docs.nodes where kind == \"x\""), &[], "r").is_none());
587 let mut q = parse("from docs where path == \"a.md\"");
589 q.from.push(Expr::Ident {
590 name: "nodes".to_owned(),
591 span: oqx::Span::EMPTY,
592 });
593 assert!(compile(&q, &[], "r").is_none());
594 }
595
596 #[test]
599 fn a_residual_that_could_raise_declines_the_whole_query() {
600 let declined = |src: &str| {
601 assert!(
602 compile(&parse(src), &[], "r").is_none(),
603 "should decline: {src}"
604 );
605 };
606 let planned = |src: &str| {
607 assert!(
608 compile(&parse(src), &[], "r").is_some(),
609 "should plan: {src}"
610 );
611 };
612 planned("from docs where path == \"x\" && path == \"nope.md\"");
615 planned("from docs where path == \"nope.md\" && !raw");
616 planned("from docs where path == \"nope.md\" && updated_at");
617 planned("from docs where path == \"nope.md\" && tags");
618 planned("from blocks where path == \"x\" && doc.path == \"y\"");
620 planned("from blocks where path == \"x\" && nodes exists { where doc.path == \"y\" }");
621 declined("from docs where path.matches(\"[\") && path == \"nope.md\"");
623 declined("from docs where nope(\"x\") && path == \"nope.md\"");
624 declined("from docs where path == \"x\" && size(tags) > 1");
625 declined("from docs where path == \"x\" && nodes exists { where name.lower() == \"a\" }");
626 declined(
627 "from docs where path == \"x\" && nodes count { where kind == \"a\" order by size(name) } > 1",
628 );
629 declined("from docs where path == \"x\" && ^slug == \"y\"");
631 declined("from docs where path == \"x\" && nodes exists { where name == ^title }");
632 declined("from docs where path == \"x\" && nodes collect { ^first_task: name }");
633 declined(
635 "from docs where path == \"x\" && blocks exists { select t: nodes single { where kind == \"md:task\" } }",
636 );
637 planned("from docs where path.startsWith(\"lab/\") && path == \"x\"");
639 planned("from docs where path == \"x\" && era in 800..1680");
641 planned("from docs where path == \"x\" && !verified");
642 planned("from docs where path == \"x\" && (layer == \"a\" || layer == \"b\")");
643 planned("from docs where path == \"x\" && verified == true");
644 planned("from docs where path == \"x\" && nodes exists { where kind == \"md:task\" }");
645 planned(
646 "from docs where path == \"x\" && nodes count { where kind == \"md:task\" limit 5 } > 1",
647 );
648 planned("from blocks where path == \"x\" && !path");
651 planned("from docs where path == \"x\" && nodes exists { where path == \"y\" }");
652 planned("from docs where path == \"x\" && frontmatter.path == \"y\"");
653 let fm = compile(
657 &parse("from docs where frontmatter.layer == \"canon\""),
658 &[],
659 "r",
660 )
661 .expect("planned");
662 assert!(fm.sql.contains("p.source = 'frontmatter'"), "{}", fm.sql);
663 let data =
664 compile(&parse("from docs where data.layer == \"canon\""), &[], "r").expect("planned");
665 assert!(!data.sql.contains("p.source"), "{}", data.sql);
666 planned("from blocks where doc.inline.element == \"fire\"");
667 planned("from docs where path == \"x\" && frontmatter == null");
668 declined("from docs where path == \"x\" && entries(frontmatter) exists { }");
669 }
670
671 #[test]
672 fn a_top_level_not_or_or_is_the_whole_residual_and_declines() {
673 assert!(compile(&parse("from docs where !(path == \"a\")"), &[], "r").is_none());
676 assert!(
677 compile(
678 &parse("from docs where (path == \"a\" || path == \"b\") && layer == \"canon\""),
679 &[],
680 "r"
681 )
682 .is_some()
683 );
684 }
685}