inillucent_sql/dml.rs
1//! Binding INSERT, UPDATE and DELETE.
2//!
3//! Invariant: a bound DML statement names every value it will write, in table
4//! column order, before anything is compiled. A column the statement did not
5//! mention is not left to be filled in later by whoever runs it - it carries
6//! its `DEFAULT`, or a NULL, as an expression like any other. That is what
7//! makes `INSERT INTO t(b) VALUES(1)` and `INSERT INTO t VALUES(NULL, 1)`
8//! compile to the same shape, and it is why the constraint checks can be
9//! written once against a row image rather than twice against two.
10//!
11//! Constraints are bound here too, out of the `CREATE TABLE` text the file
12//! stores. The catalog keeps them as source, because the catalog sits below
13//! the binder and cannot bind anything; the binder parses that source against
14//! the table it belongs to and gets an ordinary expression back. A CHECK is
15//! therefore evaluated by exactly the machinery that evaluates a WHERE clause,
16//! which is the only way to be sure the two agree about what `x > 0` means
17//! when `x` is text.
18
19use inillucent_base::limits::Limits;
20use inillucent_value::Collation;
21
22use crate::ast::{self, ConflictAction};
23use crate::bind::{
24 no_such_column, refused, unsupported, Binder, BoundExpr, BoundOrderTerm, BoundResultColumn,
25 BoundSelect, BoundSource,
26};
27use crate::catalog_view::{IndexInfo, TableInfo, TableKind, TriggerEventInfo, TriggerInfo};
28use crate::diagnostic::ParseError;
29use crate::lexer::Span;
30use crate::parser::parse_expression;
31
32/// The internal tables an application may write, as SQLite allows.
33///
34/// **Four, and two of them are the schema.** Every table whose
35/// name begins with `sqlite_` used to be refused, which is wrong for all four,
36/// because writing them is the documented way to use them:
37///
38/// - `sqlite_schema`, and `sqlite_master` which is its other name, are what
39/// `PRAGMA writable_schema` is for, and `.dump` emits
40/// `INSERT INTO sqlite_schema(type,name,tbl_name,rootpage,sql)VALUES(...)`
41/// for a virtual table - which is the only way a dump can restore one
42/// without building empty shadow tables over the ones it is about to fill
43/// (task-1979, R2). Whether the pragma is on is the *engine's* question and
44/// not the binder's: `ImportedDatabase::refuse_schema_write` refuses the
45/// statement when it is off, the way `refuse_shadow_write` refuses a write a
46/// defensive connection may not make.
47///
48/// - `sqlite_sequence` holds one row per `AUTOINCREMENT` table, and
49/// `UPDATE sqlite_sequence SET seq = 0 WHERE name = 't'` is how the counter is
50/// reset. `DELETE FROM sqlite_sequence` is how it is reset for every table at
51/// once. Refusing them left no way at all to do either.
52/// - `sqlite_stat1` is what `ANALYZE` writes, and `.dump` emits
53/// `INSERT INTO sqlite_stat1 VALUES(...)` for it - so a dump this engine
54/// produced could not be replayed into it.
55///
56/// They are ordinary tables in every other respect: the rows are what they are,
57/// and a value written into one is used exactly as `ANALYZE` or the rowid
58/// allocator would have used the one it replaced.
59const WRITABLE_INTERNAL: [&[u8]; 4] = [
60 b"sqlite_sequence",
61 b"sqlite_stat1",
62 b"sqlite_schema",
63 b"sqlite_master",
64];
65
66/// Where one column's value comes from in an INSERT.
67#[derive(Clone, Debug, PartialEq)]
68pub enum ColumnSource {
69 /// The value at this position of the source row.
70 Row(usize),
71 /// An expression evaluated once per row, which is what a `DEFAULT` is.
72 Expr(BoundExpr),
73 /// A generated column, computed from the rest of the row rather than from
74 /// anything the statement supplied.
75 ///
76 /// It is its own variant because it is evaluated at a different *time*: a
77 /// `DEFAULT` is a value like any other, while a generated column reads the
78 /// row it is part of and so cannot be computed until the rest of it is.
79 Generated(BoundExpr),
80}
81
82/// What an INSERT inserts.
83#[derive(Clone, Debug, PartialEq)]
84pub enum BoundInsertSource {
85 /// Literal rows, each already bound.
86 Values(Vec<Vec<BoundExpr>>),
87 /// A query, whose result columns feed the target columns in order.
88 Select(Box<BoundSelect>),
89}
90
91/// One `CHECK` constraint, bound against its table.
92#[derive(Clone, Debug, PartialEq)]
93pub struct BoundCheck {
94 /// The constraint's name, when it was written with one.
95 pub name: Option<Vec<u8>>,
96 /// The predicate.
97 pub expr: BoundExpr,
98}
99
100/// A `NOT NULL` column's `DEFAULT`, bound so a `REPLACE` can stand it in.
101///
102/// **REPLACE's rule for a `NOT NULL` violation is to substitute the column's
103/// default, and to fall back to `ABORT` only when there is no default.** So
104/// `UPDATE OR REPLACE t SET c = NULL` on `c TEXT NOT NULL DEFAULT 'd'` stores
105/// `'d'`, and this engine used to refuse the statement instead.
106///
107/// The write path cannot bind one for itself: a default is schema text, and by
108/// the time a row is being checked the parser is long out of scope. The binder
109/// already binds one for every column a statement *omits*; these are the same
110/// expressions bound for the columns it supplies, which is where a NULL that
111/// needs replacing can come from.
112///
113/// Only the columns that can need it are here - `NOT NULL` and with a default -
114/// so an ordinary table carries an empty vector and the write path skips the
115/// whole apparatus.
116#[derive(Clone, Debug, PartialEq)]
117pub struct BoundDefault {
118 /// The column the default belongs to.
119 pub column: u16,
120 /// The default expression, bound.
121 pub expr: BoundExpr,
122}
123
124/// The expressions one index needs evaluated per row to be maintained.
125///
126/// **An index is usually just columns of the row, and then it needs none of
127/// this.** A partial index holds only the rows its predicate accepts, and an
128/// index on an expression holds a value no column carries - so for those two,
129/// maintaining the index means evaluating something per row rather than
130/// copying a slot. They travel on the bound statement for the same reason the
131/// table's `CHECK` predicates do: the binder is what can turn schema text into
132/// a `BoundExpr`, and the write path is what runs it.
133///
134/// The list holds only the indexes that need it, so a table with neither kind
135/// leaves it empty and the write path's loop runs zero times - which is every
136/// table the gate measures.
137#[derive(Clone, Debug, PartialEq)]
138pub struct BoundIndexExprs {
139 /// The index's position in the table's `indexes`.
140 pub position: usize,
141 /// The partial-index predicate, when it has one.
142 pub predicate: Option<BoundExpr>,
143 /// One per key column: the expression it indexes, or `None` for a column.
144 pub keys: Vec<Option<BoundExpr>>,
145}
146
147/// One statement of a trigger body, bound.
148///
149/// The four the grammar allows and no more. A trigger body is not a general
150/// statement list: it cannot create objects, cannot open transactions, and
151/// cannot return rows to the caller, so a variant for anything else would be a
152/// shape the binder is required to refuse.
153#[derive(Clone, Debug, PartialEq)]
154pub enum BoundTriggerStatement {
155 /// `INSERT`.
156 Insert(Box<BoundInsert>),
157 /// `UPDATE`.
158 Update(Box<BoundUpdate>),
159 /// `DELETE`.
160 Delete(Box<BoundDelete>),
161 /// `SELECT`, which a body runs for its side effects - in practice for the
162 /// `RAISE()` inside it.
163 Select(Box<BoundSelect>),
164}
165
166/// A trigger, bound against the write that fires it.
167///
168/// It is bound per statement rather than once per schema because the body's
169/// FROM terms take statement-wide source numbers, and those only exist relative
170/// to the statement they are inlined into.
171#[derive(Clone, Debug, PartialEq)]
172pub struct BoundTrigger {
173 /// The trigger's name, for the diagnostic when its body fails.
174 pub name: Vec<u8>,
175 /// The folded name of the table it is attached to.
176 ///
177 /// Read by the executor to decide whether a body statement is writing the
178 /// trigger's *own* table, which is what `PRAGMA recursive_triggers` is
179 /// about: with it on, such a write fires this trigger again.
180 pub table: Vec<u8>,
181 /// Whether it fires before or after the row is written.
182 pub time: ast::TriggerTime,
183 /// The `WHEN` guard, when one was written.
184 pub when: Option<BoundExpr>,
185 /// The body statements, in written order.
186 pub body: Vec<BoundTriggerStatement>,
187 /// Whether the binder synthesised this from a `REFERENCES` clause rather
188 /// than reading it from a `CREATE TRIGGER`.
189 ///
190 /// **Read by `DROP TABLE` (task-1979, F6).** Dropping a table with foreign
191 /// keys on runs an implicit `DELETE FROM` first, so the keys that reference
192 /// it are enforced - and SQLite's rule is that the implicit delete fires no
193 /// triggers of its own while still performing every foreign key action. A
194 /// delete bound for that purpose keeps the triggers this flag marks and
195 /// drops the rest.
196 pub foreign_key: bool,
197 /// Whether the foreign key this enforces has one table as both its child
198 /// and its parent.
199 ///
200 /// **Also read by `DROP TABLE` (task-1979, F6).** The implicit delete keeps
201 /// the foreign key triggers and drops this one, because emptying a table
202 /// cannot leave a row of that same table pointing at nothing - see
203 /// `ForeignKeyTrigger::self_referencing`, which is where the value comes
204 /// from. Always false on a trigger the schema wrote.
205 pub self_referencing: bool,
206}
207
208/// A bound `INSERT`.
209#[derive(Clone, Debug, PartialEq)]
210pub struct BoundInsert {
211 /// The table being written.
212 pub table: TableInfo,
213 /// The statement-wide number of the FROM term being written.
214 ///
215 /// It used to be implicitly zero, because a DML statement had exactly one
216 /// source. A trigger body is compiled into the statement that fires it, so
217 /// its target takes the next number after the firing statement's - and a
218 /// compiler that assumed zero read the wrong cursor for every fire after
219 /// the first.
220 pub target_source: usize,
221 /// Where each table column's value comes from, in column order.
222 pub columns: Vec<ColumnSource>,
223 /// Where the rowid comes from, when the statement supplies one.
224 pub rowid: Option<ColumnSource>,
225 /// Which value of the supplied row is the rowid, when the statement named
226 /// it outright.
227 ///
228 /// `INSERT INTO t(rowid, a) VALUES (7, 'x')` is legal on any rowid table,
229 /// including one with no `INTEGER PRIMARY KEY` to alias it and including a
230 /// virtual table. It is recorded separately from `rowid` because it is not
231 /// a column: nothing writes it into the record.
232 pub named_rowid: Option<usize>,
233 /// The rows.
234 pub source: BoundInsertSource,
235 /// How many values each source row supplies.
236 pub arity: usize,
237 /// The statement's conflict algorithm, when it wrote one.
238 pub on_conflict: Option<ConflictAction>,
239 /// The table's `CHECK` constraints.
240 pub checks: Vec<BoundCheck>,
241 /// The `DEFAULT`s a `REPLACE` may stand in for a NULL, by column.
242 pub not_null_defaults: Vec<BoundDefault>,
243 /// The expressions the table's partial and expression indexes need.
244 pub index_exprs: Vec<BoundIndexExprs>,
245 /// The `ON CONFLICT ... DO UPDATE` clause, when there is one.
246 pub upsert: Vec<BoundUpsert>,
247 /// `sqlite_sequence`'s root page, when the target is `AUTOINCREMENT`.
248 ///
249 /// Resolved here rather than in the compiler because it is a fact about the
250 /// catalog, and the catalog is what the binder holds. It is zero for every
251 /// other table, which is also what it reads as before the first
252 /// `AUTOINCREMENT` table in a database is created.
253 pub sequence_root: u32,
254 /// The `RETURNING` columns.
255 pub returning: Vec<BoundResultColumn>,
256 /// The triggers this write fires, in schema order.
257 pub triggers: Vec<BoundTrigger>,
258 /// The foreign-key actions a `REPLACE` fires for the row it removes.
259 ///
260 /// A `REPLACE` that deletes a row to make room for another is a delete,
261 /// and the keys pointing at that row have to be told. Written `DELETE`
262 /// triggers are *not* fired - that is SQLite's rule with its default
263 /// `recursive_triggers = off` - so these are only the ones a key implies.
264 pub replace_triggers: Vec<BoundTrigger>,
265}
266
267/// A bound `ON CONFLICT ... DO UPDATE` clause.
268#[derive(Clone, Debug, PartialEq)]
269pub struct BoundUpsert {
270 /// The conflict target columns, when written; empty means any constraint.
271 ///
272 /// Sorted, because a conflict target names a *set* of columns and
273 /// `ON CONFLICT(a,b)` and `ON CONFLICT(b,a)` name the same one. Matching
274 /// them against an index's columns is a set comparison, and sorting here
275 /// is what makes it one comparison rather than a search per column.
276 pub target: Vec<u16>,
277 /// The assignments, or empty for `DO NOTHING`.
278 pub assignments: Vec<BoundAssignment>,
279 /// Whether the action is `DO UPDATE`.
280 pub do_update: bool,
281 /// The `WHERE` on the `DO UPDATE`.
282 pub filter: Option<BoundExpr>,
283}
284
285/// One `SET` assignment.
286#[derive(Clone, Debug, PartialEq)]
287pub struct BoundAssignment {
288 /// The column being assigned, as a declared position.
289 pub column: u16,
290 /// Whether the assignment names the row's own rowid rather than a declared
291 /// column, in which case `column` says nothing.
292 ///
293 /// **`UPDATE t SET rowid = 100` was `no such column: rowid` (task-1979,
294 /// F9).** An assignment target was looked up with `column_position`, which
295 /// only knows the columns the table declares, and a table with no INTEGER
296 /// PRIMARY KEY declares none for its rowid. SQLite accepts all three
297 /// spellings of the rowid on either kind of table and moves the row to the
298 /// new key.
299 pub rowid: bool,
300 /// The new value.
301 pub value: BoundExpr,
302}
303
304/// A bound `UPDATE`.
305#[derive(Clone, Debug, PartialEq)]
306pub struct BoundUpdate {
307 /// The table being written.
308 pub table: TableInfo,
309 /// The statement-wide number of the FROM term being written.
310 ///
311 /// It used to be implicitly zero, because a DML statement had exactly one
312 /// source. A trigger body is compiled into the statement that fires it, so
313 /// its target takes the next number after the firing statement's - and a
314 /// compiler that assumed zero read the wrong cursor for every fire after
315 /// the first.
316 pub source: usize,
317 /// The extra FROM terms of an `UPDATE ... FROM`, in written order.
318 ///
319 /// **The rows being updated come from a join.** `UPDATE t SET v = s.v FROM s
320 /// WHERE s.a = t.a` is the shape a migration writes to copy a column across
321 /// tables, and the values it assigns are not expressions over the target
322 /// row: they read a *different* row, one the join found. So the query that
323 /// finds the keys carries these terms too, and projects the assigned values
324 /// beside the key; see [`BoundUpdate::from`], which is this field.
325 ///
326 /// Empty for every ordinary `UPDATE`, which is what keeps the wider row off
327 /// the path the gate's `txn.large` measures.
328 pub from: Vec<crate::bind::BoundSource>,
329 /// The assignments, in table column order with duplicates already refused.
330 pub assignments: Vec<BoundAssignment>,
331 /// The `STORED` generated columns, recomputed after the assignments.
332 ///
333 /// **A stored generated column is part of the row, so a row that is
334 /// rewritten rewrites it (task-1913).** It is never named in a `SET`, so
335 /// an `UPDATE` used to leave whatever was written when the row was
336 /// inserted: `c GENERATED ALWAYS AS (a + 1) STORED` still read 2 after
337 /// `UPDATE g SET a = 5`, where SQLite reads 6. The wrong value is on the
338 /// disk rather than in an answer, so a later read of the same file is
339 /// wrong too, and an index on the column indexes the stale value.
340 ///
341 /// A `VIRTUAL` column is not here: it has no slot in the record and is
342 /// computed when it is read, which is why only this half needed fixing.
343 ///
344 /// These are evaluated against the row *after* the assignments, which is
345 /// the one difference from [`BoundUpdate::assignments`] - those read the
346 /// before image so `SET a = b, b = a` swaps.
347 pub generated: Vec<BoundAssignment>,
348 /// The `WHERE` clause.
349 pub filter: Option<BoundExpr>,
350 /// The statement's conflict algorithm, when it wrote one.
351 pub on_conflict: Option<ConflictAction>,
352 /// The table's `CHECK` constraints.
353 pub checks: Vec<BoundCheck>,
354 /// The `DEFAULT`s a `REPLACE` may stand in for a NULL, by column.
355 pub not_null_defaults: Vec<BoundDefault>,
356 /// The expressions the table's partial and expression indexes need.
357 pub index_exprs: Vec<BoundIndexExprs>,
358 /// `INDEXED BY` or `NOT INDEXED` on the target, which the query that finds
359 /// the rows to change obeys; `inillucent_exec::dml::hint_target` puts it there.
360 pub index_hint: crate::bind::IndexChoice,
361 /// The `RETURNING` columns.
362 pub returning: Vec<BoundResultColumn>,
363 /// The `ORDER BY` that decides which rows a `LIMIT` keeps.
364 ///
365 /// Empty unless the statement wrote one, and then always with a `LIMIT`,
366 /// because the binder refuses an order with nothing to limit. It goes onto
367 /// the query that finds the rows to change, which is where SQLite puts it
368 /// too: a limited write is `WHERE rowid IN (SELECT rowid ... ORDER BY ...
369 /// LIMIT ...)` there.
370 pub order_by: Vec<BoundOrderTerm>,
371 /// The `LIMIT`.
372 pub limit: Option<BoundExpr>,
373 /// The `OFFSET`.
374 pub offset: Option<BoundExpr>,
375 /// The triggers this write fires, in schema order.
376 pub triggers: Vec<BoundTrigger>,
377 /// The rows to fire an `INSTEAD OF` trigger for, when the target is a view.
378 ///
379 /// A view has no rows of its own, so `OLD` has to come from running the
380 /// view. This is that query, with the statement's `WHERE` on it and one
381 /// result column per view column.
382 pub view_rows: Option<Box<BoundSelect>>,
383}
384
385/// A bound `DELETE`.
386#[derive(Clone, Debug, PartialEq)]
387pub struct BoundDelete {
388 /// The table being written.
389 pub table: TableInfo,
390 /// The expressions the table's partial and expression indexes need.
391 ///
392 /// A delete needs them too: an entry only comes out of a partial index if
393 /// the row was in it, and a key the index computed has to be recomputed to
394 /// be found.
395 pub index_exprs: Vec<BoundIndexExprs>,
396 /// `INDEXED BY` or `NOT INDEXED` on the target, as on [`BoundUpdate`].
397 pub index_hint: crate::bind::IndexChoice,
398 /// The statement-wide number of the FROM term being written.
399 ///
400 /// It used to be implicitly zero, because a DML statement had exactly one
401 /// source. A trigger body is compiled into the statement that fires it, so
402 /// its target takes the next number after the firing statement's - and a
403 /// compiler that assumed zero read the wrong cursor for every fire after
404 /// the first.
405 pub source: usize,
406 /// The `WHERE` clause.
407 pub filter: Option<BoundExpr>,
408 /// The `RETURNING` columns.
409 pub returning: Vec<BoundResultColumn>,
410 /// The `ORDER BY` that decides which rows a `LIMIT` keeps.
411 ///
412 /// Empty unless the statement wrote one, and then always with a `LIMIT`,
413 /// because the binder refuses an order with nothing to limit. It goes onto
414 /// the query that finds the rows to change, which is where SQLite puts it
415 /// too: a limited write is `WHERE rowid IN (SELECT rowid ... ORDER BY ...
416 /// LIMIT ...)` there.
417 pub order_by: Vec<BoundOrderTerm>,
418 /// The `LIMIT`.
419 pub limit: Option<BoundExpr>,
420 /// The `OFFSET`.
421 pub offset: Option<BoundExpr>,
422 /// The triggers this write fires, in schema order.
423 pub triggers: Vec<BoundTrigger>,
424 /// The rows to fire an `INSTEAD OF` trigger for, when the target is a view.
425 pub view_rows: Option<Box<BoundSelect>>,
426}
427
428/// Reports whether an `INSERT` can resolve a conflict by deleting a row.
429///
430/// Either the statement said so, or one of the table's own constraints did.
431/// It is asked before the delete's keys are bound, because binding them costs
432/// a parse and a bind each and the answer is no for almost every insert.
433fn can_replace(table: &TableInfo, statement: Option<ConflictAction>) -> bool {
434 if statement == Some(ConflictAction::Replace) {
435 return true;
436 }
437 table
438 .indexes
439 .iter()
440 .any(|index| index.conflict == Some(ConflictAction::Replace))
441 || table.columns.iter().any(|column| {
442 column.not_null_conflict == Some(ConflictAction::Replace)
443 || column.primary_key_conflict == Some(ConflictAction::Replace)
444 })
445}
446
447/// Reports whether an unusable key's fault is one this write has to report.
448///
449/// A child's write reports a missing parent; a parent's write reports a
450/// mismatch. A statement that touches neither side of the broken key does not
451/// have to care, which is why the fault is carried rather than raised when the
452/// schema was read.
453fn fault_applies(
454 planned: &crate::catalog_view::ForeignKeyTrigger,
455 event: &TriggerEventInfo,
456) -> bool {
457 match event {
458 TriggerEventInfo::Insert => planned.is_check,
459 TriggerEventInfo::Delete => !planned.is_check,
460 TriggerEventInfo::Update(_) => true,
461 }
462}
463
464/// Marks a synthesised body's aborts as the foreign key's rather than a
465/// trigger's.
466///
467/// The generated text says `RAISE(ABORT, ...)` because that is what a person
468/// would have written, and what a person writes reports
469/// `SQLITE_CONSTRAINT_TRIGGER`. A foreign key reports its own code, and the
470/// only difference between the two is which constraint asked - so it is set
471/// here, on the bodies this binder generated, and nowhere else.
472fn report_as_foreign_key(trigger: &mut BoundTrigger) {
473 trigger.foreign_key = true;
474 for statement in &mut trigger.body {
475 let BoundTriggerStatement::Select(select) = statement else {
476 continue;
477 };
478 for column in &mut select.columns {
479 if let BoundExpr::Raise { foreign_key, .. } = &mut column.expr {
480 *foreign_key = true;
481 }
482 }
483 }
484}
485
486/// Returns whether a view has an `INSTEAD OF` trigger for one event.
487fn has_instead_of(table: &TableInfo, event: &TriggerEventInfo) -> bool {
488 table
489 .triggers
490 .iter()
491 .any(|trigger| trigger.time == ast::TriggerTime::InsteadOf && trigger.fires_for(event, &[]))
492}
493
494/// The target position that stands for the rowid rather than a column.
495///
496/// A table cannot have this many columns - SQLite's limit is two thousand - so
497/// there is no position it can collide with, and one sentinel is cheaper than
498/// a parallel `Option` threaded through every target list.
499const ROWID_TARGET: u16 = u16::MAX;
500
501/// Returns whether a name is one of the rowid's three spellings.
502fn is_rowid_name(folded: &[u8]) -> bool {
503 matches!(folded, b"rowid" | b"oid" | b"_rowid_")
504}
505
506/// How deep one write may drive triggers firing other triggers.
507///
508/// SQLite's own limit is `SQLITE_MAX_TRIGGER_DEPTH`, enforced when the frame is
509/// pushed. Trigger bodies are inlined here rather than run as frames, so the
510/// same limit is enforced where the inlining happens - and it has to be, or a
511/// schema in which two triggers write each other's tables would compile until
512/// the compiler ran out of memory.
513///
514/// **This is one number now, and it is the one `.limit` reports.** There used
515/// to be two constants of this name: this one at 32, which was the number
516/// actually enforced, and `inillucent-exec`'s at 1000, checked at run time over
517/// a tree the binder had already capped at 32 - so that check could never fire.
518/// `crates/inillucent-base/manifests/limits.toml` advertised 1000 and
519/// `inillucent diagnose` printed 1000, and a chain of forty distinct triggers
520/// that the oracle ran was refused here (task-1946, H3). The binder reads
521/// `Limit::TriggerDepth` from the connection now, which `.limit trigger_depth`
522/// and the driver both set; this constant is what a binder built without limits
523/// falls back to, and it is the manifest's default.
524pub const MAX_TRIGGER_DEPTH: usize = 1000;
525
526/// How deep one chain of foreign-key actions may go.
527///
528/// A cascade reaches this only when the keys form a cycle, which in practice
529/// means a table whose parent column points at itself. SQLite's own limit is a
530/// run-time recursion depth; this one is a compile-time inlining depth, and it
531/// is smaller for that reason.
532pub const MAX_FOREIGN_KEY_DEPTH: usize = 64;
533
534/// How many foreign-key action bodies one statement may inline in total.
535///
536/// The depth limit alone is not enough: a table with three keys that all cycle
537/// would inline three bodies per level, so the limit that matters is the total.
538/// A chain, which is what a self-referencing tree produces, spends one per
539/// level and reaches the depth limit first.
540pub const MAX_FOREIGN_KEY_STATEMENTS: usize = 256;
541
542impl<'a> Binder<'a> {
543 /// Binds an `INSERT` or `REPLACE`.
544 pub fn bind_insert(&mut self, insert: &ast::Insert) -> Result<BoundInsert, ParseError> {
545 // **A `WITH` on a DML statement is the same `WITH` a `SELECT` has.** The
546 // CTEs are in scope for the whole statement - the source query of an
547 // `INSERT`, the `WHERE` of an `UPDATE` or `DELETE` - and the binder's
548 // CTE stack already handles nesting, so pushing them here is all it
549 // takes. They were refused rather than bound, which is what a migration
550 // script written for SQLite hits first.
551 let pushed = self.push_ctes(&insert.with)?;
552 let bound = self.bind_insert_body(insert);
553 if pushed {
554 self.pop_ctes();
555 }
556 bound
557 }
558
559 /// Binds an `INSERT` with its CTEs already in scope.
560 fn bind_insert_body(&mut self, insert: &ast::Insert) -> Result<BoundInsert, ParseError> {
561 let table = self.writable_target(
562 insert.database,
563 insert.table,
564 Span::default(),
565 &TriggerEventInfo::Insert,
566 )?;
567 let alias = match insert.alias {
568 Some(alias) => self.ast.text(alias).to_vec(),
569 None => table.name.clone(),
570 };
571 let target_source = self.push_write_source(table.clone(), alias);
572 // `DEFAULT VALUES` supplies nothing, so every column takes its default
573 // - which is what an empty target list means here. The grammar does
574 // not allow a column list with it, so there is none to honour.
575 let targets = match insert.source {
576 ast::InsertSource::DefaultValues => Vec::new(),
577 ast::InsertSource::Select(_) => self.insert_targets(&table, &insert.columns)?,
578 };
579 let (source, arity) = self.bind_insert_source(&insert.source, &table, &targets)?;
580 if arity != targets.len() {
581 return Err(refused(
582 format!("{} values for {} columns", arity, targets.len()),
583 Span::default(),
584 ));
585 }
586 let (columns, rowid) = self.column_sources(&table, &targets)?;
587 let named_rowid = targets.iter().position(|target| *target == ROWID_TARGET);
588 let checks = self.bind_checks(&table)?;
589 let not_null_defaults = self.bind_not_null_defaults(&table)?;
590 let index_exprs = self.bind_index_exprs(&table)?;
591 let upsert = self.bind_upsert(&table, insert)?;
592 let returning = self.bind_returning(&insert.returning)?;
593 let mut triggers = self.bind_triggers(&table, TriggerEventInfo::Insert, &[])?;
594 triggers.extend(self.bind_foreign_keys(&table, TriggerEventInfo::Insert, &[])?);
595 let replace_triggers = if can_replace(&table, insert.on_conflict) {
596 self.bind_foreign_keys(&table, TriggerEventInfo::Delete, &[])?
597 } else {
598 Vec::new()
599 };
600 let sequence_root = if table.autoincrement {
601 self.catalog
602 .find_table(None, b"sqlite_sequence")
603 .map_or(0, |sequence| sequence.root)
604 } else {
605 0
606 };
607 Ok(BoundInsert {
608 table,
609 index_exprs,
610 target_source,
611 columns,
612 rowid,
613 named_rowid,
614 source,
615 arity,
616 on_conflict: insert.on_conflict,
617 checks,
618 not_null_defaults,
619 upsert,
620 sequence_root,
621 returning,
622 triggers,
623 replace_triggers,
624 })
625 }
626
627 /// Binds an `UPDATE`.
628 pub fn bind_update(&mut self, update: &ast::Update) -> Result<BoundUpdate, ParseError> {
629 let pushed = self.push_ctes(&update.with)?;
630 let bound = self.bind_update_body(update);
631 if pushed {
632 self.pop_ctes();
633 }
634 bound
635 }
636
637 /// Binds an `UPDATE` with its CTEs already in scope.
638 fn bind_update_body(&mut self, update: &ast::Update) -> Result<BoundUpdate, ParseError> {
639 if let Some(refusal) = order_without_limit(update.limited_at, update.limit, "UPDATE") {
640 return Err(refusal);
641 }
642 let (table, source) =
643 self.write_target_from_term(update.target, &TriggerEventInfo::Update(Vec::new()))?;
644 // **The `FROM` terms are bound after the target**, so the target keeps
645 // the lowest source number and every reference to an unqualified column
646 // resolves to it first - which is SQLite's rule and the reason
647 // `UPDATE t SET v = v + 1 FROM s` means the target's `v`.
648 let before = self.sources.len();
649 for term in &update.from {
650 self.bind_from_term(*term)?;
651 }
652 let joined: Vec<crate::bind::BoundSource> =
653 self.sources.get(before..).unwrap_or(&[]).to_vec();
654 let mut assignments = Vec::new();
655 for (names, value) in &update.assignments {
656 let bound = self.bind_expr(*value)?;
657 for name in names {
658 let folded = self.ast.folded(*name).to_vec();
659 // `rowid`, `oid` and `_rowid_` name the row's key rather than a
660 // declared column, unless the table declares a column by one of
661 // those names - which is what `is_rowid_name` decides.
662 if table.is_rowid_name(&folded) {
663 if assignments.iter().any(|held: &BoundAssignment| held.rowid) {
664 return Err(refused(
665 format!(
666 "column {} is assigned twice",
667 String::from_utf8_lossy(self.ast.text(*name))
668 ),
669 Span::default(),
670 ));
671 }
672 assignments.push(BoundAssignment {
673 column: 0,
674 rowid: true,
675 value: bound.clone(),
676 });
677 continue;
678 }
679 let Some(position) = table.column_position(&folded) else {
680 return Err(no_such_column(self.ast.text(*name), Span::default()));
681 };
682 // **An assignment to a generated column is refused, not
683 // ignored (task-1913).** SQLite answers `cannot UPDATE
684 // generated column "c"`; this accepted the statement, reported
685 // it as a success, and wrote nothing the caller asked for -
686 // either the record took the value and the column stopped
687 // agreeing with its own expression, or the recompute above put
688 // it back and the assignment was silently dropped. `INSERT`
689 // already refused the same thing.
690 self.refuse_generated(&table, position, "UPDATE", Span::default())?;
691 if assignments
692 .iter()
693 .any(|existing: &BoundAssignment| existing.column == position)
694 {
695 return Err(refused(
696 format!(
697 "column {} is assigned twice",
698 String::from_utf8_lossy(self.ast.text(*name))
699 ),
700 Span::default(),
701 ));
702 }
703 assignments.push(BoundAssignment {
704 column: position,
705 rowid: false,
706 value: bound.clone(),
707 });
708 }
709 }
710 // The rowid assignment sorts with the declared columns rather than
711 // ahead of them, because `column` says nothing for it and the order
712 // only has to be stable.
713 assignments.sort_by_key(|assignment| (assignment.rowid, assignment.column));
714 let filter = match update.filter {
715 Some(expr) => Some(self.bind_expr(expr)?),
716 None => None,
717 };
718 let generated = self.bind_stored_generated(&table)?;
719 let checks = self.bind_checks(&table)?;
720 let not_null_defaults = self.bind_not_null_defaults(&table)?;
721 let index_exprs = self.bind_index_exprs(&table)?;
722 let returning = self.bind_returning(&update.returning)?;
723 // Bound as expressions, the way an aggregate's own `ORDER BY` is: a
724 // write has no result columns, so a bare integer names no ordinal.
725 let order_by = self.bind_aggregate_order(&update.order_by)?;
726 let limit = match update.limit {
727 Some(expr) => Some(self.bind_expr(expr)?),
728 None => None,
729 };
730 let offset = match update.offset {
731 Some(expr) => Some(self.bind_expr(expr)?),
732 None => None,
733 };
734 // The rowid is not a declared column, so no `UPDATE OF` trigger and no
735 // foreign key can be keyed on it and it contributes no name here.
736 let changed: Vec<Vec<u8>> = assignments
737 .iter()
738 .filter(|assignment| !assignment.rowid)
739 .filter_map(|assignment| table.column(assignment.column))
740 .map(|column| column.folded.clone())
741 .collect();
742 let mut triggers =
743 self.bind_triggers(&table, TriggerEventInfo::Update(Vec::new()), &changed)?;
744 triggers.extend(self.bind_foreign_keys(
745 &table,
746 TriggerEventInfo::Update(Vec::new()),
747 &changed,
748 )?);
749 let view_rows = self
750 .view_rows(&table, filter.clone())
751 .map(|rows| limit_view_rows(rows, &order_by, &limit, &offset));
752 let index_hint = self.write_hint(source, &index_exprs, filter.as_ref(), &joined)?;
753 Ok(BoundUpdate {
754 table,
755 index_exprs,
756 index_hint,
757 source,
758 from: joined,
759 assignments,
760 generated,
761 filter,
762 on_conflict: update.on_conflict,
763 checks,
764 not_null_defaults,
765 returning,
766 order_by,
767 limit,
768 offset,
769 triggers,
770 view_rows,
771 })
772 }
773
774 /// Binds a `DELETE`.
775 pub fn bind_delete(&mut self, delete: &ast::Delete) -> Result<BoundDelete, ParseError> {
776 let pushed = self.push_ctes(&delete.with)?;
777 let bound = self.bind_delete_body(delete);
778 if pushed {
779 self.pop_ctes();
780 }
781 bound
782 }
783
784 /// Binds a `DELETE` with its CTEs already in scope.
785 fn bind_delete_body(&mut self, delete: &ast::Delete) -> Result<BoundDelete, ParseError> {
786 if let Some(refusal) = order_without_limit(delete.limited_at, delete.limit, "DELETE") {
787 return Err(refusal);
788 }
789 let (table, source) =
790 self.write_target_from_term(delete.target, &TriggerEventInfo::Delete)?;
791 let index_exprs = self.bind_index_exprs(&table)?;
792 let filter = match delete.filter {
793 Some(expr) => Some(self.bind_expr(expr)?),
794 None => None,
795 };
796 let returning = self.bind_returning(&delete.returning)?;
797 let order_by = self.bind_aggregate_order(&delete.order_by)?;
798 let limit = match delete.limit {
799 Some(expr) => Some(self.bind_expr(expr)?),
800 None => None,
801 };
802 let offset = match delete.offset {
803 Some(expr) => Some(self.bind_expr(expr)?),
804 None => None,
805 };
806 let mut triggers = self.bind_triggers(&table, TriggerEventInfo::Delete, &[])?;
807 triggers.extend(self.bind_foreign_keys(&table, TriggerEventInfo::Delete, &[])?);
808 let view_rows = self
809 .view_rows(&table, filter.clone())
810 .map(|rows| limit_view_rows(rows, &order_by, &limit, &offset));
811 let index_hint = self.write_hint(source, &index_exprs, filter.as_ref(), &[])?;
812 Ok(BoundDelete {
813 table,
814 index_exprs,
815 index_hint,
816 source,
817 filter,
818 returning,
819 order_by,
820 limit,
821 offset,
822 triggers,
823 view_rows,
824 })
825 }
826
827 /// Binds the triggers one write fires, bodies and all.
828 ///
829 /// The bodies are bound here, into the same binder, so their FROM terms take
830 /// statement-wide source numbers alongside the write's own. That is what
831 /// lets the compiler inline them: a trigger body is not a separate program
832 /// with a separate cursor space, it is more of this statement.
833 ///
834 /// A trigger already being bound is skipped rather than bound again, which
835 /// is SQLite's behaviour with its default `recursive_triggers = off` and is
836 /// also the only reason inlining terminates.
837 ///
838 /// **Walked newest first.** `live.triggers` is in the order
839 /// `inillucent_catalog::paged::tables_from_entries` appended them while
840 /// reading `sqlite_schema` - the order the triggers were created in - and
841 /// SQLite fires two triggers of the same timing and event in the opposite
842 /// order: it keeps each table's trigger list with the most recently
843 /// created one first, so that one fires first.
844 /// `dml_differential.rs`'s `row_triggers_match_sqlite` has two `AFTER
845 /// INSERT` triggers on one table - `t_ai`, created first, and `t_high`,
846 /// created after it - and the pinned reference fires `t_high` before
847 /// `t_ai` on every insert. Reversing the walk here, once, at the one place
848 /// that reads `live.triggers` into a statement's own trigger list, is
849 /// enough: nothing downstream reorders it again.
850 fn bind_triggers(
851 &mut self,
852 table: &TableInfo,
853 event: TriggerEventInfo,
854 changed: &[Vec<u8>],
855 ) -> Result<Vec<BoundTrigger>, ParseError> {
856 // The catalog reference is copied out of `self` first: the trigger's
857 // arena has to outlive the binder for the body to be bound in place,
858 // and a borrow taken through `&self` would end at the first `&mut self`.
859 let catalog = self.catalog;
860 let database = catalog.database_name(table.database).to_vec();
861 let Some(live) = catalog.find_table(Some(database.as_slice()), &table.folded) else {
862 return Ok(Vec::new());
863 };
864 let (old, new) = match event {
865 TriggerEventInfo::Insert => (false, true),
866 TriggerEventInfo::Delete => (true, false),
867 TriggerEventInfo::Update(_) => (true, true),
868 };
869 let mut bound = Vec::new();
870 for trigger in live.triggers.iter().rev() {
871 if !trigger.fires_for(&event, changed) {
872 continue;
873 }
874 if self.firing.contains(&trigger.folded) {
875 continue;
876 }
877 if self.firing.len() >= self.trigger_depth {
878 // The number is in the message because a settable limit that
879 // refuses without saying what it was leaves a reader guessing
880 // between the default and whatever `.limit` last set.
881 return Err(refused(
882 format!(
883 "too many levels of trigger recursion: the limit is {}",
884 self.trigger_depth
885 ),
886 Span::default(),
887 ));
888 }
889 self.firing.push(trigger.folded.clone());
890 let saved_ast = self.ast;
891 let saved_scopes = core::mem::take(&mut self.scopes);
892 let saved_aliases = self.row_aliases.take();
893 let saved_target = self.view_target.take();
894 // A trigger body is schema text: the statements in it were written
895 // by whoever wrote the file, and they run because a write happened
896 // rather than because anybody submitted them.
897 let saved_site = self.call_site;
898 self.call_site = crate::function::CallSite::Schema;
899 self.ast = &trigger.ast;
900 self.row_aliases = Some(crate::bind::RowAliases {
901 table: table.clone(),
902 old,
903 new,
904 });
905 let result = self.bind_trigger_body(trigger, table);
906 self.call_site = saved_site;
907 self.ast = saved_ast;
908 self.scopes = saved_scopes;
909 self.row_aliases = saved_aliases;
910 self.view_target = saved_target;
911 self.firing.pop();
912 bound.push(result?);
913 }
914 Ok(bound)
915 }
916
917 /// Binds the triggers this write's foreign keys imply.
918 ///
919 /// The triggers themselves were generated when the schema was read - both
920 /// directions of every key, since nothing in the file records the reverse
921 /// one. What is decided here is which of them apply: whether keys are
922 /// enforced at all, whether a check waits for the commit, and whether this
923 /// particular write touches the columns a check is about.
924 fn bind_foreign_keys(
925 &mut self,
926 table: &TableInfo,
927 event: TriggerEventInfo,
928 changed: &[Vec<u8>],
929 ) -> Result<Vec<BoundTrigger>, ParseError> {
930 if !self.foreign_keys || table.kind != TableKind::Table {
931 return Ok(Vec::new());
932 }
933 let catalog = self.catalog;
934 let database = catalog.database_name(table.database).to_vec();
935 let Some(live) = catalog.find_table(Some(database.as_slice()), &table.folded) else {
936 return Ok(Vec::new());
937 };
938 let mut bound = Vec::new();
939 for planned in &live.foreign_key_triggers {
940 if planned.is_check && (planned.deferred || self.defer_foreign_keys) {
941 continue;
942 }
943 let Some(trigger) = planned.trigger.as_ref() else {
944 if fault_applies(planned, &event) {
945 return Err(crate::bind::schema_refused(
946 String::from_utf8_lossy(&planned.fault).into_owned(),
947 Span::default(),
948 ));
949 }
950 continue;
951 };
952 if !trigger.fires_for(&event, changed) {
953 continue;
954 }
955 if self.firing_foreign_keys.contains(&trigger.folded) {
956 continue;
957 }
958 let mut one = self.bind_foreign_key_trigger(table, trigger, &event)?;
959 one.self_referencing = planned.self_referencing;
960 bound.push(one);
961 }
962 Ok(bound)
963 }
964
965 /// Binds one synthesised trigger, inside the recursion budget.
966 ///
967 /// The budget is spent here rather than where the trigger was generated,
968 /// because what a cascade costs is the *bound* body: one copy per level it
969 /// can reach, and it can reach itself only when the keys form a cycle.
970 fn bind_foreign_key_trigger(
971 &mut self,
972 table: &TableInfo,
973 trigger: &'a TriggerInfo,
974 event: &TriggerEventInfo,
975 ) -> Result<BoundTrigger, ParseError> {
976 if self.foreign_key_depth >= MAX_FOREIGN_KEY_DEPTH || self.foreign_key_budget == 0 {
977 return Err(refused(
978 "too many levels of foreign key recursion",
979 Span::default(),
980 ));
981 }
982 self.foreign_key_depth = self.foreign_key_depth.saturating_add(1);
983 self.foreign_key_budget = self.foreign_key_budget.saturating_sub(1);
984 self.firing_foreign_keys.push(trigger.folded.clone());
985 let (old, new) = match event {
986 TriggerEventInfo::Insert => (false, true),
987 TriggerEventInfo::Delete => (true, false),
988 TriggerEventInfo::Update(_) => (true, true),
989 };
990 let saved_ast = self.ast;
991 let saved_scopes = core::mem::take(&mut self.scopes);
992 let saved_aliases = self.row_aliases.take();
993 let saved_target = self.view_target.take();
994 // A synthesised key action is generated from a `REFERENCES` clause the
995 // schema wrote, so it is schema too - the same site a written trigger
996 // gets, because the binder turns both into the same text.
997 let saved_site = self.call_site;
998 self.call_site = crate::function::CallSite::Schema;
999 self.ast = &trigger.ast;
1000 self.row_aliases = Some(crate::bind::RowAliases {
1001 table: table.clone(),
1002 old,
1003 new,
1004 });
1005 let result = self.bind_trigger_body(trigger, table);
1006 self.call_site = saved_site;
1007 self.ast = saved_ast;
1008 self.scopes = saved_scopes;
1009 self.row_aliases = saved_aliases;
1010 self.view_target = saved_target;
1011 self.foreign_key_depth = self.foreign_key_depth.saturating_sub(1);
1012 self.firing_foreign_keys.pop();
1013 let mut bound = result?;
1014 report_as_foreign_key(&mut bound);
1015 Ok(bound)
1016 }
1017
1018 /// Binds one trigger's guard and body statements.
1019 fn bind_trigger_body(
1020 &mut self,
1021 trigger: &TriggerInfo,
1022 table: &TableInfo,
1023 ) -> Result<BoundTrigger, ParseError> {
1024 let when = match trigger.when {
1025 Some(expr) => Some(self.bind_expr(expr)?),
1026 None => None,
1027 };
1028 let mut body = Vec::new();
1029 for statement in &trigger.body {
1030 // Each statement gets a fresh scope stack. A body statement's names
1031 // resolve against its own tables and against OLD and NEW, never
1032 // outward into the statement that fired it.
1033 let saved = core::mem::take(&mut self.scopes);
1034 let one = self.bind_trigger_statement(statement);
1035 self.scopes = saved;
1036 body.push(one?);
1037 }
1038 Ok(BoundTrigger {
1039 name: trigger.name.clone(),
1040 table: table.folded.clone(),
1041 time: trigger.time,
1042 when,
1043 body,
1044 foreign_key: false,
1045 self_referencing: false,
1046 })
1047 }
1048
1049 /// Binds one statement of a trigger body.
1050 pub(crate) fn bind_trigger_statement(
1051 &mut self,
1052 statement: &ast::Statement,
1053 ) -> Result<BoundTriggerStatement, ParseError> {
1054 match statement {
1055 ast::Statement::Insert(insert) => {
1056 if !insert.returning.is_empty() {
1057 return Err(refused(
1058 "RETURNING is not allowed on a trigger body statement",
1059 Span::default(),
1060 ));
1061 }
1062 Ok(BoundTriggerStatement::Insert(Box::new(
1063 self.bind_insert(insert)?,
1064 )))
1065 }
1066 ast::Statement::Update(update) => {
1067 if !update.returning.is_empty() {
1068 return Err(refused(
1069 "RETURNING is not allowed on a trigger body statement",
1070 Span::default(),
1071 ));
1072 }
1073 Ok(BoundTriggerStatement::Update(Box::new(
1074 self.bind_update(update)?,
1075 )))
1076 }
1077 ast::Statement::Delete(delete) => {
1078 if !delete.returning.is_empty() {
1079 return Err(refused(
1080 "RETURNING is not allowed on a trigger body statement",
1081 Span::default(),
1082 ));
1083 }
1084 Ok(BoundTriggerStatement::Delete(Box::new(
1085 self.bind_delete(delete)?,
1086 )))
1087 }
1088 ast::Statement::Select(select) => Ok(BoundTriggerStatement::Select(Box::new(
1089 self.bind_select(*select)?,
1090 ))),
1091 _ => Err(unsupported(
1092 "that statement in a trigger body",
1093 Span::default(),
1094 )),
1095 }
1096 }
1097
1098 /// Resolves a write target and refuses the things that cannot be written.
1099 fn writable_target(
1100 &mut self,
1101 database: Option<ast::NameId>,
1102 name: ast::NameId,
1103 span: Span,
1104 event: &TriggerEventInfo,
1105 ) -> Result<TableInfo, ParseError> {
1106 let qualifier = database.map(|id| self.ast.folded(id).to_vec());
1107 let folded = self.ast.folded(name).to_vec();
1108 let Some(table) = self
1109 .catalog
1110 .find_table(qualifier.as_deref(), &folded)
1111 .cloned()
1112 else {
1113 return Err(crate::bind::no_such_table(self.ast.text(name), span));
1114 };
1115 match table.kind {
1116 TableKind::View => {
1117 // A view is writable exactly when it has an `INSTEAD OF`
1118 // trigger for this event: the trigger *is* the write, and the
1119 // view itself is never touched.
1120 if !has_instead_of(&table, event) {
1121 return Err(unsupported("writing to a view", span));
1122 }
1123 let expanded = self.expanded_view(&table, span)?;
1124 self.record_write_dependency(table.database);
1125 return Ok(expanded);
1126 }
1127 TableKind::Virtual => {
1128 // A module decides whether it can be written; a module that
1129 // cannot refuses the call rather than the statement, because
1130 // "this table is read-only" is the module's fact and not the
1131 // binder's. What the binder still checks is that the table has
1132 // a module at all - a virtual table this build has no module
1133 // for has no columns either, and nothing can be written to it.
1134 if table.columns.is_empty() {
1135 return Err(unsupported("that virtual table's module", span));
1136 }
1137 self.record_write_dependency(table.database);
1138 return Ok(table);
1139 }
1140 TableKind::Subquery => return Err(unsupported("writing to a subquery", span)),
1141 TableKind::Table => {}
1142 }
1143 if table.folded.starts_with(b"sqlite_")
1144 && !WRITABLE_INTERNAL.contains(&table.folded.as_slice())
1145 {
1146 return Err(unsupported(
1147 "writing to a table whose name begins with sqlite_",
1148 span,
1149 ));
1150 }
1151 self.record_write_dependency(table.database);
1152 Ok(table)
1153 }
1154
1155 /// Resolves the target of an UPDATE or DELETE, which is a FROM term.
1156 fn write_target_from_term(
1157 &mut self,
1158 id: ast::FromTermId,
1159 event: &TriggerEventInfo,
1160 ) -> Result<(TableInfo, usize), ParseError> {
1161 let Some(term) = self.ast.from_term(id) else {
1162 return Err(unsupported("missing target", Span::default()));
1163 };
1164 let ast::FromSource::Table {
1165 database,
1166 name,
1167 indexed_by,
1168 ..
1169 } = term.source
1170 else {
1171 return Err(unsupported("a target that is not a table", term.span));
1172 };
1173 let table = self.writable_target(database, name, term.span, event)?;
1174 // The same rule as a SELECT's: an `INDEXED BY` that names no index of
1175 // the table is refused rather than ignored (task-1979, F7). This path
1176 // has the table in hand rather than a bound source, so it asks the
1177 // table directly.
1178 if let ast::IndexHint::IndexedBy(index) = indexed_by {
1179 let folded = self.ast.folded(index).to_vec();
1180 if !table.indexes.iter().any(|held| held.folded == folded) {
1181 return Err(crate::bind::no_such_index(self.ast.text(index), term.span));
1182 }
1183 }
1184 let alias = match term.alias {
1185 Some(alias) => self.ast.text(alias).to_vec(),
1186 None => table.name.clone(),
1187 };
1188 if table.kind == TableKind::View {
1189 // The view goes in as an ordinary nested query, so the statement's
1190 // WHERE and SET bind against the view's own columns and against the
1191 // term the block producing OLD will iterate. Binding first and
1192 // re-pointing afterwards would be two chances to disagree.
1193 let inner = self.view_query(&table, term.span)?;
1194 let source = BoundSource {
1195 index_hint: crate::bind::IndexChoice::Any,
1196 id: self.sources.len(),
1197 rows: crate::bind::SourceRows::Subquery(Box::new(inner)),
1198 table: std::rc::Rc::new(table.clone()),
1199 alias,
1200 join: ast::JoinKind::Comma,
1201 constraint: None,
1202 suppressed: Vec::new(),
1203 index_exprs: Vec::new(),
1204 };
1205 self.view_target = Some(source.id);
1206 let scope = source.id;
1207 self.sources.push(source);
1208 self.scopes.push(vec![scope]);
1209 return Ok((table, scope));
1210 }
1211 let scope = self.push_write_source(table.clone(), alias);
1212 let choice = self.index_choice(indexed_by);
1213 if let Some(source) = self.sources.get_mut(scope) {
1214 source.index_hint = choice;
1215 }
1216 Ok((table, scope))
1217 }
1218
1219 /// Returns a view's `TableInfo` with the columns its body produces.
1220 ///
1221 /// A view's catalog entry carries no column list - its columns are whatever
1222 /// binding its `SELECT` says they are - so a statement that writes one needs
1223 /// the body bound before `new.column` can resolve to anything at all.
1224 pub(crate) fn expanded_view(
1225 &mut self,
1226 table: &TableInfo,
1227 span: Span,
1228 ) -> Result<TableInfo, ParseError> {
1229 let bound = self.view_query(table, span)?;
1230 let mut expanded = table.clone();
1231 expanded.columns = crate::bind::subquery_columns(&bound, &[]);
1232 Ok(expanded)
1233 }
1234
1235 /// Binds a view's body, out of the arena the catalog snapshot holds.
1236 fn view_query(&mut self, table: &TableInfo, span: Span) -> Result<BoundSelect, ParseError> {
1237 let catalog = self.catalog;
1238 let database = catalog.database_name(table.database).to_vec();
1239 let Some(live) = catalog.find_table(Some(database.as_slice()), &table.folded) else {
1240 return Err(crate::bind::no_such_table(&table.name, span));
1241 };
1242 let Some(body) = live.view.as_ref() else {
1243 return Err(unsupported(
1244 "a view whose definition could not be parsed",
1245 span,
1246 ));
1247 };
1248 let names = body.columns.clone();
1249 let saved_ast = self.ast;
1250 let saved_scopes = core::mem::take(&mut self.scopes);
1251 self.ast = &body.ast;
1252 let bound = self.bind_select(body.select);
1253 self.ast = saved_ast;
1254 self.scopes = saved_scopes;
1255 let mut bound = bound?;
1256 // `CREATE VIEW v (a, b)` renames the body's columns, and those are the
1257 // names `new.a` resolves against.
1258 for (position, name) in names.iter().enumerate() {
1259 if let Some(column) = bound.columns.get_mut(position) {
1260 column.name = name.clone();
1261 }
1262 }
1263 Ok(bound)
1264 }
1265
1266 /// Builds the block whose rows an `INSTEAD OF UPDATE` or `DELETE` fires for.
1267 ///
1268 /// It reads the term `write_target_from_term` already pushed, so the filter
1269 /// handed in here - bound against that same term - needs no adjustment.
1270 fn view_rows(
1271 &mut self,
1272 table: &TableInfo,
1273 filter: Option<BoundExpr>,
1274 ) -> Option<Box<BoundSelect>> {
1275 // The kind is checked before the target is taken. A trigger body's own
1276 // UPDATE binds through here too, and taking first meant the body's
1277 // statement - whose target is an ordinary table - consumed the view
1278 // target belonging to the statement that fired it, which then compiled
1279 // as a write to a view's root page of zero.
1280 if table.kind != TableKind::View {
1281 return None;
1282 }
1283 let id = self.view_target.take()?;
1284 let source = self.sources.get(id)?.clone();
1285 let columns = table
1286 .columns
1287 .iter()
1288 .enumerate()
1289 .map(|(position, column)| BoundResultColumn {
1290 expr: BoundExpr::Column {
1291 source: id,
1292 column: position as u16,
1293 slot: position as u16,
1294 affinity: column.affinity,
1295 collation: Collation::from_name(
1296 core::str::from_utf8(&column.collation).unwrap_or("BINARY"),
1297 )
1298 .unwrap_or(Collation::Binary),
1299 },
1300 name: column.name.clone(),
1301 origin: None,
1302 declared_type: column.declared_type.clone(),
1303 })
1304 .collect();
1305 Some(Box::new(crate::bind::block_over(source, filter, columns)))
1306 }
1307
1308 /// Returns the target's index hint, or refuses a write whose `INDEXED BY`
1309 /// index cannot find its rows.
1310 ///
1311 /// The same rule and the same test a `SELECT` gets from
1312 /// `crate::bind::refuse_unanswerable_hints`, asked of the query the write
1313 /// will run to find its rows: the target, any `UPDATE ... FROM` terms, and
1314 /// the statement's `WHERE`. The pinned 3.53.4 shell refuses
1315 /// `DELETE FROM h INDEXED BY h_part WHERE a = 1`, where `h_part` is declared
1316 /// `WHERE c > 3`, with `no query solution`.
1317 /// @param source - the target's statement-wide number
1318 /// @param index_exprs - the target's bound index expressions
1319 /// @param filter - the statement's `WHERE`
1320 /// @param joined - the `UPDATE ... FROM` terms, empty for a `DELETE`
1321 fn write_hint(
1322 &self,
1323 source: usize,
1324 index_exprs: &[BoundIndexExprs],
1325 filter: Option<&BoundExpr>,
1326 joined: &[BoundSource],
1327 ) -> Result<crate::bind::IndexChoice, ParseError> {
1328 let Some(target) = self.sources.get(source) else {
1329 return Ok(crate::bind::IndexChoice::Any);
1330 };
1331 if target.index_hint == crate::bind::IndexChoice::Any {
1332 return Ok(crate::bind::IndexChoice::Any);
1333 }
1334 let mut probe = target.clone();
1335 probe.index_exprs = index_exprs.to_vec();
1336 let mut block = crate::bind::block_over(probe, filter.cloned(), Vec::new());
1337 block.sources.extend(joined.iter().cloned());
1338 if crate::plan::unanswerable_index_hint(&block).is_some() {
1339 return Err(crate::bind::no_query_solution(Span::default()));
1340 }
1341 Ok(target.index_hint.clone())
1342 }
1343
1344 /// Makes the target table the statement's one visible source.
1345 ///
1346 /// It opens a scope holding just the target, so every name in the
1347 /// statement's `SET`, `WHERE` and `RETURNING` resolves against the table
1348 /// being written and nothing else.
1349 fn push_write_source(&mut self, table: TableInfo, alias: Vec<u8>) -> usize {
1350 let id = self.sources.len();
1351 self.sources.push(BoundSource {
1352 index_hint: crate::bind::IndexChoice::Any,
1353 id,
1354 rows: crate::bind::SourceRows::Table,
1355 table: std::rc::Rc::new(table),
1356 alias,
1357 join: ast::JoinKind::Comma,
1358 constraint: None,
1359 suppressed: Vec::new(),
1360 index_exprs: Vec::new(),
1361 });
1362 self.scopes.push(vec![id]);
1363 id
1364 }
1365
1366 /// Refuses an attempt to write a generated column.
1367 ///
1368 /// SQLite's message names the column, because the usual cause is a script
1369 /// that inserts every column of a table one of whose columns has since been
1370 /// made generated. It names the statement too - `INSERT` or `UPDATE` - and
1371 /// so does this.
1372 ///
1373 /// @param table - the table being written
1374 /// @param position - the column the statement named
1375 /// @param verb - `INSERT into` or `UPDATE`, as SQLite writes it
1376 /// @param span - where the name was written
1377 fn refuse_generated(
1378 &self,
1379 table: &TableInfo,
1380 position: u16,
1381 verb: &str,
1382 span: Span,
1383 ) -> Result<(), ParseError> {
1384 let Some(column) = table.column(position) else {
1385 return Ok(());
1386 };
1387 if !column.generated {
1388 return Ok(());
1389 }
1390 Err(refused(
1391 format!(
1392 "cannot {verb} generated column \"{}\"",
1393 String::from_utf8_lossy(&column.name)
1394 ),
1395 span,
1396 ))
1397 }
1398
1399 /// Returns the target column positions an INSERT writes, in source order.
1400 ///
1401 /// With no column list the targets are every column in declaration order,
1402 /// which is why adding a column to a table changes what a positional
1403 /// INSERT means - SQLite's behaviour, and the reason the column list is
1404 /// worth writing.
1405 fn insert_targets(
1406 &self,
1407 table: &TableInfo,
1408 columns: &[ast::NameId],
1409 ) -> Result<Vec<u16>, ParseError> {
1410 if columns.is_empty() {
1411 // A bare `INSERT INTO t VALUES (...)` supplies the columns a person
1412 // can write, which is every column that is not generated - so a
1413 // table with a generated column takes fewer values than it has
1414 // columns, exactly as SQLite counts them.
1415 // A hidden column is not one of them either: a module's arguments
1416 // and its `rank` are named by an application that wants them, and
1417 // an `INSERT INTO fts VALUES ('a', 'b')` supplies the two indexed
1418 // columns and nothing else.
1419 return Ok((0..table.columns.len() as u16)
1420 .filter(|position| {
1421 table
1422 .column(*position)
1423 .is_some_and(|column| !column.generated && !column.hidden)
1424 })
1425 .collect());
1426 }
1427 let mut targets = Vec::with_capacity(columns.len());
1428 for name in columns {
1429 let folded = self.ast.folded(*name).to_vec();
1430 let position = match table.column_position(&folded) {
1431 Some(position) => position,
1432 // A rowid table lets the statement name its rowid, under any
1433 // of its three spellings, and that is not a column: it is the
1434 // key. A declared column of the same name wins, which is why
1435 // this is the fallback rather than the first thing tried.
1436 None if table.has_rowid() && is_rowid_name(&folded) => ROWID_TARGET,
1437 None => return Err(no_such_column(self.ast.text(*name), Span::default())),
1438 };
1439 if targets.contains(&position) {
1440 return Err(refused(
1441 format!(
1442 "column {} is named twice",
1443 String::from_utf8_lossy(self.ast.text(*name))
1444 ),
1445 Span::default(),
1446 ));
1447 }
1448 if position != ROWID_TARGET {
1449 self.refuse_generated(table, position, "INSERT into", Span::default())?;
1450 }
1451 targets.push(position);
1452 }
1453 Ok(targets)
1454 }
1455
1456 /// Binds the rows an INSERT supplies.
1457 fn bind_insert_source(
1458 &mut self,
1459 source: &ast::InsertSource,
1460 table: &TableInfo,
1461 targets: &[u16],
1462 ) -> Result<(BoundInsertSource, usize), ParseError> {
1463 match source {
1464 ast::InsertSource::DefaultValues => {
1465 let _ = (table, targets);
1466 Ok((BoundInsertSource::Values(vec![Vec::new()]), 0))
1467 }
1468 ast::InsertSource::Select(id) => {
1469 // The target table is source zero while the rows are bound, so
1470 // that `INSERT INTO t SELECT ... FROM u` resolves `u`'s columns
1471 // and not `t`'s. Binding a SELECT replaces the source list, and
1472 // the target is pushed back afterwards.
1473 // The scope stack is emptied rather than pushed to, because a
1474 // pushed scope would still be searched *outward* into the
1475 // target's, and `INSERT INTO t SELECT a FROM u` would then
1476 // resolve `a` against `t` when `u` has no such column.
1477 let saved = core::mem::take(&mut self.scopes);
1478 let select = self.bind_select(*id);
1479 let bound = match select {
1480 Ok(bound) => bound,
1481 Err(error) => {
1482 self.scopes = saved;
1483 return Err(error);
1484 }
1485 };
1486 self.scopes = saved;
1487 if bound.values.is_empty() {
1488 let arity = bound.columns.len();
1489 return Ok((BoundInsertSource::Select(Box::new(bound)), arity));
1490 }
1491 let arity = bound.values.first().map_or(0, Vec::len);
1492 for row in &bound.values {
1493 if row.len() != arity {
1494 return Err(unsupported(
1495 "all VALUES rows must have the same number of columns",
1496 Span::default(),
1497 ));
1498 }
1499 }
1500 Ok((BoundInsertSource::Values(bound.values), arity))
1501 }
1502 }
1503 }
1504
1505 /// Works out where every table column's value comes from.
1506 ///
1507 /// A column the statement named takes its value from the source row; a
1508 /// column it did not takes its `DEFAULT`, and a column with no default
1509 /// takes NULL. The rowid is separated out here rather than in the
1510 /// compiler, because an `INTEGER PRIMARY KEY` column *is* the rowid and
1511 /// writing it into the record as well would store a duplicate that SQLite
1512 /// does not.
1513 fn column_sources(
1514 &mut self,
1515 table: &TableInfo,
1516 targets: &[u16],
1517 ) -> Result<(Vec<ColumnSource>, Option<ColumnSource>), ParseError> {
1518 let mut columns = Vec::with_capacity(table.columns.len());
1519 for position in 0..table.columns.len() as u16 {
1520 if let Some(expr) = self.generated_expr(table, position)? {
1521 columns.push(ColumnSource::Generated(expr));
1522 continue;
1523 }
1524 let source = match targets.iter().position(|target| *target == position) {
1525 Some(index) => ColumnSource::Row(index),
1526 None => ColumnSource::Expr(self.default_expr(table, position)?),
1527 };
1528 columns.push(source);
1529 }
1530 let rowid = match table.rowid_alias {
1531 Some(position) => columns.get(position as usize).cloned(),
1532 None => None,
1533 };
1534 Ok((columns, rowid))
1535 }
1536
1537 /// Binds a generated column's expression, when the column is one.
1538 fn generated_expr(
1539 &mut self,
1540 table: &TableInfo,
1541 position: u16,
1542 ) -> Result<Option<BoundExpr>, ParseError> {
1543 let Some(column) = table.column(position) else {
1544 return Ok(None);
1545 };
1546 if !column.generated {
1547 return Ok(None);
1548 }
1549 let Some(sql) = column.generated_sql.clone() else {
1550 return Ok(Some(BoundExpr::Null));
1551 };
1552 Ok(Some(self.bind_schema_expr(&sql)?))
1553 }
1554
1555 /// Binds every `STORED` generated column's expression.
1556 ///
1557 /// Returns them as assignments, because that is what they are on the write
1558 /// path: a value the statement did not write and the row has to carry. See
1559 /// [`BoundUpdate::generated`] for why an `UPDATE` needs them and a
1560 /// `VIRTUAL` column does not.
1561 ///
1562 /// @param table - the table being written
1563 fn bind_stored_generated(
1564 &mut self,
1565 table: &TableInfo,
1566 ) -> Result<Vec<BoundAssignment>, ParseError> {
1567 let mut generated = Vec::new();
1568 for position in 0..table.columns.len() as u16 {
1569 let Some(column) = table.column(position) else {
1570 continue;
1571 };
1572 if !column.generated || !column.stored {
1573 continue;
1574 }
1575 let Some(expr) = self.generated_expr(table, position)? else {
1576 continue;
1577 };
1578 generated.push(BoundAssignment {
1579 column: position,
1580 rowid: false,
1581 value: expr,
1582 });
1583 }
1584 Ok(generated)
1585 }
1586
1587 /// Binds a column's `DEFAULT`, or NULL when it has none.
1588 fn default_expr(&mut self, table: &TableInfo, position: u16) -> Result<BoundExpr, ParseError> {
1589 let Some(column) = table.column(position) else {
1590 return Ok(BoundExpr::Null);
1591 };
1592 let Some(sql) = column.default_sql.as_ref() else {
1593 return Ok(BoundExpr::Null);
1594 };
1595 if sql.is_empty() {
1596 return Ok(BoundExpr::Null);
1597 }
1598 self.bind_schema_expr(sql)
1599 }
1600
1601 /// Binds the `DEFAULT` of every `NOT NULL` column that declares one.
1602 ///
1603 /// What `REPLACE` substitutes for a NULL in such a column - see
1604 /// [`BoundDefault`]. A column with no default is left out, which is what
1605 /// makes the write path's fallback to `ABORT` the absence of an entry
1606 /// rather than a second test.
1607 ///
1608 /// The rowid alias is left out too: the row image carries the key the
1609 /// statement is about to allocate, and the write path does not check it.
1610 ///
1611 /// @param table - the table being written
1612 fn bind_not_null_defaults(
1613 &mut self,
1614 table: &TableInfo,
1615 ) -> Result<Vec<BoundDefault>, ParseError> {
1616 let mut defaults = Vec::new();
1617 for (position, column) in table.columns.iter().enumerate() {
1618 if !column.not_null || Some(position as u16) == table.rowid_alias {
1619 continue;
1620 }
1621 let Some(sql) = column.default_sql.as_ref() else {
1622 continue;
1623 };
1624 if sql.is_empty() {
1625 continue;
1626 }
1627 let expr = self.bind_schema_expr(&sql.clone())?;
1628 defaults.push(BoundDefault {
1629 column: position as u16,
1630 expr,
1631 });
1632 }
1633 Ok(defaults)
1634 }
1635
1636 /// Binds every `CHECK` the table declares.
1637 fn bind_checks(&mut self, table: &TableInfo) -> Result<Vec<BoundCheck>, ParseError> {
1638 let mut checks = Vec::with_capacity(table.checks.len());
1639 for check in &table.checks {
1640 checks.push(BoundCheck {
1641 name: check.name.clone(),
1642 expr: self.bind_schema_expr(&check.expr_sql)?,
1643 });
1644 }
1645 Ok(checks)
1646 }
1647
1648 /// Binds the expressions the table's indexes need per row.
1649 ///
1650 /// Only the indexes that need any: a partial one, and one with an
1651 /// expression key. Everything else is a slot of the row and needs nothing.
1652 ///
1653 /// @param table - the table being written
1654 fn bind_index_exprs(&mut self, table: &TableInfo) -> Result<Vec<BoundIndexExprs>, ParseError> {
1655 let mut bound = Vec::new();
1656 for (position, index) in table.indexes.iter().enumerate() {
1657 let needs = index.partial_sql.is_some()
1658 || index.columns.iter().any(|key| key.expr_sql.is_some());
1659 if !needs {
1660 continue;
1661 }
1662 let predicate = match index.partial_sql.as_ref() {
1663 Some(sql) => Some(self.bind_schema_expr(sql)?),
1664 None => None,
1665 };
1666 let mut keys = Vec::with_capacity(index.columns.len());
1667 for key in &index.columns {
1668 keys.push(match key.expr_sql.as_ref() {
1669 Some(sql) => Some(self.bind_schema_expr(sql)?),
1670 None => None,
1671 });
1672 }
1673 bound.push(BoundIndexExprs {
1674 position,
1675 predicate,
1676 keys,
1677 });
1678 }
1679 Ok(bound)
1680 }
1681
1682 /// Parses and binds an expression that was written in the schema.
1683 ///
1684 /// It is parsed into its own arena and bound against the statement's
1685 /// current sources, so the result is an ordinary `BoundExpr` that refers to
1686 /// the target table by position and carries no reference to the schema
1687 /// text it came from.
1688 pub fn bind_schema_expr(&mut self, sql: &[u8]) -> Result<BoundExpr, ParseError> {
1689 let limits = Limits::default();
1690 let (ast, expr) = parse_expression(sql, &limits)?;
1691 let mut nested = Binder::new(self.catalog, &ast, self.authorizer);
1692 nested.trigger_depth = self.trigger_depth;
1693 // **This is where a `DEFAULT`, a `CHECK`, a generated column, an index
1694 // expression and a partial-index predicate all become a bound tree, so
1695 // it is where all five are told they are a schema (task-1972).** The
1696 // nested binder also inherits the connection's registrations and
1697 // collations, which it did not before: without the registrations
1698 // `bind_external_call` never sees the call at all, because the name
1699 // does not resolve to a registered function and the expression fails as
1700 // "no such function" - an error for the wrong reason, and one that
1701 // disappears the moment an application registers the same name at a
1702 // different arity.
1703 nested.externals = self.externals;
1704 nested.collations = self.collations;
1705 nested.trusted_schema = self.trusted_schema;
1706 nested.call_site = crate::function::CallSite::Schema;
1707 nested.sources = self.sources.clone();
1708 nested.scopes = self.scopes.clone();
1709 let bound = nested.bind_expr(expr)?;
1710 Ok(bound)
1711 }
1712
1713 /// Binds an `ON CONFLICT` clause.
1714 fn bind_upsert(
1715 &mut self,
1716 table: &TableInfo,
1717 insert: &ast::Insert,
1718 ) -> Result<Vec<BoundUpsert>, ParseError> {
1719 if insert.upserts.is_empty() {
1720 return Ok(Vec::new());
1721 }
1722 // **Every clause is bound, in written order.** A statement may carry
1723 // several - `ON CONFLICT(k) DO UPDATE ... ON CONFLICT(id) DO UPDATE ...`
1724 // - and which one runs is decided at *run time*, by which constraint
1725 // the row actually collided with. Binding only the first was the whole
1726 // of the old refusal.
1727 for upsert in &insert.upserts {
1728 if upsert.target_filter.is_some() {
1729 return Err(unsupported(
1730 "a partial-index conflict target",
1731 Span::default(),
1732 ));
1733 }
1734 }
1735 // A clause with no conflict target matches any constraint, so anything
1736 // written after it could never run. SQLite refuses that rather than
1737 // accepting a clause it will never reach.
1738 if let Some(position) = insert
1739 .upserts
1740 .iter()
1741 .position(|upsert| upsert.target.is_empty())
1742 {
1743 if position + 1 < insert.upserts.len() {
1744 return Err(crate::bind::schema_refused(
1745 "ON CONFLICT clause with no conflict target must be last",
1746 Span::default(),
1747 ));
1748 }
1749 }
1750 // `excluded` is in scope for the assignments and the WHERE, and only
1751 // there. Setting it around the binding rather than pushing a second
1752 // FROM term keeps unqualified names resolving to the target row, which
1753 // is what SQLite does and what a second source would have made
1754 // ambiguous - every column of the target is also a column of
1755 // `excluded`.
1756 self.excluded = Some(table.clone());
1757 let mut bound = Vec::with_capacity(insert.upserts.len());
1758 for upsert in &insert.upserts {
1759 match self.bind_upsert_body(table, upsert) {
1760 Ok(Some(one)) => bound.push(one),
1761 Ok(None) => {}
1762 Err(error) => {
1763 self.excluded = None;
1764 return Err(error);
1765 }
1766 }
1767 }
1768 self.excluded = None;
1769 Ok(bound)
1770 }
1771
1772 /// Binds an upsert's target, assignments and filter.
1773 fn bind_upsert_body(
1774 &mut self,
1775 table: &TableInfo,
1776 upsert: &ast::Upsert,
1777 ) -> Result<Option<BoundUpsert>, ParseError> {
1778 let mut target = Vec::new();
1779 for column in &upsert.target {
1780 let Some(name) = bare_indexed_column(self.ast, column) else {
1781 return Err(unsupported(
1782 "an expression in a conflict target",
1783 Span::default(),
1784 ));
1785 };
1786 let Some(position) = table.column_position(&name) else {
1787 return Err(no_such_column(&name, Span::default()));
1788 };
1789 target.push(position);
1790 }
1791 target.sort_unstable();
1792 let mut assignments = Vec::new();
1793 for (names, value) in &upsert.assignments {
1794 let bound = self.bind_expr(*value)?;
1795 for name in names {
1796 let folded = self.ast.folded(*name).to_vec();
1797 let Some(position) = table.column_position(&folded) else {
1798 return Err(no_such_column(self.ast.text(*name), Span::default()));
1799 };
1800 assignments.push(BoundAssignment {
1801 column: position,
1802 rowid: false,
1803 value: bound.clone(),
1804 });
1805 }
1806 }
1807 assignments.sort_by_key(|assignment| assignment.column);
1808 let filter = match upsert.filter {
1809 Some(expr) => Some(self.bind_expr(expr)?),
1810 None => None,
1811 };
1812 Ok(Some(BoundUpsert {
1813 target,
1814 assignments,
1815 do_update: upsert.do_update,
1816 filter,
1817 }))
1818 }
1819
1820 /// Binds a `RETURNING` list, which is a result-column list over the row
1821 /// that was written.
1822 fn bind_returning(
1823 &mut self,
1824 columns: &[ast::ResultColumn],
1825 ) -> Result<Vec<BoundResultColumn>, ParseError> {
1826 if columns.is_empty() {
1827 return Ok(Vec::new());
1828 }
1829 self.bind_result_columns_public(columns)
1830 }
1831}
1832
1833/// Returns an indexed column's bare folded name, when it names a column.
1834fn bare_indexed_column(ast: &crate::Ast, column: &ast::IndexedColumn) -> Option<Vec<u8>> {
1835 match ast.expr(column.expr) {
1836 Some(ast::Expr::Column {
1837 table: None,
1838 column: name,
1839 ..
1840 }) => Some(ast.folded(*name).to_vec()),
1841 _ => None,
1842 }
1843}
1844
1845/// The extended result codes a rejected write reports.
1846///
1847/// The numbers are SQLite's own extended codes. They are written out rather
1848/// than derived because an application matches on them, and a code that was
1849/// computed from an enum's discriminant would change the day the enum did.
1850///
1851/// They live here, beside the binder that decides which constraint a statement
1852/// can violate, because **both** engines report them: the virtual machine
1853/// compiles them into a `HaltError` and the vectorised executor returns them
1854/// from its write path. Two copies would agree until one of them was corrected.
1855pub mod codes {
1856 /// `SQLITE_CONSTRAINT_CHECK`.
1857 pub const CHECK: i32 = 275;
1858 /// `SQLITE_CONSTRAINT_DATATYPE`, which a STRICT table reports.
1859 pub const DATATYPE: i32 = 3091;
1860 /// `SQLITE_CONSTRAINT_NOTNULL`.
1861 pub const NOT_NULL: i32 = 1299;
1862 /// `SQLITE_CONSTRAINT_PRIMARYKEY`.
1863 pub const PRIMARY_KEY: i32 = 1555;
1864 /// `SQLITE_CONSTRAINT_UNIQUE`.
1865 pub const UNIQUE: i32 = 2067;
1866 /// `SQLITE_CONSTRAINT_ROWID`.
1867 pub const ROWID: i32 = 2579;
1868 /// `SQLITE_MISMATCH`, which an `INTEGER PRIMARY KEY` reports for a value
1869 /// that is not an integer.
1870 pub const MISMATCH: i32 = 20;
1871 /// `SQLITE_CONSTRAINT_TRIGGER`, which `RAISE()` reports.
1872 pub const TRIGGER: i32 = 1811;
1873 /// `SQLITE_CONSTRAINT_FOREIGNKEY`.
1874 pub const FOREIGN_KEY: i32 = 787;
1875}
1876
1877/// Returns the message a unique-index violation reports.
1878///
1879/// SQLite names every column of the index, comma separated, which is what an
1880/// application parses to find out which key collided.
1881///
1882/// @param table - the table the index belongs to
1883/// @param index - the index whose key collided
1884pub fn unique_message(table: &TableInfo, index: &IndexInfo) -> String {
1885 let names: Vec<String> = index
1886 .columns
1887 .iter()
1888 .filter_map(|key| key.column)
1889 .filter_map(|column| table.column(column))
1890 .map(|column| {
1891 format!(
1892 "{}.{}",
1893 String::from_utf8_lossy(&table.name),
1894 String::from_utf8_lossy(&column.name)
1895 )
1896 })
1897 .collect();
1898 format!("UNIQUE constraint failed: {}", names.join(", "))
1899}
1900
1901/// Returns the message a duplicate rowid reports, and its extended code.
1902///
1903/// SQLite names the aliasing column when the table has an `INTEGER PRIMARY
1904/// KEY` - and reports `SQLITE_CONSTRAINT_PRIMARYKEY` for it - and names the
1905/// hidden `rowid` under `SQLITE_CONSTRAINT_ROWID` when it does not.
1906///
1907/// **A `WITHOUT ROWID` table has no rowid to name.** Its own key *is* its
1908/// primary key, held in the one index whose root is the table's, so a collision
1909/// reports every column of that key under `SQLITE_CONSTRAINT_PRIMARYKEY` -
1910/// `UNIQUE constraint failed: t.a, t.b`. It used to answer `t.rowid`, naming a
1911/// column the table does not have, on `INSERT` as well as `UPDATE`.
1912///
1913/// @param table - the table whose key collided
1914pub fn rowid_message(table: &TableInfo) -> (i32, String) {
1915 if table.without_rowid {
1916 if let Some(index) = table.indexes.iter().find(|index| index.root == table.root) {
1917 return (codes::PRIMARY_KEY, unique_message(table, index));
1918 }
1919 }
1920 match table.rowid_alias.and_then(|column| table.column(column)) {
1921 Some(column) => (
1922 codes::PRIMARY_KEY,
1923 format!(
1924 "UNIQUE constraint failed: {}.{}",
1925 String::from_utf8_lossy(&table.name),
1926 String::from_utf8_lossy(&column.name)
1927 ),
1928 ),
1929 None => (
1930 codes::ROWID,
1931 format!(
1932 "UNIQUE constraint failed: {}.rowid",
1933 String::from_utf8_lossy(&table.name)
1934 ),
1935 ),
1936 }
1937}
1938
1939/// Puts a limited write's order, limit and offset on the query that finds a
1940/// view's rows.
1941///
1942/// A write through a view's `INSTEAD OF` trigger fires once per row the view
1943/// produces under the statement's `WHERE`, so a `LIMIT` on the write limits
1944/// that query. SQLite's `sqlite3MaterializeView` is handed the same three
1945/// clauses for the same reason.
1946///
1947/// @param rows - the query over the view the binder built
1948/// @param order_by - the statement's bound `ORDER BY`
1949/// @param limit - the statement's bound `LIMIT`
1950/// @param offset - the statement's bound `OFFSET`
1951fn limit_view_rows(
1952 mut rows: Box<BoundSelect>,
1953 order_by: &[BoundOrderTerm],
1954 limit: &Option<BoundExpr>,
1955 offset: &Option<BoundExpr>,
1956) -> Box<BoundSelect> {
1957 rows.order_by = order_by.to_vec();
1958 rows.limit = limit.clone();
1959 rows.offset = offset.clone();
1960 rows
1961}
1962
1963/// Returns the refusal a `DELETE` or `UPDATE` with `ORDER BY` and no `LIMIT`
1964/// earns, or `None` when the clause is allowed.
1965///
1966/// **`ORDER BY` and `LIMIT` on a write are run, not refused (task-2120).** They
1967/// were refused in the pinned reference's words, `near "ORDER": syntax error`,
1968/// because that build is not compiled with `SQLITE_ENABLE_UPDATE_DELETE_LIMIT`
1969/// and has no grammar for the clause. But the builds applications actually link
1970/// often are - Apple's is - and `DELETE FROM t WHERE ... LIMIT 1000` in a loop
1971/// is the ordinary way to trim a large table without one large transaction. A
1972/// consumer probing 0.1.8 against the macOS `sqlite3` reported the refusal as a
1973/// real gap, which it was.
1974///
1975/// What remains is the one rule a build compiled with the option enforces:
1976/// an order with nothing to limit is refused, in SQLite's own words, because
1977/// sorting the rows a statement changes all of changes nothing.
1978///
1979/// @param limited - which of the two words came first and where, from the parser
1980/// @param limit - the statement's `LIMIT`, when it wrote one
1981/// @param statement - `DELETE` or `UPDATE`, for the message
1982fn order_without_limit(
1983 limited: Option<(ast::Limited, Span)>,
1984 limit: Option<ast::ExprId>,
1985 statement: &str,
1986) -> Option<ParseError> {
1987 let (word, span) = limited?;
1988 if word != ast::Limited::OrderBy || limit.is_some() {
1989 return None;
1990 }
1991 Some(refused(
1992 format!("ORDER BY without LIMIT on {statement}"),
1993 span,
1994 ))
1995}
1996
1997#[cfg(test)]
1998mod tests {
1999 use super::*;
2000 use crate::catalog_view::{
2001 ColumnInfo, IndexColumnInfo, IndexInfo, IndexOrigin, TableInfo, TableKind,
2002 };
2003 use inillucent_value::Affinity;
2004
2005 /// Returns one plain column.
2006 ///
2007 /// @param name - the column's name
2008 fn a_column(name: &str) -> ColumnInfo {
2009 ColumnInfo {
2010 name: name.as_bytes().to_vec(),
2011 folded: name.to_ascii_lowercase().into_bytes(),
2012 declared_type: b"INTEGER".to_vec(),
2013 affinity: Affinity::Integer,
2014 collation: b"binary".to_vec(),
2015 not_null: false,
2016 not_null_conflict: None,
2017 primary_key_conflict: None,
2018 default_sql: None,
2019 primary_key_position: None,
2020 hidden: false,
2021 generated: false,
2022 stored: false,
2023 generated_sql: None,
2024 }
2025 }
2026
2027 /// Returns a rowid table with the columns named.
2028 ///
2029 /// @param name - the table's name
2030 /// @param columns - the column names, in declaration order
2031 fn a_table(name: &str, columns: &[&str]) -> TableInfo {
2032 TableInfo {
2033 name: name.as_bytes().to_vec(),
2034 folded: name.to_ascii_lowercase().into_bytes(),
2035 database: 0,
2036 root: 2,
2037 columns: columns.iter().map(|held| a_column(held)).collect(),
2038 rowid_alias: None,
2039 without_rowid: false,
2040 strict: false,
2041 autoincrement: false,
2042 kind: TableKind::Table,
2043 create_sql: Vec::new(),
2044 indexes: Vec::new(),
2045 view: None,
2046 triggers: Vec::new(),
2047 analysed_rows: None,
2048 foreign_key_triggers: Vec::new(),
2049 foreign_keys: Vec::new(),
2050 checks: Vec::new(),
2051 module: None,
2052 }
2053 }
2054
2055 /// Returns an index over the table columns named.
2056 ///
2057 /// @param name - the index's name
2058 /// @param root - its own tree, or the table's for a `WITHOUT ROWID` key
2059 /// @param columns - the table columns it keys on
2060 fn an_index(name: &str, root: u32, columns: &[u16]) -> IndexInfo {
2061 IndexInfo {
2062 name: name.as_bytes().to_vec(),
2063 folded: name.to_ascii_lowercase().into_bytes(),
2064 root,
2065 unique: true,
2066 columns: columns
2067 .iter()
2068 .map(|held| IndexColumnInfo {
2069 column: Some(*held),
2070 expr_sql: None,
2071 collation: b"binary".to_vec(),
2072 descending: false,
2073 declared_descending: false,
2074 })
2075 .collect(),
2076 partial_sql: None,
2077 origin: IndexOrigin::Unique,
2078 conflict: None,
2079 prefix_rows: Vec::new(),
2080 analysed_rows: None,
2081 metric: None,
2082 }
2083 }
2084
2085 /// The three spellings of the rowid are the three SQLite accepts.
2086 ///
2087 /// **A fourth would be a column name a table could not have (T3,
2088 /// task-1962).** `rowid`, `oid` and `_rowid_` all name the hidden key, and
2089 /// a table that declares a column called any of them shadows it - so the
2090 /// list decides which names a `SELECT rowid` can mean.
2091 #[test]
2092 fn the_rowid_has_three_names() {
2093 assert!(is_rowid_name(b"rowid"));
2094 assert!(is_rowid_name(b"oid"));
2095 assert!(is_rowid_name(b"_rowid_"));
2096 assert!(!is_rowid_name(b"row_id"));
2097 assert!(!is_rowid_name(b"id"));
2098 assert!(
2099 !is_rowid_name(b"ROWID"),
2100 "the argument is already folded, so an unfolded name is not one this asks about"
2101 );
2102 }
2103
2104 /// A unique violation names every column of the index, table-qualified.
2105 ///
2106 /// **The message is what an application matches on.** SQLite's wording is
2107 /// `UNIQUE constraint failed: t.a, t.b`, and a library that switched on it
2108 /// would stop recognising a collision if the columns were listed any other
2109 /// way.
2110 #[test]
2111 fn a_unique_violation_names_every_column_of_the_index() {
2112 let table = a_table("t", &["a", "b", "c"]);
2113 let one = an_index("by_a", 3, &[0]);
2114 assert_eq!(
2115 unique_message(&table, &one),
2116 "UNIQUE constraint failed: t.a"
2117 );
2118 let two = an_index("by_a_b", 4, &[0, 1]);
2119 assert_eq!(
2120 unique_message(&table, &two),
2121 "UNIQUE constraint failed: t.a, t.b",
2122 "both columns, in key order, separated the way the reference separates them"
2123 );
2124 }
2125
2126 /// A rowid collision names the aliasing column when there is one, and the
2127 /// hidden `rowid` when there is not.
2128 ///
2129 /// The extended code differs with it: `SQLITE_CONSTRAINT_PRIMARYKEY` for an
2130 /// `INTEGER PRIMARY KEY` and `SQLITE_CONSTRAINT_ROWID` for the hidden one.
2131 #[test]
2132 fn a_rowid_collision_names_the_column_that_aliases_it() {
2133 let hidden = a_table("t", &["a"]);
2134 assert_eq!(
2135 rowid_message(&hidden),
2136 (
2137 codes::ROWID,
2138 "UNIQUE constraint failed: t.rowid".to_string()
2139 )
2140 );
2141 let mut aliased = a_table("t", &["id", "a"]);
2142 aliased.rowid_alias = Some(0);
2143 assert_eq!(
2144 rowid_message(&aliased),
2145 (
2146 codes::PRIMARY_KEY,
2147 "UNIQUE constraint failed: t.id".to_string()
2148 )
2149 );
2150 }
2151
2152 /// A `WITHOUT ROWID` table has no rowid to name, so it names its key.
2153 ///
2154 /// **It used to answer `t.rowid`, naming a column the table does not
2155 /// have.** Its own key *is* its primary key, held in the one index whose
2156 /// root is the table's.
2157 #[test]
2158 fn a_without_rowid_collision_names_the_primary_key() {
2159 let mut table = a_table("t", &["a", "b"]);
2160 table.without_rowid = true;
2161 table.indexes = vec![an_index("sqlite_autoindex_t_1", table.root, &[0, 1])];
2162 assert_eq!(
2163 rowid_message(&table),
2164 (
2165 codes::PRIMARY_KEY,
2166 "UNIQUE constraint failed: t.a, t.b".to_string()
2167 )
2168 );
2169 }
2170
2171 /// A constraint's own `ON CONFLICT REPLACE` makes a statement able to
2172 /// replace, with no `OR REPLACE` written anywhere.
2173 #[test]
2174 fn a_constraint_can_make_a_plain_insert_replace() {
2175 let plain = a_table("t", &["a"]);
2176 assert!(!can_replace(&plain, None));
2177 assert!(can_replace(&plain, Some(ConflictAction::Replace)));
2178
2179 let mut on_the_index = a_table("t", &["a"]);
2180 let mut index = an_index("by_a", 3, &[0]);
2181 index.conflict = Some(ConflictAction::Replace);
2182 on_the_index.indexes = vec![index];
2183 assert!(
2184 can_replace(&on_the_index, None),
2185 "`a UNIQUE ON CONFLICT REPLACE` replaces without the statement saying so"
2186 );
2187
2188 let mut on_the_column = a_table("t", &["a"]);
2189 if let Some(column) = on_the_column.columns.first_mut() {
2190 column.not_null_conflict = Some(ConflictAction::Replace);
2191 }
2192 assert!(can_replace(&on_the_column, None));
2193 }
2194}