Skip to main content

spg_sql/
parser.rs

1//! Recursive-descent parser with a Pratt (precedence-climbing) sub-parser for
2//! expressions.
3//!
4//! Precedence (lowest → highest binding):
5//! `OR` (1) `<` `AND` (2) `<` `NOT` unary (3) `<`
6//! comparisons `=` `<>` `<` `<=` `>` `>=` (4) `<`
7//! `+` `-` (5) `<` `*` `/` (6) `<` unary `-` (7) `<` parens / atom.
8//!
9//! This matches PG's behaviour for the operators we support — e.g. `NOT a = b`
10//! parses as `NOT (a = b)` and `-a * b` as `(-a) * b`.
11
12use alloc::boxed::Box;
13use alloc::format;
14use alloc::string::{String, ToString};
15use alloc::vec;
16use alloc::vec::Vec;
17use core::fmt;
18use core::mem;
19
20use crate::ast::{
21    BinOp, CastTarget, ColumnDef, ColumnName, ColumnTypeName, CreateIndexStatement,
22    CreatePublicationStatement, CreateSubscriptionStatement, CreateTableStatement, Expr,
23    ExtractField, FkAction, ForeignKeyConstraint, FrameBound, FrameKind, FromClause, FromJoin,
24    IndexMethod, InsertStatement, JoinKind, Literal, NullTreatment, OrderBy, PublicationScope,
25    SelectItem, SelectStatement, Statement, TableRef, UnOp, UnionKind, VecEncoding, WindowFrame,
26};
27use crate::lexer::{self, LexError, Token};
28
29/// v7.9.22 — recognise pgvector / SPG vector-index opclass names
30/// in CREATE INDEX. SPG's HNSW already routes by query operator;
31/// the opclass is accepted for `pg_dump` compatibility (mailrs
32/// migration follow-up G5).
33fn is_vector_opclass_name(name: &str) -> bool {
34    let lc = name.to_ascii_lowercase();
35    matches!(
36        lc.as_str(),
37        "vector_cosine_ops"
38            | "vector_l2_ops"
39            | "vector_ip_ops"
40            | "halfvec_cosine_ops"
41            | "halfvec_l2_ops"
42            | "halfvec_ip_ops"
43            | "sq8_cosine_ops"
44            | "sq8_l2_ops"
45            | "sq8_ip_ops"
46    )
47}
48
49#[derive(Debug, Clone, PartialEq, Eq)]
50pub struct ParseError {
51    pub message: String,
52    /// Index into the token stream where parsing tripped. Not a byte offset.
53    pub token_pos: usize,
54}
55
56impl fmt::Display for ParseError {
57    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
58        write!(
59            f,
60            "parse error at token #{}: {}",
61            self.token_pos, self.message
62        )
63    }
64}
65
66impl From<LexError> for ParseError {
67    fn from(e: LexError) -> Self {
68        Self {
69            message: format!("lex: {e}"),
70            token_pos: 0,
71        }
72    }
73}
74
75/// v7.9.30 — parse a single expression (no trailing junk). Used by
76/// the engine to re-hydrate stored partial-index / unique-index
77/// predicates from their canonical Display form. The same Pratt
78/// parser the statement path uses; this entry point just skips the
79/// statement dispatch.
80pub fn parse_expression(input: &str) -> Result<Expr, ParseError> {
81    let tokens = lexer::tokenize(input)?;
82    let mut p = Parser::new(tokens);
83    let expr = p.parse_expr(0)?;
84    p.expect_eof()?;
85    Ok(expr)
86}
87
88/// Parse exactly one statement, swallow an optional trailing `;`, and require
89/// the token stream to end there.
90pub fn parse_statement(input: &str) -> Result<Statement, ParseError> {
91    let tokens = lexer::tokenize(input)?;
92    let mut p = Parser::new(tokens);
93    let stmt = p.parse_one_statement()?;
94    if matches!(p.peek(), Token::Semicolon) {
95        p.advance();
96    }
97    p.expect_eof()?;
98    Ok(stmt)
99}
100
101struct Parser {
102    tokens: Vec<Token>,
103    pos: usize,
104}
105
106impl Parser {
107    fn new(tokens: Vec<Token>) -> Self {
108        Self { tokens, pos: 0 }
109    }
110
111    fn peek(&self) -> &Token {
112        // tokens always ends with Eof; pos is clamped in advance().
113        &self.tokens[self.pos]
114    }
115
116    fn advance(&mut self) -> Token {
117        let t = mem::replace(&mut self.tokens[self.pos], Token::Eof);
118        if self.pos + 1 < self.tokens.len() {
119            self.pos += 1;
120        }
121        t
122    }
123
124    fn err(&self, message: String) -> ParseError {
125        ParseError {
126            message,
127            token_pos: self.pos,
128        }
129    }
130
131    fn expect_eof(&self) -> Result<(), ParseError> {
132        if matches!(self.peek(), Token::Eof) {
133            Ok(())
134        } else {
135            Err(self.err(format!("expected end of input, got {:?}", self.peek())))
136        }
137    }
138
139    fn expect_ident_like(&mut self) -> Result<String, ParseError> {
140        match self.advance() {
141            Token::Ident(s) | Token::QuotedIdent(s) => Ok(s),
142            other => Err(ParseError {
143                message: format!("expected identifier, got {other:?}"),
144                token_pos: self.pos.saturating_sub(1),
145            }),
146        }
147    }
148
149    #[allow(clippy::too_many_lines)]
150    fn parse_one_statement(&mut self) -> Result<Statement, ParseError> {
151        match self.peek() {
152            Token::Select => self.parse_select_stmt(),
153            // v7.9.27 — `DO $$ … $$ [LANGUAGE plpgsql]`. PG-only;
154            // SPG has no PL/pgSQL so the body is consumed (lexer
155            // already turned it into a Token::String) and the whole
156            // DO statement returns CommandOk no-op. mailrs H1 +
157            // pg_dump compat.
158            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("do") => {
159                self.advance();
160                // Body — single string token (dollar-quoted or
161                // ordinary).
162                match self.advance() {
163                    Token::String(_) => {}
164                    other => {
165                        return Err(self.err(alloc::format!(
166                            "expected dollar-quoted body after DO, got {other:?}"
167                        )));
168                    }
169                }
170                // Optional `LANGUAGE <name>` trailer (idents only).
171                if matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("language")) {
172                    self.advance();
173                    let _ = self.expect_ident_like()?;
174                }
175                Ok(Statement::DoBlock)
176            }
177            // v4.11: `WITH name AS (SELECT ...) [, ...] SELECT ...`.
178            // WITH isn't a reserved token in our lexer — comes through
179            // as `Token::Ident("with")` (case-insensitive).
180            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("with") => {
181                self.advance();
182                self.parse_with_cte_then_select()
183            }
184            // v4.26: `EXPLAIN [ANALYZE] <select>`. Comes through as
185            // an identifier — not a reserved keyword.
186            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("explain") => {
187                self.advance();
188                let mut analyze = false;
189                let mut suggest = false;
190                // v6.8.3 — `EXPLAIN (SUGGEST)` opt-in.
191                if matches!(self.peek(), Token::LParen) {
192                    self.advance();
193                    let opt = match self.peek().clone() {
194                        Token::Ident(s) | Token::QuotedIdent(s) => s,
195                        other => {
196                            return Err(self.err(format!(
197                                "expected option keyword inside EXPLAIN (…), got {other:?}"
198                            )));
199                        }
200                    };
201                    if !opt.eq_ignore_ascii_case("suggest") {
202                        return Err(self.err(format!(
203                            "unknown EXPLAIN option {opt:?}; v6.8.3 supports SUGGEST"
204                        )));
205                    }
206                    self.advance();
207                    if !matches!(self.peek(), Token::RParen) {
208                        return Err(self.err(format!(
209                            "expected ')' after EXPLAIN option, got {:?}",
210                            self.peek()
211                        )));
212                    }
213                    self.advance();
214                    suggest = true;
215                } else if let Token::Ident(s) | Token::QuotedIdent(s) = self.peek()
216                    && (s.eq_ignore_ascii_case("analyze") || s.eq_ignore_ascii_case("analyse"))
217                {
218                    self.advance();
219                    analyze = true;
220                }
221                let inner = self.parse_select_stmt()?;
222                let Statement::Select(s) = inner else {
223                    return Err(self.err(format!("EXPLAIN body must be a SELECT, got {inner:?}")));
224                };
225                Ok(Statement::Explain(crate::ast::ExplainStatement {
226                    analyze,
227                    inner: Box::new(s),
228                    suggest,
229                }))
230            }
231            Token::Create => self.parse_create_stmt(),
232            Token::Insert => self.parse_insert_stmt(),
233            Token::Begin => {
234                self.advance();
235                Ok(Statement::Begin)
236            }
237            Token::Commit => {
238                self.advance();
239                Ok(Statement::Commit)
240            }
241            Token::Rollback => {
242                self.advance();
243                // `ROLLBACK TO [SAVEPOINT] <name>` returns to that
244                // savepoint without ending the transaction. Bare
245                // `ROLLBACK` drops the whole TX.
246                if matches!(self.peek(), Token::To) {
247                    self.advance();
248                    if matches!(self.peek(), Token::Savepoint) {
249                        self.advance();
250                    }
251                    let name = self.expect_ident_like()?;
252                    Ok(Statement::RollbackToSavepoint(name))
253                } else {
254                    Ok(Statement::Rollback)
255                }
256            }
257            Token::Savepoint => {
258                self.advance();
259                let name = self.expect_ident_like()?;
260                Ok(Statement::Savepoint(name))
261            }
262            Token::Release => {
263                self.advance();
264                // `RELEASE [SAVEPOINT] <name>` — the `SAVEPOINT` keyword
265                // is optional in standard SQL.
266                if matches!(self.peek(), Token::Savepoint) {
267                    self.advance();
268                }
269                let name = self.expect_ident_like()?;
270                Ok(Statement::ReleaseSavepoint(name))
271            }
272            Token::Show => {
273                self.advance();
274                // `SHOW TABLES` / `SHOW USERS` / `SHOW COLUMNS FROM <table>`.
275                // v6.1.2 promoted TABLES to a reserved keyword (for
276                // `CREATE PUBLICATION … FOR ALL TABLES`), so it now
277                // arrives as `Token::Tables` rather than a bare ident.
278                // USERS / COLUMNS remain bare idents.
279                let target = match self.advance() {
280                    Token::Tables => "tables".to_string(),
281                    Token::Ident(s) | Token::QuotedIdent(s) => s.to_ascii_lowercase(),
282                    other => {
283                        return Err(self.err(format!(
284                            "expected SHOW target, got {other:?}"
285                        )));
286                    }
287                };
288                match target.as_str() {
289                    "tables" => Ok(Statement::ShowTables),
290                    "users" => Ok(Statement::ShowUsers),
291                    // v6.1.3 — PUBLICATIONS plural is NOT a reserved
292                    // keyword on its own; it lands here as a bare
293                    // ident. Returning all publications + their
294                    // scope summary.
295                    "publications" => Ok(Statement::ShowPublications),
296                    // v6.1.4 — same shape for SUBSCRIPTIONS plural.
297                    "subscriptions" => Ok(Statement::ShowSubscriptions),
298                    "columns" => {
299                        if !matches!(self.peek(), Token::From) {
300                            return Err(self.err(format!(
301                                "expected FROM after SHOW COLUMNS, got {:?}",
302                                self.peek()
303                            )));
304                        }
305                        self.advance();
306                        let table = self.expect_ident_like()?;
307                        Ok(Statement::ShowColumns(table))
308                    }
309                    other => Err(self.err(format!(
310                        "unknown SHOW target {other:?}; supported: TABLES, COLUMNS, USERS, PUBLICATIONS"
311                    ))),
312                }
313            }
314            // v6.1.2: `DROP` is now a reserved keyword (it dispatches
315            // to DROP USER and DROP PUBLICATION today; DROP TABLE /
316            // DROP INDEX are still SHOW-shaped admin ops). Pre-6.1.2
317            // arrived as a bare ident; tokenising it dedicatedly
318            // keeps the dispatch tree small.
319            Token::Drop => {
320                self.advance();
321                match self.peek() {
322                    Token::Publication => {
323                        self.advance();
324                        let name = self.expect_ident_or_string()?;
325                        Ok(Statement::DropPublication(name))
326                    }
327                    Token::Subscription => {
328                        self.advance();
329                        let name = self.expect_ident_or_string()?;
330                        Ok(Statement::DropSubscription(name))
331                    }
332                    Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("user") => {
333                        self.advance();
334                        let name = self.expect_ident_or_string()?;
335                        Ok(Statement::DropUser(name))
336                    }
337                    other => Err(self.err(format!(
338                        "expected USER / PUBLICATION / SUBSCRIPTION after DROP, got {other:?}"
339                    ))),
340                }
341            }
342            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("update") => {
343                self.advance();
344                self.parse_update_after_keyword()
345            }
346            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("delete") => {
347                self.advance();
348                self.parse_delete_after_keyword()
349            }
350            // v6.0.4: ALTER INDEX <name> REBUILD [WITH (encoding = ...)].
351            // ALTER is not a reserved keyword in the lexer — handled
352            // as a bare ident here.
353            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("alter") => {
354                self.advance();
355                self.parse_alter_after_keyword()
356            }
357            // v6.1.7: WAIT FOR WAL POSITION <pos> [WITH TIMEOUT <ms>].
358            // WAIT / POSITION / TIMEOUT are bare idents — no lexer
359            // additions needed.
360            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("wait") => {
361                self.advance();
362                self.parse_wait_after_keyword()
363            }
364            // v6.2.0: ANALYZE [<table>]. ANALYZE is a bare ident.
365            // Bare ANALYZE → analyse every user table; ANALYZE
366            // <name> → re-stats one. The argument is an optional
367            // ident (or quoted ident); anything else is a parse
368            // error.
369            // v6.7.3 — `COMPACT COLD SEGMENTS`. No arguments, no
370            // `WHERE` filter (carved out per V6_7_DESIGN.md
371            // STABILITY). Lex order: identifier "compact" → "cold"
372            // → "segments". Anything else after `COMPACT` is a
373            // parse error.
374            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("compact") => {
375                self.advance();
376                let next = self.peek().clone();
377                let cold = match next {
378                    Token::Ident(s) | Token::QuotedIdent(s) => s,
379                    _ => {
380                        return Err(
381                            self.err(format!("expected COLD after COMPACT, got {:?}", self.peek()))
382                        );
383                    }
384                };
385                if !cold.eq_ignore_ascii_case("cold") {
386                    return Err(self.err(format!("expected COLD after COMPACT, got {cold:?}")));
387                }
388                self.advance();
389                let next = self.peek().clone();
390                let segments = match next {
391                    Token::Ident(s) | Token::QuotedIdent(s) => s,
392                    _ => {
393                        return Err(self.err(format!(
394                            "expected SEGMENTS after COMPACT COLD, got {:?}",
395                            self.peek()
396                        )));
397                    }
398                };
399                if !segments.eq_ignore_ascii_case("segments") {
400                    return Err(self.err(format!(
401                        "expected SEGMENTS after COMPACT COLD, got {segments:?}"
402                    )));
403                }
404                self.advance();
405                Ok(Statement::CompactColdSegments)
406            }
407            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("analyze") => {
408                self.advance();
409                let target = match self.peek() {
410                    Token::Eof | Token::Semicolon => None,
411                    Token::Ident(_) | Token::QuotedIdent(_) => {
412                        Some(self.expect_ident_like()?)
413                    }
414                    other => {
415                        return Err(self.err(format!(
416                            "expected table name or end of statement after ANALYZE, got {other:?}"
417                        )));
418                    }
419                };
420                Ok(Statement::Analyze(target))
421            }
422            other => Err(self.err(format!(
423                "expected SELECT / CREATE / DROP / INSERT / UPDATE / DELETE / ALTER / BEGIN / COMMIT / \
424                 ROLLBACK / SAVEPOINT / RELEASE / SHOW at start of statement, got {other:?}"
425            ))),
426        }
427    }
428
429    fn parse_create_stmt(&mut self) -> Result<Statement, ParseError> {
430        debug_assert!(matches!(self.peek(), Token::Create));
431        self.advance();
432        match self.peek() {
433            Token::Table => self.parse_create_table_stmt_after_create(),
434            Token::Index => self.parse_create_index_stmt_after_create(false),
435            // v7.9.29 — `CREATE UNIQUE INDEX … [WHERE pred]`.
436            // The `UNIQUE` modifier turns a partial index into a
437            // partial-uniqueness invariant (only rows matching the
438            // WHERE predicate are checked for duplicates). mailrs
439            // K1 (3 hits: email_templates default, calendar_events
440            // master, calendar_events instance).
441            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("unique") => {
442                self.advance();
443                if !matches!(self.peek(), Token::Index) {
444                    return Err(self.err(alloc::format!(
445                        "expected INDEX after CREATE UNIQUE, got {:?}",
446                        self.peek()
447                    )));
448                }
449                self.parse_create_index_stmt_after_create(true)
450            }
451            Token::Publication => {
452                self.advance();
453                self.parse_create_publication_after_keyword()
454            }
455            Token::Subscription => {
456                self.advance();
457                self.parse_create_subscription_after_keyword()
458            }
459            // v4.1: CREATE USER 'name' WITH PASSWORD 'pw' [ROLE 'role'].
460            // USER isn't a reserved keyword — we look for the bare
461            // identifier so the lexer doesn't have to grow a token.
462            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("user") => {
463                self.advance();
464                self.parse_create_user_after_keyword()
465            }
466            // v7.9.15 — `CREATE EXTENSION [IF NOT EXISTS] <name>
467            // [WITH SCHEMA …] [VERSION '…'] [CASCADE]` as a
468            // no-op. mailrs follow-up F3.
469            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("extension") => {
470                self.advance();
471                self.parse_create_extension_after_keyword()
472            }
473            other => Err(self.err(format!(
474                "expected TABLE / INDEX / USER / EXTENSION / PUBLICATION / SUBSCRIPTION after CREATE, got {other:?}"
475            ))),
476        }
477    }
478
479    /// v7.9.15 — accept and discard `CREATE EXTENSION` DDL.
480    /// SPG doesn't have a registry; pgvector / similar are
481    /// either builtin (VECTOR(N) ↔ pgvector) or n/a. Parsing
482    /// the syntax lets dual-target schemas keep the line.
483    fn parse_create_extension_after_keyword(&mut self) -> Result<Statement, ParseError> {
484        // Optional `IF NOT EXISTS`.
485        self.consume_if_not_exists();
486        let name = self.expect_ident_like()?;
487        // Drain optional WITH SCHEMA <ident> / VERSION '<v>' /
488        // CASCADE / FROM '<v>' clauses; we don't model them.
489        loop {
490            match self.peek() {
491                Token::Ident(s) if s.eq_ignore_ascii_case("with") => {
492                    self.advance();
493                    continue;
494                }
495                Token::Ident(s) if s.eq_ignore_ascii_case("schema") => {
496                    self.advance();
497                    let _ = self.expect_ident_like()?;
498                    continue;
499                }
500                Token::Ident(s) if s.eq_ignore_ascii_case("version") => {
501                    self.advance();
502                    // String or ident literal.
503                    let _ = self.advance();
504                    continue;
505                }
506                Token::Ident(s) if s.eq_ignore_ascii_case("from") => {
507                    self.advance();
508                    let _ = self.advance();
509                    continue;
510                }
511                Token::Ident(s) if s.eq_ignore_ascii_case("cascade") => {
512                    self.advance();
513                    continue;
514                }
515                _ => break,
516            }
517        }
518        Ok(Statement::CreateExtension(name))
519    }
520
521    /// v6.1.2 → v6.1.3 — `CREATE PUBLICATION <name>` body. Accepts:
522    ///   - (no clause) → implicit `FOR ALL TABLES`
523    ///   - `FOR ALL TABLES`
524    ///   - `FOR ALL TABLES EXCEPT t1, t2, …` (v6.1.3)
525    ///   - `FOR TABLE t1, t2, …` (v6.1.3) — `FOR TABLES …` also
526    ///     accepted (PG accepts both forms in PG 19).
527    fn parse_create_publication_after_keyword(&mut self) -> Result<Statement, ParseError> {
528        let name = self.expect_ident_or_string()?;
529        // Bare DDL maps to FOR ALL TABLES — matches the v6.1.2
530        // shape so existing publications keep parsing identically.
531        let scope = if matches!(self.peek(), Token::For) {
532            self.advance();
533            if matches!(self.peek(), Token::All) {
534                self.advance();
535                if !matches!(self.peek(), Token::Tables) {
536                    return Err(self.err(format!(
537                        "expected TABLES after FOR ALL, got {:?}",
538                        self.peek()
539                    )));
540                }
541                self.advance();
542                if matches!(self.peek(), Token::Except) {
543                    self.advance();
544                    let tables = self.parse_publication_table_list()?;
545                    PublicationScope::AllTablesExcept(tables)
546                } else {
547                    PublicationScope::AllTables
548                }
549            } else if matches!(self.peek(), Token::Table | Token::Tables) {
550                // PG 19 accepts both `FOR TABLE …` (singular) and
551                // `FOR TABLES …` (plural); SPG matches.
552                self.advance();
553                let tables = self.parse_publication_table_list()?;
554                PublicationScope::ForTables(tables)
555            } else {
556                return Err(self.err(format!(
557                    "expected ALL TABLES or TABLE <list> after FOR, got {:?}",
558                    self.peek()
559                )));
560            }
561        } else {
562            PublicationScope::AllTables
563        };
564        Ok(Statement::CreatePublication(CreatePublicationStatement {
565            name,
566            scope,
567        }))
568    }
569
570    /// v6.1.3 — Comma-separated identifier list for the publication
571    /// FOR-clause. Requires at least one entry; empty list is a
572    /// parse error (PG behaviour). Quoted idents are accepted; the
573    /// names round-trip through `Display` as `quote_ident(name)`.
574    fn parse_publication_table_list(&mut self) -> Result<Vec<String>, ParseError> {
575        let first = self.expect_ident_like()?;
576        let mut out = alloc::vec![first];
577        while matches!(self.peek(), Token::Comma) {
578            self.advance();
579            out.push(self.expect_ident_like()?);
580        }
581        Ok(out)
582    }
583
584    /// v6.1.4 — `CREATE SUBSCRIPTION <name>
585    ///                 CONNECTION '<conn>'
586    ///                 PUBLICATION <pub> [, <pub> ...]`.
587    ///
588    /// The clause order is fixed (CONNECTION first, then
589    /// PUBLICATION) to match PG. No WITH-options accepted in
590    /// v6.1.4 — `enabled` defaults to true, no other knobs ship.
591    fn parse_create_subscription_after_keyword(&mut self) -> Result<Statement, ParseError> {
592        let name = self.expect_ident_or_string()?;
593        if !matches!(self.peek(), Token::Connection) {
594            return Err(self.err(format!(
595                "expected CONNECTION after CREATE SUBSCRIPTION <name>, got {:?}",
596                self.peek()
597            )));
598        }
599        self.advance();
600        let conn_str = self.expect_string_literal()?;
601        if !matches!(self.peek(), Token::Publication) {
602            return Err(self.err(format!(
603                "expected PUBLICATION after CONNECTION '<conn>', got {:?}",
604                self.peek()
605            )));
606        }
607        self.advance();
608        // Reuse the publication FOR-list parser shape: at least one
609        // identifier, comma-separated.
610        let first = self.expect_ident_like()?;
611        let mut publications = alloc::vec![first];
612        while matches!(self.peek(), Token::Comma) {
613            self.advance();
614            publications.push(self.expect_ident_like()?);
615        }
616        Ok(Statement::CreateSubscription(
617            CreateSubscriptionStatement {
618                name,
619                conn_str,
620                publications,
621            },
622        ))
623    }
624
625    /// v6.1.7 — `WAIT FOR WAL POSITION <pos> [WITH TIMEOUT <ms>]`.
626    /// All keywords after `WAIT` are bare idents in v6.1.x; no
627    /// lexer churn. Both `<pos>` and `<ms>` are positive integers
628    /// that fit `u64`.
629    fn parse_wait_after_keyword(&mut self) -> Result<Statement, ParseError> {
630        // FOR is a v6.1.2-reserved keyword (Token::For). The
631        // other two are bare idents — they've never needed lexer
632        // support and we keep it that way.
633        if !matches!(self.peek(), Token::For) {
634            return Err(self.err(format!(
635                "expected FOR after WAIT, got {:?}",
636                self.peek()
637            )));
638        }
639        self.advance();
640        self.expect_keyword_ident("wal")?;
641        self.expect_keyword_ident("position")?;
642        let pos = self.expect_u64_literal()?;
643        let timeout_ms = if matches!(self.peek(), Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("with"))
644        {
645            self.advance();
646            self.expect_keyword_ident("timeout")?;
647            Some(self.expect_u64_literal()?)
648        } else {
649            None
650        };
651        Ok(Statement::WaitForWalPosition { pos, timeout_ms })
652    }
653
654    /// v6.1.7 helper — consume a `Token::Integer` and check it
655    /// fits `u64`. WAL positions and millisecond timeouts are
656    /// non-negative.
657    fn expect_u64_literal(&mut self) -> Result<u64, ParseError> {
658        match self.advance() {
659            Token::Integer(n) if n >= 0 => Ok(n as u64),
660            Token::Integer(n) => Err(ParseError {
661                message: format!("expected non-negative integer, got {n}"),
662                token_pos: self.pos.saturating_sub(1),
663            }),
664            other => Err(ParseError {
665                message: format!("expected integer literal, got {other:?}"),
666                token_pos: self.pos.saturating_sub(1),
667            }),
668        }
669    }
670
671    /// `CREATE USER` body — name + WITH PASSWORD '<pw>' + optional
672    /// ROLE '<role>' (defaults to readonly). All string slots accept
673    /// either a quoted ident or a quoted string literal.
674    fn parse_create_user_after_keyword(&mut self) -> Result<Statement, ParseError> {
675        let name = self.expect_ident_or_string()?;
676        self.expect_keyword_ident("with")?;
677        self.expect_keyword_ident("password")?;
678        let password = self.expect_string_literal()?;
679        let role = if let Token::Ident(s) | Token::QuotedIdent(s) = self.peek()
680            && s.eq_ignore_ascii_case("role")
681        {
682            self.advance();
683            self.expect_string_literal()?
684        } else {
685            "readonly".to_string()
686        };
687        Ok(Statement::CreateUser(crate::ast::CreateUserStatement {
688            name,
689            password,
690            role,
691        }))
692    }
693
694    /// v4.4 `UPDATE <table> SET col = expr [, col = expr]* [WHERE cond]`.
695    /// Caller already consumed the leading `UPDATE` ident.
696    fn parse_update_after_keyword(&mut self) -> Result<Statement, ParseError> {
697        let table = self.expect_ident_like()?;
698        self.expect_keyword_ident("set")?;
699        let mut assignments = Vec::new();
700        loop {
701            let col = self.expect_ident_like()?;
702            if !matches!(self.peek(), Token::Eq) {
703                return Err(self.err(format!(
704                    "expected `=` after column name in UPDATE SET, got {:?}",
705                    self.peek()
706                )));
707            }
708            self.advance();
709            let value = self.parse_expr(0)?;
710            assignments.push((col, value));
711            if matches!(self.peek(), Token::Comma) {
712                self.advance();
713                continue;
714            }
715            break;
716        }
717        let where_ = if matches!(self.peek(), Token::Where) {
718            self.advance();
719            Some(self.parse_expr(0)?)
720        } else {
721            None
722        };
723        let returning = self.parse_optional_returning()?;
724        Ok(Statement::Update(crate::ast::UpdateStatement {
725            table,
726            assignments,
727            where_,
728            returning,
729        }))
730    }
731
732    /// v4.4 `DELETE FROM <table> [WHERE cond]`. Caller already consumed
733    /// the leading `DELETE` ident.
734    fn parse_delete_after_keyword(&mut self) -> Result<Statement, ParseError> {
735        if !matches!(self.peek(), Token::From) {
736            return Err(self.err(format!("expected FROM after DELETE, got {:?}", self.peek())));
737        }
738        self.advance();
739        let table = self.expect_ident_like()?;
740        let where_ = if matches!(self.peek(), Token::Where) {
741            self.advance();
742            Some(self.parse_expr(0)?)
743        } else {
744            None
745        };
746        let returning = self.parse_optional_returning()?;
747        Ok(Statement::Delete(crate::ast::DeleteStatement {
748            table,
749            where_,
750            returning,
751        }))
752    }
753
754    /// v7.9.4 — parse the optional trailing `RETURNING <projection>`
755    /// clause on INSERT / UPDATE / DELETE. Same projection grammar
756    /// as SELECT, so `RETURNING *`, `RETURNING col`,
757    /// `RETURNING expr AS alias`, and `RETURNING a, b, c` all work.
758    fn parse_optional_returning(&mut self) -> Result<Option<Vec<crate::ast::SelectItem>>, ParseError> {
759        let is_returning_kw = matches!(
760            self.peek(),
761            Token::Ident(s) if s.eq_ignore_ascii_case("returning")
762        );
763        if !is_returning_kw {
764            return Ok(None);
765        }
766        self.advance();
767        let mut items = Vec::new();
768        loop {
769            items.push(self.parse_select_item()?);
770            if matches!(self.peek(), Token::Comma) {
771                self.advance();
772                continue;
773            }
774            break;
775        }
776        Ok(Some(items))
777    }
778
779    /// v6.0.4 — parse the tail of an ALTER statement after the
780    /// leading `ALTER` keyword has been consumed. Only one form is
781    /// supported in v6.0.4:
782    ///
783    /// ```text
784    /// ALTER INDEX <name> REBUILD [WITH (encoding = <enc>)]
785    /// ```
786    fn parse_alter_after_keyword(&mut self) -> Result<Statement, ParseError> {
787        // ALTER INDEX <name> ... | ALTER TABLE <name> SET hot_tier_bytes = <n>
788        match self.advance() {
789            Token::Index => {}
790            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("index") => {}
791            // v6.7.2 — ALTER TABLE t SET hot_tier_bytes = X
792            Token::Table => return self.parse_alter_table_after_keyword(),
793            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("table") => {
794                return self.parse_alter_table_after_keyword();
795            }
796            other => {
797                return Err(self.err(format!("expected INDEX or TABLE after ALTER, got {other:?}")));
798            }
799        }
800        let name = self.expect_ident_like()?;
801        // REBUILD
802        self.expect_keyword_ident("rebuild")?;
803        // Optional: WITH (encoding = <enc>)
804        let encoding = if matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("with")) {
805            self.advance();
806            if !matches!(self.peek(), Token::LParen) {
807                return Err(self.err(format!(
808                    "expected '(' after WITH in ALTER INDEX REBUILD, got {:?}",
809                    self.peek()
810                )));
811            }
812            self.advance();
813            self.expect_keyword_ident("encoding")?;
814            if !matches!(self.peek(), Token::Eq) {
815                return Err(self.err(format!(
816                    "expected '=' after encoding in ALTER INDEX REBUILD, got {:?}",
817                    self.peek()
818                )));
819            }
820            self.advance();
821            let enc_ident = match self.advance() {
822                Token::Ident(s) | Token::QuotedIdent(s) => s,
823                other => {
824                    return Err(self.err(format!("expected encoding name after =, got {other:?}")));
825                }
826            };
827            let enc = match enc_ident.to_ascii_lowercase().as_str() {
828                "f32" => VecEncoding::F32,
829                "sq8" => VecEncoding::Sq8,
830                "half" => VecEncoding::F16,
831                other => {
832                    return Err(self.err(format!(
833                        "unknown vector encoding {other:?} in ALTER INDEX REBUILD; supported: F32, SQ8, HALF"
834                    )));
835                }
836            };
837            if !matches!(self.peek(), Token::RParen) {
838                return Err(self.err(format!(
839                    "expected ')' after encoding value, got {:?}",
840                    self.peek()
841                )));
842            }
843            self.advance();
844            Some(enc)
845        } else {
846            None
847        };
848        Ok(Statement::AlterIndex(crate::ast::AlterIndexStatement {
849            name,
850            target: crate::ast::AlterIndexTarget::Rebuild { encoding },
851        }))
852    }
853
854    /// v6.7.2 — `ALTER TABLE <name> SET hot_tier_bytes = <n>`. The
855    /// only `SET` form currently supported; future v6.7.x can add
856    /// more SET subjects without changing the dispatch shape.
857    fn parse_alter_table_after_keyword(&mut self) -> Result<Statement, ParseError> {
858        let table_name = self.expect_ident_like()?;
859        // v7.6.8 — dispatch on the next keyword: SET / ADD / DROP.
860        // SET kept identical to v6.7.x. ADD / DROP CONSTRAINT routes
861        // to FK installation / removal.
862        match self.peek() {
863            Token::Ident(s) if s.eq_ignore_ascii_case("set") => {
864                self.advance();
865                let setting = self.expect_ident_like()?;
866                if !setting.eq_ignore_ascii_case("hot_tier_bytes") {
867                    return Err(self.err(alloc::format!(
868                        "ALTER TABLE SET: unknown setting {setting:?}; supported: hot_tier_bytes"
869                    )));
870                }
871                if !matches!(self.peek(), Token::Eq) {
872                    return Err(self.err(alloc::format!(
873                        "expected '=' after hot_tier_bytes, got {:?}",
874                        self.peek()
875                    )));
876                }
877                self.advance();
878                let n = self.expect_u64_literal()?;
879                Ok(Statement::AlterTable(crate::ast::AlterTableStatement {
880                    name: table_name,
881                    target: crate::ast::AlterTableTarget::SetHotTierBytes(n),
882                }))
883            }
884            Token::Ident(s) if s.eq_ignore_ascii_case("add") => {
885                self.advance();
886                // Optional `CONSTRAINT <name>` prefix, then the same
887                // FK clause shape as table-level CREATE TABLE FK.
888                let fk = self.parse_table_level_fk()?;
889                Ok(Statement::AlterTable(crate::ast::AlterTableStatement {
890                    name: table_name,
891                    target: crate::ast::AlterTableTarget::AddForeignKey(fk),
892                }))
893            }
894            Token::Drop => {
895                self.advance();
896                match self.advance() {
897                    Token::Ident(s) if s.eq_ignore_ascii_case("constraint") => {}
898                    other => {
899                        return Err(self.err(alloc::format!(
900                            "expected CONSTRAINT after DROP in ALTER TABLE, got {other:?}"
901                        )));
902                    }
903                }
904                let cname = self.expect_ident_like()?;
905                Ok(Statement::AlterTable(crate::ast::AlterTableStatement {
906                    name: table_name,
907                    target: crate::ast::AlterTableTarget::DropForeignKey(cname),
908                }))
909            }
910            other => Err(self.err(alloc::format!(
911                "expected SET / ADD / DROP in ALTER TABLE, got {other:?}"
912            ))),
913        }
914    }
915
916    /// Consume a bare ident if its lowercase matches `kw`, else err.
917    fn expect_keyword_ident(&mut self, kw: &str) -> Result<(), ParseError> {
918        match self.advance() {
919            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case(kw) => Ok(()),
920            other => Err(ParseError {
921                message: format!("expected {kw:?}, got {other:?}"),
922                token_pos: self.pos.saturating_sub(1),
923            }),
924        }
925    }
926
927    /// Accept either a quoted identifier (`"foo"`) or a quoted string
928    /// literal (`'foo'`) — same shape used by CREATE USER for the
929    /// username slot.
930    fn expect_ident_or_string(&mut self) -> Result<String, ParseError> {
931        match self.advance() {
932            Token::Ident(s) | Token::QuotedIdent(s) | Token::String(s) => Ok(s),
933            other => Err(ParseError {
934                message: format!("expected identifier or string, got {other:?}"),
935                token_pos: self.pos.saturating_sub(1),
936            }),
937        }
938    }
939
940    fn expect_string_literal(&mut self) -> Result<String, ParseError> {
941        match self.advance() {
942            Token::String(s) => Ok(s),
943            other => Err(ParseError {
944                message: format!("expected quoted string, got {other:?}"),
945                token_pos: self.pos.saturating_sub(1),
946            }),
947        }
948    }
949
950    fn parse_select_stmt(&mut self) -> Result<Statement, ParseError> {
951        // Caller dispatches on Token::Select; the inner helper handles
952        // the rest. ORDER BY / LIMIT bind at this top level; UNION peers
953        // get a fresh bare-select parse and may not have their own ORDER
954        // BY / LIMIT.
955        let mut head = self.parse_bare_select()?;
956        while matches!(self.peek(), Token::Union) {
957            self.advance();
958            let kind = if matches!(self.peek(), Token::All) {
959                self.advance();
960                UnionKind::All
961            } else {
962                UnionKind::Distinct
963            };
964            let peer = self.parse_bare_select()?;
965            head.unions.push((kind, peer));
966        }
967        head.order_by = if matches!(self.peek(), Token::Order) {
968            self.advance();
969            if !matches!(self.peek(), Token::By) {
970                return Err(self.err(format!("expected BY after ORDER, got {:?}", self.peek())));
971            }
972            self.advance();
973            // v6.4.0 — multi-key ORDER BY. Loop over comma-separated
974            // `<expr> [ASC|DESC]` items.
975            let mut keys = Vec::new();
976            loop {
977                let expr = self.parse_expr(0)?;
978                let desc = if matches!(self.peek(), Token::Desc) {
979                    self.advance();
980                    true
981                } else if matches!(self.peek(), Token::Asc) {
982                    self.advance();
983                    false
984                } else {
985                    false
986                };
987                keys.push(OrderBy { expr, desc });
988                if matches!(self.peek(), Token::Comma) {
989                    self.advance();
990                } else {
991                    break;
992                }
993            }
994            keys
995        } else {
996            Vec::new()
997        };
998        head.limit = if matches!(self.peek(), Token::Limit) {
999            self.advance();
1000            Some(self.parse_limit_expr("LIMIT")?)
1001        } else {
1002            None
1003        };
1004        head.offset = if matches!(self.peek(), Token::Offset) {
1005            self.advance();
1006            Some(self.parse_limit_expr("OFFSET")?)
1007        } else {
1008            None
1009        };
1010        Ok(Statement::Select(head))
1011    }
1012
1013    /// v7.9.24 — accept `LIMIT <int>` or `LIMIT $N`. mailrs H2.
1014    /// Bind value gets resolved during prepared-statement Execute;
1015    /// the Pratt expression parser would over-accept here (e.g.
1016    /// `LIMIT 5 + 5`), so we narrowly accept only the two PG forms.
1017    fn parse_limit_expr(&mut self, label: &str) -> Result<crate::ast::LimitExpr, ParseError> {
1018        match self.advance() {
1019            Token::Integer(n) if n >= 0 => u32::try_from(n)
1020                .map(crate::ast::LimitExpr::Literal)
1021                .map_err(|_| ParseError {
1022                    message: alloc::format!("{label} value too large: {n}"),
1023                    token_pos: self.pos.saturating_sub(1),
1024                }),
1025            Token::Placeholder(n) => Ok(crate::ast::LimitExpr::Placeholder(n)),
1026            other => Err(ParseError {
1027                message: alloc::format!(
1028                    "expected non-negative integer or $N placeholder after {label}, got {other:?}"
1029                ),
1030                token_pos: self.pos.saturating_sub(1),
1031            }),
1032        }
1033    }
1034
1035    fn expect_u32_literal(&mut self, label: &str) -> Result<u32, ParseError> {
1036        match self.advance() {
1037            Token::Integer(n) if n >= 0 => u32::try_from(n).map_err(|_| ParseError {
1038                message: format!("{label} value too large: {n}"),
1039                token_pos: self.pos.saturating_sub(1),
1040            }),
1041            other => Err(ParseError {
1042                message: format!("expected non-negative integer after {label}, got {other:?}"),
1043                token_pos: self.pos.saturating_sub(1),
1044            }),
1045        }
1046    }
1047
1048    /// Parse one SELECT block without ORDER BY / LIMIT / UNION chaining —
1049    /// just `[DISTINCT] items [FROM] [WHERE] [GROUP BY]`. Returned with
1050    /// `unions` empty and `order_by` / `limit` `None`; the top-level
1051    /// `parse_select_stmt` is responsible for filling those in.
1052    fn parse_bare_select(&mut self) -> Result<SelectStatement, ParseError> {
1053        if !matches!(self.peek(), Token::Select) {
1054            return Err(self.err(format!(
1055                "expected SELECT to start a query block, got {:?}",
1056                self.peek()
1057            )));
1058        }
1059        self.advance();
1060        let distinct = if matches!(self.peek(), Token::Distinct) {
1061            self.advance();
1062            true
1063        } else {
1064            false
1065        };
1066        let items = self.parse_select_list()?;
1067        let from = if matches!(self.peek(), Token::From) {
1068            self.advance();
1069            Some(self.parse_from_clause()?)
1070        } else {
1071            None
1072        };
1073        let where_ = if matches!(self.peek(), Token::Where) {
1074            self.advance();
1075            Some(self.parse_expr(0)?)
1076        } else {
1077            None
1078        };
1079        let mut group_by_all = false;
1080        let group_by = if matches!(self.peek(), Token::Group) {
1081            self.advance();
1082            if !matches!(self.peek(), Token::By) {
1083                return Err(self.err(format!("expected BY after GROUP, got {:?}", self.peek())));
1084            }
1085            self.advance();
1086            // v6.4.1 — `GROUP BY ALL` shortcut. Planner expands to
1087            // every non-aggregate SELECT-list item later.
1088            if matches!(self.peek(), Token::All) {
1089                self.advance();
1090                group_by_all = true;
1091                None
1092            } else {
1093                let mut groups = Vec::new();
1094                loop {
1095                    groups.push(self.parse_expr(0)?);
1096                    if matches!(self.peek(), Token::Comma) {
1097                        self.advance();
1098                    } else {
1099                        break;
1100                    }
1101                }
1102                Some(groups)
1103            }
1104        } else {
1105            None
1106        };
1107        let having = if matches!(self.peek(), Token::Having) {
1108            self.advance();
1109            Some(self.parse_expr(0)?)
1110        } else {
1111            None
1112        };
1113        Ok(SelectStatement {
1114            ctes: Vec::new(),
1115            distinct,
1116            items,
1117            from,
1118            where_,
1119            group_by,
1120            group_by_all,
1121            having,
1122            unions: Vec::new(),
1123            order_by: Vec::new(),
1124            limit: None,
1125            offset: None,
1126        })
1127    }
1128
1129    fn parse_create_table_stmt_after_create(&mut self) -> Result<Statement, ParseError> {
1130        // Caller already consumed CREATE; we're sitting on TABLE.
1131        debug_assert!(matches!(self.peek(), Token::Table));
1132        self.advance();
1133        let if_not_exists = self.consume_if_not_exists();
1134        let name = self.expect_ident_like()?;
1135        if !matches!(self.peek(), Token::LParen) {
1136            return Err(self.err(format!(
1137                "expected '(' after table name, got {:?}",
1138                self.peek()
1139            )));
1140        }
1141        self.advance();
1142        let mut columns = Vec::new();
1143        let mut foreign_keys: Vec<ForeignKeyConstraint> = Vec::new();
1144        let mut table_constraints: Vec<crate::ast::TableConstraint> = Vec::new();
1145        loop {
1146            // v7.6.0 / v7.9.18 — distinguish table-level constraint
1147            // clauses from column definitions. Constraints start
1148            // with `CONSTRAINT <name> …`, `FOREIGN KEY (…)`,
1149            // `PRIMARY KEY (…)`, or `UNIQUE (…)`. Anything else is
1150            // a column.
1151            if self.peek_table_level_pk_start() {
1152                table_constraints.push(self.parse_table_level_primary_key()?);
1153            } else if self.peek_table_level_unique_start() {
1154                table_constraints.push(self.parse_table_level_unique()?);
1155            } else if self.peek_constraint_or_fk_start() {
1156                foreign_keys.push(self.parse_table_level_fk()?);
1157            } else {
1158                let (col, col_level_fk) = self.parse_column_def_with_fk()?;
1159                columns.push(col);
1160                if let Some(fk) = col_level_fk {
1161                    foreign_keys.push(fk);
1162                }
1163            }
1164            match self.peek() {
1165                Token::Comma => {
1166                    self.advance();
1167                }
1168                Token::RParen => {
1169                    self.advance();
1170                    break;
1171                }
1172                other => {
1173                    return Err(
1174                        self.err(format!("expected ',' or ')' in column list, got {other:?}"))
1175                    );
1176                }
1177            }
1178        }
1179        if columns.is_empty() {
1180            return Err(self.err("CREATE TABLE requires at least one column".into()));
1181        }
1182        Ok(Statement::CreateTable(CreateTableStatement {
1183            name,
1184            columns,
1185            if_not_exists,
1186            foreign_keys,
1187            table_constraints,
1188        }))
1189    }
1190
1191    /// v7.9.18 — true when the next tokens are `PRIMARY KEY (…)`.
1192    /// PRIMARY and KEY are bare idents; we look-ahead 2 to be
1193    /// sure (otherwise a column literally named `primary` would
1194    /// be mistaken).
1195    fn peek_table_level_pk_start(&self) -> bool {
1196        let cur = self.peek();
1197        let nxt = self.tokens.get(self.pos + 1);
1198        let nxt2 = self.tokens.get(self.pos + 2);
1199        let is_primary = matches!(cur, Token::Ident(s) if s.eq_ignore_ascii_case("primary"));
1200        let is_key = matches!(nxt, Some(Token::Ident(s)) if s.eq_ignore_ascii_case("key"));
1201        let is_lparen = matches!(nxt2, Some(Token::LParen));
1202        is_primary && is_key && is_lparen
1203    }
1204
1205    /// v7.9.18 — true when the next tokens are `UNIQUE (…)`.
1206    fn peek_table_level_unique_start(&self) -> bool {
1207        let cur = self.peek();
1208        let nxt = self.tokens.get(self.pos + 1);
1209        let is_unique = matches!(cur, Token::Ident(s) if s.eq_ignore_ascii_case("unique"));
1210        let is_lparen = matches!(nxt, Some(Token::LParen));
1211        is_unique && is_lparen
1212    }
1213
1214    fn parse_table_level_primary_key(
1215        &mut self,
1216    ) -> Result<crate::ast::TableConstraint, ParseError> {
1217        self.advance(); // PRIMARY
1218        self.advance(); // KEY
1219        let columns = self.parse_paren_ident_list("PRIMARY KEY")?;
1220        Ok(crate::ast::TableConstraint::PrimaryKey {
1221            name: None,
1222            columns,
1223        })
1224    }
1225
1226    fn parse_table_level_unique(
1227        &mut self,
1228    ) -> Result<crate::ast::TableConstraint, ParseError> {
1229        self.advance(); // UNIQUE
1230        let columns = self.parse_paren_ident_list("UNIQUE")?;
1231        Ok(crate::ast::TableConstraint::Unique {
1232            name: None,
1233            columns,
1234        })
1235    }
1236
1237    fn parse_paren_ident_list(
1238        &mut self,
1239        ctx: &str,
1240    ) -> Result<Vec<String>, ParseError> {
1241        if !matches!(self.peek(), Token::LParen) {
1242            return Err(self.err(alloc::format!(
1243                "expected '(' after {ctx}, got {:?}",
1244                self.peek()
1245            )));
1246        }
1247        self.advance();
1248        let mut out = Vec::new();
1249        loop {
1250            out.push(self.expect_ident_like()?);
1251            match self.peek() {
1252                Token::Comma => {
1253                    self.advance();
1254                }
1255                Token::RParen => {
1256                    self.advance();
1257                    break;
1258                }
1259                other => {
1260                    return Err(self.err(alloc::format!(
1261                        "expected ',' or ')' in {ctx} list, got {other:?}"
1262                    )));
1263                }
1264            }
1265        }
1266        if out.is_empty() {
1267            return Err(self.err(alloc::format!("{ctx} requires at least one column")));
1268        }
1269        Ok(out)
1270    }
1271
1272    /// v7.6.0 — true when the next tokens are `CONSTRAINT <name>
1273    /// FOREIGN KEY` or bare `FOREIGN KEY`. Both introduce a
1274    /// table-level FK; a column def never starts with either keyword
1275    /// (column names are not in this reserved set).
1276    fn peek_constraint_or_fk_start(&self) -> bool {
1277        let is_constraint_kw = matches!(
1278            self.peek(),
1279            Token::Ident(s) if s.eq_ignore_ascii_case("constraint")
1280        );
1281        let is_foreign_kw = matches!(
1282            self.peek(),
1283            Token::Ident(s) if s.eq_ignore_ascii_case("foreign")
1284        );
1285        is_constraint_kw || is_foreign_kw
1286    }
1287
1288    /// v7.6.0 — parse a table-level FK clause:
1289    /// `[CONSTRAINT <name>] FOREIGN KEY (<col>[,<col>]*) REFERENCES
1290    /// <tbl> [(<pcol>[,<pcol>]*)] [ON DELETE <action>] [ON UPDATE <action>]`.
1291    fn parse_table_level_fk(&mut self) -> Result<ForeignKeyConstraint, ParseError> {
1292        let mut name: Option<String> = None;
1293        if matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("constraint")) {
1294            self.advance();
1295            name = Some(self.expect_ident_like()?);
1296        }
1297        // `FOREIGN`
1298        match self.advance() {
1299            Token::Ident(s) if s.eq_ignore_ascii_case("foreign") => {}
1300            other => return Err(self.err(format!("expected FOREIGN, got {other:?}"))),
1301        }
1302        // `KEY`
1303        match self.advance() {
1304            Token::Ident(s) if s.eq_ignore_ascii_case("key") => {}
1305            other => return Err(self.err(format!("expected KEY after FOREIGN, got {other:?}"))),
1306        }
1307        // `(col, col, ...)`
1308        if !matches!(self.peek(), Token::LParen) {
1309            return Err(self.err(format!("expected '(' after FOREIGN KEY, got {:?}", self.peek())));
1310        }
1311        self.advance();
1312        let mut columns = Vec::new();
1313        loop {
1314            columns.push(self.expect_ident_like()?);
1315            match self.peek() {
1316                Token::Comma => {
1317                    self.advance();
1318                }
1319                Token::RParen => {
1320                    self.advance();
1321                    break;
1322                }
1323                other => return Err(self.err(format!("expected ',' or ')' in FK column list, got {other:?}"))),
1324            }
1325        }
1326        if columns.is_empty() {
1327            return Err(self.err("FOREIGN KEY requires at least one column".into()));
1328        }
1329        let (parent_table, parent_columns, on_delete, on_update) =
1330            self.parse_references_tail(columns.len())?;
1331        Ok(ForeignKeyConstraint {
1332            name,
1333            columns,
1334            parent_table,
1335            parent_columns,
1336            on_delete,
1337            on_update,
1338        })
1339    }
1340
1341    /// v7.6.0 — parse the tail `REFERENCES <tbl> [(<pcol>...)] [ON
1342    /// DELETE <action>] [ON UPDATE <action>]`. `expected_arity` is
1343    /// the local column count, used to default the parent column
1344    /// list when omitted (SQL spec: parent's PK is implied).
1345    fn parse_references_tail(
1346        &mut self,
1347        expected_arity: usize,
1348    ) -> Result<(String, Vec<String>, FkAction, FkAction), ParseError> {
1349        match self.advance() {
1350            Token::Ident(s) if s.eq_ignore_ascii_case("references") => {}
1351            other => return Err(self.err(format!("expected REFERENCES, got {other:?}"))),
1352        }
1353        let parent_table = self.expect_ident_like()?;
1354        let mut parent_columns: Vec<String> = Vec::new();
1355        if matches!(self.peek(), Token::LParen) {
1356            self.advance();
1357            loop {
1358                parent_columns.push(self.expect_ident_like()?);
1359                match self.peek() {
1360                    Token::Comma => {
1361                        self.advance();
1362                    }
1363                    Token::RParen => {
1364                        self.advance();
1365                        break;
1366                    }
1367                    other => return Err(self.err(format!("expected ',' or ')' in REFERENCES column list, got {other:?}"))),
1368                }
1369            }
1370        }
1371        if !parent_columns.is_empty() && parent_columns.len() != expected_arity {
1372            return Err(self.err(format!(
1373                "FK arity mismatch: {} local column(s) vs {} parent column(s)",
1374                expected_arity,
1375                parent_columns.len()
1376            )));
1377        }
1378        // v7.6.7 — accept and reject `[NOT] DEFERRABLE [INITIALLY
1379        // {DEFERRED | IMMEDIATE}]` so existing PG dumps don't fail
1380        // at parse time. SPG's single-writer model has no deferred
1381        // constraint window, so we surface this as a clean
1382        // unsupported-feature error rather than a syntax error.
1383        loop {
1384            if matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("deferrable")) {
1385                return Err(self.err(
1386                    "DEFERRABLE constraints are not supported (SPG is single-writer; \
1387                     constraints are always evaluated immediately at commit)"
1388                        .into(),
1389                ));
1390            }
1391            if matches!(self.peek(), Token::Not) {
1392                let look = self.tokens.get(self.pos + 1);
1393                if matches!(look, Some(Token::Ident(s)) if s.eq_ignore_ascii_case("deferrable")) {
1394                    // NOT DEFERRABLE — accept as the SPG default
1395                    // and consume both tokens silently.
1396                    self.advance();
1397                    self.advance();
1398                    // Optional `INITIALLY IMMEDIATE` clause.
1399                    if matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("initially"))
1400                    {
1401                        self.advance();
1402                        match self.advance() {
1403                            Token::Ident(s) if s.eq_ignore_ascii_case("immediate") => {}
1404                            other => {
1405                                return Err(self.err(format!(
1406                                    "expected IMMEDIATE after INITIALLY for NOT DEFERRABLE, \
1407                                     got {other:?}"
1408                                )));
1409                            }
1410                        }
1411                    }
1412                    continue;
1413                }
1414                break;
1415            }
1416            break;
1417        }
1418        // Optional `ON DELETE <action>` and `ON UPDATE <action>` in
1419        // either order, each at most once.
1420        let mut on_delete = FkAction::Restrict;
1421        let mut on_update = FkAction::Restrict;
1422        let mut seen_on_delete = false;
1423        let mut seen_on_update = false;
1424        loop {
1425            if !matches!(self.peek(), Token::On) {
1426                break;
1427            }
1428            self.advance();
1429            let which = self.advance();
1430            let action = self.parse_fk_action()?;
1431            match which {
1432                Token::Ident(ref s) if s.eq_ignore_ascii_case("delete") => {
1433                    if seen_on_delete {
1434                        return Err(self.err("ON DELETE specified twice".into()));
1435                    }
1436                    seen_on_delete = true;
1437                    on_delete = action;
1438                }
1439                Token::Ident(ref s) if s.eq_ignore_ascii_case("update") => {
1440                    if seen_on_update {
1441                        return Err(self.err("ON UPDATE specified twice".into()));
1442                    }
1443                    seen_on_update = true;
1444                    on_update = action;
1445                }
1446                other => {
1447                    return Err(self.err(format!(
1448                        "expected DELETE or UPDATE after ON, got {other:?}"
1449                    )));
1450                }
1451            }
1452        }
1453        Ok((parent_table, parent_columns, on_delete, on_update))
1454    }
1455
1456    /// v7.6.0 — parse `CASCADE | RESTRICT | SET NULL | SET DEFAULT |
1457    /// NO ACTION`.
1458    fn parse_fk_action(&mut self) -> Result<FkAction, ParseError> {
1459        match self.advance() {
1460            Token::Ident(s) if s.eq_ignore_ascii_case("cascade") => Ok(FkAction::Cascade),
1461            Token::Ident(s) if s.eq_ignore_ascii_case("restrict") => Ok(FkAction::Restrict),
1462            Token::Ident(s) if s.eq_ignore_ascii_case("set") => {
1463                match self.advance() {
1464                    Token::Null => Ok(FkAction::SetNull),
1465                    Token::Default => Ok(FkAction::SetDefault),
1466                    other => Err(self.err(format!(
1467                        "expected NULL or DEFAULT after SET in FK action, got {other:?}"
1468                    ))),
1469                }
1470            }
1471            Token::Ident(s) if s.eq_ignore_ascii_case("no") => {
1472                match self.advance() {
1473                    Token::Ident(s) if s.eq_ignore_ascii_case("action") => Ok(FkAction::NoAction),
1474                    other => Err(self.err(format!(
1475                        "expected ACTION after NO in FK action, got {other:?}"
1476                    ))),
1477                }
1478            }
1479            other => Err(self.err(format!(
1480                "expected CASCADE | RESTRICT | SET NULL | SET DEFAULT | NO ACTION, got {other:?}"
1481            ))),
1482        }
1483    }
1484
1485    /// Recognise the optional `IF NOT EXISTS` prefix shared by `CREATE
1486    /// TABLE` and `CREATE INDEX`. Returns `true` if consumed.
1487    fn consume_if_not_exists(&mut self) -> bool {
1488        // `IF` arrives as a bare Ident (we don't reserve it because it
1489        // also appears mid-expression in PG, though we don't support
1490        // those forms yet).
1491        let looks_like_if = matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("if"));
1492        if !looks_like_if {
1493            return false;
1494        }
1495        // Peek one ahead before committing: only consume IF when it's
1496        // actually `IF NOT EXISTS`.
1497        if !matches!(self.tokens.get(self.pos + 1), Some(Token::Not)) {
1498            return false;
1499        }
1500        if !matches!(
1501            self.tokens.get(self.pos + 2),
1502            Some(Token::Ident(s)) if s.eq_ignore_ascii_case("exists")
1503        ) {
1504            return false;
1505        }
1506        self.advance(); // IF
1507        self.advance(); // NOT
1508        self.advance(); // EXISTS
1509        true
1510    }
1511
1512    /// v7.9.14 — consume `ASC | DESC | NULLS FIRST | NULLS LAST`
1513    /// qualifiers after an index column ref. ASC / DESC are
1514    /// reserved tokens; NULLS / FIRST / LAST are bare idents.
1515    /// We accept and discard them since single-column BTree
1516    /// stores rows in natural key order today.
1517    fn consume_optional_index_column_qualifiers(&mut self) {
1518        loop {
1519            match self.peek() {
1520                Token::Asc | Token::Desc => {
1521                    self.advance();
1522                }
1523                Token::Ident(s) if s.eq_ignore_ascii_case("nulls") => {
1524                    let look = self.tokens.get(self.pos + 1);
1525                    if matches!(
1526                        look,
1527                        Some(Token::Ident(k)) if k.eq_ignore_ascii_case("first")
1528                            || k.eq_ignore_ascii_case("last")
1529                    ) {
1530                        self.advance();
1531                        self.advance();
1532                    } else {
1533                        break;
1534                    }
1535                }
1536                _ => break,
1537            }
1538        }
1539    }
1540
1541    fn parse_create_index_stmt_after_create(
1542        &mut self,
1543        is_unique: bool,
1544    ) -> Result<Statement, ParseError> {
1545        // Caller consumed CREATE (and the optional UNIQUE); we're on INDEX.
1546        debug_assert!(matches!(self.peek(), Token::Index));
1547        self.advance();
1548        let if_not_exists = self.consume_if_not_exists();
1549        let name = self.expect_ident_like()?;
1550        if !matches!(self.peek(), Token::On) {
1551            return Err(self.err(format!(
1552                "expected ON after CREATE INDEX <name>, got {:?}",
1553                self.peek()
1554            )));
1555        }
1556        self.advance();
1557        let table = self.expect_ident_like()?;
1558        // Optional `USING <method>` — only recognised method in v2.0 is
1559        // `hnsw` (a single-layer NSW graph for kNN). `USING` is the bare
1560        // ident `using` (we don't promote it to a reserved keyword
1561        // because it isn't reserved anywhere else in our SQL surface).
1562        let method = if matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("using")) {
1563            self.advance();
1564            let m = self.expect_ident_like()?;
1565            match m.to_ascii_lowercase().as_str() {
1566                "hnsw" => IndexMethod::Hnsw,
1567                "btree" => IndexMethod::BTree,
1568                "brin" => IndexMethod::Brin,
1569                // v7.9.26b — PG `pg_dump` emits `USING gin` /
1570                // `USING gist` / `USING spgist` / `USING hash` for
1571                // their built-in index AMs. SPG doesn't have a
1572                // matching implementation; degrade to BTree on the
1573                // leading column so the schema loads + the index
1574                // catalogue stays consistent. Operator pays the
1575                // planner cost only for the queries that would have
1576                // used the specialised AM.
1577                "gin" | "gist" | "spgist" | "hash" => IndexMethod::BTree,
1578                other => {
1579                    return Err(self.err(alloc::format!(
1580                        "unknown index method {other:?}; supported: hnsw, btree, brin (gin/gist/spgist/hash accepted as BTree fallback)"
1581                    )));
1582                }
1583            }
1584        } else {
1585            IndexMethod::BTree
1586        };
1587        if !matches!(self.peek(), Token::LParen) {
1588            return Err(self.err(format!(
1589                "expected '(' before indexed column, got {:?}",
1590                self.peek()
1591            )));
1592        }
1593        self.advance();
1594        // v6.8.2 — accept either a bare column ident (legacy) or
1595        // an expression `fn(col, …)` for expression indexes.
1596        // Distinguish by peeking the token *after* the current
1597        // ident: `ident )` is the legacy column-only path;
1598        // anything else triggers the Pratt expression parser.
1599        // (`advance()` uses `mem::replace` to nil out the current
1600        // slot, so we can't save+rewind cleanly — peek-ahead via
1601        // direct index avoids the mutation.)
1602        let (column, expression): (String, Option<Expr>) = match self.peek().clone() {
1603            // Single column with `)` immediately after — fast path.
1604            // v7.9.29 — also: bare column followed by `,` (the
1605            // multi-column form `(a, b, c)`). Without this branch
1606            // the leading ident gets pulled into `parse_expr`
1607            // which then sets `expression = Some(Column(a))` and
1608            // breaks Display round-trip on the multi-column shape.
1609            Token::Ident(s) | Token::QuotedIdent(s)
1610                if matches!(
1611                    self.tokens.get(self.pos + 1),
1612                    Some(Token::RParen | Token::Comma)
1613                ) =>
1614            {
1615                self.advance();
1616                (s, None)
1617            }
1618            // v7.9.22 — single column followed by a pgvector
1619            // opclass ident: `(col vector_cosine_ops)`. mailrs G5.
1620            // SPG's HNSW currently picks its distance metric from
1621            // the query's operator (`<->` / `<#>` / `<=>`), so the
1622            // opclass is informational — accepted and discarded.
1623            // Recognised opclasses: vector_cosine_ops, vector_l2_ops,
1624            // vector_ip_ops, halfvec_*_ops, sq8_*_ops.
1625            Token::Ident(s) | Token::QuotedIdent(s)
1626                if matches!(
1627                    self.tokens.get(self.pos + 1),
1628                    Some(Token::Ident(op) | Token::QuotedIdent(op))
1629                        if is_vector_opclass_name(op)
1630                ) =>
1631            {
1632                self.advance(); // column name
1633                self.advance(); // opclass ident — drop
1634                (s, None)
1635            }
1636            Token::Ident(_) | Token::QuotedIdent(_) => {
1637                let key_expr = self.parse_expr(0)?;
1638                let primary = extract_first_column(&key_expr).ok_or_else(|| {
1639                    self.err(
1640                        "expression index key must reference at least one column".into(),
1641                    )
1642                })?;
1643                (primary, Some(key_expr))
1644            }
1645            other => {
1646                return Err(self.err(format!(
1647                    "expected column ident or expression, got {other:?}"
1648                )));
1649            }
1650        };
1651        // v7.9.14 — accept extra comma-separated columns inside
1652        // the index key parens (`CREATE INDEX … (a, b, c)`).
1653        // mailrs F2. Each extra column may carry an optional
1654        // `ASC` / `DESC` / `NULLS FIRST` / `NULLS LAST` clause
1655        // — parsed and discarded; SPG doesn't honour direction
1656        // on a BTree index today (column ordering is intrinsic
1657        // to the storage). v7.10 will widen to genuine composite
1658        // index keys.
1659        let mut extra_columns: Vec<String> = Vec::new();
1660        // The leading column may also have ASC/DESC after it.
1661        self.consume_optional_index_column_qualifiers();
1662        while matches!(self.peek(), Token::Comma) {
1663            self.advance();
1664            let extra = self.expect_ident_like()?;
1665            self.consume_optional_index_column_qualifiers();
1666            extra_columns.push(extra);
1667        }
1668        if !matches!(self.peek(), Token::RParen) {
1669            return Err(self.err(format!(
1670                "expected ')' after indexed column / expression, got {:?}",
1671                self.peek()
1672            )));
1673        }
1674        self.advance();
1675        // v6.8.0 — optional `INCLUDE (col1, col2, …)` clause for
1676        // index-only-scan annotation. Bare ident (not a reserved
1677        // keyword) so we test by case-insensitive string match.
1678        let included_columns =
1679            if matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("include")) {
1680                self.advance();
1681                if !matches!(self.peek(), Token::LParen) {
1682                    return Err(self.err(format!(
1683                        "expected '(' after INCLUDE, got {:?}",
1684                        self.peek()
1685                    )));
1686                }
1687                self.advance();
1688                let mut cols = Vec::new();
1689                loop {
1690                    cols.push(self.expect_ident_like()?);
1691                    match self.peek() {
1692                        Token::Comma => {
1693                            self.advance();
1694                        }
1695                        Token::RParen => {
1696                            self.advance();
1697                            break;
1698                        }
1699                        other => {
1700                            return Err(self.err(format!(
1701                                "expected ',' or ')' in INCLUDE list, got {other:?}"
1702                            )));
1703                        }
1704                    }
1705                }
1706                cols
1707            } else {
1708                Vec::new()
1709            };
1710        // v6.8.1 — optional `WHERE <expr>` partial-index predicate.
1711        let partial_predicate = if matches!(self.peek(), Token::Where) {
1712            self.advance();
1713            Some(self.parse_expr(0)?)
1714        } else {
1715            None
1716        };
1717        // v7.9.29 — UNIQUE on a vector index (HNSW) makes no
1718        // sense: uniqueness over an ANN structure has no clean
1719        // semantics. Reject early. (BRIN UNIQUE is similarly
1720        // meaningless — block both.)
1721        if is_unique && !matches!(method, IndexMethod::BTree) {
1722            return Err(self.err(alloc::format!(
1723                "UNIQUE is only supported on BTree indexes, got USING {:?}",
1724                method
1725            )));
1726        }
1727        Ok(Statement::CreateIndex(CreateIndexStatement {
1728            name,
1729            table,
1730            column,
1731            method,
1732            if_not_exists,
1733            included_columns,
1734            partial_predicate,
1735            extra_columns: extra_columns.clone(),
1736            expression,
1737            is_unique,
1738        }))
1739    }
1740
1741    /// v7.6.0 — wraps `parse_column_def` and consumes an optional
1742    /// column-level `REFERENCES ...` clause. The trailing FK is
1743    /// normalised into table-level shape (single-element columns +
1744    /// parent_columns) so the engine sees one uniform constraint list.
1745    fn parse_column_def_with_fk(
1746        &mut self,
1747    ) -> Result<(ColumnDef, Option<ForeignKeyConstraint>), ParseError> {
1748        let col = self.parse_column_def()?;
1749        // Inline form: `col INT REFERENCES tbl(pcol) [ON DELETE ...] [ON UPDATE ...]`.
1750        let inline_references = matches!(
1751            self.peek(),
1752            Token::Ident(s) if s.eq_ignore_ascii_case("references")
1753        );
1754        if !inline_references {
1755            return Ok((col, None));
1756        }
1757        let (parent_table, parent_columns, on_delete, on_update) =
1758            self.parse_references_tail(1)?;
1759        let fk = ForeignKeyConstraint {
1760            name: None,
1761            columns: vec![col.name.clone()],
1762            parent_table,
1763            parent_columns,
1764            on_delete,
1765            on_update,
1766        };
1767        Ok((col, Some(fk)))
1768    }
1769
1770    fn parse_column_def(&mut self) -> Result<ColumnDef, ParseError> {
1771        let name = self.expect_ident_like()?;
1772        // Type keyword arrives as a bare Ident (we did not promote type names
1773        // to keyword tokens — see lexer rationale).
1774        let ty_ident = match self.advance() {
1775            Token::Ident(s) => s,
1776            other => {
1777                return Err(ParseError {
1778                    message: format!("expected column type, got {other:?}"),
1779                    token_pos: self.pos.saturating_sub(1),
1780                });
1781            }
1782        };
1783        // v7.9.6 — PG `SERIAL` / `BIGSERIAL` shorthand for
1784        // `INT/BIGINT NOT NULL AUTO_INCREMENT`. PG also defines
1785        // SMALLSERIAL → SMALLINT; we accept that too. The implicit
1786        // NOT NULL + AUTO_INCREMENT flags get baked in after the
1787        // type tag so the rest of the constraint-loop parser sees
1788        // them as if user-supplied (rejecting duplicates).
1789        let mut implied_auto_increment = false;
1790        let mut implied_not_null = false;
1791        let ty = match ty_ident.as_str() {
1792            // PG SERIAL family. Implies NOT NULL + AUTO_INCREMENT.
1793            "smallserial" | "serial2" => {
1794                implied_auto_increment = true;
1795                implied_not_null = true;
1796                ColumnTypeName::SmallInt
1797            }
1798            "serial" | "serial4" => {
1799                implied_auto_increment = true;
1800                implied_not_null = true;
1801                ColumnTypeName::Int
1802            }
1803            "bigserial" | "serial8" => {
1804                implied_auto_increment = true;
1805                implied_not_null = true;
1806                ColumnTypeName::BigInt
1807            }
1808            // MySQL flavours we accept by aliasing to the closest SPG
1809            // type. TINYINT covers MySQL's i8 — held inside SMALLINT
1810            // since SPG doesn't have a dedicated i8. MEDIUMINT (MySQL
1811            // 24-bit) → INT. UNSIGNED modifiers are consumed below
1812            // without semantic effect.
1813            "smallint" | "tinyint" => ColumnTypeName::SmallInt,
1814            // INTEGER is MySQL's spelling for INT; MEDIUMINT widens up.
1815            "int" | "integer" | "mediumint" => ColumnTypeName::Int,
1816            "bigint" => ColumnTypeName::BigInt,
1817            // DOUBLE / REAL are 64-bit IEEE — same as our FLOAT.
1818            "float" | "double" | "real" => ColumnTypeName::Float,
1819            "text" => ColumnTypeName::Text,
1820            "bool" | "boolean" => ColumnTypeName::Bool,
1821            "varchar" => ColumnTypeName::Varchar(self.parse_paren_size("VARCHAR")?),
1822            "char" => ColumnTypeName::Char(self.parse_paren_size("CHAR")?),
1823            "vector" => {
1824                let dim = self.parse_paren_size("VECTOR")?;
1825                let encoding = self.parse_optional_vector_encoding()?;
1826                ColumnTypeName::Vector { dim, encoding }
1827            }
1828            "numeric" => {
1829                let (precision, scale) = self.parse_optional_numeric_params()?;
1830                ColumnTypeName::Numeric(precision, scale)
1831            }
1832            "date" => ColumnTypeName::Date,
1833            // MySQL's `DATETIME` is the same domain as standard
1834            // `TIMESTAMP` — accept both spellings.
1835            "timestamp" | "datetime" => ColumnTypeName::Timestamp,
1836            // v7.9.2 — `TIMESTAMPTZ` and full PG spelling
1837            // `TIMESTAMP WITH TIME ZONE`. Same storage as TIMESTAMP;
1838            // only PG-wire OID differs.
1839            "timestamptz" => ColumnTypeName::Timestamptz,
1840            // v4.9: JSON / JSONB. Stored as raw text — no parse-time
1841            // validation. We accept the JSONB spelling too because
1842            // most PG clients default to it; SPG doesn't distinguish
1843            // the two (no path-operator perf advantage to model).
1844            "json" => ColumnTypeName::Json,
1845            "jsonb" => ColumnTypeName::Jsonb,
1846            other => {
1847                return Err(ParseError {
1848                    message: format!("unsupported column type {other:?}"),
1849                    token_pos: self.pos.saturating_sub(1),
1850                });
1851            }
1852        };
1853        // MySQL's `UNSIGNED` modifier sits right after the type
1854        // keyword. SPG doesn't carry a separate unsigned variant —
1855        // accepting the keyword keeps existing schemas compatible
1856        // without changing semantics. Drop it silently.
1857        if matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("unsigned")) {
1858            self.advance();
1859        }
1860        // Column constraints: `DEFAULT <expr>`, `NOT NULL`, and the
1861        // MySQL-flavoured `AUTO_INCREMENT` may appear in any order;
1862        // each at most once.
1863        let mut default: Option<Expr> = None;
1864        let mut nullable = !implied_not_null;
1865        let mut nullability_seen = implied_not_null;
1866        let mut auto_increment = implied_auto_increment;
1867        let mut is_primary_key = false;
1868        loop {
1869            if matches!(self.peek(), Token::Default) {
1870                if default.is_some() {
1871                    return Err(self.err("DEFAULT specified twice".into()));
1872                }
1873                self.advance();
1874                default = Some(self.parse_expr(0)?);
1875                continue;
1876            }
1877            if matches!(self.peek(), Token::Not) {
1878                if nullability_seen {
1879                    return Err(self.err("NOT NULL specified twice".into()));
1880                }
1881                self.advance();
1882                if !matches!(self.peek(), Token::Null) {
1883                    return Err(self.err(format!(
1884                        "expected NULL after NOT in column def, got {:?}",
1885                        self.peek()
1886                    )));
1887                }
1888                self.advance();
1889                nullable = false;
1890                nullability_seen = true;
1891                continue;
1892            }
1893            // `AUTO_INCREMENT` or its abbreviated form `AUTOINCREMENT`
1894            // arrives as a bare Ident. Match either, case-insensitive.
1895            if let Token::Ident(s) = self.peek()
1896                && (s.eq_ignore_ascii_case("auto_increment")
1897                    || s.eq_ignore_ascii_case("autoincrement"))
1898            {
1899                if auto_increment {
1900                    return Err(self.err("AUTO_INCREMENT specified twice".into()));
1901                }
1902                self.advance();
1903                auto_increment = true;
1904                continue;
1905            }
1906            // v7.9.13 — inline `PRIMARY KEY` column constraint
1907            // (mailrs F1). Implies `NOT NULL`. The engine creates
1908            // a BTree index for the PK column at CREATE TABLE time
1909            // so FK parent-side index lookups resolve.
1910            if let Token::Ident(s) = self.peek()
1911                && s.eq_ignore_ascii_case("primary")
1912            {
1913                if is_primary_key {
1914                    return Err(self.err("PRIMARY KEY specified twice".into()));
1915                }
1916                // Peek-ahead for the required `KEY` token.
1917                let next = self.tokens.get(self.pos + 1);
1918                let next_is_key = matches!(
1919                    next,
1920                    Some(Token::Ident(k)) if k.eq_ignore_ascii_case("key")
1921                );
1922                if !next_is_key {
1923                    return Err(self.err(format!(
1924                        "expected KEY after PRIMARY in column def, got {:?}",
1925                        next
1926                    )));
1927                }
1928                self.advance(); // PRIMARY
1929                self.advance(); // KEY
1930                is_primary_key = true;
1931                if nullability_seen && nullable {
1932                    return Err(self.err(
1933                        "column declared NULL but inline PRIMARY KEY implies NOT NULL".into(),
1934                    ));
1935                }
1936                nullable = false;
1937                nullability_seen = true;
1938                continue;
1939            }
1940            break;
1941        }
1942        Ok(ColumnDef {
1943            name,
1944            ty,
1945            nullable,
1946            default,
1947            auto_increment,
1948            is_primary_key,
1949        })
1950    }
1951
1952    /// `NUMERIC` may appear without parameters, with one (precision
1953    /// only, scale=0), or with both. Returns `(precision, scale)` with
1954    /// 0 = unspecified for the bare form.
1955    fn parse_optional_numeric_params(&mut self) -> Result<(u8, u8), ParseError> {
1956        if !matches!(self.peek(), Token::LParen) {
1957            // Bare `NUMERIC` — PG treats this as "unlimited precision";
1958            // we surface it as precision=0 to mean "unconstrained" so
1959            // the engine doesn't need a separate variant.
1960            return Ok((0, 0));
1961        }
1962        self.advance();
1963        let precision = match self.advance() {
1964            Token::Integer(n) if (1..=38).contains(&n) => u8::try_from(n).expect("range-checked"),
1965            other => {
1966                return Err(ParseError {
1967                    message: format!(
1968                        "NUMERIC precision must be an integer in 1..=38, got {other:?}"
1969                    ),
1970                    token_pos: self.pos.saturating_sub(1),
1971                });
1972            }
1973        };
1974        let scale = if matches!(self.peek(), Token::Comma) {
1975            self.advance();
1976            match self.advance() {
1977                Token::Integer(n) if (0..=i64::from(precision)).contains(&n) => {
1978                    u8::try_from(n).expect("range-checked")
1979                }
1980                other => {
1981                    return Err(ParseError {
1982                        message: format!(
1983                            "NUMERIC scale must be a non-negative integer ≤ precision, got {other:?}"
1984                        ),
1985                        token_pos: self.pos.saturating_sub(1),
1986                    });
1987                }
1988            }
1989        } else {
1990            0
1991        };
1992        if !matches!(self.peek(), Token::RParen) {
1993            return Err(self.err(format!(
1994                "expected ')' to close NUMERIC params, got {:?}",
1995                self.peek()
1996            )));
1997        }
1998        self.advance();
1999        Ok((precision, scale))
2000    }
2001
2002    /// Parse `(N)` where `N` is a positive integer literal — used by the
2003    /// `VARCHAR`/`CHAR`/`VECTOR` column types. `label` is the type name
2004    /// for the error message.
2005    /// v6.0.1: parse the optional `USING <encoding>` clause that
2006    /// follows `VECTOR(N)` in a column definition. Missing clause
2007    /// → `VecEncoding::F32` (pre-v6 default). Unknown encoding
2008    /// ident → `ParseError` listing the encodings recognised today.
2009    fn parse_optional_vector_encoding(&mut self) -> Result<VecEncoding, ParseError> {
2010        if !matches!(self.peek(), Token::Ident(s) if s.eq_ignore_ascii_case("using")) {
2011            return Ok(VecEncoding::F32);
2012        }
2013        self.advance();
2014        let enc_ident = match self.advance() {
2015            Token::Ident(s) => s,
2016            other => {
2017                return Err(self.err(format!(
2018                    "expected vector encoding after USING, got {other:?}"
2019                )));
2020            }
2021        };
2022        match enc_ident.to_ascii_lowercase().as_str() {
2023            "sq8" => Ok(VecEncoding::Sq8),
2024            // v6.0.3: `HALF` (pgvector convention) selects IEEE-754
2025            // binary16 per-element storage.
2026            "half" => Ok(VecEncoding::F16),
2027            other => Err(self.err(format!(
2028                "unknown vector encoding {other:?}; supported: SQ8, HALF"
2029            ))),
2030        }
2031    }
2032
2033    fn parse_paren_size(&mut self, label: &str) -> Result<u32, ParseError> {
2034        if !matches!(self.peek(), Token::LParen) {
2035            return Err(self.err(format!("{label} type requires (N), got {:?}", self.peek())));
2036        }
2037        self.advance();
2038        let n = match self.advance() {
2039            Token::Integer(n) if n > 0 => u32::try_from(n).map_err(|_| ParseError {
2040                message: format!("{label} size too large: {n}"),
2041                token_pos: self.pos.saturating_sub(1),
2042            })?,
2043            other => {
2044                return Err(ParseError {
2045                    message: format!("expected positive integer {label} size, got {other:?}"),
2046                    token_pos: self.pos.saturating_sub(1),
2047                });
2048            }
2049        };
2050        if !matches!(self.peek(), Token::RParen) {
2051            return Err(self.err(format!(
2052                "expected ')' after {label} size, got {:?}",
2053                self.peek()
2054            )));
2055        }
2056        self.advance();
2057        Ok(n)
2058    }
2059
2060    fn parse_insert_stmt(&mut self) -> Result<Statement, ParseError> {
2061        debug_assert!(matches!(self.peek(), Token::Insert));
2062        self.advance();
2063        if !matches!(self.peek(), Token::Into) {
2064            return Err(self.err(format!("expected INTO after INSERT, got {:?}", self.peek())));
2065        }
2066        self.advance();
2067        let table = self.expect_ident_like()?;
2068        // Optional column list — `INSERT INTO t (a, b) VALUES ...`.
2069        let columns = if matches!(self.peek(), Token::LParen) {
2070            self.advance();
2071            let mut names = Vec::new();
2072            loop {
2073                names.push(self.expect_ident_like()?);
2074                match self.peek() {
2075                    Token::Comma => {
2076                        self.advance();
2077                    }
2078                    Token::RParen => {
2079                        self.advance();
2080                        break;
2081                    }
2082                    other => {
2083                        return Err(self.err(format!(
2084                            "expected ',' or ')' in INSERT column list, got {other:?}"
2085                        )));
2086                    }
2087                }
2088            }
2089            Some(names)
2090        } else {
2091            None
2092        };
2093        if !matches!(self.peek(), Token::Values) {
2094            return Err(self.err(format!(
2095                "expected VALUES after table name, got {:?}",
2096                self.peek()
2097            )));
2098        }
2099        self.advance();
2100        if !matches!(self.peek(), Token::LParen) {
2101            return Err(self.err(format!("expected '(' after VALUES, got {:?}", self.peek())));
2102        }
2103        let mut rows = Vec::new();
2104        loop {
2105            // Each iteration consumes one `(expr, expr, …)` tuple.
2106            if !matches!(self.peek(), Token::LParen) {
2107                return Err(self.err(format!(
2108                    "expected '(' for next VALUES tuple, got {:?}",
2109                    self.peek()
2110                )));
2111            }
2112            self.advance();
2113            let mut tuple = Vec::new();
2114            loop {
2115                tuple.push(self.parse_expr(0)?);
2116                match self.peek() {
2117                    Token::Comma => {
2118                        self.advance();
2119                    }
2120                    Token::RParen => {
2121                        self.advance();
2122                        break;
2123                    }
2124                    other => {
2125                        return Err(self.err(format!(
2126                            "expected ',' or ')' in VALUES tuple, got {other:?}"
2127                        )));
2128                    }
2129                }
2130            }
2131            if tuple.is_empty() {
2132                return Err(self.err("INSERT VALUES tuple requires at least one value".into()));
2133            }
2134            rows.push(tuple);
2135            // Continue with comma-separated tuples.
2136            if matches!(self.peek(), Token::Comma) {
2137                self.advance();
2138            } else {
2139                break;
2140            }
2141        }
2142        let on_conflict = self.parse_optional_on_conflict()?;
2143        let returning = self.parse_optional_returning()?;
2144        Ok(Statement::Insert(InsertStatement {
2145            table,
2146            columns,
2147            rows,
2148            on_conflict,
2149            returning,
2150        }))
2151    }
2152
2153    /// v7.9.7 — parse the optional `ON CONFLICT (cols) DO …`
2154    /// clause sitting between the INSERT body and the trailing
2155    /// RETURNING. All keywords come in as bare idents; `ON` is
2156    /// a reserved Token though.
2157    fn parse_optional_on_conflict(
2158        &mut self,
2159    ) -> Result<Option<crate::ast::OnConflictClause>, ParseError> {
2160        if !matches!(self.peek(), Token::On) {
2161            return Ok(None);
2162        }
2163        // Peek further: we want exactly "ON CONFLICT ...". If the
2164        // next ident isn't "conflict", let some other parser handle.
2165        let next_is_conflict = matches!(
2166            self.tokens.get(self.pos + 1),
2167            Some(Token::Ident(s) | Token::QuotedIdent(s)) if s.eq_ignore_ascii_case("conflict")
2168        );
2169        if !next_is_conflict {
2170            return Ok(None);
2171        }
2172        self.advance(); // ON
2173        self.advance(); // CONFLICT
2174        // Optional `(col [, col]*)` target list.
2175        let mut target_columns: Vec<String> = Vec::new();
2176        if matches!(self.peek(), Token::LParen) {
2177            self.advance();
2178            loop {
2179                target_columns.push(self.expect_ident_like()?);
2180                match self.peek() {
2181                    Token::Comma => {
2182                        self.advance();
2183                    }
2184                    Token::RParen => {
2185                        self.advance();
2186                        break;
2187                    }
2188                    other => {
2189                        return Err(self.err(alloc::format!(
2190                            "expected ',' or ')' in ON CONFLICT target list, got {other:?}"
2191                        )));
2192                    }
2193                }
2194            }
2195        }
2196        // Required `DO`.
2197        match self.advance() {
2198            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("do") => {}
2199            other => {
2200                return Err(self.err(alloc::format!(
2201                    "expected DO after ON CONFLICT [(…)], got {other:?}"
2202                )));
2203            }
2204        }
2205        // Action: NOTHING | UPDATE SET …
2206        let action = match self.advance() {
2207            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("nothing") => {
2208                crate::ast::OnConflictAction::Nothing
2209            }
2210            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("update") => {
2211                self.parse_on_conflict_update_action()?
2212            }
2213            other => {
2214                return Err(self.err(alloc::format!(
2215                    "expected NOTHING or UPDATE after ON CONFLICT DO, got {other:?}"
2216                )));
2217            }
2218        };
2219        Ok(Some(crate::ast::OnConflictClause {
2220            target_columns,
2221            action,
2222        }))
2223    }
2224
2225    /// v7.9.7 — tail of `ON CONFLICT … DO UPDATE`: parse
2226    /// `SET col = expr [, …] [WHERE cond]`. Caller already
2227    /// consumed `UPDATE`.
2228    fn parse_on_conflict_update_action(
2229        &mut self,
2230    ) -> Result<crate::ast::OnConflictAction, ParseError> {
2231        // `SET`
2232        match self.advance() {
2233            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("set") => {}
2234            other => {
2235                return Err(self.err(alloc::format!(
2236                    "expected SET after ON CONFLICT DO UPDATE, got {other:?}"
2237                )));
2238            }
2239        }
2240        let mut assignments: Vec<(String, Expr)> = Vec::new();
2241        loop {
2242            let col = self.expect_ident_like()?;
2243            if !matches!(self.peek(), Token::Eq) {
2244                return Err(self.err(alloc::format!(
2245                    "expected `=` after column in ON CONFLICT DO UPDATE SET, got {:?}",
2246                    self.peek()
2247                )));
2248            }
2249            self.advance();
2250            let value = self.parse_expr(0)?;
2251            assignments.push((col, value));
2252            if matches!(self.peek(), Token::Comma) {
2253                self.advance();
2254                continue;
2255            }
2256            break;
2257        }
2258        let where_ = if matches!(self.peek(), Token::Where) {
2259            self.advance();
2260            Some(self.parse_expr(0)?)
2261        } else {
2262            None
2263        };
2264        Ok(crate::ast::OnConflictAction::Update {
2265            assignments,
2266            where_,
2267        })
2268    }
2269
2270    fn parse_select_list(&mut self) -> Result<Vec<SelectItem>, ParseError> {
2271        let mut items = Vec::new();
2272        loop {
2273            items.push(self.parse_select_item()?);
2274            if matches!(self.peek(), Token::Comma) {
2275                self.advance();
2276            } else {
2277                break;
2278            }
2279        }
2280        Ok(items)
2281    }
2282
2283    fn parse_select_item(&mut self) -> Result<SelectItem, ParseError> {
2284        if matches!(self.peek(), Token::Star) {
2285            self.advance();
2286            return Ok(SelectItem::Wildcard);
2287        }
2288        let expr = self.parse_expr(0)?;
2289        let alias = self.parse_optional_alias();
2290        Ok(SelectItem::Expr { expr, alias })
2291    }
2292
2293    fn parse_table_ref(&mut self) -> Result<TableRef, ParseError> {
2294        let name = self.expect_ident_like()?;
2295        // v6.10.2 — optional `AS OF SEGMENT '<id>'` cold-tier
2296        // time-travel clause. Parse BEFORE the alias so the
2297        // alias can still ride at the tail (`tbl AS OF SEGMENT
2298        // '5' alias`). `AS` is a reserved keyword token, while
2299        // `OF` and `SEGMENT` are bare idents.
2300        let as_of_segment = if matches!(self.peek(), Token::As)
2301            && matches!(self.tokens.get(self.pos + 1), Some(Token::Ident(s) | Token::QuotedIdent(s)) if s.eq_ignore_ascii_case("of"))
2302        {
2303            self.advance(); // AS
2304            self.advance(); // OF
2305            let kw = match self.peek().clone() {
2306                Token::Ident(s) | Token::QuotedIdent(s) => s,
2307                other => {
2308                    return Err(self.err(format!(
2309                        "expected SEGMENT after AS OF, got {other:?}"
2310                    )));
2311                }
2312            };
2313            if !kw.eq_ignore_ascii_case("segment") {
2314                return Err(self.err(format!(
2315                    "expected SEGMENT after AS OF, got {kw:?}; v6.10.2 supports SEGMENT only"
2316                )));
2317            }
2318            self.advance();
2319            // Segment id literal — accept either a string or
2320            // integer for operator ergonomics.
2321            let id = match self.advance() {
2322                Token::String(s) => s
2323                    .parse::<u32>()
2324                    .map_err(|e| self.err(format!("AS OF SEGMENT id parse: {e}")))?,
2325                Token::Integer(n) => u32::try_from(n).map_err(|e| {
2326                    self.err(format!("AS OF SEGMENT id parse: {e}"))
2327                })?,
2328                other => {
2329                    return Err(self.err(format!(
2330                        "expected segment id literal after AS OF SEGMENT, got {other:?}"
2331                    )));
2332                }
2333            };
2334            Some(id)
2335        } else {
2336            None
2337        };
2338        let alias = self.parse_optional_alias();
2339        Ok(TableRef {
2340            name,
2341            alias,
2342            as_of_segment,
2343        })
2344    }
2345
2346    /// FROM-clause: a primary table reference plus zero-or-more joined
2347    /// peers expressed via either `, <table>` (cross-product, no ON) or
2348    /// `[INNER|LEFT [OUTER]|CROSS] JOIN <table> [ON expr]`. v1.10 keeps
2349    /// the join list flat (left-associative nested-loop semantics).
2350    fn parse_from_clause(&mut self) -> Result<FromClause, ParseError> {
2351        let primary = self.parse_table_ref()?;
2352        let mut joins = Vec::new();
2353        loop {
2354            // `, <table>` — cross-product with no ON.
2355            if matches!(self.peek(), Token::Comma) {
2356                self.advance();
2357                let table = self.parse_table_ref()?;
2358                joins.push(FromJoin {
2359                    kind: JoinKind::Cross,
2360                    table,
2361                    on: None,
2362                });
2363                continue;
2364            }
2365            // Explicit JOIN syntax. Accept INNER JOIN, LEFT [OUTER] JOIN,
2366            // CROSS JOIN, and bare JOIN (defaults to INNER).
2367            let kind =
2368                match self.peek() {
2369                    Token::Inner => {
2370                        self.advance();
2371                        if !matches!(self.peek(), Token::Join) {
2372                            return Err(self
2373                                .err(format!("expected JOIN after INNER, got {:?}", self.peek())));
2374                        }
2375                        self.advance();
2376                        JoinKind::Inner
2377                    }
2378                    Token::Left => {
2379                        self.advance();
2380                        if matches!(self.peek(), Token::Outer) {
2381                            self.advance();
2382                        }
2383                        if !matches!(self.peek(), Token::Join) {
2384                            return Err(self.err(format!(
2385                                "expected JOIN after LEFT [OUTER], got {:?}",
2386                                self.peek()
2387                            )));
2388                        }
2389                        self.advance();
2390                        JoinKind::Left
2391                    }
2392                    Token::Cross => {
2393                        self.advance();
2394                        if !matches!(self.peek(), Token::Join) {
2395                            return Err(self
2396                                .err(format!("expected JOIN after CROSS, got {:?}", self.peek())));
2397                        }
2398                        self.advance();
2399                        JoinKind::Cross
2400                    }
2401                    Token::Join => {
2402                        self.advance();
2403                        JoinKind::Inner
2404                    }
2405                    _ => break,
2406                };
2407            let table = self.parse_table_ref()?;
2408            let on = if matches!(self.peek(), Token::On) {
2409                self.advance();
2410                Some(self.parse_expr(0)?)
2411            } else if kind == JoinKind::Cross {
2412                None
2413            } else {
2414                return Err(self.err(format!(
2415                    "expected ON after {:?} JOIN, got {:?}",
2416                    kind,
2417                    self.peek()
2418                )));
2419            };
2420            joins.push(FromJoin { kind, table, on });
2421        }
2422        Ok(FromClause { primary, joins })
2423    }
2424
2425    /// Optional alias after an expression or table:
2426    /// `AS <ident>` is unambiguous; a bare `<ident>` directly after is also
2427    /// accepted (PG-style implicit alias). Returns `None` if the next token
2428    /// is not alias-shaped (e.g. comma, FROM, WHERE, semicolon, EOF, operator).
2429    fn parse_optional_alias(&mut self) -> Option<String> {
2430        if matches!(self.peek(), Token::As) {
2431            self.advance();
2432            // After AS, the next token MUST be an identifier-like — if not,
2433            // we still return None and let the caller surface the error on the
2434            // next expectation. v0.2 keeps the alias path forgiving; the
2435            // corpus tests don't exercise the malformed case.
2436            if let Token::Ident(_) | Token::QuotedIdent(_) = self.peek() {
2437                return self.expect_ident_like().ok();
2438            }
2439            return None;
2440        }
2441        if let Token::Ident(_) | Token::QuotedIdent(_) = self.peek() {
2442            return self.expect_ident_like().ok();
2443        }
2444        None
2445    }
2446
2447    /// Pratt loop. `min_prec` is the minimum binary-op precedence we'll accept.
2448    fn parse_expr(&mut self, min_prec: u8) -> Result<Expr, ParseError> {
2449        let mut lhs = self.parse_unary()?;
2450        while let Some((op, prec)) = binop_from(self.peek()) {
2451            if prec < min_prec {
2452                break;
2453            }
2454            self.advance();
2455            let rhs = self.parse_expr(prec + 1)?;
2456            lhs = Expr::Binary {
2457                lhs: Box::new(lhs),
2458                op,
2459                rhs: Box::new(rhs),
2460            };
2461        }
2462        Ok(lhs)
2463    }
2464
2465    fn parse_unary(&mut self) -> Result<Expr, ParseError> {
2466        match self.peek() {
2467            Token::Not => {
2468                self.advance();
2469                // NOT sits between AND (2) and comparisons (4) — bind everything
2470                // ≥3, which leaves AND/OR outside.
2471                let e = self.parse_expr(3)?;
2472                Ok(Expr::Unary {
2473                    op: UnOp::Not,
2474                    expr: Box::new(e),
2475                })
2476            }
2477            Token::Minus => {
2478                self.advance();
2479                // Unary minus binds tighter than `*`/`/` (now at prec 7 after
2480                // `<->` slotted into 5 and arithmetic shifted up).
2481                let e = self.parse_expr(8)?;
2482                Ok(Expr::Unary {
2483                    op: UnOp::Neg,
2484                    expr: Box::new(e),
2485                })
2486            }
2487            _ => self.parse_atom(),
2488        }
2489    }
2490
2491    fn parse_atom(&mut self) -> Result<Expr, ParseError> {
2492        let tok_pos = self.pos;
2493        match self.advance() {
2494            Token::Integer(n) => Ok(Expr::Literal(Literal::Integer(n))),
2495            Token::Float(x) => Ok(Expr::Literal(Literal::Float(x))),
2496            Token::String(s) => Ok(Expr::Literal(Literal::String(s))),
2497            Token::True => Ok(Expr::Literal(Literal::Bool(true))),
2498            Token::False => Ok(Expr::Literal(Literal::Bool(false))),
2499            Token::Null => Ok(Expr::Literal(Literal::Null)),
2500            // v6.1.1 — `$N` placeholder. The actual Value lookup
2501            // happens in the engine eval path against the prepared-
2502            // statement bind buffer.
2503            Token::Placeholder(n) => Ok(Expr::Placeholder(n)),
2504            Token::LParen => {
2505                // v4.10: `(SELECT ...)` in expression position is a
2506                // scalar subquery; otherwise it's a parenthesised
2507                // expression. Peek for SELECT keyword to dispatch.
2508                if matches!(self.peek(), Token::Select) {
2509                    let inner = self.parse_select_stmt()?;
2510                    match self.advance() {
2511                        Token::RParen => {
2512                            let Statement::Select(s) = inner else {
2513                                unreachable!("parse_select_stmt returns Select")
2514                            };
2515                            Ok(Expr::ScalarSubquery(Box::new(s)))
2516                        }
2517                        other => Err(ParseError {
2518                            message: format!("expected ')' after scalar subquery, got {other:?}"),
2519                            token_pos: self.pos.saturating_sub(1),
2520                        }),
2521                    }
2522                } else {
2523                    let e = self.parse_expr(0)?;
2524                    match self.advance() {
2525                        Token::RParen => Ok(e),
2526                        other => Err(ParseError {
2527                            message: format!("expected ')', got {other:?}"),
2528                            token_pos: self.pos.saturating_sub(1),
2529                        }),
2530                    }
2531                }
2532            }
2533            Token::LBracket => self.parse_vector_literal_body(),
2534            Token::Extract => self.parse_extract_atom(),
2535            Token::Interval => self.parse_interval_atom(),
2536            // v4.10: EXISTS / NOT EXISTS. EXISTS isn't a reserved
2537            // token; we match on the bare ident. NOT is a token
2538            // (consumed in the comparison rung), but `EXISTS (...)`
2539            // at the top of an expression starts here.
2540            Token::Ident(s) | Token::QuotedIdent(s) if s.eq_ignore_ascii_case("exists") => {
2541                self.parse_exists_atom(false)
2542            }
2543            Token::Ident(s) | Token::QuotedIdent(s) => self.finish_ident_atom(s),
2544            other => Err(ParseError {
2545                message: format!("unexpected token {other:?} in expression"),
2546                token_pos: tok_pos,
2547            }),
2548        }
2549        // After parsing the atom, fold any postfix `::vector` casts.
2550        .and_then(|atom| self.finish_postfix_casts(atom))
2551    }
2552
2553    /// Postfix operators on an atom: `::TYPE` cast and `IS [NOT] NULL`.
2554    /// Both bind tighter than any binary op.
2555    fn finish_postfix_casts(&mut self, mut expr: Expr) -> Result<Expr, ParseError> {
2556        loop {
2557            if matches!(self.peek(), Token::DoubleColon) {
2558                self.advance();
2559                // v7.9.25 / v7.9.26 — broaden the postfix `::` cast
2560                // target set to include INTERVAL (reserved Token),
2561                // TIMESTAMPTZ, and PG catalog regtype / regclass.
2562                // mailrs follow-up H3a + H3b.
2563                let target = match self.advance() {
2564                    Token::Ident(s) => match s.to_ascii_lowercase().as_str() {
2565                        "int" | "integer" | "int4" => CastTarget::Int,
2566                        "bigint" | "int8" => CastTarget::BigInt,
2567                        "float" | "double" | "real" => CastTarget::Float,
2568                        "text" => CastTarget::Text,
2569                        "bool" | "boolean" => CastTarget::Bool,
2570                        "vector" => CastTarget::Vector,
2571                        "date" => CastTarget::Date,
2572                        "timestamp" | "datetime" => CastTarget::Timestamp,
2573                        "timestamptz" => CastTarget::Timestamptz,
2574                        "interval" => CastTarget::Interval,
2575                        "json" => CastTarget::Json,
2576                        "jsonb" => CastTarget::Jsonb,
2577                        "regtype" => CastTarget::RegType,
2578                        "regclass" => CastTarget::RegClass,
2579                        other => {
2580                            return Err(ParseError {
2581                                message: format!("unsupported cast target `::{other}`"),
2582                                token_pos: self.pos.saturating_sub(1),
2583                            });
2584                        }
2585                    },
2586                    Token::Interval => CastTarget::Interval,
2587                    other => {
2588                        return Err(ParseError {
2589                            message: format!("expected type ident after `::`, got {other:?}"),
2590                            token_pos: self.pos.saturating_sub(1),
2591                        });
2592                    }
2593                };
2594                expr = Expr::Cast {
2595                    expr: Box::new(expr),
2596                    target,
2597                };
2598                continue;
2599            }
2600            if matches!(self.peek(), Token::Is) {
2601                self.advance();
2602                let negated = if matches!(self.peek(), Token::Not) {
2603                    self.advance();
2604                    true
2605                } else {
2606                    false
2607                };
2608                // v7.9.27b — `IS [NOT] DISTINCT FROM <rhs>`.
2609                // mailrs pg_dump.
2610                if matches!(self.peek(), Token::Distinct) {
2611                    self.advance();
2612                    if !matches!(self.peek(), Token::From) {
2613                        return Err(self.err(format!(
2614                            "expected FROM after IS{} DISTINCT, got {:?}",
2615                            if negated { " NOT" } else { "" },
2616                            self.peek()
2617                        )));
2618                    }
2619                    self.advance();
2620                    // Right-hand side: parse at the same precedence
2621                    // tier as comparison so `x IS DISTINCT FROM a + b`
2622                    // groups as `x IS DISTINCT FROM (a + b)`.
2623                    let rhs = self.parse_expr(20)?;
2624                    let op = if negated {
2625                        BinOp::IsNotDistinctFrom
2626                    } else {
2627                        BinOp::IsDistinctFrom
2628                    };
2629                    expr = Expr::Binary {
2630                        op,
2631                        lhs: Box::new(expr),
2632                        rhs: Box::new(rhs),
2633                    };
2634                    continue;
2635                }
2636                if !matches!(self.peek(), Token::Null) {
2637                    return Err(self.err(format!(
2638                        "expected NULL or DISTINCT after IS{}, got {:?}",
2639                        if negated { " NOT" } else { "" },
2640                        self.peek()
2641                    )));
2642                }
2643                self.advance();
2644                expr = Expr::IsNull {
2645                    expr: Box::new(expr),
2646                    negated,
2647                };
2648                continue;
2649            }
2650            // `x [NOT] BETWEEN a AND b`, `x [NOT] IN (...)`, `x [NOT] LIKE p`.
2651            // Look one token ahead so a stray `NOT` not followed by any of
2652            // these flows through to the early return below untouched.
2653            let negated = if matches!(self.peek(), Token::Not) {
2654                let next = self.tokens.get(self.pos + 1);
2655                matches!(next, Some(Token::Between | Token::In | Token::Like))
2656            } else {
2657                false
2658            };
2659            if negated {
2660                self.advance();
2661            }
2662            if matches!(self.peek(), Token::Between) {
2663                expr = self.parse_between_tail(expr, negated)?;
2664                continue;
2665            }
2666            if matches!(self.peek(), Token::In) {
2667                expr = self.parse_in_tail(expr, negated)?;
2668                continue;
2669            }
2670            if matches!(self.peek(), Token::Like) {
2671                self.advance();
2672                // Pattern at the same precedence as other comparison RHSes —
2673                // 5 leaves AND/OR alone so `a LIKE 'x%' AND b` parses right.
2674                let pattern = self.parse_expr(5)?;
2675                expr = Expr::Like {
2676                    expr: Box::new(expr),
2677                    pattern: Box::new(pattern),
2678                    negated,
2679                };
2680                continue;
2681            }
2682            return Ok(expr);
2683        }
2684    }
2685
2686    /// `x BETWEEN low AND high`  →  `(x >= low) AND (x <= high)`, wrapped in
2687    /// `NOT` when `negated`. Bounds parse at precedence 5 so the trailing
2688    /// `AND` is not swallowed.
2689    fn parse_between_tail(&mut self, expr: Expr, negated: bool) -> Result<Expr, ParseError> {
2690        self.advance(); // BETWEEN
2691        let low = self.parse_expr(5)?;
2692        if !matches!(self.peek(), Token::And) {
2693            return Err(self.err(format!(
2694                "expected AND after BETWEEN low bound, got {:?}",
2695                self.peek()
2696            )));
2697        }
2698        self.advance();
2699        let high = self.parse_expr(5)?;
2700        let target = Box::new(expr);
2701        let combined = Expr::Binary {
2702            lhs: Box::new(Expr::Binary {
2703                lhs: target.clone(),
2704                op: BinOp::GtEq,
2705                rhs: Box::new(low),
2706            }),
2707            op: BinOp::And,
2708            rhs: Box::new(Expr::Binary {
2709                lhs: target,
2710                op: BinOp::LtEq,
2711                rhs: Box::new(high),
2712            }),
2713        };
2714        Ok(maybe_not(combined, negated))
2715    }
2716
2717    /// `x IN (a, b, c)`  →  chained OR of equalities. Empty list collapses
2718    /// to FALSE (TRUE under NOT IN), matching standard SQL semantics.
2719    /// v4.11: parse `WITH name AS (SELECT ...) [, ...] SELECT ...`.
2720    /// Caller already consumed the leading `WITH` ident.
2721    fn parse_with_cte_then_select(&mut self) -> Result<Statement, ParseError> {
2722        // v4.22: WITH RECURSIVE — optional keyword right after WITH.
2723        // Comes through as an identifier; consume it if present and
2724        // mark every CTE in the clause as recursive (PG semantics —
2725        // the flag is per-WITH, not per-CTE).
2726        let mut recursive = false;
2727        if let Token::Ident(s) | Token::QuotedIdent(s) = self.peek()
2728            && s.eq_ignore_ascii_case("recursive")
2729        {
2730            self.advance();
2731            recursive = true;
2732        }
2733        let mut ctes = Vec::new();
2734        loop {
2735            let name = self.expect_ident_like()?;
2736            // v4.22: optional column-name list — `WITH t(a,b,c) AS ...`.
2737            // PG uses these to rename the body's output columns; we
2738            // do the same below by overriding `columns[i].name`.
2739            let column_overrides: Vec<String> = if matches!(self.peek(), Token::LParen) {
2740                self.advance();
2741                let mut names = Vec::new();
2742                loop {
2743                    names.push(self.expect_ident_like()?);
2744                    if matches!(self.peek(), Token::Comma) {
2745                        self.advance();
2746                        continue;
2747                    }
2748                    break;
2749                }
2750                if !matches!(self.peek(), Token::RParen) {
2751                    return Err(self.err(format!(
2752                        "expected ')' to close CTE column list, got {:?}",
2753                        self.peek()
2754                    )));
2755                }
2756                self.advance();
2757                names
2758            } else {
2759                Vec::new()
2760            };
2761            // AS is a reserved Token::As (used by SELECT-item / FROM
2762            // aliasing) — handle it specially rather than as a bare
2763            // ident.
2764            if !matches!(self.peek(), Token::As) {
2765                return Err(self.err(format!(
2766                    "expected AS after CTE name {name:?}, got {:?}",
2767                    self.peek()
2768                )));
2769            }
2770            self.advance();
2771            if !matches!(self.peek(), Token::LParen) {
2772                return Err(self.err(format!(
2773                    "expected '(' after AS in WITH clause, got {:?}",
2774                    self.peek()
2775                )));
2776            }
2777            self.advance();
2778            if !matches!(self.peek(), Token::Select) {
2779                return Err(self.err(format!("WITH body must be a SELECT, got {:?}", self.peek())));
2780            }
2781            let inner = self.parse_select_stmt()?;
2782            if !matches!(self.peek(), Token::RParen) {
2783                return Err(self.err(format!(
2784                    "expected ')' after CTE body, got {:?}",
2785                    self.peek()
2786                )));
2787            }
2788            self.advance();
2789            let Statement::Select(body) = inner else {
2790                unreachable!("parse_select_stmt returns Select")
2791            };
2792            ctes.push(crate::ast::Cte {
2793                name,
2794                body,
2795                recursive,
2796                column_overrides,
2797            });
2798            if matches!(self.peek(), Token::Comma) {
2799                self.advance();
2800                continue;
2801            }
2802            break;
2803        }
2804        // The body SELECT follows. Must start with SELECT.
2805        if !matches!(self.peek(), Token::Select) {
2806            return Err(self.err(format!(
2807                "expected SELECT after WITH clause, got {:?}",
2808                self.peek()
2809            )));
2810        }
2811        let body_stmt = self.parse_select_stmt()?;
2812        let Statement::Select(mut body) = body_stmt else {
2813            unreachable!()
2814        };
2815        body.ctes = ctes;
2816        Ok(Statement::Select(body))
2817    }
2818
2819    /// v4.10: parse `EXISTS (SELECT ...)`. Caller (`parse_atom`)
2820    /// already consumed the leading `EXISTS` ident via
2821    /// `self.advance()`.
2822    fn parse_exists_atom(&mut self, negated: bool) -> Result<Expr, ParseError> {
2823        if !matches!(self.peek(), Token::LParen) {
2824            return Err(self.err(format!("expected '(' after EXISTS, got {:?}", self.peek())));
2825        }
2826        self.advance();
2827        let inner = self.parse_select_stmt()?;
2828        if !matches!(self.peek(), Token::RParen) {
2829            return Err(self.err(format!(
2830                "expected ')' after EXISTS-subquery, got {:?}",
2831                self.peek()
2832            )));
2833        }
2834        self.advance();
2835        let Statement::Select(s) = inner else {
2836            unreachable!("parse_select_stmt returns Select")
2837        };
2838        Ok(Expr::Exists {
2839            subquery: Box::new(s),
2840            negated,
2841        })
2842    }
2843
2844    fn parse_in_tail(&mut self, expr: Expr, negated: bool) -> Result<Expr, ParseError> {
2845        self.advance(); // IN
2846        if !matches!(self.peek(), Token::LParen) {
2847            return Err(self.err(format!("expected '(' after IN, got {:?}", self.peek())));
2848        }
2849        self.advance();
2850        // v4.10: `IN (SELECT ...)` — subquery branch.
2851        if matches!(self.peek(), Token::Select) {
2852            let inner = self.parse_select_stmt()?;
2853            if !matches!(self.peek(), Token::RParen) {
2854                return Err(self.err(format!(
2855                    "expected ')' after IN-subquery, got {:?}",
2856                    self.peek()
2857                )));
2858            }
2859            self.advance();
2860            let Statement::Select(s) = inner else {
2861                unreachable!("parse_select_stmt always returns Statement::Select")
2862            };
2863            return Ok(Expr::InSubquery {
2864                expr: Box::new(expr),
2865                subquery: Box::new(s),
2866                negated,
2867            });
2868        }
2869        let mut elements = Vec::new();
2870        if !matches!(self.peek(), Token::RParen) {
2871            loop {
2872                elements.push(self.parse_expr(0)?);
2873                match self.peek() {
2874                    Token::Comma => {
2875                        self.advance();
2876                    }
2877                    Token::RParen => break,
2878                    other => {
2879                        return Err(
2880                            self.err(format!("expected ',' or ')' in IN list, got {other:?}"))
2881                        );
2882                    }
2883                }
2884            }
2885        }
2886        self.advance(); // ')'
2887        let target = Box::new(expr);
2888        let combined = if elements.is_empty() {
2889            Expr::Literal(Literal::Bool(false))
2890        } else {
2891            let mut iter = elements.into_iter();
2892            let first = iter.next().unwrap();
2893            let mut acc = Expr::Binary {
2894                lhs: target.clone(),
2895                op: BinOp::Eq,
2896                rhs: Box::new(first),
2897            };
2898            for elt in iter {
2899                acc = Expr::Binary {
2900                    lhs: Box::new(acc),
2901                    op: BinOp::Or,
2902                    rhs: Box::new(Expr::Binary {
2903                        lhs: target.clone(),
2904                        op: BinOp::Eq,
2905                        rhs: Box::new(elt),
2906                    }),
2907                };
2908            }
2909            acc
2910        };
2911        Ok(maybe_not(combined, negated))
2912    }
2913
2914    /// Parse a pgvector array literal `[ x1, x2, ... ]`. The opening `[` is
2915    /// already consumed by the caller. Elements must be numeric literals
2916    /// (with optional unary `-`); any compound expression is rejected at
2917    /// parse time so the runtime never needs to evaluate inside a vector.
2918    /// `EXTRACT(<field> FROM <source>)`. The dispatching `parse_atom`
2919    /// has already consumed the `EXTRACT` token before calling us —
2920    /// we pick up at the opening `(`.
2921    fn parse_extract_atom(&mut self) -> Result<Expr, ParseError> {
2922        if !matches!(self.peek(), Token::LParen) {
2923            return Err(self.err(format!("expected '(' after EXTRACT, got {:?}", self.peek())));
2924        }
2925        self.advance();
2926        let field_name = self.expect_ident_like()?;
2927        let field = match field_name.to_ascii_lowercase().as_str() {
2928            "year" => ExtractField::Year,
2929            "month" => ExtractField::Month,
2930            "day" => ExtractField::Day,
2931            "hour" => ExtractField::Hour,
2932            "minute" => ExtractField::Minute,
2933            "second" => ExtractField::Second,
2934            "microsecond" | "microseconds" => ExtractField::Microsecond,
2935            other => {
2936                return Err(self.err(format!(
2937                    "unknown EXTRACT field {other:?}; \
2938                     supported: YEAR, MONTH, DAY, HOUR, MINUTE, SECOND, MICROSECOND"
2939                )));
2940            }
2941        };
2942        if !matches!(self.peek(), Token::From) {
2943            return Err(self.err(format!(
2944                "expected FROM after EXTRACT field, got {:?}",
2945                self.peek()
2946            )));
2947        }
2948        self.advance();
2949        let source = self.parse_expr(0)?;
2950        if !matches!(self.peek(), Token::RParen) {
2951            return Err(self.err(format!(
2952                "expected ')' to close EXTRACT, got {:?}",
2953                self.peek()
2954            )));
2955        }
2956        self.advance();
2957        Ok(Expr::Extract {
2958            field,
2959            source: Box::new(source),
2960        })
2961    }
2962
2963    /// `INTERVAL '<n> <unit> [<n> <unit> ...]'` — the `INTERVAL` keyword
2964    /// is already consumed; we expect a single string literal next and
2965    /// resolve it into `Literal::Interval` at parse time so the engine
2966    /// never has to re-tokenise inside the string.
2967    fn parse_interval_atom(&mut self) -> Result<Expr, ParseError> {
2968        let tok = self.advance();
2969        let Token::String(text) = tok else {
2970            return Err(self.err(format!(
2971                "expected string literal after INTERVAL, got {tok:?}"
2972            )));
2973        };
2974        let (months, micros) = parse_interval_text(&text).ok_or_else(|| ParseError {
2975            message: format!(
2976                "cannot parse INTERVAL {text:?}; \
2977                     expected `<n> <unit> [<n> <unit> ...]` with units \
2978                     microsecond[s], millisecond[s], second[s], minute[s], \
2979                     hour[s], day[s], week[s], month[s], year[s]"
2980            ),
2981            token_pos: self.pos.saturating_sub(1),
2982        })?;
2983        Ok(Expr::Literal(Literal::Interval {
2984            months,
2985            micros,
2986            text,
2987        }))
2988    }
2989
2990    fn parse_vector_literal_body(&mut self) -> Result<Expr, ParseError> {
2991        let mut elems = Vec::new();
2992        if matches!(self.peek(), Token::RBracket) {
2993            self.advance();
2994            return Ok(Expr::Literal(Literal::Vector(elems)));
2995        }
2996        loop {
2997            let e = self.parse_expr(0)?;
2998            let x = extract_numeric_literal(&e).ok_or_else(|| ParseError {
2999                message: format!("vector element must be a numeric literal, got {e:?}"),
3000                token_pos: self.pos,
3001            })?;
3002            elems.push(x);
3003            match self.peek() {
3004                Token::Comma => {
3005                    self.advance();
3006                }
3007                Token::RBracket => {
3008                    self.advance();
3009                    break;
3010                }
3011                other => {
3012                    return Err(self.err(format!("expected ',' or ']' in vector, got {other:?}")));
3013                }
3014            }
3015        }
3016        Ok(Expr::Literal(Literal::Vector(elems)))
3017    }
3018
3019    /// Atom that started with an identifier: could be `t.col`, `col`, or
3020    /// `func(arg, ...)`. Detect each shape by looking at the next token.
3021    /// v4.12: parse `(PARTITION BY expr, ... ORDER BY expr [DESC]
3022    /// [, ...])`. Caller has already consumed `OVER`. Either clause
3023    /// is optional; an empty `()` is also legal (PG semantics).
3024    /// v6.4.2 — consume an optional `IGNORE NULLS` / `RESPECT NULLS`
3025    /// modifier between `name(args)` and `OVER (...)`. Default is
3026    /// `Respect`. Unrecognised idents leave the stream unchanged.
3027    fn parse_null_treatment_modifier(&mut self) -> NullTreatment {
3028        let Token::Ident(s) = self.peek().clone() else {
3029            return NullTreatment::Respect;
3030        };
3031        let is_ignore = s.eq_ignore_ascii_case("ignore");
3032        let is_respect = s.eq_ignore_ascii_case("respect");
3033        if !is_ignore && !is_respect {
3034            return NullTreatment::Respect;
3035        }
3036        // Lookahead for NULLS — only consume both tokens together.
3037        // pos+1 must hold a "nulls" ident.
3038        if self.pos + 1 < self.tokens.len()
3039            && let Token::Ident(s2) = &self.tokens[self.pos + 1]
3040            && s2.eq_ignore_ascii_case("nulls")
3041        {
3042            self.advance();
3043            self.advance();
3044            return if is_ignore {
3045                NullTreatment::Ignore
3046            } else {
3047                NullTreatment::Respect
3048            };
3049        }
3050        NullTreatment::Respect
3051    }
3052
3053    /// No frame clause is supported.
3054    #[allow(clippy::type_complexity)] // (partitions, ordered-keys-with-desc) is the natural shape
3055    fn parse_over_clause(
3056        &mut self,
3057    ) -> Result<(Vec<Expr>, Vec<(Expr, bool)>, Option<WindowFrame>), ParseError> {
3058        if !matches!(self.peek(), Token::LParen) {
3059            return Err(self.err(format!("expected '(' after OVER, got {:?}", self.peek())));
3060        }
3061        self.advance();
3062        let mut partition_by = Vec::new();
3063        let mut order_by = Vec::new();
3064        // PARTITION BY ?
3065        if let Token::Ident(s) | Token::QuotedIdent(s) = self.peek()
3066            && s.eq_ignore_ascii_case("partition")
3067        {
3068            self.advance();
3069            if !matches!(self.peek(), Token::By) {
3070                return Err(self.err(format!(
3071                    "expected BY after PARTITION, got {:?}",
3072                    self.peek()
3073                )));
3074            }
3075            self.advance();
3076            loop {
3077                partition_by.push(self.parse_expr(0)?);
3078                if matches!(self.peek(), Token::Comma) {
3079                    self.advance();
3080                    continue;
3081                }
3082                break;
3083            }
3084        }
3085        // ORDER BY ?
3086        if matches!(self.peek(), Token::Order) {
3087            self.advance();
3088            if !matches!(self.peek(), Token::By) {
3089                return Err(self.err(format!("expected BY after ORDER, got {:?}", self.peek())));
3090            }
3091            self.advance();
3092            loop {
3093                let e = self.parse_expr(0)?;
3094                let desc = if matches!(self.peek(), Token::Desc) {
3095                    self.advance();
3096                    true
3097                } else if matches!(self.peek(), Token::Asc) {
3098                    self.advance();
3099                    false
3100                } else {
3101                    false
3102                };
3103                order_by.push((e, desc));
3104                if matches!(self.peek(), Token::Comma) {
3105                    self.advance();
3106                    continue;
3107                }
3108                break;
3109            }
3110        }
3111        // v4.20: optional explicit frame, `ROWS ...` / `RANGE ...`.
3112        // Both keywords come through the lexer as identifiers; match
3113        // case-insensitively.
3114        let mut frame: Option<WindowFrame> = None;
3115        if let Token::Ident(s) | Token::QuotedIdent(s) = self.peek() {
3116            let kind = if s.eq_ignore_ascii_case("rows") {
3117                Some(FrameKind::Rows)
3118            } else if s.eq_ignore_ascii_case("range") {
3119                Some(FrameKind::Range)
3120            } else {
3121                None
3122            };
3123            if let Some(kind) = kind {
3124                self.advance();
3125                frame = Some(self.parse_frame_tail(kind)?);
3126            }
3127        }
3128        if !matches!(self.peek(), Token::RParen) {
3129            return Err(self.err(format!(
3130                "expected ')' to close OVER clause, got {:?}",
3131                self.peek()
3132            )));
3133        }
3134        self.advance();
3135        Ok((partition_by, order_by, frame))
3136    }
3137
3138    /// v4.20: parse the tail of an explicit frame, given the `ROWS`
3139    /// or `RANGE` keyword was just consumed. Accepts both
3140    /// `BETWEEN <bound> AND <bound>` and the single-bound shorthand
3141    /// (`ROWS UNBOUNDED PRECEDING`, `ROWS 5 PRECEDING`, etc.) which
3142    /// PG normalises to `BETWEEN <bound> AND CURRENT ROW`.
3143    fn parse_frame_tail(&mut self, kind: FrameKind) -> Result<WindowFrame, ParseError> {
3144        if matches!(self.peek(), Token::Between) {
3145            self.advance();
3146            let start = self.parse_frame_bound()?;
3147            if !matches!(self.peek(), Token::And) {
3148                return Err(self.err(format!("expected AND in frame spec, got {:?}", self.peek())));
3149            }
3150            self.advance();
3151            let end = self.parse_frame_bound()?;
3152            Ok(WindowFrame {
3153                kind,
3154                start,
3155                end: Some(end),
3156            })
3157        } else {
3158            let start = self.parse_frame_bound()?;
3159            Ok(WindowFrame {
3160                kind,
3161                start,
3162                end: None,
3163            })
3164        }
3165    }
3166
3167    /// Parse one frame bound: `UNBOUNDED PRECEDING`, `<n> PRECEDING`,
3168    /// `CURRENT ROW`, `<n> FOLLOWING`, `UNBOUNDED FOLLOWING`.
3169    fn parse_frame_bound(&mut self) -> Result<FrameBound, ParseError> {
3170        // Number-led: "<n> PRECEDING" / "<n> FOLLOWING".
3171        if let Token::Integer(n) = *self.peek() {
3172            self.advance();
3173            let n: u64 = u64::try_from(n).map_err(|_| {
3174                self.err(format!(
3175                    "invalid frame offset {n} — expected non-negative integer"
3176                ))
3177            })?;
3178            let dir = self.expect_ident_like()?;
3179            return if dir.eq_ignore_ascii_case("preceding") {
3180                Ok(FrameBound::OffsetPreceding(n))
3181            } else if dir.eq_ignore_ascii_case("following") {
3182                Ok(FrameBound::OffsetFollowing(n))
3183            } else {
3184                Err(self.err(format!(
3185                    "expected PRECEDING or FOLLOWING after offset, got {dir:?}"
3186                )))
3187            };
3188        }
3189        let first = self.expect_ident_like()?;
3190        if first.eq_ignore_ascii_case("unbounded") {
3191            let dir = self.expect_ident_like()?;
3192            return if dir.eq_ignore_ascii_case("preceding") {
3193                Ok(FrameBound::UnboundedPreceding)
3194            } else if dir.eq_ignore_ascii_case("following") {
3195                Ok(FrameBound::UnboundedFollowing)
3196            } else {
3197                Err(self.err(format!(
3198                    "expected PRECEDING or FOLLOWING after UNBOUNDED, got {dir:?}"
3199                )))
3200            };
3201        }
3202        if first.eq_ignore_ascii_case("current") {
3203            let row = self.expect_ident_like()?;
3204            if !row.eq_ignore_ascii_case("row") {
3205                return Err(self.err(format!("expected ROW after CURRENT, got {row:?}")));
3206            }
3207            return Ok(FrameBound::CurrentRow);
3208        }
3209        Err(self.err(format!(
3210            "expected frame bound (UNBOUNDED/CURRENT/<n>), got {first:?}"
3211        )))
3212    }
3213
3214    fn finish_ident_atom(&mut self, first: String) -> Result<Expr, ParseError> {
3215        if matches!(self.peek(), Token::Dot) {
3216            self.advance();
3217            let name = self.expect_ident_like()?;
3218            return Ok(Expr::Column(ColumnName {
3219                qualifier: Some(first),
3220                name,
3221            }));
3222        }
3223        if matches!(self.peek(), Token::LParen) {
3224            self.advance();
3225            // `COUNT(*)` — special-cased here because `*` isn't a normal
3226            // expression token. Lower-case match on `first` since the lexer
3227            // folds identifiers.
3228            if first.eq_ignore_ascii_case("count") && matches!(self.peek(), Token::Star) {
3229                self.advance();
3230                if !matches!(self.peek(), Token::RParen) {
3231                    return Err(self.err(format!(
3232                        "expected ')' after COUNT(*), got {:?}",
3233                        self.peek()
3234                    )));
3235                }
3236                self.advance();
3237                // v4.12: COUNT(*) OVER (...) — same window tail.
3238                let null_treatment = self.parse_null_treatment_modifier();
3239                if let Token::Ident(s) | Token::QuotedIdent(s) = self.peek()
3240                    && s.eq_ignore_ascii_case("over")
3241                {
3242                    self.advance();
3243                    let (partition_by, order_by, frame) = self.parse_over_clause()?;
3244                    return Ok(Expr::WindowFunction {
3245                        name: "count_star".into(),
3246                        args: Vec::new(),
3247                        partition_by,
3248                        order_by,
3249                        frame,
3250                        null_treatment,
3251                    });
3252                }
3253                return Ok(Expr::FunctionCall {
3254                    name: "count_star".into(),
3255                    args: Vec::new(),
3256                });
3257            }
3258            // Function call. PG-style: zero-or-more comma-separated args.
3259            let mut args = Vec::new();
3260            if !matches!(self.peek(), Token::RParen) {
3261                loop {
3262                    args.push(self.parse_expr(0)?);
3263                    match self.peek() {
3264                        Token::Comma => {
3265                            self.advance();
3266                        }
3267                        Token::RParen => break,
3268                        other => {
3269                            return Err(self.err(format!(
3270                                "expected ',' or ')' in function args, got {other:?}"
3271                            )));
3272                        }
3273                    }
3274                }
3275            }
3276            self.advance(); // consume ')'
3277            // v4.12: window-function tail — `name(args) OVER (...)`.
3278            // Promotes the just-parsed FunctionCall into a
3279            // WindowFunction node carrying partition + order.
3280            // v6.4.2: also accepts `name(args) IGNORE NULLS OVER (...)`
3281            // / `RESPECT NULLS OVER (...)` between the closing paren
3282            // and `OVER`.
3283            let null_treatment = self.parse_null_treatment_modifier();
3284            if let Token::Ident(s) | Token::QuotedIdent(s) = self.peek()
3285                && s.eq_ignore_ascii_case("over")
3286            {
3287                self.advance();
3288                let (partition_by, order_by, frame) = self.parse_over_clause()?;
3289                return Ok(Expr::WindowFunction {
3290                    name: first,
3291                    args,
3292                    partition_by,
3293                    order_by,
3294                    frame,
3295                    null_treatment,
3296                });
3297            }
3298            return Ok(Expr::FunctionCall { name: first, args });
3299        }
3300        // v7.9.20 — SQL-standard parenless keyword expressions
3301        // (PG treats these as functions called without parens).
3302        // Resolve to a synthetic FunctionCall so the engine's
3303        // eval path reuses the existing function-call routing.
3304        // mailrs G3.
3305        let lc = first.to_ascii_lowercase();
3306        if matches!(
3307            lc.as_str(),
3308            "current_date"
3309                | "current_time"
3310                | "current_timestamp"
3311                | "localtimestamp"
3312                | "localtime"
3313        ) {
3314            return Ok(Expr::FunctionCall {
3315                name: lc,
3316                args: Vec::new(),
3317            });
3318        }
3319        Ok(Expr::Column(ColumnName {
3320            qualifier: None,
3321            name: first,
3322        }))
3323    }
3324}
3325
3326/// v6.8.2 — walk an expression tree and return the first column
3327/// reference's bare name. Used by `parse_create_index_stmt_after_create`
3328/// to derive `CreateIndexStatement.column` from an expression
3329/// key (so downstream planner code resolving a primary column
3330/// position keeps working with expression indexes). Returns
3331/// `None` when the expression has no column ref at all — caller
3332/// surfaces that as a parse error.
3333fn extract_first_column(expr: &Expr) -> Option<String> {
3334    match expr {
3335        Expr::Column(cn) => Some(cn.name.clone()),
3336        Expr::FunctionCall { args, .. } => args.iter().find_map(extract_first_column),
3337        Expr::Binary { lhs, rhs, .. } => {
3338            extract_first_column(lhs).or_else(|| extract_first_column(rhs))
3339        }
3340        Expr::Unary { expr: e, .. } => extract_first_column(e),
3341        _ => None,
3342    }
3343}
3344
3345fn maybe_not(expr: Expr, negated: bool) -> Expr {
3346    if negated {
3347        Expr::Unary {
3348            op: UnOp::Not,
3349            expr: Box::new(expr),
3350        }
3351    } else {
3352        expr
3353    }
3354}
3355
3356fn binop_from(tok: &Token) -> Option<(BinOp, u8)> {
3357    let pair = match tok {
3358        Token::Or => (BinOp::Or, 1),
3359        Token::And => (BinOp::And, 2),
3360        Token::Eq => (BinOp::Eq, 4),
3361        Token::NotEq => (BinOp::NotEq, 4),
3362        Token::Lt => (BinOp::Lt, 4),
3363        Token::LtEq => (BinOp::LtEq, 4),
3364        Token::Gt => (BinOp::Gt, 4),
3365        Token::GtEq => (BinOp::GtEq, 4),
3366        // pgvector distance ops all sit on the same rung — tighter than
3367        // comparisons (4) so `col <-> v < threshold` parses correctly.
3368        Token::L2Distance => (BinOp::L2Distance, 5),
3369        Token::InnerProduct => (BinOp::InnerProduct, 5),
3370        Token::CosineDistance => (BinOp::CosineDistance, 5),
3371        Token::Plus => (BinOp::Add, 6),
3372        Token::Minus => (BinOp::Sub, 6),
3373        // `||` sits beside `+`/`-` (matches PG conceptually — concat groups
3374        // by the same level as binary additive arithmetic).
3375        Token::Concat => (BinOp::Concat, 6),
3376        Token::Star => (BinOp::Mul, 7),
3377        Token::Slash => (BinOp::Div, 7),
3378        // v4.14: JSON path ops bind tighter than comparisons (4)
3379        // and additive (6) so `doc->'k' = 'v'` parses correctly.
3380        // Same rung as the multiplicative ops.
3381        Token::JsonGet => (BinOp::JsonGet, 7),
3382        Token::JsonGetText => (BinOp::JsonGetText, 7),
3383        Token::JsonGetPath => (BinOp::JsonGetPath, 7),
3384        Token::JsonGetPathText => (BinOp::JsonGetPathText, 7),
3385        Token::JsonContains => (BinOp::JsonContains, 7),
3386        _ => return None,
3387    };
3388    Some(pair)
3389}
3390
3391#[allow(clippy::cast_possible_truncation, clippy::cast_precision_loss)]
3392// `as f32` here is intentional: vector elements widen / narrow into f32 on
3393// purpose. i64 → f32 loses precision past 2^24, f64 → f32 loses precision
3394// past ~15 decimal digits — both are acceptable for a fixed-precision
3395// pgvector column.
3396fn extract_numeric_literal(e: &Expr) -> Option<f32> {
3397    match e {
3398        Expr::Literal(Literal::Integer(n)) => Some(*n as f32),
3399        Expr::Literal(Literal::Float(x)) => Some(*x as f32),
3400        Expr::Unary {
3401            op: UnOp::Neg,
3402            expr,
3403        } => extract_numeric_literal(expr).map(|x| -x),
3404        _ => None,
3405    }
3406}
3407
3408/// Parse the text inside `INTERVAL '...'` into `(months, micros)`. Accepts
3409/// one or more `<n> <unit>` pairs separated by whitespace. `<n>` may be
3410/// negative. Returns `None` if any pair fails to parse or no pair is found.
3411///
3412/// Recognised units (case-insensitive, optional trailing `s`):
3413/// `microsecond`, `millisecond`, `second`, `minute`, `hour`, `day`, `week`,
3414/// `month`, `year`. `week` widens to 7 days; `year` widens to 12 months.
3415pub fn parse_interval_text(s: &str) -> Option<(i32, i64)> {
3416    let parts: Vec<&str> = s.split_whitespace().collect();
3417    if parts.is_empty() || !parts.len().is_multiple_of(2) {
3418        return None;
3419    }
3420    let mut months: i32 = 0;
3421    let mut micros: i64 = 0;
3422    let mut i = 0;
3423    while i < parts.len() {
3424        let n: i64 = parts[i].parse().ok()?;
3425        let unit = parts[i + 1].to_ascii_lowercase();
3426        let unit_stripped = unit.strip_suffix('s').unwrap_or(&unit);
3427        match unit_stripped {
3428            "microsecond" => micros = micros.checked_add(n)?,
3429            "millisecond" => micros = micros.checked_add(n.checked_mul(1_000)?)?,
3430            "second" => micros = micros.checked_add(n.checked_mul(1_000_000)?)?,
3431            "minute" => micros = micros.checked_add(n.checked_mul(60_000_000)?)?,
3432            "hour" => micros = micros.checked_add(n.checked_mul(3_600_000_000)?)?,
3433            "day" => micros = micros.checked_add(n.checked_mul(86_400_000_000)?)?,
3434            "week" => micros = micros.checked_add(n.checked_mul(604_800_000_000)?)?,
3435            "month" => {
3436                let n32 = i32::try_from(n).ok()?;
3437                months = months.checked_add(n32)?;
3438            }
3439            "year" => {
3440                let n32 = i32::try_from(n).ok()?;
3441                months = months.checked_add(n32.checked_mul(12)?)?;
3442            }
3443            _ => return None,
3444        }
3445        i += 2;
3446    }
3447    Some((months, micros))
3448}
3449
3450#[cfg(test)]
3451mod tests {
3452    use super::*;
3453    use alloc::string::ToString;
3454
3455    fn parse(s: &str) -> Statement {
3456        parse_statement(s).expect("parse ok")
3457    }
3458
3459    fn lit_int(n: i64) -> Expr {
3460        Expr::Literal(Literal::Integer(n))
3461    }
3462
3463    fn col(name: &str) -> Expr {
3464        Expr::Column(ColumnName {
3465            qualifier: None,
3466            name: name.into(),
3467        })
3468    }
3469
3470    #[test]
3471    fn select_single_integer() {
3472        let s = parse("SELECT 1");
3473        let Statement::Select(s) = s else {
3474            panic!("expected SELECT")
3475        };
3476        assert_eq!(s.items.len(), 1);
3477        assert!(s.from.is_none());
3478        assert!(s.where_.is_none());
3479    }
3480
3481    #[test]
3482    fn select_multiple_literal_kinds() {
3483        let s = parse("SELECT 1, 'hi', NULL, TRUE, 1.5");
3484        let Statement::Select(s) = s else {
3485            panic!("expected SELECT")
3486        };
3487        assert_eq!(s.items.len(), 5);
3488    }
3489
3490    #[test]
3491    fn select_wildcard_from_table() {
3492        let s = parse("SELECT * FROM users");
3493        let Statement::Select(s) = s else {
3494            panic!("expected SELECT")
3495        };
3496        assert!(matches!(s.items[..], [SelectItem::Wildcard]));
3497        assert_eq!(s.from.as_ref().unwrap().primary.name, "users");
3498    }
3499
3500    #[test]
3501    fn select_with_table_alias() {
3502        let s = parse("SELECT * FROM users AS u");
3503        let Statement::Select(s) = s else {
3504            panic!("expected SELECT")
3505        };
3506        let t = &s.from.as_ref().unwrap().primary;
3507        assert_eq!(t.name, "users");
3508        assert_eq!(t.alias.as_deref(), Some("u"));
3509    }
3510
3511    #[test]
3512    fn select_with_where_eq() {
3513        let s = parse("SELECT a FROM t WHERE a = 1");
3514        let Statement::Select(s) = s else {
3515            panic!("expected SELECT")
3516        };
3517        let w = s.where_.unwrap();
3518        assert_eq!(
3519            w,
3520            Expr::Binary {
3521                lhs: Box::new(col("a")),
3522                op: BinOp::Eq,
3523                rhs: Box::new(lit_int(1)),
3524            }
3525        );
3526    }
3527
3528    #[test]
3529    fn arithmetic_precedence() {
3530        let s = parse("SELECT 1 + 2 * 3");
3531        let Statement::Select(s) = s else {
3532            panic!("expected SELECT")
3533        };
3534        let SelectItem::Expr { expr, .. } = &s.items[0] else {
3535            panic!("wildcard?")
3536        };
3537        assert_eq!(
3538            expr,
3539            &Expr::Binary {
3540                lhs: Box::new(lit_int(1)),
3541                op: BinOp::Add,
3542                rhs: Box::new(Expr::Binary {
3543                    lhs: Box::new(lit_int(2)),
3544                    op: BinOp::Mul,
3545                    rhs: Box::new(lit_int(3)),
3546                }),
3547            }
3548        );
3549    }
3550
3551    #[test]
3552    fn parentheses_override_precedence() {
3553        let s = parse("SELECT (1 + 2) * 3");
3554        let Statement::Select(s) = s else {
3555            panic!("expected SELECT")
3556        };
3557        let SelectItem::Expr { expr, .. } = &s.items[0] else {
3558            panic!()
3559        };
3560        assert_eq!(
3561            expr,
3562            &Expr::Binary {
3563                lhs: Box::new(Expr::Binary {
3564                    lhs: Box::new(lit_int(1)),
3565                    op: BinOp::Add,
3566                    rhs: Box::new(lit_int(2)),
3567                }),
3568                op: BinOp::Mul,
3569                rhs: Box::new(lit_int(3)),
3570            }
3571        );
3572    }
3573
3574    #[test]
3575    fn not_binds_below_comparison() {
3576        // `NOT a = 1` should parse as `NOT (a = 1)`.
3577        let s = parse("SELECT NOT a = 1 FROM t");
3578        let Statement::Select(s) = s else {
3579            panic!("expected SELECT")
3580        };
3581        let SelectItem::Expr { expr, .. } = &s.items[0] else {
3582            panic!()
3583        };
3584        assert_eq!(
3585            expr,
3586            &Expr::Unary {
3587                op: UnOp::Not,
3588                expr: Box::new(Expr::Binary {
3589                    lhs: Box::new(col("a")),
3590                    op: BinOp::Eq,
3591                    rhs: Box::new(lit_int(1)),
3592                }),
3593            }
3594        );
3595    }
3596
3597    #[test]
3598    fn unary_minus_binds_above_multiplication() {
3599        // `-a * 2` should be `(-a) * 2`.
3600        let s = parse("SELECT -a * 2 FROM t");
3601        let Statement::Select(s) = s else {
3602            panic!("expected SELECT")
3603        };
3604        let SelectItem::Expr { expr, .. } = &s.items[0] else {
3605            panic!()
3606        };
3607        assert_eq!(
3608            expr,
3609            &Expr::Binary {
3610                lhs: Box::new(Expr::Unary {
3611                    op: UnOp::Neg,
3612                    expr: Box::new(col("a")),
3613                }),
3614                op: BinOp::Mul,
3615                rhs: Box::new(lit_int(2)),
3616            }
3617        );
3618    }
3619
3620    #[test]
3621    fn qualified_column() {
3622        let s = parse("SELECT t.col FROM t");
3623        let Statement::Select(s) = s else {
3624            panic!("expected SELECT")
3625        };
3626        let SelectItem::Expr { expr, .. } = &s.items[0] else {
3627            panic!()
3628        };
3629        assert_eq!(
3630            expr,
3631            &Expr::Column(ColumnName {
3632                qualifier: Some("t".into()),
3633                name: "col".into()
3634            })
3635        );
3636    }
3637
3638    #[test]
3639    fn select_item_alias_with_as() {
3640        let s = parse("SELECT a AS y FROM t");
3641        let Statement::Select(s) = s else {
3642            panic!("expected SELECT")
3643        };
3644        let SelectItem::Expr { alias, .. } = &s.items[0] else {
3645            panic!()
3646        };
3647        assert_eq!(alias.as_deref(), Some("y"));
3648    }
3649
3650    #[test]
3651    fn trailing_semicolon_accepted() {
3652        let s = parse("SELECT 1;");
3653        let Statement::Select(s) = s else {
3654            panic!("expected SELECT")
3655        };
3656        assert_eq!(s.items.len(), 1);
3657    }
3658
3659    #[test]
3660    fn boolean_chain_with_and_or_not() {
3661        // (NOT a) OR (b AND (NOT c))
3662        let s = parse("SELECT NOT a OR b AND NOT c FROM t");
3663        let Statement::Select(s) = s else {
3664            panic!("expected SELECT")
3665        };
3666        let SelectItem::Expr { expr, .. } = &s.items[0] else {
3667            panic!()
3668        };
3669        let expected = Expr::Binary {
3670            lhs: Box::new(Expr::Unary {
3671                op: UnOp::Not,
3672                expr: Box::new(col("a")),
3673            }),
3674            op: BinOp::Or,
3675            rhs: Box::new(Expr::Binary {
3676                lhs: Box::new(col("b")),
3677                op: BinOp::And,
3678                rhs: Box::new(Expr::Unary {
3679                    op: UnOp::Not,
3680                    expr: Box::new(col("c")),
3681                }),
3682            }),
3683        };
3684        assert_eq!(expr, &expected);
3685    }
3686
3687    #[test]
3688    fn empty_input_errors() {
3689        let err = parse_statement("").unwrap_err();
3690        assert!(err.message.contains("SELECT"));
3691    }
3692
3693    #[test]
3694    fn unmatched_paren_errors() {
3695        assert!(parse_statement("SELECT (1 + 2").is_err());
3696    }
3697
3698    #[test]
3699    fn display_round_trip_simple_select() {
3700        let original = parse("SELECT a + 1 FROM t WHERE a > 0");
3701        let text = original.to_string();
3702        let again = parse_statement(&text).expect("re-parse");
3703        assert_eq!(original, again);
3704    }
3705
3706    // --- CREATE TABLE & INSERT (v0.3) ---------------------------------------
3707
3708    #[test]
3709    fn create_table_single_column() {
3710        let s = parse("CREATE TABLE foo (a INT)");
3711        let Statement::CreateTable(c) = s else {
3712            panic!("expected CreateTable")
3713        };
3714        assert_eq!(c.name, "foo");
3715        assert_eq!(c.columns.len(), 1);
3716        assert_eq!(c.columns[0].name, "a");
3717        assert_eq!(c.columns[0].ty, ColumnTypeName::Int);
3718        assert!(c.columns[0].nullable);
3719    }
3720
3721    #[test]
3722    fn create_table_multi_column_with_not_null_mix() {
3723        let s = parse("CREATE TABLE u (id INT NOT NULL, name TEXT, score FLOAT NOT NULL, ok BOOL)");
3724        let Statement::CreateTable(c) = s else {
3725            panic!()
3726        };
3727        assert_eq!(c.columns.len(), 4);
3728        assert_eq!(c.columns[0].ty, ColumnTypeName::Int);
3729        assert!(!c.columns[0].nullable);
3730        assert_eq!(c.columns[1].ty, ColumnTypeName::Text);
3731        assert!(c.columns[1].nullable);
3732        assert_eq!(c.columns[2].ty, ColumnTypeName::Float);
3733        assert!(!c.columns[2].nullable);
3734        assert_eq!(c.columns[3].ty, ColumnTypeName::Bool);
3735    }
3736
3737    #[test]
3738    fn create_table_bigint_supported() {
3739        let s = parse("CREATE TABLE accounts (id BIGINT NOT NULL)");
3740        let Statement::CreateTable(c) = s else {
3741            panic!()
3742        };
3743        assert_eq!(c.columns[0].ty, ColumnTypeName::BigInt);
3744    }
3745
3746    #[test]
3747    fn create_table_vector_default_is_f32() {
3748        let s = parse("CREATE TABLE t (v VECTOR(128))");
3749        let Statement::CreateTable(c) = s else {
3750            panic!()
3751        };
3752        assert_eq!(
3753            c.columns[0].ty,
3754            ColumnTypeName::Vector {
3755                dim: 128,
3756                encoding: VecEncoding::F32,
3757            },
3758        );
3759    }
3760
3761    #[test]
3762    fn create_table_vector_using_sq8() {
3763        // v6.0.1: `USING SQ8` selects scalar-quantised encoding.
3764        // Case-insensitive on both `USING` and the encoding name.
3765        for sql in [
3766            "CREATE TABLE t (v VECTOR(128) USING SQ8)",
3767            "CREATE TABLE t (v VECTOR(128) using sq8)",
3768        ] {
3769            let s = parse(sql);
3770            let Statement::CreateTable(c) = s else {
3771                panic!()
3772            };
3773            assert_eq!(
3774                c.columns[0].ty,
3775                ColumnTypeName::Vector {
3776                    dim: 128,
3777                    encoding: VecEncoding::Sq8,
3778                },
3779                "{sql}",
3780            );
3781        }
3782    }
3783
3784    #[test]
3785    fn create_table_vector_using_unknown_errors() {
3786        let err = parse_statement("CREATE TABLE t (v VECTOR(8) USING PQ8)").unwrap_err();
3787        assert!(
3788            err.message.contains("unknown vector encoding"),
3789            "got: {}",
3790            err.message
3791        );
3792    }
3793
3794    #[test]
3795    fn vector_using_sq8_display_roundtrips() {
3796        // The Display impl must produce text that re-parses to the
3797        // same AST. Guard for the v6.0.1 `USING SQ8` suffix.
3798        let s = parse("CREATE TABLE t (v VECTOR(64) USING SQ8)");
3799        let Statement::CreateTable(c) = s else {
3800            panic!()
3801        };
3802        assert_eq!(c.columns[0].ty.to_string(), "VECTOR(64) USING SQ8");
3803    }
3804
3805    #[test]
3806    fn parser_recognises_placeholders() {
3807        use crate::ast::{Expr, SelectItem, Statement};
3808        // $N in expression position parses as Expr::Placeholder(N).
3809        let s = parse("SELECT $1, $2 + 1 FROM t WHERE x = $3");
3810        let Statement::Select(sel) = s else { panic!() };
3811        assert!(matches!(
3812            sel.items[0],
3813            SelectItem::Expr {
3814                expr: Expr::Placeholder(1),
3815                alias: None
3816            }
3817        ));
3818        // $2 + 1
3819        let SelectItem::Expr {
3820            expr: Expr::Binary { lhs, rhs, .. },
3821            ..
3822        } = &sel.items[1]
3823        else {
3824            panic!()
3825        };
3826        assert!(matches!(**lhs, Expr::Placeholder(2)));
3827        assert!(matches!(**rhs, Expr::Literal(Literal::Integer(1))));
3828        // WHERE x = $3
3829        let Some(Expr::Binary { rhs, .. }) = sel.where_.as_ref() else {
3830            panic!()
3831        };
3832        assert!(matches!(**rhs, Expr::Placeholder(3)));
3833    }
3834
3835    #[test]
3836    fn parser_rejects_dollar_zero() {
3837        // $0 is not valid in PG; the lexer rejects it.
3838        assert!(parse_statement("SELECT $0").is_err());
3839    }
3840
3841    #[test]
3842    fn placeholder_display_roundtrips() {
3843        // The Display impl must produce text that re-lexes to the
3844        // same Placeholder token.
3845        let s = parse("SELECT $42 FROM t");
3846        let printed = s.to_string();
3847        assert!(printed.contains("$42"));
3848        let again = parse(&printed);
3849        assert_eq!(s, again);
3850    }
3851
3852    #[test]
3853    fn alter_index_rebuild_bare() {
3854        use crate::ast::{AlterIndexTarget, Statement};
3855        let s = parse("ALTER INDEX my_idx REBUILD");
3856        let Statement::AlterIndex(a) = s else {
3857            panic!("expected AlterIndex, got {s:?}")
3858        };
3859        assert_eq!(a.name, "my_idx");
3860        assert_eq!(a.target, AlterIndexTarget::Rebuild { encoding: None });
3861    }
3862
3863    #[test]
3864    fn alter_index_rebuild_with_encoding() {
3865        use crate::ast::{AlterIndexTarget, Statement};
3866        for (sql, want) in [
3867            (
3868                "ALTER INDEX my_idx REBUILD WITH (encoding = F32)",
3869                VecEncoding::F32,
3870            ),
3871            (
3872                "ALTER INDEX my_idx REBUILD WITH (encoding = sq8)",
3873                VecEncoding::Sq8,
3874            ),
3875            (
3876                "ALTER INDEX my_idx REBUILD WITH (encoding = HALF)",
3877                VecEncoding::F16,
3878            ),
3879        ] {
3880            let s = parse(sql);
3881            let Statement::AlterIndex(a) = s else {
3882                panic!("{sql}: expected AlterIndex")
3883            };
3884            assert_eq!(a.name, "my_idx");
3885            assert_eq!(
3886                a.target,
3887                AlterIndexTarget::Rebuild {
3888                    encoding: Some(want)
3889                },
3890                "{sql}"
3891            );
3892        }
3893    }
3894
3895    #[test]
3896    fn alter_index_rebuild_unknown_encoding_errors() {
3897        let err = parse_statement("ALTER INDEX my_idx REBUILD WITH (encoding = PQ8)").unwrap_err();
3898        assert!(
3899            err.message.contains("unknown vector encoding"),
3900            "got: {}",
3901            err.message
3902        );
3903    }
3904
3905    #[test]
3906    fn alter_index_rebuild_display_roundtrips() {
3907        for (input, want) in [
3908            ("ALTER INDEX my_idx REBUILD", "ALTER INDEX my_idx REBUILD"),
3909            (
3910                "ALTER INDEX my_idx REBUILD WITH (encoding = SQ8)",
3911                "ALTER INDEX my_idx REBUILD WITH (encoding = SQ8)",
3912            ),
3913            (
3914                "ALTER INDEX my_idx REBUILD WITH (encoding = HALF)",
3915                "ALTER INDEX my_idx REBUILD WITH (encoding = HALF)",
3916            ),
3917        ] {
3918            let s = parse(input);
3919            assert_eq!(s.to_string(), want);
3920        }
3921    }
3922
3923    #[test]
3924    fn create_table_unknown_type_errors() {
3925        // v4.9: JSON is now real; pick an actually unsupported keyword
3926        // (XML never landed and isn't planned).
3927        let err = parse_statement("CREATE TABLE x (a xml)").unwrap_err();
3928        assert!(err.message.contains("unsupported column type"));
3929    }
3930
3931    #[test]
3932    fn create_table_missing_table_keyword_errors() {
3933        assert!(parse_statement("CREATE x (a INT)").is_err());
3934    }
3935
3936    #[test]
3937    fn insert_single_value() {
3938        let s = parse("INSERT INTO foo VALUES (42)");
3939        let Statement::Insert(i) = s else {
3940            panic!("expected Insert")
3941        };
3942        assert_eq!(i.table, "foo");
3943        assert_eq!(i.rows.len(), 1);
3944        assert_eq!(i.rows[0].len(), 1);
3945        assert!(matches!(i.rows[0][0], Expr::Literal(Literal::Integer(42))));
3946    }
3947
3948    #[test]
3949    fn insert_multi_value_with_mixed_literals() {
3950        let s = parse("INSERT INTO foo VALUES (1, 'hi', 3.14, TRUE, NULL)");
3951        let Statement::Insert(i) = s else { panic!() };
3952        assert_eq!(i.rows.len(), 1);
3953        assert_eq!(i.rows[0].len(), 5);
3954    }
3955
3956    #[test]
3957    fn insert_missing_into_errors() {
3958        assert!(parse_statement("INSERT foo VALUES (1)").is_err());
3959    }
3960
3961    #[test]
3962    fn create_table_round_trip() {
3963        let original =
3964            parse("CREATE TABLE foo (id BIGINT NOT NULL, label TEXT, score FLOAT NOT NULL)");
3965        let text = original.to_string();
3966        let again = parse_statement(&text).expect("re-parse");
3967        assert_eq!(original, again);
3968    }
3969
3970    #[test]
3971    fn insert_round_trip_with_negation_and_string() {
3972        let original = parse("INSERT INTO t VALUES (-1, 'it''s', NULL)");
3973        let text = original.to_string();
3974        let again = parse_statement(&text).expect("re-parse");
3975        assert_eq!(original, again);
3976    }
3977
3978    #[test]
3979    fn unknown_keyword_at_statement_start_errors() {
3980        // v4.4: UPDATE is real SQL now. Use a fabricated keyword so
3981        // the top-level dispatch still has no branch to take.
3982        let err = parse_statement("FROBNICATE foo SET x = 1").unwrap_err();
3983        assert!(err.message.contains("expected SELECT"));
3984    }
3985
3986    // --- v0.8 CREATE INDEX --------------------------------------------------
3987
3988    #[test]
3989    fn create_index_basic() {
3990        let s = parse("CREATE INDEX idx_id ON users (id)");
3991        let Statement::CreateIndex(c) = s else {
3992            panic!("expected CreateIndex")
3993        };
3994        assert_eq!(c.name, "idx_id");
3995        assert_eq!(c.table, "users");
3996        assert_eq!(c.column, "id");
3997    }
3998
3999    #[test]
4000    fn create_index_missing_on_errors() {
4001        assert!(parse_statement("CREATE INDEX foo users (id)").is_err());
4002    }
4003
4004    #[test]
4005    fn create_index_missing_paren_errors() {
4006        assert!(parse_statement("CREATE INDEX foo ON users id").is_err());
4007    }
4008
4009    #[test]
4010    fn create_index_round_trip() {
4011        let original = parse("CREATE INDEX by_name ON users (name)");
4012        let again = parse_statement(&original.to_string()).unwrap();
4013        assert_eq!(original, again);
4014    }
4015
4016    // --- v7.9.29 CREATE UNIQUE INDEX [WHERE pred] (mailrs K1) -------------
4017
4018    #[test]
4019    fn create_unique_index_basic() {
4020        let s = parse("CREATE UNIQUE INDEX uq_x ON t (a)");
4021        let Statement::CreateIndex(c) = s else {
4022            panic!("expected CreateIndex");
4023        };
4024        assert!(c.is_unique);
4025        assert_eq!(c.column, "a");
4026        assert!(c.partial_predicate.is_none());
4027    }
4028
4029    #[test]
4030    fn create_unique_index_partial() {
4031        // mailrs's email_templates "one default per user" shape.
4032        let s = parse(
4033            "CREATE UNIQUE INDEX idx_email_templates_user_default \
4034             ON email_templates (user_address) WHERE is_default = true",
4035        );
4036        let Statement::CreateIndex(c) = s else {
4037            panic!("expected CreateIndex");
4038        };
4039        assert!(c.is_unique);
4040        assert_eq!(c.table, "email_templates");
4041        assert_eq!(c.column, "user_address");
4042        assert!(c.partial_predicate.is_some());
4043    }
4044
4045    #[test]
4046    fn create_unique_index_composite_with_predicate() {
4047        // mailrs's calendar_events instance: composite columns.
4048        let s = parse(
4049            "CREATE UNIQUE INDEX uq_calendar_events_instance \
4050             ON calendar_events (calendar_id, uid, recurrence_id) \
4051             WHERE recurrence_id IS NOT NULL",
4052        );
4053        let Statement::CreateIndex(c) = s else {
4054            panic!("expected CreateIndex");
4055        };
4056        assert!(c.is_unique);
4057        assert_eq!(c.column, "calendar_id");
4058        assert_eq!(c.extra_columns, vec!["uid".to_string(), "recurrence_id".to_string()]);
4059        assert!(c.partial_predicate.is_some());
4060    }
4061
4062    #[test]
4063    fn create_unique_index_using_btree_ok() {
4064        let s = parse("CREATE UNIQUE INDEX uq_x ON t USING btree (a)");
4065        assert!(matches!(s, Statement::CreateIndex(ref c) if c.is_unique));
4066    }
4067
4068    #[test]
4069    fn create_unique_index_using_hnsw_rejected() {
4070        let err = parse_statement(
4071            "CREATE UNIQUE INDEX uq_v ON t USING hnsw (embedding)",
4072        )
4073        .unwrap_err();
4074        assert!(err.message.contains("UNIQUE"), "{}", err.message);
4075    }
4076
4077    #[test]
4078    fn create_unique_index_round_trip() {
4079        let original = parse(
4080            "CREATE UNIQUE INDEX uq_calendar_events_master \
4081             ON calendar_events (calendar_id, uid) WHERE recurrence_id IS NULL",
4082        );
4083        let again = parse_statement(&original.to_string()).unwrap();
4084        assert_eq!(original, again);
4085    }
4086
4087    #[test]
4088    fn create_unique_without_index_errors() {
4089        let err = parse_statement("CREATE UNIQUE TABLE t (a INT)").unwrap_err();
4090        assert!(err.message.contains("INDEX"), "{}", err.message);
4091    }
4092
4093    // --- v0.9 transactions -------------------------------------------------
4094
4095    #[test]
4096    fn begin_commit_rollback_parse_as_unit_variants() {
4097        assert_eq!(parse("BEGIN"), Statement::Begin);
4098        assert_eq!(parse("COMMIT"), Statement::Commit);
4099        assert_eq!(parse("ROLLBACK"), Statement::Rollback);
4100        // Trailing semicolons accepted too.
4101        assert_eq!(parse("BEGIN;"), Statement::Begin);
4102    }
4103
4104    // --- v1.2: pgvector distance ops + ::vector cast --------------------
4105
4106    #[test]
4107    fn inner_product_binop_parses() {
4108        let s = parse("SELECT v <#> [1.0, 2.0] FROM t");
4109        let Statement::Select(s) = s else { panic!() };
4110        let SelectItem::Expr { expr, .. } = &s.items[0] else {
4111            panic!()
4112        };
4113        assert!(matches!(
4114            expr,
4115            Expr::Binary {
4116                op: BinOp::InnerProduct,
4117                ..
4118            }
4119        ));
4120    }
4121
4122    #[test]
4123    fn cosine_distance_binop_parses() {
4124        let s = parse("SELECT v <=> [1.0, 2.0] FROM t");
4125        let Statement::Select(s) = s else { panic!() };
4126        let SelectItem::Expr { expr, .. } = &s.items[0] else {
4127            panic!()
4128        };
4129        assert!(matches!(
4130            expr,
4131            Expr::Binary {
4132                op: BinOp::CosineDistance,
4133                ..
4134            }
4135        ));
4136    }
4137
4138    #[test]
4139    fn vector_cast_postfix_wraps_string_literal() {
4140        let s = parse("SELECT '[1,2,3]'::vector FROM t");
4141        let Statement::Select(s) = s else { panic!() };
4142        let SelectItem::Expr { expr, .. } = &s.items[0] else {
4143            panic!()
4144        };
4145        assert!(matches!(
4146            expr,
4147            Expr::Cast {
4148                target: CastTarget::Vector,
4149                ..
4150            }
4151        ));
4152    }
4153
4154    #[test]
4155    fn unsupported_cast_target_errors() {
4156        // `::numeric` isn't in the v1.3 cast target set.
4157        let err = parse_statement("SELECT 1::numeric FROM t").unwrap_err();
4158        assert!(err.message.contains("unsupported cast target"));
4159    }
4160
4161    #[test]
4162    fn tx_statements_round_trip() {
4163        for q in ["BEGIN", "COMMIT", "ROLLBACK"] {
4164            let original = parse(q);
4165            let again = parse_statement(&original.to_string()).unwrap();
4166            assert_eq!(original, again);
4167        }
4168    }
4169
4170    #[test]
4171    fn interval_text_parsing_units() {
4172        // Single unit.
4173        assert_eq!(parse_interval_text("1 day"), Some((0, 86_400_000_000)));
4174        assert_eq!(parse_interval_text("1 second"), Some((0, 1_000_000)));
4175        assert_eq!(parse_interval_text("1 month"), Some((1, 0)));
4176        assert_eq!(parse_interval_text("2 years"), Some((24, 0)));
4177        // Compound spans accumulate.
4178        assert_eq!(parse_interval_text("1 year 6 months"), Some((18, 0)));
4179        assert_eq!(
4180            parse_interval_text("1 day 2 hours"),
4181            Some((0, 86_400_000_000 + 7_200_000_000))
4182        );
4183        // Negative numbers carry through.
4184        assert_eq!(parse_interval_text("-1 day"), Some((0, -86_400_000_000)));
4185        // Bad shapes return None.
4186        assert_eq!(parse_interval_text(""), None);
4187        assert_eq!(parse_interval_text("garbage"), None);
4188        assert_eq!(parse_interval_text("1 fortnight"), None);
4189        assert_eq!(parse_interval_text("1"), None);
4190    }
4191
4192    #[test]
4193    fn interval_literal_roundtrips_via_display() {
4194        let parsed = parse("SELECT INTERVAL '1 day 2 hours'");
4195        let s = parsed.to_string();
4196        // Display preserves the original text verbatim.
4197        assert!(s.contains("INTERVAL '1 day 2 hours'"), "got: {s}");
4198        // And re-parsing yields a structurally equal statement.
4199        let again = parse_statement(&s).unwrap();
4200        assert_eq!(parsed, again);
4201    }
4202
4203    // ── v6.1.2: CREATE / DROP PUBLICATION ────────────────────
4204
4205    #[test]
4206    fn parser_recognises_create_publication_bare() {
4207        let s = parse("CREATE PUBLICATION pub_a");
4208        let Statement::CreatePublication(p) = s else {
4209            panic!("expected CreatePublication, got {s:?}")
4210        };
4211        assert_eq!(p.name, "pub_a");
4212        assert_eq!(p.scope, PublicationScope::AllTables);
4213    }
4214
4215    #[test]
4216    fn parser_recognises_create_publication_for_all_tables() {
4217        let s = parse("CREATE PUBLICATION pub_a FOR ALL TABLES");
4218        let Statement::CreatePublication(p) = s else {
4219            panic!("expected CreatePublication, got {s:?}")
4220        };
4221        assert_eq!(p.name, "pub_a");
4222        assert_eq!(p.scope, PublicationScope::AllTables);
4223    }
4224
4225    #[test]
4226    fn parser_recognises_drop_publication() {
4227        let s = parse("DROP PUBLICATION pub_a");
4228        let Statement::DropPublication(name) = s else {
4229            panic!("expected DropPublication, got {s:?}")
4230        };
4231        assert_eq!(name, "pub_a");
4232    }
4233
4234    #[test]
4235    fn parser_recognises_for_table_list() {
4236        let s = parse("CREATE PUBLICATION pub_a FOR TABLE t1, t2, t3");
4237        let Statement::CreatePublication(p) = s else {
4238            panic!("expected CreatePublication, got {s:?}")
4239        };
4240        assert_eq!(p.name, "pub_a");
4241        let PublicationScope::ForTables(ts) = p.scope else {
4242            panic!("expected ForTables scope")
4243        };
4244        assert_eq!(ts, alloc::vec!["t1", "t2", "t3"]);
4245    }
4246
4247    #[test]
4248    fn parser_recognises_for_tables_plural() {
4249        // PG 19 accepts both `FOR TABLE` and `FOR TABLES` — match.
4250        let s = parse("CREATE PUBLICATION pub_a FOR TABLES t1, t2");
4251        let Statement::CreatePublication(p) = s else {
4252            panic!("expected CreatePublication, got {s:?}")
4253        };
4254        let PublicationScope::ForTables(ts) = p.scope else {
4255            panic!("expected ForTables")
4256        };
4257        assert_eq!(ts, alloc::vec!["t1", "t2"]);
4258    }
4259
4260    #[test]
4261    fn parser_recognises_for_all_tables_except_list() {
4262        let s = parse("CREATE PUBLICATION p FOR ALL TABLES EXCEPT t1, t2");
4263        let Statement::CreatePublication(p) = s else {
4264            panic!()
4265        };
4266        let PublicationScope::AllTablesExcept(ts) = p.scope else {
4267            panic!("expected AllTablesExcept")
4268        };
4269        assert_eq!(ts, alloc::vec!["t1", "t2"]);
4270    }
4271
4272    #[test]
4273    fn parser_rejects_for_table_with_empty_list() {
4274        // `FOR TABLE` with nothing after is a parse error.
4275        let err = parse_statement("CREATE PUBLICATION p FOR TABLE")
4276            .expect_err("must error on empty list");
4277        // No specific message asserted — the call falls through to
4278        // expect_ident_like which yields "expected identifier, got …".
4279        assert!(!err.message.is_empty());
4280    }
4281
4282    #[test]
4283    fn parser_recognises_show_publications() {
4284        // v6.1.3 — SHOW PUBLICATIONS lands here. PUBLICATIONS is a
4285        // bare ident in this position, NOT a reserved keyword.
4286        let s = parse("SHOW PUBLICATIONS");
4287        assert!(matches!(s, Statement::ShowPublications));
4288    }
4289
4290    // ── v6.1.4: CREATE / DROP SUBSCRIPTION + SHOW SUBSCRIPTIONS ─
4291
4292    #[test]
4293    fn parser_recognises_create_subscription_single_publication() {
4294        let s = parse("CREATE SUBSCRIPTION sub_a CONNECTION 'host=127.0.0.1 port=20002' PUBLICATION pub_a");
4295        let Statement::CreateSubscription(c) = s else {
4296            panic!("expected CreateSubscription, got {s:?}")
4297        };
4298        assert_eq!(c.name, "sub_a");
4299        assert_eq!(c.conn_str, "host=127.0.0.1 port=20002");
4300        assert_eq!(c.publications, alloc::vec!["pub_a"]);
4301    }
4302
4303    #[test]
4304    fn parser_recognises_create_subscription_multi_publication() {
4305        let s = parse(
4306            "CREATE SUBSCRIPTION sub_a CONNECTION 'host=h' PUBLICATION p1, p2, p3",
4307        );
4308        let Statement::CreateSubscription(c) = s else {
4309            panic!()
4310        };
4311        assert_eq!(c.publications, alloc::vec!["p1", "p2", "p3"]);
4312    }
4313
4314    #[test]
4315    fn parser_rejects_create_subscription_missing_connection() {
4316        let err = parse_statement("CREATE SUBSCRIPTION s PUBLICATION p")
4317            .expect_err("must error on missing CONNECTION");
4318        assert!(err.message.contains("CONNECTION"), "got: {}", err.message);
4319    }
4320
4321    #[test]
4322    fn parser_rejects_create_subscription_missing_publication() {
4323        let err = parse_statement("CREATE SUBSCRIPTION s CONNECTION 'host=x'")
4324            .expect_err("must error on missing PUBLICATION");
4325        assert!(err.message.contains("PUBLICATION"), "got: {}", err.message);
4326    }
4327
4328    #[test]
4329    fn parser_recognises_drop_subscription() {
4330        let s = parse("DROP SUBSCRIPTION sub_a");
4331        let Statement::DropSubscription(name) = s else {
4332            panic!("expected DropSubscription, got {s:?}")
4333        };
4334        assert_eq!(name, "sub_a");
4335    }
4336
4337    #[test]
4338    fn parser_recognises_show_subscriptions() {
4339        let s = parse("SHOW SUBSCRIPTIONS");
4340        assert!(matches!(s, Statement::ShowSubscriptions));
4341    }
4342
4343    #[test]
4344    fn parser_recognises_wait_for_wal_position_no_timeout() {
4345        let s = parse("WAIT FOR WAL POSITION 12345");
4346        let Statement::WaitForWalPosition { pos, timeout_ms } = s else {
4347            panic!("expected WaitForWalPosition, got {s:?}")
4348        };
4349        assert_eq!(pos, 12345);
4350        assert!(timeout_ms.is_none());
4351    }
4352
4353    #[test]
4354    fn parser_recognises_wait_for_wal_position_with_timeout() {
4355        let s = parse("WAIT FOR WAL POSITION 67890 WITH TIMEOUT 5000");
4356        let Statement::WaitForWalPosition { pos, timeout_ms } = s else {
4357            panic!()
4358        };
4359        assert_eq!(pos, 67890);
4360        assert_eq!(timeout_ms, Some(5000));
4361    }
4362
4363    #[test]
4364    fn parser_rejects_wait_with_negative_position() {
4365        // The lexer treats `-` as a token; `expect_u64_literal`
4366        // only sees the Integer that follows, so the negative
4367        // arrives as a unary-minus expression at higher levels.
4368        // Bare `WAIT FOR WAL POSITION -1` thus surfaces as a
4369        // parse error one way or another.
4370        let err = parse_statement("WAIT FOR WAL POSITION -1").unwrap_err();
4371        assert!(!err.message.is_empty());
4372    }
4373
4374    #[test]
4375    fn parser_recognises_bare_analyze() {
4376        let s = parse("ANALYZE");
4377        assert!(matches!(s, Statement::Analyze(None)));
4378    }
4379
4380    #[test]
4381    fn parser_recognises_analyze_with_table() {
4382        let s = parse("ANALYZE users");
4383        let Statement::Analyze(Some(name)) = s else {
4384            panic!("expected Analyze, got {s:?}")
4385        };
4386        assert_eq!(name, "users");
4387    }
4388
4389    #[test]
4390    fn parser_recognises_analyze_with_quoted_table() {
4391        let s = parse("ANALYZE \"Mixed Case\"");
4392        let Statement::Analyze(Some(name)) = s else {
4393            panic!()
4394        };
4395        assert_eq!(name, "Mixed Case");
4396    }
4397
4398    #[test]
4399    fn parser_rejects_analyze_with_garbage_token() {
4400        let err = parse_statement("ANALYZE 42").expect_err("must error");
4401        assert!(!err.message.is_empty());
4402    }
4403
4404    #[test]
4405    fn analyze_display_roundtrips() {
4406        for sql in ["ANALYZE", "ANALYZE users"] {
4407            let s = parse(sql);
4408            let printed = s.to_string();
4409            let again = parse_statement(&printed)
4410                .unwrap_or_else(|e| panic!("re-parse failed for {printed:?}: {e}"));
4411            assert_eq!(s, again);
4412        }
4413    }
4414
4415    #[test]
4416    fn wait_for_display_roundtrips() {
4417        for sql in [
4418            "WAIT FOR WAL POSITION 12345",
4419            "WAIT FOR WAL POSITION 67890 WITH TIMEOUT 5000",
4420        ] {
4421            let s = parse(sql);
4422            let printed = s.to_string();
4423            let again = parse_statement(&printed)
4424                .unwrap_or_else(|e| panic!("re-parse failed for {printed:?}: {e}"));
4425            assert_eq!(s, again, "round-trip mismatch for {sql:?}");
4426        }
4427    }
4428
4429    #[test]
4430    fn subscription_ddl_display_roundtrips() {
4431        for sql in [
4432            "CREATE SUBSCRIPTION sub_a CONNECTION 'host=h port=20002' PUBLICATION pub_a",
4433            "CREATE SUBSCRIPTION sub_b CONNECTION 'host=h' PUBLICATION p1, p2",
4434            "DROP SUBSCRIPTION sub_a",
4435            "SHOW SUBSCRIPTIONS",
4436        ] {
4437            let s = parse(sql);
4438            let printed = s.to_string();
4439            let again = parse_statement(&printed)
4440                .unwrap_or_else(|e| panic!("re-parse failed for {printed:?}: {e}"));
4441            assert_eq!(s, again, "round-trip mismatch for {sql:?}");
4442        }
4443    }
4444
4445    #[test]
4446    fn parser_drop_dispatches_user_vs_publication() {
4447        // Pre-v6.1.2 DROP USER took the bare-ident path; v6.1.2
4448        // tokenises DROP. Both targets must still parse.
4449        let s = parse("DROP USER 'alice'");
4450        let Statement::DropUser(name) = s else {
4451            panic!("expected DropUser, got {s:?}")
4452        };
4453        assert_eq!(name, "alice");
4454        // And DROP PUBLICATION lands the new variant.
4455        let s = parse("DROP PUBLICATION p1");
4456        assert!(matches!(s, Statement::DropPublication(_)));
4457    }
4458
4459    #[test]
4460    fn publication_ddl_display_roundtrips() {
4461        // Every CREATE PUBLICATION variant must Display → parse →
4462        // same AST. v6.1.3 covers all three scope shapes.
4463        for sql in [
4464            "CREATE PUBLICATION pub_a",
4465            "CREATE PUBLICATION pub_a FOR ALL TABLES",
4466            "CREATE PUBLICATION pub_a FOR TABLE t1, t2",
4467            "CREATE PUBLICATION pub_a FOR ALL TABLES EXCEPT t1",
4468            "DROP PUBLICATION pub_a",
4469            "SHOW PUBLICATIONS",
4470        ] {
4471            let s = parse(sql);
4472            let printed = s.to_string();
4473            let again = parse_statement(&printed)
4474                .unwrap_or_else(|e| panic!("re-parse failed for {printed:?}: {e}"));
4475            assert_eq!(s, again, "round-trip mismatch for {sql:?}");
4476        }
4477    }
4478}