Skip to main content

rudb_parse/
ast.rs

1//! rudb's abstract syntax tree.
2//!
3//! The parse tree the matcher produces is DuckDB's grammar, faithfully. That is the point of it and
4//! it is also why nothing downstream should read it: a bump of the vendored grammar is allowed to
5//! rename `BetweenInLikeExpression`, and if the binder is matching on that name then the bump is a
6//! rewrite. This module is the boundary. It is ours, it changes when we decide it changes, and
7//! `transform` is the one place that knows both shapes.
8//!
9//! Everything is an arena with `u32` indices, per `spec/04-architecture.md` section 4.5. There is
10//! no `Box` and no `Vec` inside a node. A list of children is a [`Slice`] into a side vector, which
11//! means a node is a fixed size, the whole tree is a handful of allocations, and walking it is a
12//! sequential read rather than a pointer chase per node. It also means an `Ast` is `Clone` and
13//! `Send` without any thought, and that a subtree can be addressed by a `u32` in a plan or an
14//! error without borrowing anything.
15//!
16//! The one cost is that you cannot hold a reference to a node and index the arena at the same time,
17//! so the code reads a node out by value first. Nodes are small and `Copy`, so that is a register
18//! move.
19
20use rudb_common::Span;
21
22use crate::matcher::NONE;
23
24/// A run of items in one of the side vectors.
25///
26/// Empty is `len == 0`, and `start` is then meaningless rather than wrong. There is no `Option`
27/// wrapper because an absent list and an empty list are the same thing everywhere this is used.
28#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
29pub struct Slice {
30    /// The first item.
31    pub start: u32,
32    /// How many items.
33    pub len: u32,
34}
35
36impl Slice {
37    /// Whether the run is empty.
38    pub const fn is_empty(self) -> bool {
39        self.len == 0
40    }
41
42    /// The run as a range, for indexing the backing vector.
43    pub const fn range(self) -> std::ops::Range<usize> {
44        self.start as usize..(self.start + self.len) as usize
45    }
46}
47
48/// An index into `Ast::strings`.
49pub type StrRef = u32;
50/// An index into `Ast::exprs`.
51pub type ExprRef = u32;
52/// An index into `Ast::sources`.
53pub type SourceRef = u32;
54/// An index into `Ast::queries`.
55pub type QueryRef = u32;
56/// An index into `Ast::selects`.
57pub type SelectRef = u32;
58/// An index into `Ast::create_tables`.
59pub type CreateTableRef = u32;
60/// An index into `Ast::create_views`.
61pub type CreateViewRef = u32;
62/// An index into `Ast::drop_tables`.
63pub type DropTableRef = u32;
64/// Index into [`Ast::schemas`].
65pub type SchemaRef = u32;
66/// Index into [`Ast::sequences`].
67pub type SequenceRef = u32;
68/// An index into `Ast::inserts`.
69pub type InsertRef = u32;
70/// An index into `Ast::settings`.
71pub type SettingRef = u32;
72/// An index into `Ast::windows`.
73pub type WindowRef = u32;
74
75/// One statement.
76///
77/// Seven of the twenty seven the grammar reaches. The rest are a transform error naming the rule
78/// rather than a variant that nothing fills in, so that adding one is a compile error somewhere
79/// useful rather than a silent `todo!()`.
80#[derive(Debug, Clone, Copy, PartialEq, Eq)]
81pub enum Statement {
82    /// A query, meaning a `SELECT` or a set operation over two of them.
83    Query(QueryRef),
84    /// `CREATE TABLE`.
85    CreateTable(CreateTableRef),
86    /// `CREATE VIEW`.
87    CreateView(CreateViewRef),
88    /// `DROP TABLE` or `DROP VIEW`, which are one rule in the grammar and one statement here.
89    DropTable(DropTableRef),
90    /// `CREATE SCHEMA` or `DROP SCHEMA`.
91    Schema(SchemaRef),
92    /// `CREATE SEQUENCE` or `DROP SEQUENCE`.
93    Sequence(SequenceRef),
94    /// `INSERT INTO`.
95    Insert(InsertRef),
96    /// `UPDATE`, held as an [`Insert`] whose columns are the ones `SET` names and whose source is
97    /// `SELECT *, condition, value, ... FROM table`, one value per named column.
98    ///
99    /// The binder knows how wide the table is and the transform does not, so the source carries
100    /// the table's columns, whether the row matched, and the new values side by side, and the
101    /// binder picks each column's new value or its old one out of them.
102    Update(InsertRef),
103    /// `DELETE FROM` and `TRUNCATE`, held the same way as [`Statement::Update`] with no columns.
104    Delete(InsertRef),
105    /// `SET name = value`.
106    Set(SettingRef),
107    /// `RESET name`, which is the same shape with nothing on the right of it.
108    Reset(SettingRef),
109    /// `CHECKPOINT` or `FORCE CHECKPOINT`.
110    Checkpoint,
111    /// `BEGIN`, `COMMIT` or `ROLLBACK`, under any of the spellings the grammar takes for each.
112    Transaction(Transaction),
113    /// `EXPLAIN` over a query, and whether `ANALYZE` was asked for.
114    ///
115    /// The query rather than a statement, because the grammar lets every statement be explained
116    /// and a plan is the only thing there is to show. `EXPLAIN INSERT` is a refusal rather than a
117    /// plan of the source, since the source is not what the statement does.
118    ///
119    /// `ANALYZE` means the query is run and the plan is printed with what happened on it, so it is
120    /// a flag on the same statement rather than a statement of its own. Everything between the
121    /// parser and the printer is the same either way, which is the point: the analyzed plan has to
122    /// be the plan that ran.
123    ///
124    /// `STATISTICS` asks for the section that says what the planner knew, which is what
125    /// `spec/stats/05-every-query.md` section 5.1.1 asks `EXPLAIN` to print. It is a flag for the
126    /// same reason `ANALYZE` is: it changes what goes on the end of the output and nothing before
127    /// it.
128    Explain { query: QueryRef, analyze: bool, statistics: bool },
129}
130
131/// `SET name = value` and `RESET name`.
132///
133/// One struct for the two, because `RESET name` is `SET name` with no value and giving it its own
134/// arena would mean two of everything to say the same thing twice.
135#[derive(Debug, Clone, Copy, PartialEq, Eq)]
136pub struct Setting {
137    /// The setting name, as written.
138    pub name: StrRef,
139    /// The scope word, if one was written.
140    pub scope: Scope,
141    /// The value, or `NONE` for a `RESET`.
142    ///
143    /// An expression rather than text. `SET memory_limit = '1GB'` writes a string and `SET threads
144    /// = 4` writes a number, and what a setting does with either is the setting's business.
145    pub value: ExprRef,
146    /// Whether the statement was written as a bare `PRAGMA name`.
147    ///
148    /// `PRAGMA disable_optimizer` is a `SET` with the name and the value both folded into one word,
149    /// and which word means what is the catalog's business rather than the parser's, so it arrives
150    /// here as a name with no value and this flag to say that no value is not a `RESET`.
151    pub pragma: bool,
152}
153
154/// Which copy of a setting a statement means.
155#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
156pub enum Scope {
157    /// No scope word, which every setting reads as the one it has.
158    #[default]
159    Unwritten,
160    /// `GLOBAL`.
161    Global,
162    /// `SESSION`.
163    Session,
164    /// `LOCAL`.
165    Local,
166}
167
168impl Scope {
169    /// The word that was written, for the sentence an error prints.
170    #[must_use]
171    pub const fn keyword(self) -> &'static str {
172        match self {
173            Self::Unwritten => "",
174            Self::Global => "GLOBAL",
175            Self::Session => "SESSION",
176            Self::Local => "LOCAL",
177        }
178    }
179}
180
181/// `CREATE TABLE name (columns)` or `CREATE TABLE name AS query`.
182///
183/// Exactly one of `columns` and `query` says what the table is. A column list is the ordinary form
184/// and `query` is `CREATE TABLE AS`, where the columns come from what the query produced and the
185/// only thing the syntax contributes is optionally renaming them, which is `columns` with the types
186/// left as `NONE`.
187#[derive(Debug, Clone, Copy, PartialEq, Eq)]
188pub struct CreateTable {
189    /// The table name, as a run of [`Slice`] parts, outermost first.
190    pub name: Slice,
191    /// The column definitions, as a run of [`ColumnDef`].
192    pub columns: Slice,
193    /// The `AS` query, or `NONE`.
194    pub query: QueryRef,
195    /// Whether `IF NOT EXISTS` was written.
196    pub if_not_exists: bool,
197    /// Whether `OR REPLACE` was written.
198    pub or_replace: bool,
199    /// Whether `TEMP` or `TEMPORARY` was written.
200    pub temporary: bool,
201    /// The column names of each `PRIMARY KEY` and `UNIQUE`, as a run of name lists in the order
202    /// they were written, whether on a column or on the table.
203    pub keys: Slice,
204    /// Which of `keys` is the primary key, or `NONE`.
205    pub primary: u32,
206    /// Every `CHECK` expression, as a run of expressions in the order they were written, whether on
207    /// a column or on the table.
208    pub checks: Slice,
209    /// The columns of each `FOREIGN KEY`, as a run of name lists in the order written, whether on
210    /// a column or on the table.
211    pub foreign: Slice,
212    /// The table each of `foreign` references, as a run of name lists of its parts.
213    pub foreign_tables: Slice,
214    /// The referenced columns of each of `foreign`, as a run of name lists, an empty one when the
215    /// constraint named none and so means the referenced table's primary key.
216    pub foreign_referenced: Slice,
217}
218
219/// One column of a `CREATE TABLE`.
220///
221/// The type is the text as written rather than a resolved type, because resolving a type is the
222/// binder's job and this crate is syntax. `VARCHAR(10)` and `STRUCT(a INTEGER)` reach the binder
223/// as themselves.
224#[derive(Debug, Clone, Copy, PartialEq, Eq)]
225pub struct ColumnDef {
226    /// The column name.
227    pub name: StrRef,
228    /// The type as written, or `NONE` when the definition had none, which only `CREATE TABLE AS`
229    /// allows.
230    pub ty: StrRef,
231    /// Whether `NOT NULL` was written.
232    pub not_null: bool,
233    /// The `DEFAULT` expression, or `NONE` when the definition had none.
234    pub default: ExprRef,
235}
236
237/// `CREATE VIEW name (columns) AS query`.
238///
239/// The body is kept twice over, as a bound reference into this same arena and as the text that was
240/// written. Both are needed and they are needed for different things. The reference is what binds
241/// the body at creation, which is where a view over a table that is not there is refused. The text
242/// is what the catalog keeps, because a view is bound again at every reference rather than frozen
243/// at creation: a view over `SELECT * FROM t` follows `t` when a column is added to it, which was
244/// measured, and the only way to follow it is to have the query to bind again.
245#[derive(Debug, Clone, Copy, PartialEq, Eq)]
246pub struct CreateView {
247    /// The view name, as a run of [`Slice`] parts, outermost first.
248    pub name: Slice,
249    /// The column aliases, as a run of parts, empty when the statement wrote no list.
250    pub columns: Slice,
251    /// The body.
252    pub query: QueryRef,
253    /// The body as it was written, which is what the catalog keeps.
254    pub sql: StrRef,
255    /// Whether `IF NOT EXISTS` was written.
256    pub if_not_exists: bool,
257    /// Whether `OR REPLACE` was written.
258    pub or_replace: bool,
259    /// Whether `TEMP` or `TEMPORARY` was written.
260    pub temporary: bool,
261}
262
263/// `DROP TABLE a, b` or `DROP VIEW a, b`.
264#[derive(Debug, Clone, Copy, PartialEq, Eq)]
265pub struct DropTable {
266    /// The names, as a run of [`Slice`] into `Ast::name_lists`, each of which is a run of parts.
267    pub names: Slice,
268    /// Whether `IF EXISTS` was written.
269    pub if_exists: bool,
270    /// Whether `VIEW` was written where `TABLE` could have been. Dropping one as the other is an
271    /// error rather than a synonym, so which word was written has to survive the transform.
272    pub view: bool,
273}
274
275/// `CREATE SCHEMA name` or `DROP SCHEMA name`.
276#[derive(Debug, Clone, Copy, PartialEq, Eq)]
277pub struct Schema {
278    /// The name, as a run of parts, outermost first.
279    pub name: Slice,
280    /// Whether this is a `DROP` rather than a `CREATE`.
281    pub drop: bool,
282    /// Whether `IF NOT EXISTS` was written on a create or `IF EXISTS` on a drop.
283    pub quiet: bool,
284    /// Whether `OR REPLACE` was written, which only a create can have.
285    pub or_replace: bool,
286    /// Whether `TEMP` or `TEMPORARY` was written, which only a create can have.
287    pub temporary: bool,
288    /// Whether `CASCADE` was written, which only a drop can have.
289    pub cascade: bool,
290}
291
292/// `CREATE SEQUENCE name options` or `DROP SEQUENCE name`.
293///
294/// The options are settled here rather than in the binder, defaults and all, because that is where
295/// the pin settles them and every refusal of a bad combination is a parser error there.
296#[derive(Debug, Clone, Copy, PartialEq, Eq)]
297pub struct Sequence {
298    /// The name, as a run of parts, outermost first.
299    pub name: Slice,
300    /// Whether this is a `DROP` rather than a `CREATE`.
301    pub drop: bool,
302    /// Whether `IF NOT EXISTS` was written on a create or `IF EXISTS` on a drop.
303    pub quiet: bool,
304    /// Whether `OR REPLACE` was written, which only a create can have.
305    pub or_replace: bool,
306    /// Whether `TEMP` or `TEMPORARY` was written, which only a create can have.
307    pub temporary: bool,
308    /// Whether `CASCADE` was written, which only a drop can have.
309    pub cascade: bool,
310    /// What a create settled, and the defaults on a drop.
311    pub options: rudb_common::sequence::Options,
312}
313
314/// `INSERT INTO name (columns) query`.
315#[derive(Debug, Clone, Copy, PartialEq, Eq)]
316pub struct Insert {
317    /// The table name, as a run of parts, outermost first.
318    pub name: Slice,
319    /// The column list, as a run of parts, empty when the statement did not write one.
320    pub columns: Slice,
321    /// What produces the rows, which is a `VALUES` clause or any other query, or `NONE` for
322    /// `DEFAULT VALUES`, which is one row of every column's default.
323    pub source: QueryRef,
324    /// The `RETURNING` list, held as `SELECT list FROM table [AS alias]` and run over the rows the
325    /// statement wrote rather than over the table.
326    pub returning: Option<QueryRef>,
327    /// What an `INSERT` does with a row whose key the table already holds, when it said.
328    pub conflict: Option<Conflict>,
329}
330
331/// `ON CONFLICT`, `INSERT OR REPLACE` or `INSERT OR IGNORE`.
332#[derive(Debug, Clone, Copy, PartialEq, Eq)]
333pub struct Conflict {
334    /// The columns of the key the statement named, as a run of parts, empty when it named none.
335    pub target: Slice,
336    /// What happens to a row that clashes.
337    pub action: ConflictAction,
338}
339
340/// What happens to a row whose key is already held.
341#[derive(Debug, Clone, Copy, PartialEq, Eq)]
342pub enum ConflictAction {
343    /// `DO NOTHING` or `OR IGNORE`: the row is dropped.
344    Nothing,
345    /// `OR REPLACE`: the held row takes the new row's values in the columns the statement wrote.
346    Replace,
347    /// `DO UPDATE SET`, held as `SELECT values..., condition FROM table AS alias POSITIONAL JOIN
348    /// table AS excluded`, which the write runs with the held rows on the left and the new rows on
349    /// the right.
350    Update {
351        /// The columns that are set, as a run of parts, one for each value.
352        columns: Slice,
353        /// The query that works out the values and whether the row is updated at all.
354        query: QueryRef,
355    },
356}
357
358/// A `WITH name AS MATERIALIZED (query)`, which is run once and read wherever it is named.
359///
360/// Only the materialised ones are here. A plain `WITH` and a `NOT MATERIALIZED` one are put into
361/// every place they are named while the tree is being built, the way the reference binary does it,
362/// so by the time anything reads an [`Ast`] there is no name left to resolve.
363#[derive(Debug, Clone, Copy, PartialEq, Eq)]
364pub struct Cte {
365    /// The name it was written with.
366    pub name: StrRef,
367    /// What produces its rows.
368    pub query: QueryRef,
369    /// The column names from `AS name(a, b)`, as a run of [`StrRef`], empty when there were none.
370    pub columns: Slice,
371}
372
373/// A query: a body, plus the modifiers that apply to whatever the body produced.
374///
375/// The split is the grammar's, not an invention. `SelectStatementInternal <- WithClause?
376/// SelectSetOpChain ResultModifiers?` puts `ORDER BY` and `LIMIT` outside the set operator chain,
377/// which is the only place they can go and be right: `a UNION b ORDER BY x` sorts the union and not
378/// the second half of it. Hanging them off `Select` instead would have made that unrepresentable.
379#[derive(Debug, Clone, Copy, PartialEq, Eq)]
380pub struct Query {
381    /// The materialised `WITH` definitions this query introduces, as a run of indexes into
382    /// `Ast::ctes` held in `Ast::cte_lists`, outermost first.
383    ///
384    /// A list of indexes rather than a run of the arena itself, because a materialised `WITH`
385    /// inside another one is pushed while the outer one is still being built, so what one query
386    /// owns is not a contiguous stretch of the arena.
387    pub ctes: Slice,
388    /// What produces the rows.
389    pub body: QueryBody,
390    /// The `ORDER BY` list, as a run of [`OrderItem`].
391    pub order_by: Slice,
392    /// Whether the clause was `ORDER BY ALL`.
393    pub order_by_all: bool,
394    /// The `LIMIT` expression, or `NONE`.
395    pub limit: ExprRef,
396    /// Whether the limit was a percentage rather than a row count.
397    pub limit_percent: bool,
398    /// The `OFFSET` expression, or `NONE`.
399    pub offset: ExprRef,
400}
401
402impl Query {
403    /// A query with no modifiers on it.
404    pub const fn bare(body: QueryBody) -> Self {
405        Self {
406            ctes: Slice { start: 0, len: 0 },
407            body,
408            order_by: Slice { start: 0, len: 0 },
409            order_by_all: false,
410            limit: NONE,
411            limit_percent: false,
412            offset: NONE,
413        }
414    }
415}
416
417/// What produces the rows of a query.
418#[derive(Debug, Clone, Copy, PartialEq, Eq)]
419pub enum QueryBody {
420    /// One `SELECT ... FROM ... WHERE ...` block.
421    Select(SelectRef),
422    /// `UNION`, `EXCEPT` or `INTERSECT` over two queries.
423    SetOp {
424        /// Which operator.
425        op: SetOp,
426        /// Whether duplicates survive.
427        quantifier: Quantifier,
428        /// Whether the columns are matched up by name rather than by position.
429        by_name: bool,
430        /// The query on the left.
431        left: QueryRef,
432        /// The query on the right.
433        right: QueryRef,
434    },
435    /// `VALUES (1, 'a'), (2, 'b')`, as a run of [`Slice`] in `Ast::rows`.
436    ///
437    /// A row count and a column count and nothing else, so it is a query body rather than a
438    /// statement of its own. That is also what makes `INSERT INTO t VALUES (1)` and
439    /// `INSERT INTO t SELECT 1` the same shape by the time anything downstream sees them, which is
440    /// the reason the insert walker does not have two arms.
441    Values(Slice),
442    /// `DESCRIBE SELECT ...`, `DESCRIBE t` and `DESCRIBE 'file.parquet'`.
443    ///
444    /// A query body rather than a statement, because that is where the grammar puts it:
445    /// `SelectStatementType <- ... / DescribeStatement / ...`, so `FROM (DESCRIBE SELECT 1)` is a
446    /// subquery over one and needs no rule of its own. The two spellings that name something
447    /// instead of writing a query arrive here as `DESCRIBE SELECT * FROM that`, which is not a
448    /// shortcut: on the reference binary `DESCRIBE t` and `DESCRIBE SELECT * FROM t` produce the
449    /// same six columns and the same rows, down to the primary key and the default.
450    Describe(QueryRef),
451    /// `SHOW name`, resolved as a setting or a deprecated table description while binding.
452    Show { name: Slice, relation: QueryRef },
453}
454
455/// Which set operator.
456#[derive(Debug, Clone, Copy, PartialEq, Eq)]
457pub enum SetOp {
458    /// `UNION`.
459    Union,
460    /// `EXCEPT`.
461    Except,
462    /// `INTERSECT`.
463    Intersect,
464}
465
466/// Whether a set operator or an aggregate keeps duplicates.
467///
468/// `Unstated` is not the same as `All` even though the two agree for `UNION`, because they disagree
469/// for `INTERSECT` in some dialects and because an error message that says what was written is
470/// better than one that says what it was taken to mean.
471#[derive(Debug, Clone, Copy, PartialEq, Eq)]
472pub enum Quantifier {
473    /// Neither word was written.
474    Unstated,
475    /// `ALL`.
476    All,
477    /// `DISTINCT`.
478    Distinct,
479}
480
481/// What the `DISTINCT` clause of a select said.
482#[derive(Debug, Clone, Copy, PartialEq, Eq)]
483pub enum Distinct {
484    /// No clause, or the no-op `SELECT ALL`.
485    No,
486    /// `SELECT DISTINCT`.
487    Yes,
488    /// `SELECT DISTINCT ON (a, b)`, holding the expressions in the parentheses.
489    On(Slice),
490}
491
492/// One select block.
493///
494/// Every optional expression is `NONE` when it is absent rather than an `Option<u32>`, which keeps
495/// the struct at forty bytes and keeps the absent case spelled the same way it is spelled in the
496/// parse tree arena.
497#[derive(Debug, Clone, Copy, PartialEq, Eq)]
498pub struct Select {
499    /// The `DISTINCT` clause.
500    pub distinct: Distinct,
501    /// The target list, as a run of [`Target`].
502    pub targets: Slice,
503    /// The `FROM` list, as a run of [`SourceRef`]. Several entries mean a cross product.
504    pub from: Slice,
505    /// The `WHERE` expression, or `NONE`.
506    pub filter: ExprRef,
507    /// The `GROUP BY` list, as a run of [`ExprRef`].
508    pub group_by: Slice,
509    /// Whether the clause was `GROUP BY ALL`.
510    pub group_by_all: bool,
511    /// The `HAVING` expression, or `NONE`.
512    pub having: ExprRef,
513}
514
515impl Select {
516    /// An empty select, which is what the transformer fills in from.
517    pub const fn empty() -> Self {
518        Self {
519            distinct: Distinct::No,
520            targets: Slice { start: 0, len: 0 },
521            from: Slice { start: 0, len: 0 },
522            filter: NONE,
523            group_by: Slice { start: 0, len: 0 },
524            group_by_all: false,
525            having: NONE,
526        }
527    }
528}
529
530/// One entry of a target list.
531#[derive(Debug, Clone, Copy, PartialEq, Eq)]
532pub struct Target {
533    /// What is being selected.
534    pub expr: ExprRef,
535    /// The alias, or `NONE`. The binder invents one when there is none, because what it invents
536    /// depends on the expression and that is a binder question rather than a parser question.
537    pub alias: StrRef,
538}
539
540/// One entry of an order by list.
541#[derive(Debug, Clone, Copy, PartialEq, Eq)]
542pub struct OrderItem {
543    /// What to sort on.
544    pub expr: ExprRef,
545    /// The direction.
546    pub order: Order,
547    /// Where nulls go.
548    pub nulls: Nulls,
549}
550
551/// Sort direction, with the unwritten case kept apart from the default it resolves to.
552#[derive(Debug, Clone, Copy, PartialEq, Eq)]
553pub enum Order {
554    /// Nothing was written.
555    Unstated,
556    /// `ASC` or `ASCENDING`.
557    Ascending,
558    /// `DESC` or `DESCENDING`.
559    Descending,
560}
561
562/// Null placement in a sort.
563#[derive(Debug, Clone, Copy, PartialEq, Eq)]
564pub enum Nulls {
565    /// Nothing was written, so the session default applies.
566    Unstated,
567    /// `NULLS FIRST`.
568    First,
569    /// `NULLS LAST`.
570    Last,
571}
572
573/// How a window frame measures the distance to its bounds.
574#[derive(Debug, Clone, Copy, PartialEq, Eq)]
575pub enum WindowUnit {
576    /// `ROWS`, so a bound counts rows.
577    Rows,
578    /// `RANGE`, so a bound is a value offset from the current row's sort key.
579    Range,
580    /// `GROUPS`, so a bound counts runs of rows that tie on the sort key.
581    Groups,
582}
583
584/// One end of a window frame.
585#[derive(Debug, Clone, Copy, PartialEq, Eq)]
586pub enum WindowBound {
587    /// `UNBOUNDED PRECEDING`, the first row of the partition.
588    UnboundedPreceding,
589    /// `n PRECEDING`, holding the offset expression.
590    Preceding(ExprRef),
591    /// `CURRENT ROW`.
592    CurrentRow,
593    /// `n FOLLOWING`, holding the offset expression.
594    Following(ExprRef),
595    /// `UNBOUNDED FOLLOWING`, the last row of the partition.
596    UnboundedFollowing,
597}
598
599/// Which peers of the current row the frame drops once its bounds have been applied.
600#[derive(Debug, Clone, Copy, PartialEq, Eq)]
601pub enum WindowExclude {
602    /// `EXCLUDE NO OTHERS`, which is also what an unwritten clause means.
603    NoOthers,
604    /// `EXCLUDE CURRENT ROW`.
605    CurrentRow,
606    /// `EXCLUDE GROUP`, dropping the current row and everything that ties with it.
607    Group,
608    /// `EXCLUDE TIES`, dropping everything that ties with the current row but keeping it.
609    Ties,
610}
611
612/// Everything inside the parentheses of an `OVER`.
613///
614/// A named window is resolved here rather than downstream, because the resolution is a parser
615/// question on the reference binary: a reference to a window nobody defined is a `Parser Error`
616/// there, and a view written with `OVER w` comes back out of the catalog with the definition
617/// inlined. So nothing after the transform ever sees a name, and there is no window clause on
618/// [`Select`] for it to see one in.
619#[derive(Debug, Clone, Copy, PartialEq, Eq)]
620pub struct WindowSpec {
621    /// The `PARTITION BY` list, as a run of [`ExprRef`], empty when there was no clause.
622    pub partition: Slice,
623    /// The `ORDER BY` list, as a run of [`OrderItem`], empty when there was no clause.
624    pub order: Slice,
625    /// Which of the three units the bounds are measured in.
626    pub unit: WindowUnit,
627    /// Where the frame starts.
628    pub start: WindowBound,
629    /// Where the frame ends.
630    pub end: WindowBound,
631    /// Which peers the frame drops.
632    pub exclude: WindowExclude,
633}
634
635impl WindowSpec {
636    /// The frame a window with no frame clause gets, which the standard fixes and DuckDB follows.
637    pub const DEFAULT_UNIT: WindowUnit = WindowUnit::Range;
638    /// The start a window with no frame clause gets.
639    pub const DEFAULT_START: WindowBound = WindowBound::UnboundedPreceding;
640    /// The end a window with no frame clause gets.
641    pub const DEFAULT_END: WindowBound = WindowBound::CurrentRow;
642
643    /// A window with no clauses at all, which is what `OVER ()` means.
644    pub const fn empty() -> Self {
645        Self {
646            partition: Slice { start: 0, len: 0 },
647            order: Slice { start: 0, len: 0 },
648            unit: Self::DEFAULT_UNIT,
649            start: Self::DEFAULT_START,
650            end: Self::DEFAULT_END,
651            exclude: WindowExclude::NoOthers,
652        }
653    }
654
655    /// Whether the frame is the one an unwritten frame clause means.
656    ///
657    /// This is what decides whether the frame is printed, which is not a matter of taste: the
658    /// printed form is the column name a window target gets when the query wrote no alias, so
659    /// `SELECT sum(x) OVER (ORDER BY x)` has to be named without a frame in it to agree with the
660    /// reference binary.
661    pub fn frame_is_default(&self) -> bool {
662        self.unit == Self::DEFAULT_UNIT
663            && self.start == Self::DEFAULT_START
664            && self.end == Self::DEFAULT_END
665            && self.exclude == WindowExclude::NoOthers
666    }
667}
668
669/// One entry in a `FROM` clause, which is a tree because joins nest.
670#[derive(Debug, Clone, Copy, PartialEq, Eq)]
671pub enum Source {
672    /// A named table, possibly qualified by schema and catalog.
673    Table {
674        /// The name, as a run of [`StrRef`] in `Ast::parts`, outermost first.
675        name: Slice,
676        /// The alias, or `NONE`.
677        alias: StrRef,
678        /// Column aliases from `AS t(a, b)`, as a run of [`StrRef`].
679        columns: Slice,
680    },
681    /// A materialised `WITH` named where a table goes.
682    ///
683    /// Which definition it reads is settled here rather than left as a name, because shadowing is
684    /// a question about where the name was written and this is the only place that still knows.
685    Cte {
686        /// Which definition, as an index into `Ast::ctes`.
687        cte: u32,
688        /// The alias, or `NONE`, which for a bare name is the name itself.
689        alias: StrRef,
690        /// Column aliases from `AS c(a, b)`, as a run of [`StrRef`].
691        columns: Slice,
692    },
693    /// A parenthesised query in the `FROM` clause.
694    Subquery {
695        /// The query.
696        query: QueryRef,
697        /// The alias, or `NONE`.
698        alias: StrRef,
699        /// Column aliases, as a run of [`StrRef`].
700        columns: Slice,
701    },
702    /// A function call where a table goes, such as `range(10)`.
703    ///
704    /// Held with the name as a qualified run rather than a single string, because `main.range(10)`
705    /// is legal and a function in a schema that does not exist has to say so rather than being
706    /// looked up unqualified and found.
707    Function {
708        /// The name, as a run of [`StrRef`] in `Ast::parts`, outermost first.
709        name: Slice,
710        /// The arguments, as a run of [`Target`] where the alias is the parameter name and is
711        /// `NONE` for a positional one.
712        args: Slice,
713        /// The alias, or `NONE`.
714        alias: StrRef,
715        /// Column aliases from `AS t(a, b)`, as a run of [`StrRef`].
716        columns: Slice,
717        /// Whether the call was written as `PRAGMA name` rather than as a function call.
718        ///
719        /// The two are the same query, because `PRAGMA table_info('t')` is rewritten to
720        /// `SELECT * FROM pragma_table_info('t')` here the way upstream rewrites it, and the
721        /// rewritten form is what the plan and the deparser see. What the flag is for is the two
722        /// messages a bad call produces, which upstream writes in the spelling the user used:
723        /// `table_info()` rather than `pragma_table_info()`, and a candidate line reading
724        /// `PRAGMA "table_info"(VARCHAR)`. A user who wrote a pragma and is told about a function
725        /// they did not name has been handed the rewrite to debug rather than their own statement.
726        pragma: bool,
727    },
728    /// A `VALUES` in the `FROM` clause.
729    Values {
730        /// The rows, as a run of [`Slice`] in `Ast::rows`.
731        rows: Slice,
732        /// The alias, or `NONE`.
733        alias: StrRef,
734        /// Column aliases, as a run of [`StrRef`].
735        columns: Slice,
736    },
737    /// Two sources joined.
738    Join {
739        /// The left side.
740        left: SourceRef,
741        /// The right side.
742        right: SourceRef,
743        /// Which join.
744        kind: JoinKind,
745        /// Whether it was written `NATURAL`.
746        natural: bool,
747        /// The `ON` expression, or `NONE`.
748        on: ExprRef,
749        /// The `USING` column list, as a run of [`StrRef`].
750        using: Slice,
751    },
752}
753
754/// Which join.
755#[derive(Debug, Clone, Copy, PartialEq, Eq)]
756pub enum JoinKind {
757    /// `[INNER] JOIN`.
758    Inner,
759    /// `LEFT [OUTER] JOIN`.
760    Left,
761    /// `RIGHT [OUTER] JOIN`.
762    Right,
763    /// `FULL [OUTER] JOIN`.
764    Full,
765    /// `SEMI JOIN`.
766    Semi,
767    /// `ANTI JOIN`.
768    Anti,
769    /// `CROSS JOIN`.
770    Cross,
771    /// `POSITIONAL JOIN`, which is DuckDB's own and pairs rows by ordinal.
772    Positional,
773}
774
775/// One expression.
776///
777/// Twenty four bytes, which is the widest variant rounded up. The precedence chain in the grammar
778/// does not survive into here: twenty levels of `X <- Y Tail*` become one [`Expr::Binary`] tree,
779/// because the levels exist to make the grammar unambiguous and mean nothing afterwards.
780#[derive(Debug, Clone, Copy, PartialEq, Eq)]
781pub enum Expr {
782    /// `*`, or `t.*` with a qualifier.
783    Star {
784        /// The qualifier, as a run of [`StrRef`], empty for a bare star.
785        qualifier: Slice,
786        /// `REPLACE (expression AS column)`, as a run of [`Target`] where the alias is the column
787        /// being replaced, empty for a star with no replace list.
788        ///
789        /// A [`Target`] rather than a type of its own because a replacement is an expression and a
790        /// name, which is exactly what a target is, and because that puts it in the arena every
791        /// other expression and name pair already lives in.
792        replacements: Slice,
793    },
794    /// A column reference, qualified or not.
795    Column {
796        /// The name, as a run of [`StrRef`], outermost first, so `s.t.a` is three parts.
797        name: Slice,
798    },
799    /// A literal, kept as the text that was written.
800    Literal {
801        /// Which kind.
802        kind: LiteralKind,
803        /// The text, with quotes stripped and escapes resolved for a string, `NONE` for a keyword
804        /// literal like `NULL` where the kind already says everything.
805        text: StrRef,
806    },
807    /// A prefix or postfix operator.
808    Unary {
809        /// Which operator.
810        op: UnaryOp,
811        /// What it applies to.
812        operand: ExprRef,
813    },
814    /// An infix operator.
815    Binary {
816        /// Which operator.
817        op: BinaryOp,
818        /// The left operand.
819        left: ExprRef,
820        /// The right operand.
821        right: ExprRef,
822    },
823    /// A function call.
824    Function {
825        /// The name, as a run of [`StrRef`], so `main.count` is two parts.
826        name: Slice,
827        /// The arguments, as a run of [`ExprRef`].
828        args: Slice,
829        /// Whether the call said `DISTINCT`.
830        distinct: bool,
831        /// The `FILTER (WHERE ...)` predicate, or `NONE`. Kept on every call and not only on the
832        /// ones that can carry it, because which names can carry it is a question about the
833        /// function catalog and the parser does not have one.
834        filter: ExprRef,
835    },
836    /// A function call with an `OVER` on the end of it.
837    ///
838    /// Kept apart from [`Expr::Function`] rather than given an optional window, because the two
839    /// are different things by every rule that applies to them: a window call is refused in a
840    /// `WHERE` and in a `HAVING`, it may not appear inside an aggregate, and it resolves against a
841    /// different set of names. A variant that only some of the code has to remember to look at is
842    /// a variant the rest of the code gets wrong.
843    Window {
844        /// The name, as a run of [`StrRef`], so `main.sum` is two parts.
845        name: Slice,
846        /// The arguments, as a run of [`ExprRef`].
847        args: Slice,
848        /// Whether the call said `DISTINCT`.
849        distinct: bool,
850        /// The `FILTER (WHERE ...)` predicate, or `NONE`. It is written before the `OVER` and not
851        /// after it, which is a rule of the grammar rather than of the binder.
852        filter: ExprRef,
853        /// Whether the call said `IGNORE NULLS`. `RESPECT NULLS` is the default and is not kept,
854        /// because the reference binary drops it: a view written with it comes back without it.
855        ignore_nulls: bool,
856        /// The `ORDER BY` written inside the brackets, as a run of [`OrderItem`], empty when there
857        /// was none. This is the order the call reads the rows of its frame in, and it has nothing
858        /// to do with the `ORDER BY` in the `OVER`, which lays the partition out.
859        order: Slice,
860        /// The window itself, into `Ast::windows`.
861        spec: WindowRef,
862    },
863    /// `CAST(x AS t)` or `TRY_CAST(x AS t)`.
864    Cast {
865        /// What is being cast.
866        operand: ExprRef,
867        /// The target type, as the text it was written with. Parsing it is `rudb-common`'s job and
868        /// doing it here would put the type system in the parser.
869        ty: StrRef,
870        /// Whether a failure yields null rather than an error.
871        try_cast: bool,
872    },
873    /// `CASE`, searched or simple.
874    Case {
875        /// The operand of a simple `CASE x WHEN`, or `NONE` for a searched one.
876        operand: ExprRef,
877        /// The arms, as a run of [`CaseArm`].
878        arms: Slice,
879        /// The `ELSE`, or `NONE`.
880        otherwise: ExprRef,
881    },
882    /// `x BETWEEN a AND b`.
883    Between {
884        /// What is being tested.
885        operand: ExprRef,
886        /// The lower bound.
887        low: ExprRef,
888        /// The upper bound.
889        high: ExprRef,
890        /// Whether it was written `NOT BETWEEN`.
891        negated: bool,
892    },
893    /// `x IN (a, b, c)`.
894    In {
895        /// What is being tested.
896        operand: ExprRef,
897        /// The list, as a run of [`ExprRef`].
898        list: Slice,
899        /// Whether it was written `NOT IN`.
900        negated: bool,
901    },
902    /// `x IN (SELECT ...)` or its negation.
903    InSubquery {
904        /// What is being tested.
905        operand: ExprRef,
906        /// The query producing the candidates.
907        query: QueryRef,
908        /// Whether it was written `NOT IN`.
909        negated: bool,
910    },
911    /// `x op ANY (SELECT ...)` or `x op ALL (SELECT ...)`.
912    QuantifiedSubquery {
913        /// The value on the left of the comparison.
914        operand: ExprRef,
915        /// The comparison applied to each candidate.
916        op: BinaryOp,
917        /// The query producing the candidates.
918        query: QueryRef,
919        /// Whether the quantifier was `ALL` rather than `ANY`.
920        all: bool,
921    },
922    /// `DEFAULT` where a value is written, which is the column's default and only means something
923    /// as a whole item of an `INSERT`'s `VALUES` row.
924    Default,
925    /// A prepared statement parameter, written `?`, `?1`, `$1` or `$name`.
926    Parameter {
927        /// The identifier, which is the number for a positional one and the word for a named one.
928        /// A bare `?` is numbered by where it was written, so the identifier is there either way.
929        name: StrRef,
930    },
931    /// A bracketed list of expressions, `[a, b, c]`, which is a LIST value.
932    List {
933        /// The items, as a run of [`ExprRef`], in the order they were written.
934        items: Slice,
935    },
936    /// `LAMBDA x, i: body`, a function written inline as the argument of one that takes it.
937    ///
938    /// It is an expression only so that it can sit in an argument list. Anywhere else it means
939    /// nothing, and the binder says so in upstream's words rather than the parser refusing it,
940    /// because upstream's parser accepts it anywhere too.
941    Lambda {
942        /// The parameter names, as a run of [`StrRef`], in the order they were written.
943        params: Slice,
944        /// What the function computes from them.
945        body: ExprRef,
946    },
947    /// A braced struct, `{'a': 1, b: 2}`, which is a STRUCT value with the field names written.
948    Struct {
949        /// The field names, as a run of [`StrRef`], in the order they were written.
950        names: Slice,
951        /// The values, as a run of [`ExprRef`], one for each name.
952        values: Slice,
953    },
954    /// A parenthesised list of more than one expression, which is a row value.
955    Row {
956        /// The items, as a run of [`ExprRef`].
957        items: Slice,
958    },
959    /// A scalar subquery, `(SELECT ...)` where an expression is expected.
960    Subquery {
961        /// The query.
962        query: QueryRef,
963    },
964    /// `EXISTS (SELECT ...)` or its negation.
965    Exists {
966        /// The query whose cardinality is tested.
967        query: QueryRef,
968        /// Whether `NOT` was written before `EXISTS`.
969        negated: bool,
970    },
971}
972
973/// One `WHEN a THEN b`.
974#[derive(Debug, Clone, Copy, PartialEq, Eq)]
975pub struct CaseArm {
976    /// The `WHEN`.
977    pub when: ExprRef,
978    /// The `THEN`.
979    pub then: ExprRef,
980}
981
982/// What a transaction statement asks for.
983#[derive(Debug, Clone, Copy, PartialEq, Eq)]
984pub enum Transaction {
985    /// `BEGIN` or `START TRANSACTION`, and whether `READ ONLY` was written after it.
986    Begin {
987        /// Whether the transaction may not write.
988        read_only: bool,
989    },
990    /// `COMMIT` or `END`.
991    Commit,
992    /// `ROLLBACK` or `ABORT`.
993    Rollback,
994}
995
996/// Which literal.
997#[derive(Debug, Clone, Copy, PartialEq, Eq)]
998pub enum LiteralKind {
999    /// A number, kept as text because the width it wants depends on where it lands.
1000    Number,
1001    /// A string.
1002    String,
1003    /// A blob, kept as the text a blob prints as, which is the text a cast reads it back from.
1004    Blob,
1005    /// `NULL`.
1006    Null,
1007    /// `TRUE`.
1008    True,
1009    /// `FALSE`.
1010    False,
1011}
1012
1013/// A prefix or postfix operator.
1014#[derive(Debug, Clone, Copy, PartialEq, Eq)]
1015pub enum UnaryOp {
1016    /// `NOT x`.
1017    Not,
1018    /// `-x`.
1019    Negate,
1020    /// `+x`, which is a no-op that still has to survive to the binder so that `+'a'` errors.
1021    Plus,
1022    /// `~x`.
1023    BitNot,
1024    /// `x!`.
1025    Factorial,
1026    /// `x IS NULL` or `x ISNULL`.
1027    IsNull,
1028    /// `x IS NOT NULL` or `x NOTNULL`.
1029    IsNotNull,
1030    /// `x IS TRUE`.
1031    IsTrue,
1032    /// `x IS NOT TRUE`.
1033    IsNotTrue,
1034    /// `x IS FALSE`.
1035    IsFalse,
1036    /// `x IS NOT FALSE`.
1037    IsNotFalse,
1038    /// `x IS UNKNOWN`.
1039    IsUnknown,
1040    /// `x IS NOT UNKNOWN`.
1041    IsNotUnknown,
1042}
1043
1044/// An infix operator.
1045///
1046/// The list is the dialect and not a general idea of what operators are. `Named` is the one open
1047/// door, because `OperatorLiteral` in the grammar takes any run of operator characters that is not
1048/// already a token, and rejecting that here would reject SQL DuckDB accepts.
1049#[derive(Debug, Clone, Copy, PartialEq, Eq)]
1050pub enum BinaryOp {
1051    /// `OR`.
1052    Or,
1053    /// `AND`.
1054    And,
1055    /// `=` or `==`.
1056    Eq,
1057    /// `!=` or `<>`.
1058    NotEq,
1059    /// `<`.
1060    Lt,
1061    /// `>`.
1062    Gt,
1063    /// `<=`.
1064    LtEq,
1065    /// `>=`.
1066    GtEq,
1067    /// `IS DISTINCT FROM`.
1068    IsDistinctFrom,
1069    /// `IS NOT DISTINCT FROM`.
1070    IsNotDistinctFrom,
1071    /// `+`.
1072    Add,
1073    /// `-`.
1074    Subtract,
1075    /// `*`.
1076    Multiply,
1077    /// `/`.
1078    Divide,
1079    /// `//`, integer division.
1080    IntegerDivide,
1081    /// `%`.
1082    Modulo,
1083    /// `^` or `**`.
1084    Power,
1085    /// `&`.
1086    BitAnd,
1087    /// `|`.
1088    BitOr,
1089    /// `<<`.
1090    ShiftLeft,
1091    /// `>>`.
1092    ShiftRight,
1093    /// `||`.
1094    Concat,
1095    /// `LIKE` or `~~`.
1096    Like,
1097    /// `NOT LIKE` or `!~~`.
1098    NotLike,
1099    /// `ILIKE` or `~~*`.
1100    ILike,
1101    /// `NOT ILIKE` or `!~~*`.
1102    NotILike,
1103    /// `GLOB` or `~~~`.
1104    Glob,
1105    /// `SIMILAR TO`.
1106    SimilarTo,
1107    /// `NOT SIMILAR TO`.
1108    NotSimilarTo,
1109    /// `~`, a regex match.
1110    Regex,
1111    /// `!~`, a negated regex match.
1112    NotRegex,
1113    /// `~*`, a case insensitive regex match.
1114    RegexInsensitive,
1115    /// `!~*`, a negated case insensitive regex match.
1116    NotRegexInsensitive,
1117    /// `COLLATE`.
1118    Collate,
1119    /// `AT TIME ZONE`.
1120    AtTimeZone,
1121    /// `->`.
1122    Arrow,
1123    /// `->>`.
1124    LongArrow,
1125    /// `@>`, contains.
1126    Contains,
1127    /// `<@`, contained by.
1128    ContainedBy,
1129    /// `&&`, overlaps.
1130    Overlaps,
1131    /// `^@`, starts with.
1132    StartsWith,
1133    /// `<<=`, an inet operator.
1134    InetContainedByOrEq,
1135    /// `>>=`, an inet operator.
1136    InetContainsOrEq,
1137    /// An operator the dialect does not name, which DuckDB resolves as a binary function of that
1138    /// name. `a <=> b` is the shape.
1139    Named(StrRef),
1140}
1141
1142/// A parsed statement or script, with every arena it points into.
1143///
1144/// Cheap to clone, cheap to send, and self contained: no index in here refers to anything outside
1145/// it, and nothing in here borrows the query text. The text is copied into `strings` on the way in,
1146/// which costs one allocation per distinct identifier and buys an `Ast` that outlives the string it
1147/// came from.
1148#[derive(Debug, Clone, Default, PartialEq, Eq)]
1149pub struct Ast {
1150    /// The statements in the script, in order.
1151    pub statements: Vec<Statement>,
1152    /// The query arena.
1153    pub queries: Vec<Query>,
1154    /// Source ranges parallel to `queries`.
1155    pub query_spans: Vec<Span>,
1156    /// The select arena.
1157    pub selects: Vec<Select>,
1158    /// The expression arena.
1159    pub exprs: Vec<Expr>,
1160    /// Source ranges parallel to `exprs`.
1161    pub expr_spans: Vec<Span>,
1162    /// The from-item arena.
1163    pub sources: Vec<Source>,
1164    /// Interned text. Identifiers keep the case they were written in, because DuckDB does not fold
1165    /// it at any point, including for quoted identifiers.
1166    pub strings: Vec<String>,
1167    /// Backing store for every [`Slice`] of names.
1168    pub parts: Vec<StrRef>,
1169    /// Backing store for every [`Slice`] of expressions.
1170    pub expr_lists: Vec<ExprRef>,
1171    /// Backing store for every [`Slice`] of from items.
1172    pub source_lists: Vec<SourceRef>,
1173    /// Backing store for every [`Slice`] of target list entries.
1174    pub targets: Vec<Target>,
1175    /// Backing store for every [`Slice`] of order by entries.
1176    pub order_items: Vec<OrderItem>,
1177    /// Backing store for every [`Slice`] of case arms.
1178    pub case_arms: Vec<CaseArm>,
1179    /// The `CREATE TABLE` arena.
1180    pub create_tables: Vec<CreateTable>,
1181    /// The `CREATE VIEW` arena.
1182    pub create_views: Vec<CreateView>,
1183    /// The `DROP TABLE` arena.
1184    pub drop_tables: Vec<DropTable>,
1185    /// The `CREATE SCHEMA` and `DROP SCHEMA` arena.
1186    pub schemas: Vec<Schema>,
1187    /// The `CREATE SEQUENCE` and `DROP SEQUENCE` arena.
1188    pub sequences: Vec<Sequence>,
1189    /// The `INSERT` arena.
1190    pub inserts: Vec<Insert>,
1191    /// The `SET` and `RESET` arena.
1192    pub settings: Vec<Setting>,
1193    /// Backing store for every [`Slice`] of column definitions.
1194    pub column_defs: Vec<ColumnDef>,
1195    /// Backing store for every [`Slice`] of names, which is a name list rather than a name.
1196    pub name_lists: Vec<Slice>,
1197    /// Backing store for the rows of a `VALUES`, each of which is a run of expressions.
1198    pub rows: Vec<Slice>,
1199    /// The window arena, holding what was inside the parentheses of every `OVER`.
1200    pub windows: Vec<WindowSpec>,
1201    /// The materialised `WITH` arena.
1202    pub ctes: Vec<Cte>,
1203    /// Backing store for every [`Slice`] of materialised `WITH` indexes.
1204    pub cte_lists: Vec<u32>,
1205}
1206
1207impl Ast {
1208    /// The source range of an expression.
1209    pub fn expr_span(&self, expr: ExprRef) -> Span {
1210        self.expr_spans[expr as usize]
1211    }
1212
1213    /// The source range of a query.
1214    pub fn query_span(&self, query: QueryRef) -> Span {
1215        self.query_spans[query as usize]
1216    }
1217
1218    /// The text behind a [`StrRef`], or the empty string for `NONE`.
1219    pub fn string(&self, index: StrRef) -> &str {
1220        if index == NONE { "" } else { &self.strings[index as usize] }
1221    }
1222
1223    /// Every parameter identifier the statement uses, once each, in the order they were written.
1224    ///
1225    /// The arena is built as the walk goes, so its order is the written order, and a parameter used
1226    /// twice is one identifier here because it is one value to provide.
1227    pub fn parameters(&self) -> Vec<&str> {
1228        let mut found: Vec<&str> = Vec::new();
1229        for expr in &self.exprs {
1230            if let Expr::Parameter { name } = *expr {
1231                let name = self.string(name);
1232                if !found.contains(&name) {
1233                    found.push(name);
1234                }
1235            }
1236        }
1237        found
1238    }
1239
1240    /// The parts of a name, outermost first.
1241    pub fn name(&self, slice: Slice) -> impl Iterator<Item = &str> {
1242        self.parts[slice.range()].iter().map(|&part| self.string(part))
1243    }
1244
1245    /// A name written back out with dots between the parts, for error messages and tests.
1246    pub fn name_text(&self, slice: Slice) -> String {
1247        self.name(slice).collect::<Vec<_>>().join(".")
1248    }
1249
1250    /// One expression.
1251    pub fn expr(&self, index: ExprRef) -> Expr {
1252        self.exprs[index as usize]
1253    }
1254
1255    /// One from item.
1256    pub fn source(&self, index: SourceRef) -> Source {
1257        self.sources[index as usize]
1258    }
1259
1260    /// One query.
1261    pub fn query(&self, index: QueryRef) -> Query {
1262        self.queries[index as usize]
1263    }
1264
1265    /// One select block.
1266    pub fn select(&self, index: SelectRef) -> Select {
1267        self.selects[index as usize]
1268    }
1269
1270    /// One window.
1271    pub fn window(&self, index: WindowRef) -> WindowSpec {
1272        self.windows[index as usize]
1273    }
1274
1275    /// One materialised `WITH` definition.
1276    pub fn cte(&self, index: u32) -> Cte {
1277        self.ctes[index as usize]
1278    }
1279
1280    /// The materialised `WITH` definitions a query introduces, outermost first.
1281    pub fn cte_list(&self, slice: Slice) -> &[u32] {
1282        &self.cte_lists[slice.range()]
1283    }
1284
1285    /// The expressions of a list.
1286    pub fn expr_list(&self, slice: Slice) -> &[ExprRef] {
1287        &self.expr_lists[slice.range()]
1288    }
1289
1290    /// The from items of a list.
1291    pub fn source_list(&self, slice: Slice) -> &[SourceRef] {
1292        &self.source_lists[slice.range()]
1293    }
1294
1295    /// The entries of a target list.
1296    pub fn target_list(&self, slice: Slice) -> &[Target] {
1297        &self.targets[slice.range()]
1298    }
1299
1300    /// The entries of an order by list.
1301    pub fn order_list(&self, slice: Slice) -> &[OrderItem] {
1302        &self.order_items[slice.range()]
1303    }
1304
1305    /// The arms of a case.
1306    pub fn arm_list(&self, slice: Slice) -> &[CaseArm] {
1307        &self.case_arms[slice.range()]
1308    }
1309
1310    /// One `CREATE TABLE`.
1311    pub fn create_table(&self, index: CreateTableRef) -> CreateTable {
1312        self.create_tables[index as usize]
1313    }
1314
1315    /// One `CREATE VIEW`.
1316    pub fn create_view(&self, index: CreateViewRef) -> CreateView {
1317        self.create_views[index as usize]
1318    }
1319
1320    /// One `DROP TABLE`.
1321    pub fn drop_table(&self, index: DropTableRef) -> DropTable {
1322        self.drop_tables[index as usize]
1323    }
1324
1325    /// One `CREATE SCHEMA` or `DROP SCHEMA`.
1326    pub fn schema(&self, index: SchemaRef) -> Schema {
1327        self.schemas[index as usize]
1328    }
1329
1330    /// One `CREATE SEQUENCE` or `DROP SEQUENCE`.
1331    pub fn sequence(&self, index: SequenceRef) -> Sequence {
1332        self.sequences[index as usize]
1333    }
1334
1335    /// One `INSERT`.
1336    pub fn insert(&self, index: InsertRef) -> Insert {
1337        self.inserts[index as usize]
1338    }
1339
1340    /// One `SET` or `RESET`.
1341    pub fn setting(&self, index: SettingRef) -> Setting {
1342        self.settings[index as usize]
1343    }
1344
1345    /// The column definitions of a `CREATE TABLE`.
1346    pub fn column_defs(&self, slice: Slice) -> &[ColumnDef] {
1347        &self.column_defs[slice.range()]
1348    }
1349
1350    /// The names of a name list, each of which is itself a run of parts.
1351    pub fn name_list(&self, slice: Slice) -> &[Slice] {
1352        &self.name_lists[slice.range()]
1353    }
1354
1355    /// The rows of a `VALUES`, each of which is itself a run of expressions.
1356    pub fn rows(&self, slice: Slice) -> &[Slice] {
1357        &self.rows[slice.range()]
1358    }
1359
1360    /// How many nodes the whole tree is, across every arena.
1361    ///
1362    /// The number to watch when the transformer changes. A parse tree of five thousand nodes that
1363    /// becomes an AST of thirty is the twenty precedence levels being thrown away, which is the
1364    /// whole reason this module exists.
1365    pub fn node_count(&self) -> usize {
1366        self.queries.len() + self.selects.len() + self.exprs.len() + self.sources.len()
1367    }
1368}