Skip to main content

uqa_sql/ast/
from.rs

1//
2// Unified Query Algebra
3//
4// Copyright (c) 2023-2026 Cognica, Inc.
5//
6
7use serde::{Deserialize, Serialize};
8
9use super::{Expr, FunctionBinding, InternalRelationId, SelectStmt};
10
11#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
12pub enum FromClause {
13    /// `FROM <table> [AS <alias>]`.
14    Table {
15        /// Durable catalog identity, including an explicit schema when present.
16        name: String,
17        /// Relation name visible to SQL column binding before an alias is applied.
18        qualifier: String,
19        alias: Option<String>,
20        /// Positional names exposed by the range-table alias.
21        #[serde(default, skip_serializing_if = "Vec::is_empty")]
22        column_aliases: Vec<String>,
23        /// Creation-bound physical columns of a stored table source, before positional aliases. Column deletion and renaming update this list; later additions do not change its shape.
24        #[serde(default, skip_serializing_if = "Option::is_none")]
25        bound_columns: Option<Vec<String>>,
26        /// Ordinary references include inheritance children; `ONLY table`
27        /// clears this flag.
28        #[serde(default = "default_include_descendants")]
29        include_descendants: bool,
30    },
31    /// `FROM left <kind> right ON predicate`. `lateral` is true when
32    /// the right side is a LATERAL subquery / function -- the engine
33    /// re-evaluates it for every left row.
34    Join {
35        left: Box<FromClause>,
36        right: Box<FromClause>,
37        kind: JoinKind,
38        /// Boolean qualification supplied by `ON`. This is mutually
39        /// exclusive with `using` and `natural` in parser-produced trees.
40        on: Option<Expr>,
41        /// `PostgreSQL` `USING (column, ...) [AS alias]` metadata. The column
42        /// list must remain explicit until both input row types are known so
43        /// binding can validate each side and construct the merged output.
44        #[serde(default, skip_serializing_if = "Option::is_none")]
45        using: Option<JoinUsing>,
46        /// `NATURAL` derives its `USING` list from the visible columns of both
47        /// input row types at binding time.
48        #[serde(default)]
49        natural: bool,
50        /// Alias applied to the complete parenthesized JOIN result. When
51        /// present, the input relation names are hidden from the enclosing
52        /// query level.
53        #[serde(default, skip_serializing_if = "Option::is_none")]
54        alias: Option<String>,
55        /// Positional aliases for the JOIN output after USING/NATURAL shaping.
56        #[serde(default, skip_serializing_if = "Vec::is_empty")]
57        column_aliases: Vec<String>,
58        #[allow(dead_code)]
59        lateral: bool,
60    },
61    /// `FROM (VALUES (...)...) [AS <alias>(<col_aliases>)]`.
62    Values {
63        rows: Vec<Vec<Expr>>,
64        alias: Option<String>,
65        column_aliases: Vec<String>,
66        /// Opaque identity for an engine-injected, SQL-invisible VALUES row
67        /// carrier. Parser-produced VALUES sources always leave this unset.
68        #[serde(default, skip_serializing_if = "Option::is_none")]
69        #[doc(hidden)]
70        internal_relation: Option<InternalRelationId>,
71        /// Declared physical attribute types for an internal VALUES carrier;
72        /// needed even when the carrier has zero rows.
73        #[serde(default, skip_serializing_if = "Vec::is_empty")]
74        #[doc(hidden)]
75        internal_column_types: Vec<Option<super::ColumnType>>,
76    },
77    /// `FROM <fn>(<args>) [AS <alias>(<col_aliases>)]` -- e.g.
78    /// `generate_series(1, 5)`, `unnest(arr)`, `regexp_split_to_table`,
79    /// `json_each(...)`, `cypher(...) AS (col agtype, ...)`. The engine
80    /// dispatches by name.
81    Function {
82        name: String,
83        /// Exact catalog routine identity for a stored expression tree. Parser-produced trees leave this unset and catalog owners bind it before persistence.
84        #[serde(default, skip_serializing_if = "Option::is_none")]
85        binding: Option<FunctionBinding>,
86        /// Local function identifier used as `PostgreSQL`'s default output column label. Kept separate from the catalog-qualified lookup name so quoted identifiers containing `.` remain indivisible.
87        output_name: String,
88        /// Catalog relations bound to a tuple-producing operator join, kept separate from scalar arguments so each operand retains its own name-resolution, dependency, planning, and execution context.
89        #[serde(default, skip_serializing_if = "Option::is_none")]
90        relations: Option<OperatorJoinRelations>,
91        args: Vec<Expr>,
92        alias: Option<String>,
93        column_aliases: Vec<String>,
94        /// Append `PostgreSQL`'s one-based `bigint` ordinality column after the function's ordinary output columns.
95        #[serde(default, skip_serializing_if = "std::ops::Not::not")]
96        ordinality: bool,
97        /// Declared column types when the alias used a column
98        /// definition list (`AS (col agtype, n int)`); empty when the
99        /// alias only renamed columns. Type names are lowercased
100        /// `PostgreSQL` internal names (`agtype`, `int4`, `text`, ...).
101        #[serde(default)]
102        column_types: Vec<String>,
103    },
104    /// One `PostgreSQL` range-function group. This represents explicit
105    /// `ROWS FROM (...)` syntax and the parser transform of an unqualified
106    /// multi-argument `unnest(a, b, ...)` into independent unary
107    /// `pg_catalog.unnest` members. Members are evaluated independently and
108    /// their result columns are concatenated in declaration order.
109    FunctionGroup {
110        functions: Vec<TableFunction>,
111        /// Alias applied to the complete group rather than to an individual
112        /// member.
113        alias: Option<String>,
114        /// Positional aliases for the concatenated group output.
115        column_aliases: Vec<String>,
116        /// Append one group-wide, one-based `bigint` ordinality column after
117        /// every member's ordinary output columns.
118        #[serde(default, skip_serializing_if = "std::ops::Not::not")]
119        ordinality: bool,
120    },
121    /// `FROM (SELECT ...) AS <alias>` -- subquery as a relation.
122    /// The body re-runs as if a CTE; the alias renames the result
123    /// columns when supplied.
124    Subquery {
125        body: Box<SelectStmt>,
126        alias: Option<String>,
127        column_aliases: Vec<String>,
128    },
129}
130
131const fn default_include_descendants() -> bool {
132    true
133}
134
135/// One function inside a [`FromClause::FunctionGroup`].
136///
137/// A member owns its column definition list because `ROWS FROM` permits a
138/// distinct `AS (name type, ...)` clause after each call. The range item's
139/// relation alias, positional aliases, and ordinality remain on the enclosing
140/// group.
141#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
142pub struct TableFunction {
143    pub name: String,
144    /// Exact catalog routine identity for a stored expression tree. Parser-produced trees leave this unset and catalog owners bind it before persistence.
145    #[serde(default, skip_serializing_if = "Option::is_none")]
146    pub binding: Option<FunctionBinding>,
147    pub output_name: String,
148    #[serde(default, skip_serializing_if = "Option::is_none")]
149    pub relations: Option<OperatorJoinRelations>,
150    pub args: Vec<Expr>,
151    #[serde(default, skip_serializing_if = "Vec::is_empty")]
152    pub column_aliases: Vec<String>,
153    #[serde(default, skip_serializing_if = "Vec::is_empty")]
154    pub column_types: Vec<String>,
155}
156
157/// The two independently bound catalog relations consumed by an operator-join table function.
158#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
159pub struct OperatorJoinRelations {
160    pub left: String,
161    pub right: String,
162}
163
164#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
165pub struct JoinUsing {
166    pub columns: Vec<String>,
167    #[serde(default, skip_serializing_if = "Option::is_none")]
168    pub alias: Option<String>,
169}
170
171impl FromClause {
172    /// All table names referenced under this clause, in declaration
173    /// order. Used by the compiler to resolve unqualified column refs.
174    pub fn collect_tables(&self, out: &mut Vec<(String, Option<String>)>) {
175        match self {
176            FromClause::Table {
177                name,
178                qualifier,
179                alias,
180                ..
181            } => out.push((
182                name.clone(),
183                Some(alias.as_ref().unwrap_or(qualifier).clone()),
184            )),
185            FromClause::Join { left, right, .. } => {
186                left.collect_tables(out);
187                right.collect_tables(out);
188            }
189            FromClause::Values { alias, .. }
190            | FromClause::Function { alias, .. }
191            | FromClause::FunctionGroup { alias, .. }
192            | FromClause::Subquery { alias, .. } => {
193                if let Some(a) = alias {
194                    out.push((a.clone(), Some(a.clone())));
195                }
196            }
197        }
198    }
199}
200
201#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
202pub enum JoinKind {
203    Inner,
204    Left,
205    Right,
206    Full,
207    Cross,
208}