1use std::cell::RefCell;
39use std::collections::HashMap;
40
41use omgbase_properties::Bound;
42use omgbase_search::{cosine_bytes, sanitize_fts_query};
43use oqx::semantics::{builtin_function, builtin_method_with, make_range};
44use oqx::{DataContext, Object, OqxError, RegexDialect, Value};
45use rusqlite::types::{Value as SqlValue, ValueRef};
46use rusqlite::{Connection, OptionalExtension, params_from_iter};
47
48pub const TAG_KEY: &str = "__oqx_target";
50const REPO_TAG: &str = "$repo";
51
52#[derive(Clone, Copy, Debug, PartialEq, Eq)]
54pub enum Target {
55 Docs,
56 Blocks,
57 Nodes,
58 Edges,
59}
60
61impl Target {
62 #[must_use]
64 pub fn as_str(self) -> &'static str {
65 match self {
66 Target::Docs => "docs",
67 Target::Blocks => "blocks",
68 Target::Nodes => "nodes",
69 Target::Edges => "edges",
70 }
71 }
72
73 #[must_use]
75 pub fn parse(s: &str) -> Option<Self> {
76 Some(match s {
77 "docs" => Target::Docs,
78 "blocks" => Target::Blocks,
79 "nodes" => Target::Nodes,
80 "edges" => Target::Edges,
81 _ => return None,
82 })
83 }
84}
85
86const RESERVED_DOC_BASENAMES: [&str; 5] = ["id", "path", "updated_at", "content_hash", "body"];
89
90#[derive(Clone, Debug, PartialEq)]
93pub struct SemanticVec {
94 pub model: String,
95 pub vec: Vec<u8>,
96}
97
98pub struct StoreContext<'a> {
100 conn: &'a Connection,
101 repo_id: String,
102 semantic: HashMap<String, SemanticVec>,
103 root_failure: RefCell<Option<OqxError>>,
106 rows_root: Option<Vec<Value>>,
108}
109
110fn sql_value(v: ValueRef<'_>) -> Value {
111 match v {
112 ValueRef::Null => Value::Null,
113 ValueRef::Integer(i) => Value::Number(i as f64),
114 ValueRef::Real(f) => Value::Number(f),
115 ValueRef::Text(t) => Value::Str(String::from_utf8_lossy(t).into_owned()),
116 ValueRef::Blob(b) => Value::Str(omgbase_format::hash::hex(b)),
117 }
118}
119
120pub(crate) fn to_sql(v: &Value) -> SqlValue {
125 match v {
126 Value::Undefined | Value::Null | Value::Range(_) => SqlValue::Null,
127 Value::Bool(b) => SqlValue::Integer(i64::from(*b)),
128 Value::Number(n) => SqlValue::Real(*n),
129 Value::Str(s) => SqlValue::Text(s.clone()),
130 Value::Array(_) | Value::Object(_) => SqlValue::Text(v.to_string()),
131 }
132}
133
134fn js_string(v: &Value) -> String {
136 v.to_string()
137}
138
139fn arg_or_empty(args: &[Value], i: usize) -> String {
141 match args.get(i) {
142 None | Some(Value::Undefined) | Some(Value::Null) => String::new(),
143 Some(v) => js_string(v),
144 }
145}
146
147fn parse_json(v: &Value) -> Value {
149 match v {
150 Value::Str(s) => {
151 serde_json::from_str::<serde_json::Value>(s).map_or_else(|_| v.clone(), Value::from)
152 }
153 Value::Null | Value::Undefined => Value::Undefined,
154 other => other.clone(),
155 }
156}
157
158fn sql_err(e: rusqlite::Error) -> OqxError {
161 OqxError::eval(format!("sqlite: {e}"))
162}
163
164pub fn target_of(row: &Value) -> Option<Target> {
166 row.as_object()
167 .and_then(|o| o.get(TAG_KEY))
168 .and_then(Value::as_str)
169 .and_then(Target::parse)
170}
171
172fn is_repo_root(row: &Value) -> bool {
173 row.as_object()
174 .and_then(|o| o.get(TAG_KEY))
175 .and_then(Value::as_str)
176 == Some(REPO_TAG)
177}
178
179fn col<'v>(row: &'v Value, key: &str) -> &'v Value {
180 row.as_object()
181 .and_then(|o| o.get(key))
182 .unwrap_or(&Value::Undefined)
183}
184
185fn col_str(row: &Value, key: &str) -> String {
186 match col(row, key) {
187 Value::Undefined | Value::Null => String::new(),
188 v => js_string(v),
189 }
190}
191
192#[must_use]
194pub fn strip_tags(v: Value) -> Value {
195 match v {
196 Value::Object(o) => Value::Object(
197 o.into_iter()
198 .filter(|(k, _)| k != TAG_KEY)
199 .map(|(k, x)| (k, strip_tags(x)))
200 .collect(),
201 ),
202 Value::Array(a) => Value::Array(a.into_iter().map(strip_tags).collect()),
203 other => other,
204 }
205}
206
207impl<'a> StoreContext<'a> {
208 #[must_use]
211 pub fn new(
212 conn: &'a Connection,
213 repo_id: &str,
214 semantic: HashMap<String, SemanticVec>,
215 ) -> Self {
216 Self {
217 conn,
218 repo_id: repo_id.to_owned(),
219 semantic,
220 root_failure: RefCell::new(None),
221 rows_root: None,
222 }
223 }
224
225 #[must_use]
230 pub fn with_rows_root(mut self, rows: Vec<Value>) -> Self {
231 self.rows_root = Some(rows);
232 self
233 }
234
235 pub fn take_root_failure(&self) -> Option<OqxError> {
240 self.root_failure.borrow_mut().take()
241 }
242
243 fn all(&self, sql: &str, params: &[SqlValue]) -> oqx::Result<Vec<Object>> {
246 fetch_rows(self.conn, sql, params).map_err(sql_err)
247 }
248
249 fn one(&self, sql: &str, params: &[SqlValue]) -> oqx::Result<Option<Object>> {
250 Ok(self.all(sql, params)?.into_iter().next())
251 }
252
253 fn scalar(&self, sql: &str, params: &[SqlValue]) -> oqx::Result<Value> {
254 Ok(self
255 .one(sql, params)?
256 .and_then(|o| o.values().next().cloned())
257 .unwrap_or(Value::Undefined))
258 }
259
260 fn exists(&self, sql: &str, params: &[SqlValue]) -> oqx::Result<bool> {
261 let mut stmt = self.conn.prepare_cached(sql).map_err(sql_err)?;
262 stmt.exists(params_from_iter(params.iter()))
263 .map_err(sql_err)
264 }
265
266 fn tag_all(rows: Vec<Object>, t: Target) -> Value {
267 Value::Array(tag_rows(rows, t))
268 }
269
270 fn tag(row: Object, t: Target) -> Value {
271 tag_row(row, t)
272 }
273
274 fn repo_root(&self) -> Value {
275 let mut o = Object::with_capacity(1);
276 o.insert(TAG_KEY, Value::Str(REPO_TAG.to_owned()));
277 Value::Object(o)
278 }
279
280 fn root_scan(&self, t: Target) -> oqx::Result<Value> {
283 let repo = [SqlValue::Text(self.repo_id.clone())];
284 let sql = match t {
285 Target::Docs => {
286 "SELECT * FROM docs WHERE repo_id = ?1 AND deleted_commit IS NULL ORDER BY path, doc_id"
287 }
288 Target::Blocks => {
289 "SELECT b.*, d.path AS __path FROM blocks b JOIN docs d ON d.doc_id = b.doc_id
290 WHERE b.repo_id = ?1 AND b.deleted_commit IS NULL AND d.deleted_commit IS NULL
291 ORDER BY d.path, b.block_id"
292 }
293 Target::Nodes => {
294 "SELECT n.*, d.path AS __path FROM nodes n JOIN docs d ON d.doc_id = n.doc_id
295 WHERE n.repo_id = ?1 AND d.deleted_commit IS NULL ORDER BY d.path, n.node_id"
296 }
297 Target::Edges => {
298 "SELECT e.*, d.path AS __path FROM edges e JOIN docs d ON d.doc_id = e.src_doc
299 WHERE e.repo_id = ?1 AND e.to_commit IS NULL AND d.deleted_commit IS NULL
300 ORDER BY d.path, e.edge_id"
301 }
302 };
303 Ok(Self::tag_all(self.all(sql, &repo)?, t))
304 }
305
306 fn decode_prop(r: &Object) -> Value {
311 let get = |k: &str| r.get(k).cloned().unwrap_or(Value::Undefined);
312 match get("type").as_str().unwrap_or("") {
313 "number" => get("val_num"),
314 "bool" => Value::Bool(get("val_bool").truthy()),
315 "null" => Value::Null,
316 "json" => parse_json(&get("val_json")),
317 _ => get("val_text"),
318 }
319 }
320
321 fn doc_prop(&self, doc_id: &str, key: &str, source: Option<&str>) -> oqx::Result<Value> {
324 let rows = match source {
325 Some(s) => self.all(
326 "SELECT * FROM properties WHERE doc_id = ?1 AND key = ?2 AND source = ?3 AND deleted_commit IS NULL ORDER BY ord",
327 &[SqlValue::Text(doc_id.to_owned()), SqlValue::Text(key.to_owned()), SqlValue::Text(s.to_owned())],
328 )?,
329 None => self.all(
330 "SELECT * FROM properties WHERE doc_id = ?1 AND key = ?2 AND deleted_commit IS NULL ORDER BY ord",
331 &[SqlValue::Text(doc_id.to_owned()), SqlValue::Text(key.to_owned())],
332 )?,
333 };
334 if rows.is_empty() {
335 return self.doc_prop_object(doc_id, key, source);
336 }
337 if rows.len() == 1 && rows[0].get("card").and_then(Value::as_str) == Some("scalar") {
338 return Ok(Self::decode_prop(&rows[0]));
339 }
340 Ok(Value::Array(rows.iter().map(Self::decode_prop).collect()))
341 }
342
343 fn doc_prop_object(
346 &self,
347 doc_id: &str,
348 prefix: &str,
349 source: Option<&str>,
350 ) -> oqx::Result<Value> {
351 let like = SqlValue::Text(format!("{prefix}.%"));
352 let rows = match source {
353 Some(s) => self.all(
354 "SELECT * FROM properties WHERE doc_id = ?1 AND key LIKE ?2 AND source = ?3 AND deleted_commit IS NULL ORDER BY ord",
355 &[SqlValue::Text(doc_id.to_owned()), like, SqlValue::Text(s.to_owned())],
356 )?,
357 None => self.all(
358 "SELECT * FROM properties WHERE doc_id = ?1 AND key LIKE ?2 AND deleted_commit IS NULL ORDER BY ord",
359 &[SqlValue::Text(doc_id.to_owned()), like],
360 )?,
361 };
362 if rows.is_empty() {
363 return Ok(Value::Undefined);
364 }
365 let mut out = Object::new();
366 for r in &rows {
367 let key = r.get("key").and_then(Value::as_str).unwrap_or("");
368 let rest: Vec<&str> = key[(prefix.len() + 1).min(key.len())..]
369 .split('.')
370 .collect();
371 set_nested(&mut out, &rest, Self::decode_prop(r));
372 }
373 Ok(Value::Object(out))
374 }
375
376 fn doc_prop_bag(&self, doc_id: &str, source: &str) -> oqx::Result<Value> {
379 let keys = self.all(
380 "SELECT DISTINCT key FROM properties WHERE doc_id = ?1 AND source = ?2 AND deleted_commit IS NULL ORDER BY key",
381 &[SqlValue::Text(doc_id.to_owned()), SqlValue::Text(source.to_owned())],
382 )?;
383 let mut out = Object::new();
384 for k in keys {
385 let key = k.get("key").and_then(Value::as_str).unwrap_or("");
386 let top = key.split('.').next().unwrap_or("");
387 if !out.contains_key(top) {
388 let v = self.doc_prop(doc_id, top, Some(source))?;
389 out.insert(top, v);
390 }
391 }
392 Ok(Value::Object(out))
393 }
394
395 fn top_ordinal(&self, block: &Value) -> oqx::Result<Value> {
400 let ordinal = col(block, "ordinal").clone();
401 if col(block, "parent_block").is_absent() {
402 return Ok(ordinal);
403 }
404 let ap = col_str(block, "ancestor_path");
405 let Some(first) = ap.split('/').find(|s| !s.is_empty()) else {
406 return Ok(ordinal);
407 };
408 let r = self.scalar(
409 "SELECT ordinal FROM blocks WHERE doc_id = ?1 AND block_id = ?2",
410 &[
411 SqlValue::Text(col_str(block, "doc_id")),
412 SqlValue::Text(first.to_owned()),
413 ],
414 )?;
415 Ok(if r.is_absent() { ordinal } else { r })
416 }
417
418 fn doc_blocks_preorder(&self, doc_id: &str, path: &str) -> oqx::Result<Vec<Value>> {
422 let rows = self.all(
423 "SELECT b.*, ?1 AS __path FROM blocks b WHERE b.doc_id = ?2 AND b.deleted_commit IS NULL ORDER BY b.ordinal, b.block_id",
424 &[SqlValue::Text(path.to_owned()), SqlValue::Text(doc_id.to_owned())],
425 )?;
426 let ids: Vec<String> = rows
427 .iter()
428 .map(|r| {
429 r.get("block_id")
430 .and_then(Value::as_str)
431 .unwrap_or("")
432 .to_owned()
433 })
434 .collect();
435 let parent_index: Vec<Option<usize>> = rows
436 .iter()
437 .map(|r| {
438 r.get("parent_block")
439 .and_then(Value::as_str)
440 .and_then(|p| ids.iter().position(|id| id == p))
441 })
442 .collect();
443 let mut children: Vec<Vec<usize>> = vec![Vec::new(); rows.len()];
444 let mut roots = Vec::new();
445 for (i, p) in parent_index.iter().enumerate() {
446 match p {
447 Some(p) => children[*p].push(i),
448 None => roots.push(i),
449 }
450 }
451 fn walk(i: usize, children: &[Vec<usize>], order: &mut Vec<usize>) {
452 order.push(i);
453 for &c in &children[i] {
454 walk(c, children, order);
455 }
456 }
457 let mut order = Vec::with_capacity(rows.len());
458 for r in roots {
459 walk(r, &children, &mut order);
460 }
461 let mut slots: Vec<Option<Object>> = rows.into_iter().map(Some).collect();
462 Ok(order
463 .into_iter()
464 .map(|i| Self::tag(slots[i].take().expect("visited once"), Target::Blocks))
465 .collect())
466 }
467
468 fn jattr(row: &Value, k: &str) -> Value {
469 match parse_json(col(row, "attrs")) {
470 Value::Object(o) => o.get(k).cloned().unwrap_or(Value::Undefined),
471 _ => Value::Undefined,
472 }
473 }
474
475 fn relation(&self, row: &Value, t: Target, key: &str) -> oqx::Result<Option<Value>> {
477 let path = || SqlValue::Text(col_str(row, "__path"));
478 let doc_id = || SqlValue::Text(col_str(row, "doc_id"));
479 let doc_path = || SqlValue::Text(col_str(row, "path"));
480 let block_id = || SqlValue::Text(col_str(row, "block_id"));
481 let repo = || SqlValue::Text(self.repo_id.clone());
482 Ok(Some(match (t, key) {
483 (Target::Docs, "nodes") => {
484 let rows = self.all(
487 "SELECT n.*, ?1 AS __path FROM nodes n WHERE n.doc_id = ?2 ORDER BY n.node_id",
488 &[doc_path(), doc_id()],
489 )?;
490 let blocks = self.doc_blocks_preorder(&col_str(row, "doc_id"), &col_str(row, "path"))?;
491 let rank: HashMap<String, usize> = blocks
492 .iter()
493 .enumerate()
494 .map(|(i, b)| (col_str(b, "block_id"), i))
495 .collect();
496 let mut keyed: Vec<((usize, usize, f64, String), Object)> = rows
497 .into_iter()
498 .map(|r| {
499 let block = r.get("block_id").and_then(Value::as_str);
500 let (has_block, rk) = match block {
501 None => (0, 0),
502 Some(b) => (1, rank.get(b).copied().unwrap_or(usize::MAX)),
503 };
504 let span = r.get("span_start").and_then(Value::as_f64).unwrap_or(-1.0);
505 let id = r.get("node_id").and_then(Value::as_str).unwrap_or("").to_owned();
506 ((has_block, rk, span, id), r)
507 })
508 .collect();
509 keyed.sort_by(|a, b| {
510 a.0.0
511 .cmp(&b.0.0)
512 .then(a.0.1.cmp(&b.0.1))
513 .then(a.0.2.total_cmp(&b.0.2))
514 .then(a.0.3.cmp(&b.0.3))
515 });
516 Value::Array(keyed.into_iter().map(|(_, r)| Self::tag(r, Target::Nodes)).collect())
517 }
518 (Target::Docs, "blocks") => {
519 Value::Array(self.doc_blocks_preorder(&col_str(row, "doc_id"), &col_str(row, "path"))?)
520 }
521 (Target::Docs, "out") => Self::tag_all(
522 self.all(
523 "SELECT DISTINCT d2.* FROM docs d2 JOIN edges e ON e.dst_node = d2.doc_id
524 WHERE e.src_doc = ?1 AND e.to_commit IS NULL AND d2.repo_id = ?2 AND d2.deleted_commit IS NULL ORDER BY d2.path, d2.doc_id",
525 &[doc_id(), repo()],
526 )?,
527 Target::Docs,
528 ),
529 (Target::Docs, "in") => Self::tag_all(
530 self.all(
531 "SELECT DISTINCT d2.* FROM docs d2 JOIN edges e ON e.src_doc = d2.doc_id
532 WHERE e.dst_node = ?1 AND e.to_commit IS NULL AND d2.repo_id = ?2 AND d2.deleted_commit IS NULL ORDER BY d2.path, d2.doc_id",
533 &[doc_id(), repo()],
534 )?,
535 Target::Docs,
536 ),
537 (Target::Docs, "out_edges") => Self::tag_all(
538 self.all(
539 "SELECT e.*, ?1 AS __path FROM edges e WHERE e.src_doc = ?2 AND e.to_commit IS NULL ORDER BY e.predicate, e.edge_id",
540 &[doc_path(), doc_id()],
541 )?,
542 Target::Edges,
543 ),
544 (Target::Docs, "in_edges") => Self::tag_all(
545 self.all(
546 "SELECT e.*, d.path AS __path FROM edges e JOIN docs d ON d.doc_id = e.src_doc
547 WHERE e.dst_node = ?1 AND e.to_commit IS NULL AND d.deleted_commit IS NULL ORDER BY e.predicate, e.edge_id",
548 &[doc_id()],
549 )?,
550 Target::Edges,
551 ),
552 (Target::Blocks, "children") => Self::tag_all(
553 self.all(
554 "SELECT b.*, ?1 AS __path FROM blocks b WHERE b.parent_block = ?2 AND b.deleted_commit IS NULL ORDER BY b.ordinal, b.block_id",
555 &[path(), block_id()],
556 )?,
557 Target::Blocks,
558 ),
559 (Target::Blocks, "nodes") => Self::tag_all(
560 self.all(
561 "SELECT n.*, ?1 AS __path FROM nodes n WHERE n.block_id = ?2 ORDER BY n.span_start, n.node_id",
562 &[path(), block_id()],
563 )?,
564 Target::Nodes,
565 ),
566 (Target::Blocks, "out_edges") => Self::tag_all(
567 self.all(
568 "SELECT e.*, ?1 AS __path FROM edges e WHERE e.src_block = ?2 AND e.to_commit IS NULL ORDER BY e.predicate, e.edge_id",
569 &[path(), block_id()],
570 )?,
571 Target::Edges,
572 ),
573 (Target::Blocks, "section") => {
574 let top = to_sql(&self.top_ordinal(row)?);
575 Self::tag_all(
576 self.all(
577 "SELECT n.*, ?1 AS __path FROM nodes n WHERE n.doc_id = ?2 AND n.kind = 'md:section'
578 AND json_extract(n.attrs,'$.first_ordinal') <= ?3 AND json_extract(n.attrs,'$.last_ordinal') >= ?4
579 ORDER BY json_extract(n.attrs,'$.first_ordinal'), n.node_id",
580 &[path(), doc_id(), top.clone(), top],
581 )?,
582 Target::Nodes,
583 )
584 }
585 (Target::Nodes, "blocks") => {
586 let (f, l) = (Self::jattr(row, "first_ordinal"), Self::jattr(row, "last_ordinal"));
587 if f.is_absent() || l.is_absent() {
588 return Ok(Some(Value::Array(Vec::new())));
589 }
590 let (f, l) = (
591 f.as_f64().unwrap_or(f64::NAN),
592 l.as_f64().unwrap_or(f64::NAN),
593 );
594 let rows = self.doc_blocks_preorder(&col_str(row, "doc_id"), &col_str(row, "__path"))?;
595 let mut kept: Vec<Value> = Vec::new();
596 for b in rows {
597 let t = self.top_ordinal(&b)?.as_f64().unwrap_or(f64::NAN);
598 if t >= f && t <= l {
599 kept.push(b);
600 }
601 }
602 Value::Array(kept)
603 }
604 (Target::Nodes, "subsections") => {
605 let (f, l, lvl) = (
606 Self::jattr(row, "first_ordinal"),
607 Self::jattr(row, "last_ordinal"),
608 Self::jattr(row, "level"),
609 );
610 if f.is_absent() {
611 return Ok(Some(Value::Array(Vec::new())));
612 }
613 Self::tag_all(
614 self.all(
615 "SELECT n.*, ?1 AS __path FROM nodes n WHERE n.doc_id = ?2 AND n.kind = 'md:section'
616 AND json_extract(n.attrs,'$.first_ordinal') >= ?3 AND json_extract(n.attrs,'$.last_ordinal') <= ?4
617 AND json_extract(n.attrs,'$.level') > ?5 ORDER BY json_extract(n.attrs,'$.first_ordinal'), n.node_id",
618 &[path(), doc_id(), to_sql(&f), to_sql(&l), to_sql(&lvl)],
619 )?,
620 Target::Nodes,
621 )
622 }
623 (Target::Nodes, "children") => {
624 let (f, l, lvl) = (
625 Self::jattr(row, "first_ordinal"),
626 Self::jattr(row, "last_ordinal"),
627 Self::jattr(row, "level"),
628 );
629 if f.is_absent() {
630 return Ok(Some(Value::Array(Vec::new())));
631 }
632 Self::tag_all(
633 self.all(
634 "SELECT i.*, ?1 AS __path FROM nodes i WHERE i.doc_id = ?2 AND i.kind = 'md:section'
635 AND json_extract(i.attrs,'$.level') > ?3
636 AND json_extract(i.attrs,'$.first_ordinal') >= ?4 AND json_extract(i.attrs,'$.last_ordinal') <= ?5
637 AND NOT EXISTS (SELECT 1 FROM nodes m WHERE m.doc_id = i.doc_id AND m.kind = 'md:section'
638 AND json_extract(m.attrs,'$.level') > ?6 AND json_extract(m.attrs,'$.level') < json_extract(i.attrs,'$.level')
639 AND json_extract(m.attrs,'$.first_ordinal') <= json_extract(i.attrs,'$.first_ordinal')
640 AND json_extract(m.attrs,'$.last_ordinal') >= json_extract(i.attrs,'$.last_ordinal'))
641 ORDER BY json_extract(i.attrs,'$.first_ordinal'), i.node_id",
642 &[path(), doc_id(), to_sql(&lvl), to_sql(&f), to_sql(&l), to_sql(&lvl)],
643 )?,
644 Target::Nodes,
645 )
646 }
647 _ => return Ok(None),
648 }))
649 }
650
651 fn owning_doc(&self, row: &Value) -> oqx::Result<Value> {
652 let id = match col(row, "doc_id") {
653 Value::Undefined | Value::Null => col(row, "src_doc").clone(),
654 v => v.clone(),
655 };
656 Ok(self
657 .one("SELECT * FROM docs WHERE doc_id = ?1", &[to_sql(&id)])?
658 .map_or(Value::Undefined, |o| Self::tag(o, Target::Docs)))
659 }
660
661 fn owning_block(&self, row: &Value) -> oqx::Result<Value> {
662 let id = col(row, "block_id");
663 if !id.truthy() {
664 return Ok(Value::Undefined);
665 }
666 Ok(self
667 .one(
668 "SELECT b.*, d.path AS __path FROM blocks b JOIN docs d ON d.doc_id = b.doc_id WHERE b.block_id = ?1",
669 &[to_sql(id)],
670 )?
671 .map_or(Value::Undefined, |o| Self::tag(o, Target::Blocks)))
672 }
673
674 fn null_if_absent(v: Value) -> Value {
677 if v.is_absent() { Value::Null } else { v }
678 }
679
680 fn intrinsic(&self, row: &Value, t: Target, name: &str) -> oqx::Result<Value> {
681 if name == "$self" {
682 return Ok(row.clone());
683 }
684 let c = |k: &str| col(row, k).clone();
685 Ok(match (t, name) {
686 (Target::Docs, "$id") => c("doc_id"),
687 (Target::Docs, "$path") => c("path"),
688 (Target::Docs, "$content_hash") => Self::null_if_absent(c("file_hash")),
689 (Target::Docs, "$updated_at") => Self::null_if_absent(self.scalar(
690 "SELECT c.ts FROM revisions r JOIN commits c ON c.commit_id = r.commit_id WHERE r.rev_id = ?1",
691 &[to_sql(&c("current_rev"))],
692 )?),
693 (Target::Docs, "$body") => {
694 match omgbase_store::read::reconstruct(self.conn, &col_str(row, "doc_id")) {
695 Ok(Some(s)) => Value::Str(s),
696 Ok(None) => Value::Null,
697 Err(e) => return Err(OqxError::eval(e.to_string())),
698 }
699 }
700 (Target::Docs, "$title") => {
701 Self::null_if_absent(self.doc_prop(&col_str(row, "doc_id"), "$title", Some("computed"))?)
702 }
703 (Target::Docs, "$tags") => {
704 Self::null_if_absent(self.doc_prop(&col_str(row, "doc_id"), "$tags", Some("computed"))?)
705 }
706 (Target::Blocks, "$id") => c("block_id"),
707 (Target::Blocks, "$doc") => c("doc_id"),
708 (Target::Blocks, "$path") => c("__path"),
709 (Target::Blocks, "$ordinal") => c("ordinal"),
710 (Target::Blocks, "$depth") => c("depth"),
711 (Target::Blocks, "$body") => c("text"),
712 (Target::Blocks, "$content_hash") => Self::null_if_absent(c("raw_hash")),
713 (Target::Blocks, "$updated_at") => Self::null_if_absent(self.scalar(
714 "SELECT MAX(c.ts) FROM block_changes bc JOIN commits c ON c.commit_id = bc.commit_id WHERE bc.block_id = ?1",
715 &[to_sql(&c("block_id"))],
716 )?),
717 (Target::Nodes, "$id" | "$node_id") => c("node_id"),
718 (Target::Nodes, "$doc_id") => c("doc_id"),
719 (Target::Nodes, "$block_id") => c("block_id"),
720 (Target::Nodes, "$path") => c("__path"),
721 (Target::Edges, "$id") => c("edge_id"),
722 (Target::Edges, "$src") => c("src_doc"),
723 (Target::Edges, "$dst") => c("dst_node"),
724 (Target::Edges, "$src_block") => c("src_block"),
725 (Target::Edges, "$via") => c("via_node"),
726 (Target::Edges, "$from_commit") => c("from_commit"),
727 (Target::Edges, "$path") => c("__path"),
728 (Target::Edges, "$dst_path") => Self::null_if_absent(self.scalar(
729 "SELECT path FROM docs WHERE doc_id = ?1",
730 &[to_sql(&c("dst_node"))],
731 )?),
732 (Target::Edges, "$dst_uri") => Self::null_if_absent(self.scalar(
733 "SELECT uri FROM external_nodes WHERE node_id = ?1",
734 &[to_sql(&c("dst_node"))],
735 )?),
736 _ => Value::Undefined,
737 })
738 }
739
740 fn filter_invalid(msg: String) -> Option<oqx::Result<Value>> {
743 Some(Err(OqxError::eval(msg)))
744 }
745
746 fn require_target(t: Target, want: Target, name: &str) -> Option<oqx::Result<Value>> {
747 (t != want).then(|| {
748 Err(OqxError::eval(format!(
749 "{name}() is only available on the {} target",
750 want.as_str()
751 )))
752 })
753 }
754
755 fn sql_result(r: oqx::Result<bool>) -> oqx::Result<Value> {
756 r.map(Value::Bool)
757 }
758
759 fn row_method(
760 &self,
761 name: &str,
762 row: &Value,
763 t: Target,
764 args: &[Value],
765 ) -> Option<oqx::Result<Value>> {
766 let c = |k: &str| col(row, k).clone();
767 match name {
768 "text" => Some(self.text_match(t, row, &arg_or_empty(args, 0))),
769 "semantic" => Some(self.semantic_score(t, row, &arg_or_empty(args, 0))),
770 "has_anchor" => Self::require_target(t, Target::Blocks, name).or_else(|| {
771 Some(Self::sql_result(self.exists(
772 "SELECT 1 FROM edges WHERE src_block = ?1 AND anchor IS NOT NULL LIMIT 1",
773 &[to_sql(&c("block_id"))],
774 )))
775 }),
776 "child_count" => Self::require_target(t, Target::Blocks, name).or_else(|| {
777 Some(self.scalar(
778 "SELECT COUNT(*) FROM blocks WHERE parent_block = ?1 AND deleted_commit IS NULL",
779 &[to_sql(&c("block_id"))],
780 ))
781 }),
782 "parent_type" => Self::require_target(t, Target::Blocks, name).or_else(|| {
783 Some(
784 self.scalar(
785 "SELECT type FROM blocks WHERE block_id = ?1",
786 &[to_sql(&c("parent_block"))],
787 )
788 .map(Self::null_if_absent),
789 )
790 }),
791 "has_edge" => {
792 let pred = js_string(args.first().unwrap_or(&Value::Undefined));
793 let (src_col, src_val) = if t == Target::Blocks {
794 ("src_block", c("block_id"))
795 } else {
796 ("src_doc", c("doc_id"))
797 };
798 let r = if args.len() >= 2 {
799 self.exists(
800 &format!("SELECT 1 FROM edges WHERE {src_col} = ?1 AND predicate = ?2 AND to_commit IS NULL AND dst_node = ?3 LIMIT 1"),
801 &[to_sql(&src_val), SqlValue::Text(pred), to_sql(&args[1])],
802 )
803 } else {
804 self.exists(
805 &format!("SELECT 1 FROM edges WHERE {src_col} = ?1 AND predicate = ?2 AND to_commit IS NULL LIMIT 1"),
806 &[to_sql(&src_val), SqlValue::Text(pred)],
807 )
808 };
809 Some(Self::sql_result(r))
810 }
811 "under" => Self::require_target(t, Target::Blocks, name).or_else(|| {
812 let target = js_string(args.first().unwrap_or(&Value::Undefined));
813 let ap = col_str(row, "ancestor_path");
814 Some(Ok(Value::Bool(
815 ap.contains(&format!("/{target}/")) || col_str(row, "block_id") == target,
816 )))
817 }),
818 "under_heading" => Self::require_target(t, Target::Blocks, name).or_else(|| {
819 let text = js_string(args.first().unwrap_or(&Value::Undefined));
820 let top = match self.top_ordinal(row) {
821 Ok(v) => to_sql(&v),
822 Err(e) => return Some(Err(e)),
823 };
824 Some(Self::sql_result(self.exists(
825 "SELECT 1 FROM sections s JOIN blocks hb ON hb.block_id = s.heading_block
826 WHERE s.doc_id = ?1 AND lower(hb.text) LIKE '%' || lower(?2) || '%' AND s.first_ordinal <= ?3 AND s.last_ordinal >= ?4 LIMIT 1",
827 &[to_sql(&c("doc_id")), SqlValue::Text(text), top.clone(), top],
828 )))
829 }),
830 "within" => Self::require_target(t, Target::Blocks, name).or_else(|| {
831 let target = js_string(args.first().unwrap_or(&Value::Undefined));
832 if target.starts_with("d_") {
833 return Some(Ok(Value::Bool(col_str(row, "doc_id") == target)));
834 }
835 if target.contains('*') {
836 let like = glob_to_like(&target, false);
837 return Some(Self::sql_result(self.exists(
838 "SELECT 1 WHERE ?1 LIKE ?2 ESCAPE '\\'",
839 &[SqlValue::Text(col_str(row, "__path")), SqlValue::Text(like)],
840 )));
841 }
842 Some(Ok(Value::Bool(col_str(row, "__path") == target)))
843 }),
844 "under_kind" => Self::require_target(t, Target::Blocks, name).or_else(|| {
845 let kind = js_string(args.first().unwrap_or(&Value::Undefined));
846 let ap: Vec<String> = col_str(row, "ancestor_path")
847 .split('/')
848 .filter(|s| !s.is_empty())
849 .map(str::to_owned)
850 .collect();
851 if ap.is_empty() {
852 return Some(Ok(Value::Bool(false)));
853 }
854 let placeholders: Vec<String> = (1..=ap.len()).map(|i| format!("?{i}")).collect();
855 let placeholders = placeholders.join(",");
856 let mut params: Vec<SqlValue> = ap.into_iter().map(SqlValue::Text).collect();
857 let n = params.len();
858 params.push(SqlValue::Text(kind));
859 let r = match args.get(1) {
860 Some(v) if !v.is_absent() => {
861 let nm = js_string(v);
862 params.push(SqlValue::Text(nm.clone()));
863 params.push(SqlValue::Text(nm));
864 self.exists(
865 &format!(
866 "SELECT 1 FROM blocks WHERE block_id IN ({placeholders}) AND type = ?{} AND (lower(text) LIKE '%' || lower(?{}) || '%' OR json_extract(attrs,'$.key') = ?{}) LIMIT 1",
867 n + 1,
868 n + 2,
869 n + 3
870 ),
871 ¶ms,
872 )
873 }
874 _ => self.exists(
875 &format!(
876 "SELECT 1 FROM blocks WHERE block_id IN ({placeholders}) AND type = ?{} LIMIT 1",
877 n + 1
878 ),
879 ¶ms,
880 ),
881 };
882 Some(Self::sql_result(r))
883 }),
884 "yaml_path" => Self::require_target(t, Target::Blocks, name).or_else(|| {
885 Some(Ok(Self::key_path(row, &js_string(args.first().unwrap_or(&Value::Undefined)), "yaml")))
886 }),
887 "json_pointer" => Self::require_target(t, Target::Blocks, name).or_else(|| {
888 Some(Ok(Self::key_path(row, &js_string(args.first().unwrap_or(&Value::Undefined)), "json")))
889 }),
890 _ => None,
891 }
892 }
893
894 fn key_path(row: &Value, path: &str, kind: &str) -> Value {
895 let key = if kind == "json" {
896 let mut p = path;
897 p = p.strip_prefix('#').unwrap_or(p);
898 p = p.strip_prefix('/').unwrap_or(p);
899 p.split('/').collect::<Vec<_>>().join(".")
900 } else {
901 path.to_owned()
902 };
903 let leaf = key.rsplit('.').next().unwrap_or("").to_owned();
904 if !col_str(row, "type").starts_with(&format!("{kind}:")) {
905 return Value::Bool(false);
906 }
907 let k = Self::jattr(row, "key");
908 Value::Bool(k == Value::Str(leaf) || k == Value::Str(key))
909 }
910
911 fn text_match(&self, t: Target, row: &Value, terms: &str) -> oqx::Result<Value> {
912 if t == Target::Edges {
913 return Err(OqxError::eval(
914 "text(...) is not available on the edges target",
915 ));
916 }
917 let m = sanitize_fts_query(terms);
918 if m.is_empty() {
919 return Ok(Value::Bool(false));
920 }
921 let r = match t {
922 Target::Docs => self.exists(
923 "SELECT 1 FROM blocks_fts JOIN blocks b ON b.rowid = blocks_fts.rowid WHERE b.doc_id = ?1 AND blocks_fts MATCH ?2 LIMIT 1",
924 &[to_sql(col(row, "doc_id")), SqlValue::Text(m)],
925 ),
926 Target::Nodes => self.exists(
927 "SELECT 1 FROM nodes_fts WHERE rowid = (SELECT rowid FROM nodes WHERE node_id = ?1) AND nodes_fts MATCH ?2",
928 &[to_sql(col(row, "node_id")), SqlValue::Text(m)],
929 ),
930 _ => self.exists(
931 "SELECT 1 FROM blocks_fts WHERE rowid = (SELECT rowid FROM blocks WHERE block_id = ?1) AND blocks_fts MATCH ?2",
932 &[to_sql(col(row, "block_id")), SqlValue::Text(m)],
933 ),
934 };
935 Self::sql_result(r)
936 }
937
938 fn semantic_score(&self, t: Target, row: &Value, phrase: &str) -> oqx::Result<Value> {
939 if matches!(t, Target::Nodes | Target::Edges) {
940 return Err(OqxError::eval(
941 "semantic(...) is available on the docs and blocks targets",
942 ));
943 }
944 let Some(resolved) = self.semantic.get(phrase) else {
945 return Err(OqxError::eval(format!(
946 "semantic({}) needs an embedding provider; none is configured for this query",
947 serde_json::Value::String(phrase.to_owned())
948 )));
949 };
950 let vec: oqx::Result<Option<Vec<u8>>> = match t {
951 Target::Docs => self
952 .conn
953 .query_row(
954 "SELECT vec FROM doc_embeddings WHERE doc_id = ?1 AND model = ?2",
955 rusqlite::params![col_str(row, "doc_id"), resolved.model],
956 |r| r.get(0),
957 )
958 .optional()
959 .map_err(sql_err),
960 _ => omgbase_store::block_vector(self.conn, &col_str(row, "block_id"), &resolved.model)
963 .map_err(|e| OqxError::eval(e.to_string())),
964 };
965 match vec? {
966 Some(v) => Ok(Value::Number(cosine_bytes(&v, &resolved.vec))),
967 None => Ok(Value::Null),
968 }
969 }
970}
971
972pub(crate) fn fetch_rows(
975 conn: &Connection,
976 sql: &str,
977 params: &[SqlValue],
978) -> rusqlite::Result<Vec<Object>> {
979 let mut stmt = conn.prepare_cached(sql)?;
980 let names: Vec<String> = stmt
981 .column_names()
982 .iter()
983 .map(|s| (*s).to_owned())
984 .collect();
985 let rows = stmt.query_map(params_from_iter(params.iter()), |r| {
986 let mut o = Object::with_capacity(names.len());
987 for (i, name) in names.iter().enumerate() {
988 o.insert(name.as_str(), sql_value(r.get_ref(i)?));
989 }
990 Ok(o)
991 })?;
992 rows.collect()
993}
994
995pub(crate) fn tag_row(mut row: Object, t: Target) -> Value {
998 row.insert(TAG_KEY, Value::Str(t.as_str().to_owned()));
999 Value::Object(row)
1000}
1001
1002pub(crate) fn tag_rows(rows: Vec<Object>, t: Target) -> Vec<Value> {
1004 rows.into_iter().map(|r| tag_row(r, t)).collect()
1005}
1006
1007fn set_nested(out: &mut Object, path: &[&str], leaf: Value) {
1010 let Some((first, rest)) = path.split_first() else {
1011 return;
1012 };
1013 if rest.is_empty() {
1014 out.insert(*first, leaf);
1015 return;
1016 }
1017 let mut child = match out.get(first) {
1018 Some(Value::Object(o)) => o.clone(),
1019 _ => Object::new(),
1020 };
1021 set_nested(&mut child, rest, leaf);
1022 out.insert(*first, Value::Object(child));
1023}
1024
1025#[must_use]
1028pub fn glob_to_like(glob: &str, escape_backslash: bool) -> String {
1029 let mut out = String::with_capacity(glob.len() + 4);
1030 for ch in glob.chars() {
1031 match ch {
1032 '%' | '_' => {
1033 out.push('\\');
1034 out.push(ch);
1035 }
1036 '\\' if escape_backslash => out.push_str("\\\\"),
1037 '*' => out.push('%'),
1038 c => out.push(c),
1039 }
1040 }
1041 out
1042}
1043
1044impl DataContext for StoreContext<'_> {
1045 fn root(&self, name: &str) -> Value {
1046 if let Some(rows) = self.rows_root.as_ref().filter(|_| name == oqx::ROWS_ROOT) {
1047 return Value::Array(rows.clone());
1048 }
1049 if name == "$repo" {
1050 return self.repo_root();
1051 }
1052 let Some(t) = Target::parse(name) else {
1053 return Value::Undefined;
1054 };
1055 match self.root_scan(t) {
1058 Ok(rows) => rows,
1059 Err(e) => {
1060 let mut slot = self.root_failure.borrow_mut();
1061 if slot.is_none() {
1062 *slot = Some(e);
1063 }
1064 Value::Array(Vec::new())
1065 }
1066 }
1067 }
1068
1069 fn get(&self, row: &Value, key: &str) -> oqx::Result<Value> {
1070 if row.is_absent() {
1071 return Ok(Value::Undefined);
1072 }
1073 if key == "$repo" {
1076 return Ok(self.repo_root());
1077 }
1078 if is_repo_root(row) {
1079 if key == "$id" {
1080 return Ok(Value::Str(self.repo_id.clone()));
1081 }
1082 return match Target::parse(key) {
1083 Some(t) => self.root_scan(t),
1084 None => Ok(Value::Undefined),
1085 };
1086 }
1087 let Some(t) = target_of(row) else {
1088 return Ok(oqx::DefaultContext::read(row, key));
1090 };
1091 if key.starts_with('$') {
1092 return self.intrinsic(row, t, key);
1093 }
1094 match (t, key) {
1096 (Target::Docs, "doc") | (Target::Blocks, "block") | (Target::Nodes, "section") => {
1097 return Ok(row.clone());
1098 }
1099 (_, "doc") => return self.owning_doc(row),
1100 (Target::Nodes, "block") => return self.owning_block(row),
1101 _ => {}
1102 }
1103 if let Some(v) = self.relation(row, t, key)? {
1104 return Ok(v);
1105 }
1106 let c = |k: &str| col(row, k).clone();
1107 Ok(match t {
1108 Target::Docs => {
1109 if key == "format" {
1110 return Ok(c("format"));
1111 }
1112 let doc_id = col_str(row, "doc_id");
1113 if key == "frontmatter" || key == "inline" {
1114 return self.doc_prop_bag(&doc_id, key);
1115 }
1116 if RESERVED_DOC_BASENAMES.contains(&key) {
1117 return Err(OqxError::eval(format!(
1121 "bare '{key}' reads a frontmatter key; did you mean the intrinsic ${key}? (use frontmatter.{key} to force the property)"
1122 )));
1123 }
1124 return self.doc_prop(&doc_id, key, None);
1125 }
1126 Target::Blocks => match key {
1127 "type" => c("type"),
1128 "text" => c("text"),
1129 "attrs" => parse_json(&c("attrs")),
1130 _ => Self::jattr(row, key),
1131 },
1132 Target::Nodes => match key {
1133 "kind" => c("kind"),
1134 "name" => c("name"),
1135 "value" => c("value"),
1136 "attrs" => parse_json(&c("attrs")),
1137 _ => Self::jattr(row, key),
1138 },
1139 Target::Edges => match key {
1140 "predicate" | "provenance" | "dst_kind" | "anchor" | "src_field" => c(key),
1141 _ => Value::Undefined,
1142 },
1143 })
1144 }
1145
1146 fn to_rows(&self, value: &Value) -> Vec<Value> {
1147 match value {
1148 Value::Undefined | Value::Null => Vec::new(),
1149 Value::Array(a) => a.clone(),
1150 other => vec![other.clone()],
1151 }
1152 }
1153
1154 fn identity(&self, row: &Value) -> Value {
1155 match target_of(row) {
1156 Some(Target::Docs) => col(row, "doc_id").clone(),
1157 Some(Target::Blocks) => col(row, "block_id").clone(),
1158 Some(Target::Nodes) => col(row, "node_id").clone(),
1159 Some(Target::Edges) => col(row, "edge_id").clone(),
1160 None => row.clone(),
1161 }
1162 }
1163
1164 fn call_function(&self, name: &str, args: &[Value]) -> Option<oqx::Result<Value>> {
1165 if name == "range" {
1166 let x = args.first().unwrap_or(&Value::Undefined);
1167 return Some(Ok(match x {
1168 Value::Range(_) => x.clone(),
1169 Value::Str(s) => match omgbase_properties::detect_range(s) {
1170 Some(r) => {
1171 let b = |b: &Bound| match b {
1172 Bound::Open => Value::Undefined,
1173 Bound::Num(n) => Value::Number(*n),
1174 Bound::Iso(s) => Value::Str(s.clone()),
1175 };
1176 Value::from(make_range(b(&r.lo), b(&r.hi), r.exclusive_end))
1177 }
1178 None => Value::Null,
1179 },
1180 _ => Value::Null,
1181 }));
1182 }
1183 builtin_function(name, args)
1184 }
1185
1186 fn call_method(&self, name: &str, recv: &Value, args: &[Value]) -> Option<oqx::Result<Value>> {
1187 if let Some(t) = target_of(recv) {
1188 if let Some(r) = self.row_method(name, recv, t, args) {
1189 return Some(r);
1190 }
1191 } else if matches!(
1192 name,
1193 "text"
1194 | "semantic"
1195 | "under"
1196 | "under_heading"
1197 | "within"
1198 | "under_kind"
1199 | "yaml_path"
1200 | "json_pointer"
1201 | "has_edge"
1202 | "has_anchor"
1203 | "child_count"
1204 | "parent_type"
1205 ) {
1206 return Self::filter_invalid(format!("{name}() needs a docs/blocks/nodes/edges row"));
1207 }
1208 builtin_method_with(RegexDialect::Oqx, name, recv, args)
1209 }
1210}
1211
1212#[cfg(test)]
1213mod tests {
1214 use super::*;
1215
1216 #[test]
1217 fn glob_to_like_escapes() {
1218 assert_eq!(glob_to_like("a*/b_%", true), "a%/b\\_\\%");
1219 assert_eq!(glob_to_like("a\\b*", true), "a\\\\b%");
1220 assert_eq!(glob_to_like("a\\b*", false), "a\\b%");
1221 }
1222
1223 #[test]
1224 fn nested_property_objects_rebuild() {
1225 let mut o = Object::new();
1226 set_nested(&mut o, &["a", "b"], Value::Number(1.0));
1227 set_nested(&mut o, &["a", "c"], Value::Number(2.0));
1228 set_nested(&mut o, &["d"], Value::Str("x".into()));
1229 let a = o.get("a").unwrap().as_object().unwrap();
1230 assert_eq!(a.get("b"), Some(&Value::Number(1.0)));
1231 assert_eq!(a.get("c"), Some(&Value::Number(2.0)));
1232 assert_eq!(o.get("d"), Some(&Value::Str("x".into())));
1233 set_nested(&mut o, &["d", "e"], Value::Bool(true));
1235 assert!(o.get("d").unwrap().as_object().is_some());
1236 }
1237
1238 #[test]
1239 fn json_and_sql_bridges() {
1240 assert_eq!(
1241 parse_json(&Value::Str("{\"a\":1}".into()))
1242 .as_object()
1243 .unwrap()
1244 .get("a"),
1245 Some(&Value::Number(1.0))
1246 );
1247 assert_eq!(
1248 parse_json(&Value::Str("nope".into())),
1249 Value::Str("nope".into())
1250 );
1251 assert_eq!(parse_json(&Value::Null), Value::Undefined);
1252 assert_eq!(arg_or_empty(&[], 0), "");
1253 assert_eq!(arg_or_empty(&[Value::Number(2.0)], 0), "2");
1254 }
1255}