uqa_sql/ast/
expressions.rs

1//
2// Unified Query Algebra
3//
4// Copyright (c) 2023-2026 Cognica, Inc.
5//
6
7use serde::{Deserialize, Serialize};
8use std::sync::atomic::{AtomicU64, Ordering};
9use uqa_core::Value;
10
11use super::{FunctionBinding, SelectStmt};
12
13/// Query-local identity for an executor-only row source. Parser-produced SQL never contains this identity, so internal row carriers cannot collide with user relation aliases.
14#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
15#[doc(hidden)]
16pub struct InternalRelationId(u64);
17
18impl InternalRelationId {
19    /// Allocate an opaque relation identity for an engine-injected row source.
20    #[must_use]
21    pub fn allocate() -> Self {
22        static NEXT_ID: AtomicU64 = AtomicU64::new(1);
23        let id = uqa_core::atomic::try_update_u64(
24            &NEXT_ID,
25            Ordering::Relaxed,
26            Ordering::Relaxed,
27            |current| current.checked_add(1),
28        )
29        .expect("internal relation identity space exhausted");
30        Self(id)
31    }
32
33    /// Address one zero-based attribute of this internal relation.
34    #[must_use]
35    pub fn column(self, attribute: usize) -> InternalColumnRef {
36        InternalColumnRef {
37            relation: self,
38            attribute: u32::try_from(attribute).expect("internal relation attribute exceeds u32"),
39        }
40    }
41
42    #[must_use]
43    pub const fn raw(self) -> u64 {
44        self.0
45    }
46
47    #[must_use]
48    pub const fn from_raw(raw: u64) -> Self {
49        Self(raw)
50    }
51}
52
53/// Structural reference to an executor-only relation attribute. This is the UQA analogue of PostgreSQL's `Var(varno, varattno)` identity: it is never resolved through SQL text names.
54#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
55#[doc(hidden)]
56pub struct InternalColumnRef {
57    relation: InternalRelationId,
58    attribute: u32,
59}
60
61impl InternalColumnRef {
62    #[must_use]
63    pub const fn relation(self) -> InternalRelationId {
64        self.relation
65    }
66
67    #[must_use]
68    pub const fn attribute(self) -> usize {
69        self.attribute as usize
70    }
71
72    #[must_use]
73    pub const fn from_raw(relation: u64, attribute: u32) -> Self {
74        Self {
75            relation: InternalRelationId::from_raw(relation),
76            attribute,
77        }
78    }
79}
80
81#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
82pub struct Projection {
83    pub expr: Expr,
84    pub alias: Option<String>,
85}
86
87#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
88pub struct OrderBy {
89    pub expr: Expr,
90    pub descending: bool,
91    /// `NULLS FIRST` / `NULLS LAST` placement. `None` means the
92    /// SQL-standard default - `NULLS LAST` for ASC and `NULLS FIRST`
93    /// for DESC. Mirrors `PostgreSQL` semantics.
94    pub nulls: Option<NullsOrder>,
95}
96
97#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
98pub enum NullsOrder {
99    First,
100    Last,
101}
102
103/// One query-local window definition. The specification owns only its written clauses; inheritance is resolved through an earlier named definition.
104#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
105pub struct WindowDefinition<S = WindowSpec> {
106    pub name: Option<String>,
107    pub inherited: Option<usize>,
108    pub spec: S,
109}
110
111/// Compiler-only equality key for a raw window declaration; not part of stored SQL.
112#[derive(Debug, Clone, PartialEq)]
113pub struct WindowDefinitionSyntax(pub(crate) serde_json::Value);
114
115#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
116pub struct WindowSpec {
117    /// Raw syntax survives only until the compiler selects its query-local definition.
118    #[serde(skip)]
119    pub raw_definition: Option<WindowDefinitionSyntax>,
120    /// Canonical definition in the enclosing query block. Older stored inline specifications have no slot.
121    #[serde(default, skip_serializing_if = "Option::is_none")]
122    pub definition: Option<usize>,
123    /// Named window referenced by this specification while the SQL compiler resolves a `WINDOW` clause. Compiler-produced plans clear this field before lowering into the unified scalar IR.
124    #[serde(default, skip_serializing_if = "Option::is_none")]
125    pub reference: Option<WindowReference>,
126    pub partition_by: Vec<Expr>,
127    pub order_by: Vec<OrderBy>,
128    /// `ROWS` / `RANGE` frame, or `None` when not specified (defaults
129    /// to `RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW`).
130    pub frame: Option<WindowFrame>,
131}
132
133impl WindowSpec {
134    /// Scalar roots owned by this specification, excluding separately stored inherited clauses.
135    pub fn expressions_mut(&mut self) -> impl Iterator<Item = &mut Expr> {
136        self.partition_by
137            .iter_mut()
138            .chain(self.order_by.iter_mut().map(|order| &mut order.expr))
139            .chain(
140                self.frame
141                    .iter_mut()
142                    .flat_map(|frame| [&mut frame.start, &mut frame.end])
143                    .filter_map(|bound| match bound {
144                        FrameBound::Preceding(value) | FrameBound::Following(value) => {
145                            Some(value.as_mut())
146                        }
147                        _ => None,
148                    }),
149            )
150    }
151}
152
153#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
154pub struct WindowReference {
155    pub name: String,
156    pub kind: WindowReferenceKind,
157}
158
159#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
160pub enum WindowReferenceKind {
161    /// `OVER window_name` uses the named definition directly, including its frame.
162    Direct,
163    /// `OVER (window_name ...)` or `WINDOW child AS (parent ...)` copies and may extend a frameless definition.
164    Copy,
165}
166
167#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
168pub struct WindowFrame {
169    pub mode: FrameMode,
170    pub start: FrameBound,
171    pub end: FrameBound,
172    /// Whether the frame was written `BETWEEN start AND end`. A frame that names only its start, such as `ROWS UNBOUNDED PRECEDING`, ends at the current row, and the definition is deparsed as it was written.
173    #[serde(default = "super::default_true")]
174    pub between: bool,
175    #[serde(default, skip_serializing_if = "FrameExclusion::is_no_others")]
176    pub exclusion: FrameExclusion,
177}
178
179#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
180pub enum FrameMode {
181    Rows,
182    Range,
183    Groups,
184}
185
186/// The written function-call form, retained independently of overload binding. The durable `order_syntax` field keeps its original name for stored-expression compatibility.
187#[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Serialize, Deserialize)]
188pub enum FunctionCallSyntax {
189    #[default]
190    Legacy,
191    Ordinary,
192    WithinGroup,
193    Extract,
194}
195
196/// The original public name, preserved for callers and stored ordering metadata.
197pub type FunctionOrderSyntax = FunctionCallSyntax;
198
199impl FunctionCallSyntax {
200    #[must_use]
201    pub const fn is_legacy(&self) -> bool {
202        matches!(self, Self::Legacy)
203    }
204}
205
206/// The aggregate modifiers written on a window call, which `ParseFuncOrColumn` rejects after it has resolved the function. The arguments of a call written with `WITHIN GROUP` include its ordering expressions.
207#[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Serialize, Deserialize)]
208pub struct WindowCallModifiers {
209    pub distinct: bool,
210    /// An aggregate `ORDER BY` inside the argument list.
211    pub ordered: bool,
212    pub within_group: bool,
213}
214
215impl WindowCallModifiers {
216    #[must_use]
217    pub const fn is_empty(&self) -> bool {
218        !self.distinct && !self.ordered && !self.within_group
219    }
220}
221
222/// The frame exclusion clause: the rows of the frame that a window function or aggregate does not see.
223#[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Serialize, Deserialize)]
224pub enum FrameExclusion {
225    #[default]
226    NoOthers,
227    CurrentRow,
228    /// The current row and its peers.
229    Group,
230    /// The peers of the current row, but not the current row itself.
231    Ties,
232}
233
234impl FrameExclusion {
235    #[must_use]
236    pub const fn is_no_others(&self) -> bool {
237        matches!(self, Self::NoOthers)
238    }
239
240    /// The clause as `pg_get_viewdef` spells it, or `None` for `EXCLUDE NO OTHERS`, which it omits.
241    #[must_use]
242    pub const fn sql(self) -> Option<&'static str> {
243        match self {
244            Self::NoOthers => None,
245            Self::CurrentRow => Some("EXCLUDE CURRENT ROW"),
246            Self::Group => Some("EXCLUDE GROUP"),
247            Self::Ties => Some("EXCLUDE TIES"),
248        }
249    }
250}
251
252#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
253pub enum FrameBound {
254    UnboundedPreceding,
255    UnboundedFollowing,
256    CurrentRow,
257    Preceding(Box<Expr>),
258    Following(Box<Expr>),
259}
260
261/// Scalar expression nodes the compiler handles.
262#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
263pub enum Expr {
264    Star,
265    /// Relation-qualified wildcard projection (`table.*` or `alias.*`).
266    QualifiedStar(String),
267    /// `DEFAULT` in an INSERT/UPDATE assignment is a mutation marker resolved against the target column before expression evaluation.
268    Default,
269    /// Unqualified column reference (`col`).
270    Column(String),
271    /// Qualified column reference (`table.col` or `alias.col`).
272    QualifiedColumn {
273        qualifier: String,
274        column: String,
275    },
276    /// Structural column reference emitted internally rather than by SQL parsing; SQL name binding must not rewrite it.
277    #[doc(hidden)]
278    InternalColumn(InternalColumnRef),
279    Literal(Value),
280    /// An already-coerced runtime datum with its declared SQL type. Variable binding emits this leaf so reading a domain value does not repeat its constraints.
281    #[doc(hidden)]
282    TypedLiteral {
283        value: Value,
284        ty: String,
285    },
286    /// A positional bind parameter (`$1`, `$2`, ...).
287    Param(usize),
288    /// Function call dispatched through the function registry.
289    Func {
290        name: String,
291        #[serde(default, skip_serializing_if = "Option::is_none")]
292        binding: Option<FunctionBinding>,
293        args: Vec<Expr>,
294        /// `func(DISTINCT expr)` - only meaningful for aggregate
295        /// functions. Mirrors `PostgreSQL`'s `agg_distinct`.
296        distinct: bool,
297        /// Ordering expressions from either `func(expr ORDER BY ...)` or `func(expr) WITHIN GROUP (ORDER BY ...)`; `order_syntax` preserves their written location.
298        order_by: Vec<OrderBy>,
299        #[serde(default, skip_serializing_if = "FunctionOrderSyntax::is_legacy")]
300        order_syntax: FunctionOrderSyntax,
301        /// `func(...) FILTER (WHERE expr)` - aggregate-level row filter.
302        filter: Option<Box<Expr>>,
303    },
304    /// SQL array constructor.
305    Array(Vec<Expr>),
306    /// Anonymous SQL row constructor (`ROW(...)` or `(a, b)`).
307    Row(Vec<Expr>),
308    /// A stored typed row with creation-time attribute positions and already selected field conversions.
309    CompositeRow {
310        items: Vec<Expr>,
311        binding: super::CompositeRowBinding,
312    },
313    /// `lhs op rhs` - comparison or arithmetic.
314    Binary {
315        op: BinaryOp,
316        lhs: Box<Expr>,
317        rhs: Box<Expr>,
318    },
319    /// `PostgreSQL` prefix `-`, kept distinct from binary subtraction so the
320    /// operand's declared numeric width and overflow behavior survive lowering.
321    UnaryMinus(Box<Expr>),
322    /// `NOT expr`.
323    Not(Box<Expr>),
324    /// `cond_1 AND cond_2 AND ...` (n-ary).
325    And(Vec<Expr>),
326    /// `cond_1 OR cond_2 OR ...` (n-ary).
327    Or(Vec<Expr>),
328    /// `expr IS NULL` / `expr IS NOT NULL`.
329    IsNull {
330        expr: Box<Expr>,
331        negated: bool,
332    },
333    /// `expr BETWEEN low AND high`.
334    Between {
335        expr: Box<Expr>,
336        low: Box<Expr>,
337        high: Box<Expr>,
338    },
339    /// `expr IN (a, b, c)` literal list.
340    InList {
341        expr: Box<Expr>,
342        list: Vec<Expr>,
343        negated: bool,
344    },
345    /// `func(args) [FILTER (WHERE condition)] OVER (PARTITION BY ... ORDER BY ...)`. Only aggregates accept `FILTER`.
346    WindowCall {
347        name: String,
348        args: Vec<Expr>,
349        spec: Box<WindowSpec>,
350        #[serde(default, skip_serializing_if = "Option::is_none")]
351        filter: Option<Box<Expr>>,
352        #[serde(default, skip_serializing_if = "WindowCallModifiers::is_empty")]
353        modifiers: WindowCallModifiers,
354    },
355    /// `CASE [base] WHEN cond THEN result ... [ELSE default] END`.
356    /// `base` lifts simple-form `CASE expr WHEN val THEN ...` into an
357    /// optional comparison anchor; searched-form `CASE WHEN cond ...`
358    /// leaves it `None`.
359    Case {
360        base: Option<Box<Expr>>,
361        when: Vec<(Expr, Expr)>,
362        else_branch: Option<Box<Expr>>,
363    },
364    /// `CAST(expr AS type)`. The type name is preserved verbatim so
365    /// the evaluator can apply the correct coercion.
366    Cast {
367        /// Set by analysis for an implicit coercion retained in stored syntax.
368        #[serde(default, skip_serializing_if = "std::ops::Not::not")]
369        implicit: bool,
370        expr: Box<Expr>,
371        ty: String,
372    },
373    /// `(SELECT ...)` query expression: ordinary scalar consumers select one
374    /// column; a multiple-column SET target consumes the positional result.
375    ScalarSubquery(Box<SelectStmt>),
376    /// `EXISTS (SELECT ...)` -- truthy when the body produces at
377    /// least one row.
378    Exists {
379        body: Box<SelectStmt>,
380        negated: bool,
381    },
382    /// `expr [NOT] IN (SELECT ...)` set membership against a
383    /// subquery. Evaluator runs the body once per top-level
384    /// expression and tests membership.
385    InSubquery {
386        expr: Box<Expr>,
387        body: Box<SelectStmt>,
388        negated: bool,
389    },
390}
391
392impl Expr {
393    pub fn qualified_column(qualifier: impl Into<String>, column: impl Into<String>) -> Self {
394        Self::QualifiedColumn {
395            qualifier: qualifier.into(),
396            column: column.into(),
397        }
398    }
399
400    /// Upgrade compiler-owned function markers deserialized from catalogs
401    /// written by releases through 0.1.6.
402    #[doc(hidden)]
403    #[expect(
404        clippy::too_many_lines,
405        reason = "exhaustive AST migration preserves every serialized variant"
406    )]
407    pub fn upgrade_legacy_serialized_dispatches(&mut self) -> bool {
408        let mut changed = false;
409        match self {
410            Self::Func {
411                name,
412                binding,
413                args,
414                order_by,
415                filter,
416                ..
417            } => {
418                for argument in args {
419                    changed |= argument.upgrade_legacy_serialized_dispatches();
420                }
421                for order in order_by {
422                    changed |= order.expr.upgrade_legacy_serialized_dispatches();
423                }
424                if let Some(filter) = filter {
425                    changed |= filter.upgrade_legacy_serialized_dispatches();
426                }
427                changed |=
428                    super::FunctionBinding::upgrade_legacy_serialized_dispatch(name, binding);
429            }
430            Self::Array(items)
431            | Self::Row(items)
432            | Self::CompositeRow { items, .. }
433            | Self::And(items)
434            | Self::Or(items) => {
435                for item in items {
436                    changed |= item.upgrade_legacy_serialized_dispatches();
437                }
438            }
439            Self::Binary { lhs, rhs, .. } => {
440                changed |= lhs.upgrade_legacy_serialized_dispatches();
441                changed |= rhs.upgrade_legacy_serialized_dispatches();
442            }
443            Self::UnaryMinus(inner)
444            | Self::Not(inner)
445            | Self::IsNull { expr: inner, .. }
446            | Self::Cast { expr: inner, .. } => {
447                changed |= inner.upgrade_legacy_serialized_dispatches();
448            }
449            Self::Between { expr, low, high } => {
450                changed |= expr.upgrade_legacy_serialized_dispatches();
451                changed |= low.upgrade_legacy_serialized_dispatches();
452                changed |= high.upgrade_legacy_serialized_dispatches();
453            }
454            Self::InList { expr, list, .. } => {
455                changed |= expr.upgrade_legacy_serialized_dispatches();
456                for item in list {
457                    changed |= item.upgrade_legacy_serialized_dispatches();
458                }
459            }
460            Self::WindowCall {
461                args, spec, filter, ..
462            } => {
463                for argument in args {
464                    changed |= argument.upgrade_legacy_serialized_dispatches();
465                }
466                if let Some(filter) = filter {
467                    changed |= filter.upgrade_legacy_serialized_dispatches();
468                }
469                for partition in &mut spec.partition_by {
470                    changed |= partition.upgrade_legacy_serialized_dispatches();
471                }
472                for order in &mut spec.order_by {
473                    changed |= order.expr.upgrade_legacy_serialized_dispatches();
474                }
475                if let Some(frame) = &mut spec.frame {
476                    for bound in [&mut frame.start, &mut frame.end] {
477                        match bound {
478                            FrameBound::Preceding(expression)
479                            | FrameBound::Following(expression) => {
480                                changed |= expression.upgrade_legacy_serialized_dispatches();
481                            }
482                            FrameBound::UnboundedPreceding
483                            | FrameBound::UnboundedFollowing
484                            | FrameBound::CurrentRow => {}
485                        }
486                    }
487                }
488            }
489            Self::Case {
490                base,
491                when,
492                else_branch,
493            } => {
494                if let Some(base) = base {
495                    changed |= base.upgrade_legacy_serialized_dispatches();
496                }
497                for (condition, result) in when {
498                    changed |= condition.upgrade_legacy_serialized_dispatches();
499                    changed |= result.upgrade_legacy_serialized_dispatches();
500                }
501                if let Some(branch) = else_branch {
502                    changed |= branch.upgrade_legacy_serialized_dispatches();
503                }
504            }
505            Self::InSubquery { expr, body, .. } => {
506                changed |= expr.upgrade_legacy_serialized_dispatches();
507                changed |= body.upgrade_legacy_serialized_dispatches();
508            }
509            Self::ScalarSubquery(body) | Self::Exists { body, .. } => {
510                changed |= body.upgrade_legacy_serialized_dispatches();
511            }
512            Self::Default
513            | Self::Star
514            | Self::QualifiedStar(_)
515            | Self::Column(_)
516            | Self::QualifiedColumn { .. }
517            | Self::InternalColumn(_)
518            | Self::Literal(_)
519            | Self::TypedLiteral { .. }
520            | Self::Param(_) => {}
521        }
522        changed
523    }
524
525    /// True when this expression tree contains a window function call.
526    #[must_use]
527    pub fn contains_window(&self) -> bool {
528        self.any_node(&|node| matches!(node, Self::WindowCall { .. }))
529    }
530
531    /// True when this expression tree contains a built-in aggregate call.
532    #[must_use]
533    pub fn contains_aggregate(&self) -> bool {
534        self.any_node(
535            &|node| matches!(node, Self::Func { name, .. } if is_builtin_aggregate_function(name)),
536        )
537    }
538
539    /// True when this expression contains a column whose owning relation can only be determined after catalog schemas have been bound.
540    #[must_use]
541    pub fn contains_unqualified_column(&self) -> bool {
542        self.any_node(&|node| matches!(node, Self::Column(_)))
543    }
544
545    /// True when this expression contains a function whose strictness cannot be decided without an engine catalog.
546    #[must_use]
547    pub fn contains_function_with_unknown_strictness(&self) -> bool {
548        self.any_node(&|node| {
549            matches!(
550                node,
551                Self::Func {
552                    name,
553                    args,
554                    binding,
555                    ..
556                } if crate::expr::bound_scalar_function_strictness(
557                    name,
558                    binding.as_ref(),
559                    args.len(),
560                )
561                .is_none()
562            )
563        })
564    }
565
566    /// Whether `hit` matches this node or any scalar node below it. Subquery bodies are opaque because they own independent query trees.
567    #[must_use]
568    pub fn any_node(&self, hit: &dyn Fn(&Self) -> bool) -> bool {
569        if hit(self) {
570            return true;
571        }
572        match self {
573            Self::Func {
574                args,
575                order_by,
576                filter,
577                ..
578            } => {
579                args.iter().any(|arg| arg.any_node(hit))
580                    || order_by.iter().any(|order| order.expr.any_node(hit))
581                    || filter.as_deref().is_some_and(|filter| filter.any_node(hit))
582            }
583            Self::Array(items)
584            | Self::Row(items)
585            | Self::CompositeRow { items, .. }
586            | Self::And(items)
587            | Self::Or(items) => items.iter().any(|item| item.any_node(hit)),
588            Self::UnaryMinus(expr) | Self::Not(expr) | Self::Cast { expr, .. } => {
589                expr.any_node(hit)
590            }
591            Self::Binary { lhs, rhs, .. } => lhs.any_node(hit) || rhs.any_node(hit),
592            Self::IsNull { expr, .. } | Self::InSubquery { expr, .. } => expr.any_node(hit),
593            Self::Between { expr, low, high } => {
594                expr.any_node(hit) || low.any_node(hit) || high.any_node(hit)
595            }
596            Self::InList { expr, list, .. } => {
597                expr.any_node(hit) || list.iter().any(|item| item.any_node(hit))
598            }
599            Self::Case {
600                base,
601                when,
602                else_branch,
603            } => {
604                base.as_deref().is_some_and(|base| base.any_node(hit))
605                    || when
606                        .iter()
607                        .any(|(condition, result)| condition.any_node(hit) || result.any_node(hit))
608                    || else_branch
609                        .as_deref()
610                        .is_some_and(|branch| branch.any_node(hit))
611            }
612            Self::WindowCall { .. }
613            | Self::Star
614            | Self::QualifiedStar(_)
615            | Self::Default
616            | Self::Column(_)
617            | Self::QualifiedColumn { .. }
618            | Self::InternalColumn(_)
619            | Self::Literal(_)
620            | Self::TypedLiteral { .. }
621            | Self::Param(_)
622            | Self::ScalarSubquery(_)
623            | Self::Exists { .. } => false,
624        }
625    }
626}
627
628/// Return whether `name` is a built-in aggregate understood by the planner.
629#[must_use]
630pub fn is_builtin_aggregate_function(name: &str) -> bool {
631    matches!(
632        name.to_ascii_lowercase().as_str(),
633        "count"
634            | "sum"
635            | "avg"
636            | "min"
637            | "max"
638            | "string_agg"
639            | "array_agg"
640            | "bool_and"
641            | "bool_or"
642            | "stddev"
643            | "stddev_samp"
644            | "stddev_pop"
645            | "variance"
646            | "var_samp"
647            | "var_pop"
648            | "percentile_cont"
649            | "percentile_disc"
650            | "mode"
651            | "json_agg"
652            | "jsonb_agg"
653            | "json_object_agg"
654            | "jsonb_object_agg"
655    )
656}
657
658#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
659pub enum BinaryOp {
660    Equal,
661    NotEqual,
662    Less,
663    LessEqual,
664    Greater,
665    GreaterEqual,
666    Add,
667    Subtract,
668    Multiply,
669    Divide,
670}
671
672/// `Expr` restricted to value-producing forms used by `INSERT` rows.
673pub type ValueExpr = Expr;
uqa_sql/ast/expressions.rs

uqa_sql/ast/
expressions.rs
