spg_engine/eval.rs
1//! Expression evaluator. Given a parsed `Expr`, a `Row`, and the row's column
2//! schema, produce a `Value`. v0.4 implements:
3//!
4//! - literals
5//! - column lookups (bare and qualified `t.col`)
6//! - unary minus / NOT
7//! - binary arithmetic, comparison, AND, OR
8//! - numeric widening (`Int → BigInt → Float`) at evaluation time
9//! - SQL three-valued logic for NULL:
10//! * any arithmetic / comparison op with a NULL operand → NULL
11//! * `TRUE OR NULL` → TRUE, `FALSE OR NULL` → NULL,
12//! * `FALSE AND NULL` → FALSE, `TRUE AND NULL` → NULL,
13//! * `NOT NULL` → NULL
14//!
15//! v0.4 deliberately does *not* implement: function calls, string
16//! concatenation, IS NULL / IS NOT NULL, BETWEEN, IN, etc. Those come later.
17
18use alloc::borrow::Cow;
19use alloc::format;
20use alloc::string::{String, ToString};
21use alloc::vec::Vec;
22
23use spg_sql::ast::{BinOp, CastTarget, ColumnName, Expr, Literal};
24use spg_storage::{ColumnSchema, Row, Value};
25
26pub(crate) mod arity;
27pub(crate) mod binop;
28mod cast;
29pub mod compiled;
30mod datetime;
31mod encoding;
32mod encodings;
33mod format;
34pub(crate) mod functions;
35mod inet;
36pub(crate) mod math;
37mod regexp;
38mod resolve;
39mod strings;
40pub(crate) mod textsearch;
41pub(crate) mod values;
42
43pub use crate::conversions::format_money_array;
44pub(crate) use binop::{
45 add_interval_to_micros, and_3vl, apply_binary, apply_binary_by_ref, apply_binary_interval,
46};
47use binop::{apply_binary_in, apply_unary, compare, pow10_i128};
48pub use cast::{cast_to_vector, cast_value, parse_vector_text};
49pub(crate) use compiled::{
50 CompiledExpr, compile_column_pos, compile_expr, eval_compiled, eval_compiled_ref,
51 fully_compilable,
52};
53use datetime::{
54 age, date_format_mysql, date_part, date_trunc, extract_field, from_unixtime, unix_timestamp_of,
55};
56use encoding::{decode_text, encode_text};
57pub use format::{
58 days_from_civil,
59 format_bigint_array,
60 format_bool_array,
61 format_bytea_array,
62 format_bytea_hex,
63 format_date,
64 format_date_array,
65 format_float,
66 format_float_array,
67 // v7.40.0 — the five new array text forms.
68 format_inet_array,
69 format_int_array,
70 format_interval,
71 format_interval_array,
72 format_interval_kinded,
73 format_money,
74 format_numeric,
75 format_numeric_array,
76 format_numeric_kind,
77 format_real,
78 format_real_array,
79 format_smallint_array,
80 format_text_array,
81 format_time,
82 format_time_array,
83 format_timestamp,
84 format_timestamp_array,
85 format_timestamptz,
86 format_timestamptz_at,
87 format_timetz,
88 format_timetz_array,
89 format_uuid_array,
90 parse_date_literal,
91 parse_timestamp_literal,
92};
93// v7.39 (GUC knife 3) — session render styles + styled formatters.
94pub use format::{
95 DateOrder, DateStyleKind, IntervalStyleKind, RenderStyle, format_date_array_styled,
96 format_date_styled, format_float_array_styled, format_float_mysql, format_float_styled,
97 format_interval_array_styled, format_interval_styled, format_real_mysql, format_real_styled,
98 format_timestamp_array_styled, format_timestamp_styled, format_timestamptz_styled,
99 format_timestamptz_tz, parse_date_literal_ordered, parse_timestamp_literal_ordered,
100 parse_timestamp_literal_tz_ordered,
101};
102use functions::apply_function;
103use inet::{inet_host, inet_masklen, inet_network, inet_op_bool_result};
104pub(crate) use math::{f64_ceil, f64_floor, f64_sqrt};
105use math::{
106 f64_exp, f64_ln, f64_powi, f64_round_half_away, f64_trunc, prng_next_f64, prng_next_u64,
107};
108pub(crate) use regexp::{
109 CompiledRe, compile_re, compiled_is_match, regex_is_match, regexp_matches_rows,
110};
111use regexp::{regexp_matches, regexp_replace, regexp_split_to_array};
112use resolve::{
113 collation_fold_for_compare, compare_is_case_insensitive, composite_eq, eval_expr_cow,
114 is_owned_compare_value, resolve_column, resolve_column_borrowed, text_prefix_chars,
115};
116pub(crate) use resolve::{
117 column_at, column_collation, find_column_pos, is_binary_coerced, locate_column,
118};
119use strings::{
120 TrimSide, format_string, pg_quote_ident, pg_quote_literal, pg_typeof_name, string_left_right,
121 string_pad, string_trim, to_char, value_to_format_text,
122};
123pub use textsearch::{
124 decode_tsquery_external, decode_tsvector_external, format_tsquery, format_tsvector,
125};
126use textsearch::{
127 fts_phraseto_tsquery, fts_plainto_tsquery, fts_setweight, fts_to_tsquery, fts_to_tsvector,
128 fts_ts_headline, fts_ts_rank, fts_ts_rank_cd, fts_ts_rewrite, fts_tsquery_bool,
129 fts_websearch_to_tsquery, ts_match, tsvector_concat,
130};
131pub use values::gen_random_uuid_bytes;
132/// v7.39 (tz epic) — fixed-offset / abbreviation resolution, exposed
133/// for `SET timezone` validation (named zones go through the host tzdb).
134pub(crate) fn datetime_resolve_zone_offset(z: &str) -> Option<i64> {
135 datetime::resolve_zone_offset(z)
136}
137
138pub use values::value_to_text;
139pub use values::value_to_text_styled;
140pub use values::value_to_text_typed;
141pub use values::value_to_text_typed_styled;
142pub use values::value_to_text_with_fsp;
143use values::{
144 array_2d_dims, array_element_at, array_len, array_rebuild, value_cmp_for_min_max, value_to_f64,
145 values_equal_for_nullif,
146};
147
148/// Resolution context for evaluating a single row. `table_alias` is the alias
149/// (or table name) callers should accept as the qualifier on a column ref —
150/// e.g. `FROM users AS u` makes `u.name` valid and rejects `other.name`.
151#[derive(Clone)]
152#[allow(missing_debug_implementations)] // sequence_resolver is a dyn Fn — no Debug
153pub struct EvalContext<'a> {
154 pub columns: &'a [ColumnSchema],
155 pub table_alias: Option<&'a str>,
156 /// v6.1.1 — bound parameters for `$N` placeholders inside the
157 /// expression tree. Empty for simple queries; populated by the
158 /// prepared-statement Execute path with Bind values converted
159 /// to `Value`. Index N (1-based per PG) hits `params[N-1]`.
160 pub params: &'a [Value<'static>],
161 /// v7.12.1 — session text-search config (from `SET
162 /// default_text_search_config = '<name>'`). Resolved when the
163 /// engine builds an `EvalContext` and consumed by the FTS
164 /// function dispatcher when `to_tsvector(text)` /
165 /// `plainto_tsquery(text)` etc are called without an explicit
166 /// config arg. `None` falls through to `simple`.
167 pub default_text_search_config: Option<&'a str>,
168 /// v7.17.0 Phase 1.1 — `nextval` / `currval` / `setval`
169 /// resolver. The engine builds this around a `&mut Catalog`
170 /// so apply_function can mutate sequence state without
171 /// eval owning a catalog reference. When `None`, sequence
172 /// functions return an error (read-only contexts).
173 pub sequence_resolver: Option<&'a SequenceResolver<'a>>,
174 /// v7.37.16 (16.12) — read-only catalog reference for
175 /// builtins that need catalog walks (e.g. `pg_partition_root`,
176 /// `pg_partition_ancestors`). `None` falls through to the
177 /// "no catalog available" branch which returns NULL — same
178 /// shape PG returns for a non-existent OID. Most evaluation
179 /// sites don't need catalog access (row scans, projections);
180 /// they construct contexts with `catalog: None` and the
181 /// engine populates `Some(&self.catalog)` only at the engine's
182 /// top-level entry points where the borrow is unambiguous.
183 pub catalog: Option<&'a spg_storage::Catalog>,
184 /// v7.39 (round 346, M1) — is this a MySQL-dialect session? The two
185 /// dialects disagree about what counts as a truth value: MariaDB
186 /// takes any non-zero number (and a string's leading number) as
187 /// true, PG refuses anything that is not boolean. Set from the
188 /// engine by [`EvalContext::with_engine`]; a context built without
189 /// one keeps PG's stricter reading.
190 pub mysql_dialect: bool,
191 /// Session GUCs set via `SET name = value` / `set_config`, keyed by
192 /// lowercased name. `current_setting('app.foo')` reads custom
193 /// (namespaced) settings from here — the mechanism apps use for
194 /// request context / RLS. `None` in read-only contexts that have no
195 /// session; unknown names then fall through to PG defaults.
196 pub session_gucs: Option<&'a alloc::collections::BTreeMap<String, String>>,
197 /// v7.39 (read01 round 58) — the engine's role store, so
198 /// `has_table_privilege('bob', …)` can expand bob's role MEMBERSHIPS (a
199 /// grant to a group role answers `true` for its inheriting members). `None`
200 /// in a context with no engine behind it — the role then stands alone.
201 pub users: Option<&'a crate::users::UserStore>,
202 /// v7.39 (read01 round 61) — how deep we are inside USER-DEFINED function
203 /// bodies. A function's body is evaluated with a child context, and a body
204 /// may call another function, so this bounds the recursion (a function that
205 /// calls itself would otherwise blow the stack, which an embed host cannot
206 /// catch).
207 pub fn_depth: u16,
208 /// v7.39 (read01 round 63) — the ENGINE, for a user-function body that has
209 /// its own FROM (`SELECT v FROM t WHERE id = k`). Such a body has to run
210 /// through the real executor: reading `catalog`'s rows straight from eval
211 /// would bypass the row-header visibility filter, so under in-place MVCC a
212 /// function would happily read DEAD rows. `None` in a context with no
213 /// engine behind it — a body with a FROM then errors, saying so.
214 pub engine: Option<&'a crate::Engine>,
215 /// v7.38 (read01 U15) — per-scan deterministic sampler state for
216 /// `TABLESAMPLE … REPEATABLE(seed)`. A fresh cell is created before a
217 /// scan whose predicate may draw `__tsm_fract(seed)`; the cell holds
218 /// `None` until the first draw seeds it from that literal, then a
219 /// scan-local xorshift sequence (isolated from the process-global
220 /// `random()` PRNG, so it's deterministic and rescan-stable). `None`
221 /// here means no sampler is attached.
222 pub sample_rng: Option<&'a core::cell::Cell<Option<u64>>>,
223 /// v7.38 (read01 P3.25) — native-stack-overflow guard. Lazily seeded
224 /// with the stack pointer of the outermost `eval_expr` call; deeper
225 /// calls compare their own pointer against it and bail with
226 /// [`EvalError::StackDepthExceeded`] once usage crosses a safe margin,
227 /// so a pathologically nested expression errors instead of aborting the
228 /// process. Owned (not a borrowed cell) so it stays stack-local and
229 /// never touches `Engine`'s `Sync` bound.
230 pub recursion_base: core::cell::Cell<usize>,
231 /// v7.39 (GUC knife 3) — session render style (DateStyle /
232 /// IntervalStyle / extra_float_digits) for text output produced
233 /// inside expression evaluation (`::text` casts). Contexts built
234 /// away from the session (per-shard scan filters, index probes)
235 /// keep the default — they don't render text output.
236 pub render_style: crate::eval::format::RenderStyle,
237 /// v7.39 (tz epic) — host IANA timezone lookups for named zones
238 /// (session rendering, AT TIME ZONE, literal zone suffixes).
239 pub tz_offset_fn: Option<crate::TzOffsetFn>,
240 pub tz_localize_fn: Option<crate::TzLocalizeFn>,
241 pub tz_abbrev_fn: Option<crate::TzAbbrevFn>,
242 /// v7.38 (read01 P5.24) — host-provided CSPRNG (the server injects
243 /// `/dev/urandom`). Cryptographic builtins (`gen_random_bytes`,
244 /// `gen_salt`) draw from this instead of the process-static xorshift
245 /// PRNG, so their output isn't predictable. `None` (no host CSPRNG)
246 /// falls back to the PRNG — fine for the non-cryptographic `random()`.
247 pub salt_fn: Option<crate::SaltFn>,
248 /// v7.39 (read01 pgstatfuncs.c) — calling-connection identity for
249 /// pg_backend_pid(); `None` (embedded / detached contexts) → 1.
250 pub backend_pid_fn: Option<crate::BackendPidFn>,
251 /// v7.39 (round 476) — the WAL byte position, for the LSN functions.
252 pub wal_lsn_fn: Option<crate::WalLsnFn>,
253 /// v7.39 (round 318, V51) — host connection-control hook for
254 /// `pg_cancel_backend` / `pg_terminate_backend`. `None` (embedded /
255 /// detached contexts) ⇒ there is nothing to signal, so they answer
256 /// false rather than pretending the signal landed.
257 pub backend_signal_fn: Option<crate::BackendSignalFn>,
258 /// v7.38 (read01 P6.08) — host wall clock (µs since Unix epoch). `uuidv7`
259 /// uses it for the real time-ordered 48-bit millisecond prefix; `None`
260 /// (no host clock) falls back to the deterministic anchor.
261 pub clock: Option<crate::ClockFn>,
262 /// v7.38 (T24) — read-only view of the engine's transaction-version state,
263 /// so the `txid_*` / `pg_*_xact_id` / `pg_xact_status` builtins report the
264 /// real transaction ids instead of a constant stub. `None` on the scan /
265 /// join / aggregate contexts that never evaluate them.
266 pub xact: Option<XactView<'a>>,
267 /// v7.38 (T24) — PG's `txid_current()` ASSIGNS an id to a transaction that
268 /// has none. In autocommit a read-only statement has no writer version, so
269 /// the first call allocates one here and later calls in the same statement
270 /// reuse it — `SELECT txid_current(), txid_current()` must agree, as in PG.
271 pub assigned_xid: core::cell::Cell<Option<u64>>,
272}
273
274/// v7.38 (T24) — the transaction-id surface PG's `txid_*` family exposes.
275/// SPG's writer versions ARE its transaction ids (`row_header::next_version`),
276/// so no separate xid counter is needed — this is the bridge U22 was waiting
277/// on.
278#[derive(Clone, Copy, Debug)]
279pub struct XactView<'a> {
280 /// The id assigned to the current transaction (allocated at BEGIN) or, in
281 /// autocommit, to the current statement once it has written. `None` when
282 /// nothing has been assigned — `*_if_assigned` returns NULL there, as PG does.
283 pub current: Option<u64>,
284 /// Ids allocated by transactions that have neither committed nor aborted.
285 pub active: &'a alloc::collections::BTreeSet<u64>,
286 /// Ids of rolled-back transactions.
287 pub aborted: &'a alloc::collections::BTreeSet<u64>,
288}
289
290/// v7.17.0 — sequence-mutating callback used by `apply_function`
291/// for `nextval` / `currval` / `setval`. Implemented by the
292/// engine to thread `&mut Catalog` access through an immutable
293/// `&EvalContext`.
294pub type SequenceResolver<'a> = dyn Fn(SequenceOp) -> Result<i64, EvalError> + 'a;
295
296/// v7.17.0 — sequence operation requested by an Expr eval.
297#[derive(Debug, Clone)]
298pub enum SequenceOp {
299 Next(String),
300 Curr(String),
301 Set {
302 name: String,
303 value: i64,
304 is_called: bool,
305 },
306}
307
308impl<'a> EvalContext<'a> {
309 /// v7.39.3 — do these two column names refer to the same column?
310 ///
311 /// MySQL's column names are case-INSENSITIVE (measured on 9.7.2:
312 /// `mycol`, `MYCOL` and `MyCol` all resolve a column declared
313 /// `MyCol`, and a backquoted spelling resolves too). SPG compared
314 /// them byte for byte, and its lexer folds an UNQUOTED identifier —
315 /// so a table restored from a `mysqldump`, where every identifier
316 /// is backquoted and keeps its case, had every mixed-case column
317 /// unreachable from ordinary unquoted SQL:
318 ///
319 /// ```text
320 /// CREATE TABLE `Orders` (`OrderID` INT) -- from the dump
321 /// SELECT `OrderID` FROM `Orders` -- works
322 /// SELECT OrderID FROM Orders -- 1054, unknown column
323 /// ```
324 ///
325 /// That is the same "two spellings, two things" defect v7.39.1
326 /// closed for relation NAMES, still live for columns.
327 ///
328 /// PostgreSQL's rule is the opposite and stays: a quoted identifier
329 /// is case-sensitive there, and folding it would break every column
330 /// that relies on it.
331 #[must_use]
332 pub fn col_eq(&self, a: &str, b: &str) -> bool {
333 if self.mysql_dialect {
334 a.eq_ignore_ascii_case(b)
335 } else {
336 a == b
337 }
338 }
339
340 pub const fn new(columns: &'a [ColumnSchema], table_alias: Option<&'a str>) -> Self {
341 Self {
342 columns,
343 table_alias,
344 params: &[],
345 default_text_search_config: None,
346 sequence_resolver: None,
347 catalog: None,
348 mysql_dialect: false,
349 session_gucs: None,
350 users: None,
351 fn_depth: 0,
352 engine: None,
353 sample_rng: None,
354 recursion_base: core::cell::Cell::new(0),
355 render_style: crate::eval::format::RenderStyle {
356 date_style: crate::eval::format::DateStyleKind::Iso,
357 date_order: crate::eval::format::DateOrder::Mdy,
358 interval_style: crate::eval::format::IntervalStyleKind::Postgres,
359 extra_float_digits: 1,
360 bytea_escape: false,
361 mysql: false,
362 },
363 tz_offset_fn: None,
364 tz_localize_fn: None,
365 tz_abbrev_fn: None,
366 salt_fn: None,
367 backend_pid_fn: None,
368 wal_lsn_fn: None,
369 backend_signal_fn: None,
370 clock: None,
371 xact: None,
372 assigned_xid: core::cell::Cell::new(None),
373 }
374 }
375
376 /// v7.39 (GUC knife 3) — attach the session render style.
377 #[must_use]
378 pub const fn with_render_style(mut self, style: crate::eval::format::RenderStyle) -> Self {
379 self.render_style = style;
380 self
381 }
382
383 /// v7.39 (round 318, V51) — attach the host connection-control hook.
384 #[must_use]
385 pub const fn with_backend_signal_fn(mut self, f: Option<crate::BackendSignalFn>) -> Self {
386 self.backend_signal_fn = f;
387 self
388 }
389
390 /// v7.39 (round 476) — attach the WAL byte-position provider.
391 #[must_use]
392 pub const fn with_wal_lsn_fn(mut self, f: Option<crate::WalLsnFn>) -> Self {
393 self.wal_lsn_fn = f;
394 self
395 }
396
397 /// v7.39 (read01 pgstatfuncs.c) — attach the calling-connection id.
398 #[must_use]
399 pub const fn with_backend_pid_fn(mut self, f: Option<crate::BackendPidFn>) -> Self {
400 self.backend_pid_fn = f;
401 self
402 }
403
404 /// v7.39 (tz epic) — attach the host timezone lookups.
405 #[must_use]
406 pub const fn with_tz_fns(
407 mut self,
408 offset: Option<crate::TzOffsetFn>,
409 localize: Option<crate::TzLocalizeFn>,
410 abbrev: Option<crate::TzAbbrevFn>,
411 ) -> Self {
412 self.tz_offset_fn = offset;
413 self.tz_localize_fn = localize;
414 self.tz_abbrev_fn = abbrev;
415 self
416 }
417
418 /// v7.39 (tz epic) — offset (µs east) of an arbitrary zone spec at
419 /// a UTC instant: fixed forms resolve statically, named zones
420 /// through the host tzdb. None = unknown zone.
421 #[must_use]
422 pub fn zone_offset_at(&self, zone: &str, utc_micros: i64) -> Option<i64> {
423 if let Some(off) = datetime::resolve_zone_offset(zone) {
424 return Some(off);
425 }
426 self.tz_offset_fn.and_then(|f| f(zone, utc_micros))
427 }
428
429 /// v7.39 (tz epic) — the SESSION zone's offset at a UTC instant
430 /// (per-value: DST zones vary within one statement).
431 #[must_use]
432 pub fn session_tz_offset_at(&self, utc_micros: i64) -> i64 {
433 let Some(zone) = self.session_gucs.and_then(|g| g.get("timezone")) else {
434 return 0;
435 };
436 self.zone_offset_at(zone, utc_micros).unwrap_or(0)
437 }
438
439 /// v7.39 (tz epic) — the session zone's designation at an instant
440 /// (named zones only; None lets renderers spell UTC / +HH).
441 #[must_use]
442 pub fn session_tz_abbrev_at(&self, utc_micros: i64) -> Option<alloc::string::String> {
443 let zone = self.session_gucs.and_then(|g| g.get("timezone"))?;
444 if datetime::resolve_zone_offset(zone).is_some()
445 || zone.eq_ignore_ascii_case("utc")
446 || zone.eq_ignore_ascii_case("gmt")
447 {
448 return None;
449 }
450 self.tz_abbrev_fn.and_then(|f| f(zone, utc_micros))
451 }
452
453 /// v7.39 (tz epic) — local wall micros in `zone` -> UTC micros
454 /// (PG's DST disambiguation for named zones).
455 #[must_use]
456 pub fn zone_local_to_utc(&self, zone: &str, local_micros: i64) -> Option<i64> {
457 zone_local_to_utc_with(zone, local_micros, self.tz_localize_fn)
458 }
459
460 /// v7.38 (read01 P5.24) — attach the host CSPRNG so cryptographic
461 /// builtins don't fall back to the predictable PRNG.
462 #[must_use]
463 pub const fn with_salt_fn(mut self, f: Option<crate::SaltFn>) -> Self {
464 self.salt_fn = f;
465 self
466 }
467
468 /// v7.38 (read01 P6.08) — attach the host wall clock so `uuidv7` gets a
469 /// real time-ordered prefix instead of the deterministic anchor.
470 #[must_use]
471 pub const fn with_clock(mut self, f: Option<crate::ClockFn>) -> Self {
472 self.clock = f;
473 self
474 }
475
476 /// v7.38 (read01 U15) — attach a per-scan `TABLESAMPLE REPEATABLE`
477 /// sampler cell. The cell (seeded lazily on first `__tsm_fract` draw)
478 /// must outlive the context and be created fresh per scan so a rescan
479 /// re-seeds and reproduces the same sample.
480 #[must_use]
481 pub const fn with_sample_rng(mut self, cell: &'a core::cell::Cell<Option<u64>>) -> Self {
482 self.sample_rng = Some(cell);
483 self
484 }
485
486 /// Attach the session's GUC map so `current_setting` can resolve
487 /// custom (namespaced) settings written with `SET` / `set_config`.
488 #[must_use]
489 /// v7.39 (read01 round 63) — thread the engine (see `engine`).
490 pub const fn with_engine(mut self, engine: &'a crate::Engine) -> Self {
491 self.mysql_dialect = engine.speaks_mysql;
492 // v7.39 (round 368, M20 P3) — the dialect also decides how a binary
493 // string renders in a string context (latin-1 bytes vs PG `\x…`).
494 self.render_style.mysql = engine.speaks_mysql;
495 self.engine = Some(engine);
496 self
497 }
498
499 /// v7.39 (read01 round 58) — thread the role store (see `users`).
500 pub const fn with_users(mut self, users: &'a crate::users::UserStore) -> Self {
501 self.users = Some(users);
502 self
503 }
504
505 /// v7.39 (round 524) — attach a whole session bag at once. Every
506 /// write path needs the same four, and taking them one at a time is
507 /// how three of them ended up with none.
508 #[must_use]
509 pub(crate) fn with_session<'b: 'a>(mut self, s: &'b DmlSession) -> Self {
510 self.session_gucs = Some(&s.gucs);
511 self.users = Some(&s.users);
512 self.render_style = s.render_style;
513 self.tz_offset_fn = s.tz_offset_fn;
514 self.tz_localize_fn = s.tz_localize_fn;
515 self.tz_abbrev_fn = s.tz_abbrev_fn;
516 self
517 }
518
519 pub const fn with_session_gucs(
520 mut self,
521 gucs: &'a alloc::collections::BTreeMap<String, String>,
522 ) -> Self {
523 self.session_gucs = Some(gucs);
524 self
525 }
526
527 /// v7.37.16 (16.12) — attach a read-only catalog reference
528 /// so builtins like `pg_partition_root` can walk partition
529 /// roles. Defaults to None (NULL semantics).
530 #[must_use]
531 pub const fn with_catalog(mut self, catalog: &'a spg_storage::Catalog) -> Self {
532 self.catalog = Some(catalog);
533 self
534 }
535
536 /// v7.38 (T-tstz Phase 2) — the micro-offset of the session `TimeZone` GUC
537 /// (`SET TimeZone = '+09'` → +9h). A fixed offset / abbreviation resolves;
538 /// UTC and an unset GUC give 0; a named IANA zone (no tzdata) also gives 0
539 /// so a timestamptz still renders — as `+00` — rather than erroring on
540 /// every display. Timestamptz rendering / cast is the only consumer.
541 #[must_use]
542 pub fn session_tz_offset(&self) -> i64 {
543 self.session_gucs
544 .and_then(|g| g.get("timezone"))
545 .and_then(|z| datetime::resolve_zone_offset(z))
546 .unwrap_or(0)
547 }
548
549 /// v7.38 (T24) — attach the transaction-version view the `txid_*` builtins
550 /// read. Defaults to None, where they fall back to the process-wide cursor.
551 #[must_use]
552 pub const fn with_xact(mut self, xact: XactView<'a>) -> Self {
553 self.xact = Some(xact);
554 self
555 }
556
557 /// v7.17.0 — attach a sequence resolver. The engine wraps a
558 /// `&mut Catalog` in a closure that performs the requested
559 /// SequenceOp.
560 #[must_use]
561 pub const fn with_sequence_resolver(mut self, resolver: &'a SequenceResolver<'a>) -> Self {
562 self.sequence_resolver = Some(resolver);
563 self
564 }
565
566 /// v6.1.1 — attach a parameter buffer for `$N` placeholder
567 /// resolution. The slice must outlive the context; callers
568 /// construct it from the prepared statement's Bind values.
569 #[must_use]
570 pub const fn with_params(mut self, params: &'a [Value<'static>]) -> Self {
571 self.params = params;
572 self
573 }
574
575 /// v7.12.1 — attach the session's
576 /// `default_text_search_config`. Used by the FTS function
577 /// dispatcher when no explicit config arg is given.
578 #[must_use]
579 pub const fn with_default_text_search_config(mut self, cfg: Option<&'a str>) -> Self {
580 self.default_text_search_config = cfg;
581 self
582 }
583}
584
585/// v7.39 (round 523) — read a timestamp literal, reporting whether it
586/// carried an offset. Re-exported for the INSERT path, which decides
587/// there whether a value already names an instant.
588pub(crate) fn parse_timestamp_literal_tz_ordered_pub(
589 s: &str,
590 order: DateOrder,
591) -> Option<(i64, bool)> {
592 format::parse_timestamp_literal_tz_ordered(s, order)
593}
594
595/// v7.39 (round 523) — a FIXED zone's offset, when the name is one
596/// (`+09`, `UTC-5`). Named zones go through the host's tzdb instead.
597#[must_use]
598pub(crate) fn resolve_zone_offset_pub(zone: &str) -> Option<i64> {
599 datetime::resolve_zone_offset(zone)
600}
601
602/// v7.39 (round 523) — a wall-clock reading in `zone` as a UTC instant.
603///
604/// A free function because the INSERT path needs it too, and that path
605/// carries no `EvalContext`: it evaluates VALUES through a context-free
606/// literal walker. `EvalContext::zone_local_to_utc` delegates here so the
607/// two cannot drift.
608#[must_use]
609pub(crate) fn zone_local_to_utc_with(
610 zone: &str,
611 local_micros: i64,
612 localize: Option<crate::TzLocalizeFn>,
613) -> Option<i64> {
614 if let Some(off) = datetime::resolve_zone_offset(zone) {
615 return Some(local_micros - off);
616 }
617 localize.and_then(|f| f(zone, local_micros))
618}
619
620/// v7.39 (round 523) — the session zone an assignment to a timestamptz
621/// column is read in, or `None` when the session is on UTC and no shift
622/// applies.
623#[derive(Debug, Clone)]
624pub(crate) struct SessionCoercion {
625 /// The session zone, when it is not UTC. `None` leaves an instant
626 /// where it was.
627 pub zone: Option<alloc::string::String>,
628 pub localize: Option<crate::TzLocalizeFn>,
629 /// v7.39.13 — the other direction. `wall_to_utc` alone could only
630 /// serve assignments INTO a timestamptz column; PG also converts
631 /// the other way, and SPG did not.
632 pub offset: Option<crate::TzOffsetFn>,
633 /// The session's date order. A written date is ambiguous
634 /// (`01/02/2020`), and this is what resolves it.
635 pub order: DateOrder,
636}
637
638impl SessionCoercion {
639 /// The UTC instant a naive wall-clock reading names in the session
640 /// zone, or `None` when the session is on UTC.
641 #[must_use]
642 pub(crate) fn wall_to_utc(&self, wall: i64) -> Option<i64> {
643 let zone = self.zone.as_ref()?;
644 zone_local_to_utc_with(zone, wall, self.localize)
645 }
646
647 /// v7.39.13 — the wall-clock reading an instant shows in the
648 /// session zone, or `None` when the session is on UTC.
649 ///
650 /// The mirror of `wall_to_utc`, and it was missing: assigning a
651 /// `timestamptz` into a plain `timestamp` column stored the UTC
652 /// instant unchanged, where PG 18.6 renders it in the session's
653 /// zone first. Measured under `Asia/Tokyo`, one instant copied
654 /// into a `timestamp` column: SPG 1767225600, PG 1767258000.
655 #[must_use]
656 pub(crate) fn utc_to_wall(&self, utc: i64) -> Option<i64> {
657 let zone = self.zone.as_ref()?;
658 let off = self.offset.and_then(|f| f(zone, utc))?;
659 utc.checked_add(off)
660 }
661
662 /// v7.39 (round 524) — the session facts an ASSIGNMENT is read
663 /// under, from an evaluation context. `None` when both are the
664 /// defaults and nothing needs re-reading.
665 #[must_use]
666 pub(crate) fn from_ctx(ctx: &EvalContext<'_>) -> Option<Self> {
667 let zone = ctx
668 .session_gucs
669 .and_then(|g| g.get("timezone"))
670 .filter(|z| !z.eq_ignore_ascii_case("utc") && !z.eq_ignore_ascii_case("gmt"))
671 .cloned();
672 let order = ctx.render_style.date_order;
673 if zone.is_none() && order == DateOrder::Mdy {
674 return None;
675 }
676 Some(Self {
677 zone,
678 localize: ctx.tz_localize_fn,
679 offset: ctx.tz_offset_fn,
680 order,
681 })
682 }
683}
684
685/// v7.39 (round 524) — the session facts a DML evaluation context needs,
686/// cloned so the row loop can still borrow the engine mutably.
687///
688/// Every write path built a BARE `EvalContext`, so an expression in an
689/// UPDATE's SET or a DELETE's WHERE was evaluated by an engine that knew
690/// nothing about the connection. One value, built once per statement,
691/// and a grep for `dml_session` finds every path that has it.
692pub(crate) struct DmlSession {
693 pub gucs: alloc::collections::BTreeMap<String, String>,
694 pub users: crate::users::UserStore,
695 pub render_style: RenderStyle,
696 pub tz_offset_fn: Option<crate::TzOffsetFn>,
697 pub tz_localize_fn: Option<crate::TzLocalizeFn>,
698 pub tz_abbrev_fn: Option<crate::TzAbbrevFn>,
699}
700
701/// v7.39 (round 524) — read a TEXT value bound for a temporal column
702/// under the session's date order.
703///
704/// `01/02/2020` is February 1st in a DMY session and January 2nd in an
705/// MDY one, and the write path was reading every one of them as MDY: a
706/// `SELECT '01/02/2020'::date` answered PG's value while the same
707/// literal INSERTed stored the day and month swapped. Nothing errors,
708/// and once stored the two readings are indistinguishable.
709#[must_use]
710pub(crate) fn session_read_temporal_text(
711 v: Value<'static>,
712 target: spg_storage::DataType,
713 coercion: Option<&SessionCoercion>,
714) -> Value<'static> {
715 use spg_storage::DataType as D;
716 let Some(c) = coercion else { return v };
717 if c.order == DateOrder::Mdy {
718 return v;
719 }
720 let Value::Text(s) = &v else { return v };
721 match target {
722 D::Date => format::parse_date_literal_ordered(s, c.order).map_or(v, Value::Date),
723 D::Timestamp | D::Timestamptz => format::parse_timestamp_literal_tz_ordered(s, c.order)
724 .map_or(v, |(t, _)| Value::Timestamp(t)),
725 _ => v,
726 }
727}
728
729#[derive(Debug, Clone, PartialEq)]
730pub enum EvalError {
731 /// v7.39.2 — a qualified reference whose QUALIFIER resolves but
732 /// whose column does not. PostgreSQL prints it unquoted and dotted;
733 /// see the Display impl.
734 QualifiedColumnNotFound {
735 qualifier: String,
736 column: String,
737 },
738 /// v7.39.2 — a built-in called with the wrong NUMBER of arguments.
739 ///
740 /// Its own variant rather than a formatted string, because the two
741 /// wires word it differently AND number it differently, and one of
742 /// them has to tell it apart from "no such function": MySQL 9.7.2
743 /// answers 1582 `Incorrect parameter count in the call to native
744 /// function 'LOWER'` here and 1305 for a name it does not know,
745 /// where PostgreSQL gives BOTH the same sentence.
746 WrongArity {
747 name: String,
748 /// The argument types, PostgreSQL's own names, comma-joined.
749 types: String,
750 },
751 ColumnNotFound {
752 name: String,
753 },
754 UnknownQualifier {
755 qualifier: String,
756 /// v7.39.2 — the column that was written after it.
757 ///
758 /// PostgreSQL names only the missing table; MySQL 9.7.2 names
759 /// the whole reference — `Unknown column 'zz.a' in 'where
760 /// clause'` — and numbers it 1054, a COLUMN error. SPG answered
761 /// 1146, which is what a driver reads as "that table is gone".
762 /// The name has to travel for the MySQL wire to say it.
763 column: String,
764 },
765 DivisionByZero,
766 TypeMismatch {
767 detail: String,
768 },
769 /// v6.1.1 — `$N` reference past the number of bound parameters.
770 /// Either the client sent too few in Bind, or the SQL has a
771 /// placeholder the prepared statement didn't account for.
772 PlaceholderOutOfRange {
773 n: u16,
774 bound: u16,
775 },
776 /// v7.38 (read01 P3.25) — the expression tree recursed deep enough to
777 /// threaten a native stack overflow; we bail out with an error the way
778 /// PG's `check_stack_depth()` does instead of aborting the process.
779 StackDepthExceeded,
780}
781
782impl core::fmt::Display for EvalError {
783 fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
784 match self {
785 // v7.39 (read01 round 81) — PG's wording (and SQLSTATE trigger):
786 // `column "x" does not exist`, 42703. The old "column not found: x"
787 // matched none of the wire layer's `does not exist` patterns, so a
788 // missing column reached the client as the generic error class.
789 // PostgreSQL 18.6 makes no distinction between a missing
790 // function and a missing overload — `nosuchfn(text, integer,
791 // date)` and `lower(text, integer)` get the identical
792 // sentence (measured) — so this is the engine's canonical
793 // rendering and the MySQL wire re-words it.
794 Self::WrongArity { name, types } => {
795 write!(f, "function {name}({types}) does not exist")
796 }
797 Self::ColumnNotFound { name } => write!(f, "column \"{name}\" does not exist"),
798 // v7.39.2 — a QUALIFIED reference prints unquoted and
799 // dotted on PostgreSQL 18.6: `column ea.no_such does not
800 // exist`, where the bare form is `column "no_such" does not
801 // exist` (both measured). The two shapes are not
802 // interchangeable — a caller matching on `ea.no_such` finds
803 // nothing in the quoted one.
804 Self::QualifiedColumnNotFound { qualifier, column } => {
805 write!(f, "column {qualifier}.{column} does not exist")
806 }
807 // v7.39 (round 241) — PG's wording (and 42P01 trigger): a
808 // qualifier that names no table in scope is "missing
809 // FROM-clause entry for table \"x\"". The old "unknown table
810 // qualifier" matched nothing a driver branches on.
811 Self::UnknownQualifier { qualifier, .. } => {
812 write!(f, "missing FROM-clause entry for table \"{qualifier}\"")
813 }
814 Self::DivisionByZero => f.write_str("division by zero"),
815 Self::TypeMismatch { detail } => write!(f, "type mismatch: {detail}"),
816 Self::PlaceholderOutOfRange { n, bound } => write!(
817 f,
818 "parameter ${n} referenced but only {bound} bound by client"
819 ),
820 Self::StackDepthExceeded => {
821 f.write_str("stack depth limit exceeded (expression nested too deeply)")
822 }
823 }
824 }
825}
826
827/// v7.38 (read01 P3.25) — native-stack budget below the outermost
828/// `eval_expr` frame. Native stacks are typically 2–8 MB; 768 KiB leaves
829/// generous headroom while still permitting PG-class nesting depth (in a
830/// release build ~hundreds-to-thousands of frames fit under this).
831const MAX_EVAL_STACK_BYTES: usize = 768 * 1024;
832
833/// Address of a local in the current frame — a portable stand-in for the
834/// stack pointer (stacks grow downward on all supported targets).
835#[inline(never)]
836fn eval_stack_ptr() -> usize {
837 let probe = 0u8;
838 core::ptr::addr_of!(probe) as usize
839}
840
841/// v7.38 (read01 P6.40) — enforce a user DOMAIN's NOT NULL + CHECK constraints
842/// on a value being cast to it (`x::domain`). NULL fails a NOT NULL domain;
843/// otherwise every CHECK (which references the pseudo-column `VALUE`) must not
844/// evaluate to false. Returns the value unchanged when all constraints pass.
845fn apply_domain_constraints<'a>(
846 v: Value<'a>,
847 dom: &spg_storage::DomainDef,
848 name: &str,
849 cat: &spg_storage::Catalog,
850) -> Result<Value<'a>, EvalError> {
851 if matches!(v, Value::Null) {
852 // A NOT NULL anywhere in the chain rejects a NULL.
853 let mut cur = Some(dom);
854 while let Some(d) = cur {
855 if !d.nullable {
856 return Err(EvalError::TypeMismatch {
857 detail: alloc::format!("domain {name} does not allow null values"),
858 });
859 }
860 cur = d
861 .base_domain
862 .as_ref()
863 .and_then(|p| cat.domain_types().get(p.as_str()));
864 }
865 return Ok(v);
866 }
867 // v7.39 (round 259) — walk the domain chain BASE-FIRST (probed: a
868 // value violating both a parent's and the child's constraint reports
869 // the PARENT's). The message names the domain being cast TO, but the
870 // constraint that actually failed — `value for domain pchild violates
871 // check constraint "pbase_check"`.
872 let mut chain: alloc::vec::Vec<&spg_storage::DomainDef> = alloc::vec![dom];
873 let mut cur = dom;
874 while let Some(parent) = cur
875 .base_domain
876 .as_ref()
877 .and_then(|p| cat.domain_types().get(p.as_str()))
878 {
879 // A cycle cannot be created through CREATE DOMAIN (the parent must
880 // already exist), but stop defensively rather than loop forever.
881 if chain.iter().any(|d| core::ptr::eq(*d, parent)) {
882 break;
883 }
884 chain.push(parent);
885 cur = parent;
886 }
887 chain.reverse();
888 for owner in chain {
889 apply_domain_checks_of(&v, owner, name)?;
890 }
891 Ok(v)
892}
893
894/// v7.39 (round 259) — run ONE domain's own CHECK list against `v`. The
895/// error names `target` (the domain the value is being cast to) and
896/// `owner` (whose constraint failed); for a single-level domain they are
897/// the same, which is the pre-259 wording.
898fn apply_domain_checks_of(
899 v: &Value<'_>,
900 dom: &spg_storage::DomainDef,
901 target: &str,
902) -> Result<(), EvalError> {
903 let name = target;
904 for chk in &dom.checks {
905 let src = &chk.expr;
906 // v7.39 (round 260) — report the constraint that failed by NAME.
907 let owner = chk.name.as_str();
908 let expr = spg_sql::parser::parse_expression(src).map_err(|e| EvalError::TypeMismatch {
909 detail: alloc::format!("domain {name} CHECK ({src:?}) failed to re-parse: {e:?}"),
910 })?;
911 let synth_cols = alloc::vec![spg_storage::ColumnSchema::new(
912 "value",
913 dom.base_type,
914 dom.nullable,
915 )];
916 let synth_ctx = EvalContext::new(&synth_cols, None);
917 // Owned copy so the temporary row doesn't borrow `v`'s lifetime.
918 let synth_row = spg_storage::Row {
919 values: alloc::vec![v.clone().into_owned()],
920 };
921 let r = eval_expr(&expr, &synth_row, &synth_ctx)?;
922 if matches!(r, Value::Bool(false)) {
923 return Err(EvalError::TypeMismatch {
924 detail: alloc::format!(
925 "value for domain {name} violates check constraint \"{owner}\""
926 ),
927 });
928 }
929 }
930 Ok(())
931}
932
933/// v7.38 (read01 P6.67) — validate a value cast to a user ENUM: a text label
934/// must be one of the enum's members (else error, as PG does); a NULL is a
935/// valid typed null. The stored representation stays the text label.
936/// v7.39 (read01 rowtypes.c) — cast into a user composite type: parse the
937/// `(v1,"v 2",)` record text (double-quote wrapping with doubled quotes,
938/// empty field = NULL) and coerce each field to the declared type; a ROW
939/// value re-labels positionally.
940/// v7.39 (round 350/351, M7 + M11) — how MySQL reads a TEXT operand of
941/// an arithmetic or comparison operator. The identity in the PG dialect,
942/// and out-of-line so it costs the recursive `eval_expr` frame nothing.
943///
944/// Measured on MariaDB 11: `'2024-01-15' + INTERVAL 1 DAY` shifts the
945/// date; `'1abc'+0` is 1, `'abc'+0` is 0, `'2024-01-15'+0` is 2024; two
946/// strings compare as STRINGS (`'10' > '9'` is 0) while a mixed pair
947/// compares numerically (`'10' > 9` is 1).
948#[inline(never)]
949pub(crate) fn mysql_operand_reading_pair(
950 op: BinOp,
951 l: Value<'static>,
952 r: Value<'static>,
953) -> (Value<'static>, Value<'static>) {
954 if !mysql_coerces(op) {
955 return (l, r);
956 }
957 match (&l, &r) {
958 (Value::Text(t), Value::Interval { .. }) => (text_as_temporal(t).unwrap_or(l.clone()), r),
959 (Value::Interval { .. }, Value::Text(t)) => {
960 let rr = text_as_temporal(t).unwrap_or(r.clone());
961 (l, rr)
962 }
963 // v7.39 (round 353, M10) — a boolean IS an integer in MySQL, so
964 // `!1 + 1` is 1 (measured). It was `operator does not exist:
965 // boolean + integer`.
966 (Value::Bool(b), other)
967 if mysql_arith(op) && other.data_type().is_some_and(is_numeric_type) =>
968 {
969 (Value::BigInt(i64::from(*b)), r)
970 }
971 (other, Value::Bool(b))
972 if mysql_arith(op) && other.data_type().is_some_and(is_numeric_type) =>
973 {
974 let rr = Value::BigInt(i64::from(*b));
975 (l, rr)
976 }
977 (Value::Text(t), other) if other.data_type().is_some_and(is_numeric_type) => {
978 (mysql_number_of(t), r)
979 }
980 (other, Value::Text(t)) if other.data_type().is_some_and(is_numeric_type) => {
981 let rr = mysql_number_of(t);
982 (l, rr)
983 }
984 // v7.39 (round 367, M20 P2) — a binary string beside a number
985 // reads as its big-endian integer value (`0x10 + 0` = 16,
986 // `0x10 = 16` is true). Beside a Text operand it stays bytes so
987 // the byte-wise string compare (`0x61 = 'a'`) still fires.
988 (Value::Bytes(b), other) if other.data_type().is_some_and(is_numeric_type) => {
989 (mysql_bytes_as_number(b), r)
990 }
991 (other, Value::Bytes(b)) if other.data_type().is_some_and(is_numeric_type) => {
992 let rr = mysql_bytes_as_number(b);
993 (l, rr)
994 }
995 // Arithmetic between two strings is numeric; comparison is not.
996 (Value::Text(a), Value::Text(b)) if mysql_arith(op) => {
997 (mysql_number_of(a), mysql_number_of(b))
998 }
999 _ => (l, r),
1000 }
1001}
1002
1003/// Is this a mixed string/number pair, which MySQL compares numerically?
1004fn mysql_mixed_pair(l: &Value<'_>, r: &Value<'_>) -> bool {
1005 // v7.39.2 — a BINARY string beside a number is a mixed pair too: the
1006 // coercion above reads `0x10` as 16. Until now the pair reached the
1007 // owned path only because a MySQL session folded EVERY comparison, so
1008 // `compare_is_case_insensitive` said yes for an unrelated reason; the
1009 // moment a binary operand stopped folding, `0x10 = 16` fell through to
1010 // "operator does not exist: bytea = integer". The two questions are
1011 // separate and are now asked separately.
1012 matches!((l, r), (Value::Text(_) | Value::Bytes(_), o) | (o, Value::Text(_) | Value::Bytes(_))
1013 if o.data_type().is_some_and(is_numeric_type))
1014}
1015
1016const fn mysql_arith(op: BinOp) -> bool {
1017 matches!(
1018 op,
1019 BinOp::Add | BinOp::Sub | BinOp::Mul | BinOp::Div | BinOp::Mod
1020 )
1021}
1022
1023/// Does this comparison need the owned path — because a value's type
1024/// wants it, because a CI collation folds it, or (MySQL) because a mixed
1025/// string/number pair compares NUMERICALLY there while two strings
1026/// compare as strings.
1027#[inline(never)]
1028fn needs_owned_compare(
1029 lc: &Value<'_>,
1030 rc: &Value<'_>,
1031 lhs: &Expr,
1032 rhs: &Expr,
1033 ctx: &EvalContext<'_>,
1034) -> bool {
1035 is_owned_compare_value(lc)
1036 || is_owned_compare_value(rc)
1037 || compare_is_case_insensitive(lhs, rhs, ctx)
1038 || (ctx.mysql_dialect && mysql_mixed_pair(lc, rc))
1039}
1040
1041/// Which operators take MySQL's string→number reading.
1042const fn mysql_coerces(op: BinOp) -> bool {
1043 matches!(
1044 op,
1045 BinOp::Add
1046 | BinOp::Sub
1047 | BinOp::Mul
1048 | BinOp::Div
1049 | BinOp::Mod
1050 | BinOp::Eq
1051 | BinOp::NotEq
1052 | BinOp::Lt
1053 | BinOp::LtEq
1054 | BinOp::Gt
1055 | BinOp::GtEq
1056 )
1057}
1058
1059/// Does this type take part in MySQL's numeric coercion?
1060fn is_numeric_type(t: spg_storage::DataType) -> bool {
1061 use spg_storage::DataType as D;
1062 matches!(
1063 t,
1064 D::SmallInt | D::Int | D::BigInt | D::Float | D::Real | D::Numeric { .. }
1065 )
1066}
1067
1068/// v7.39 (round 364, M4 P2) — a value as it participates in a MySQL
1069/// session's default-collation comparison: text folds (accent- and
1070/// case-insensitive), everything else is itself. Used by IN and LIKE,
1071/// whose comparisons do not pass through `collation_fold_for_compare`.
1072fn mysql_collation_key(v: Value<'static>, mysql: bool, pads: bool) -> Value<'static> {
1073 match v {
1074 // v7.38.16 — BpChar too. `mysql_compare_fold` trims trailing
1075 // spaces before folding, which is the PAD SPACE half of the same
1076 // comparison, so a CHAR cell needs exactly this call and was not
1077 // getting it: `s IN ('ALPHA','BETA')` on CHAR(8) answered 1 where
1078 // MySQL 9.7.1 answers 1,2. `eval/values.rs` had the pair right
1079 // and these two sites did not.
1080 // v7.38.17 — CHAR's padding is not data, TEXT's trailing
1081 // spaces are. Two calls because they are two questions.
1082 // v7.38.18 — and the collation's padding rule. A CHAR's padding
1083 // is the type's and never counts; a TEXT's is the collation's.
1084 // `t IN ('ALPHA')` on a `utf8mb4_uca1400_ai_ci` column — what a
1085 // MariaDB dump declares — missed the row holding `'alpha '`,
1086 // which MariaDB 12.3.2 matches.
1087 Value::BpChar(s) if mysql => Value::text(spg_storage::mysql_compare_fold_char(&s)),
1088 Value::Text(s) if mysql && pads => Value::text(spg_storage::mysql_compare_fold_char(&s)),
1089 Value::Text(s) if mysql => Value::text(spg_storage::mysql_compare_fold(&s)),
1090 other => other,
1091 }
1092}
1093
1094/// A string as MySQL reads it in numeric position: an exact integer when
1095/// the leading number is one, otherwise a double.
1096#[inline(never)]
1097fn mysql_number_of(s: &str) -> Value<'static> {
1098 let n = mysql_leading_number(s);
1099 if n.fract() == 0.0 && n.abs() < 9.007_199_254_740_992e15 {
1100 #[allow(clippy::cast_possible_truncation)]
1101 Value::BigInt(n as i64)
1102 } else {
1103 Value::Float(n)
1104 }
1105}
1106
1107/// v7.39 (round 367, M20 P2) — a MySQL binary string (a `0x…` / `X'…'` /
1108/// `b'…'` literal, backed by `Value::Bytes`) reads as its bytes'
1109/// BIG-ENDIAN unsigned integer in a numeric context: `0x4142 + 0` is
1110/// 16706, `0x10 = 16` is true (measured on MariaDB 11). Only the low 16
1111/// bytes participate — a hex literal used in arithmetic is at most an
1112/// 8-byte BIGINT in practice — and a value past `i64::MAX` becomes a
1113/// NUMERIC so nothing wraps negative.
1114fn mysql_bytes_as_number(b: &[u8]) -> Value<'static> {
1115 let start = b.len().saturating_sub(16);
1116 let acc = b[start..]
1117 .iter()
1118 .fold(0u128, |a, &x| (a << 8) | u128::from(x));
1119 if acc <= i64::MAX as u128 {
1120 #[allow(clippy::cast_possible_truncation)]
1121 Value::BigInt(acc as i64)
1122 } else {
1123 crate::conversions::big_literal_to_value(&alloc::format!("{acc}"))
1124 }
1125}
1126
1127/// MySQL's `/`: a real division, and NULL on a zero divisor. `None`
1128/// when this pairing is not the integer/integer case PG and MySQL
1129/// disagree about.
1130#[inline(never)]
1131pub(crate) fn mysql_true_division(
1132 op: BinOp,
1133 l: &Value<'_>,
1134 r: &Value<'_>,
1135 text_operand: bool,
1136) -> Option<Value<'static>> {
1137 // v7.39 (round 372) — MySQL's `x % 0` / `x MOD 0` is NULL, not the PG
1138 // "division by zero" error (measured on MariaDB 11: `10%0`, `10 MOD
1139 // 0`, `10.5%0` are all NULL, matching `1/0`). A non-zero divisor takes
1140 // the normal modulo path.
1141 if matches!(op, BinOp::Mod) {
1142 return if value_is_zero(r) {
1143 Some(Value::Null)
1144 } else {
1145 None
1146 };
1147 }
1148 if !matches!(op, BinOp::Div) {
1149 return None;
1150 }
1151 // v7.39 (round 393) — MariaDB `/` on exact (int / decimal) operands is a
1152 // DECIMAL whose scale is the LEFT operand's scale + 4 (`7/2` is 3.5000,
1153 // `10.0/3` is 3.33333, `7.00/2` is 3.500000), NOT a float. A float /
1154 // double operand — or a STRING one, `'10'/'4'` is 2.5 (double) — makes
1155 // the result a float; a zero divisor is NULL.
1156 if text_operand
1157 || matches!(l, Value::Float(_) | Value::Real(_))
1158 || matches!(r, Value::Float(_) | Value::Real(_))
1159 {
1160 let f = |v: &Value<'_>| -> Option<f64> {
1161 match v {
1162 Value::Float(x) => Some(*x),
1163 Value::Real(x) => Some(f64::from(*x)),
1164 Value::SmallInt(n) => Some(f64::from(*n)),
1165 Value::Int(n) => Some(f64::from(*n)),
1166 #[allow(clippy::cast_precision_loss)]
1167 Value::BigInt(n) => Some(*n as f64),
1168 _ => None,
1169 }
1170 };
1171 let (a, b) = (f(l)?, f(r)?);
1172 return Some(if b == 0.0 {
1173 Value::Null
1174 } else {
1175 Value::Float(a / b)
1176 });
1177 }
1178 let (ls, lsc) = exact_decimal_parts(l)?;
1179 let (rs, rsc) = exact_decimal_parts(r)?;
1180 if rs == 0 {
1181 return Some(Value::Null);
1182 }
1183 let result_scale = u32::from(lsc) + 4;
1184 // result_scaled = round( ls * 10^(rsc + 4) / rs ), half away from zero.
1185 let pow = 10i128.checked_pow(u32::from(rsc) + 4)?;
1186 let num = ls.checked_mul(pow)?;
1187 let q = num / rs;
1188 let rem = num % rs;
1189 let bump = if rem.unsigned_abs() * 2 >= rs.unsigned_abs() {
1190 if (num < 0) == (rs < 0) { 1 } else { -1 }
1191 } else {
1192 0
1193 };
1194 Some(Value::numeric(q + bump, u16::try_from(result_scale).ok()?))
1195}
1196
1197/// The `(scaled, scale)` of an exact integer / NUMERIC value: an integer
1198/// has scale 0. None for a float / non-numeric (they take the float path).
1199fn exact_decimal_parts(v: &Value<'_>) -> Option<(i128, u16)> {
1200 match v {
1201 Value::SmallInt(n) => Some((i128::from(*n), 0)),
1202 Value::Int(n) => Some((i128::from(*n), 0)),
1203 Value::BigInt(n) => Some((i128::from(*n), 0)),
1204 Value::Numeric {
1205 scaled,
1206 scale,
1207 kind: spg_storage::NumericKind::Finite,
1208 } => Some((*scaled, *scale)),
1209 _ => None,
1210 }
1211}
1212
1213/// v7.39 (round 383) — the UNSIGNED 64-bit value a MySQL bitwise operand
1214/// reads as. MySQL's `& | ^ ~ << >>` all work on `BIGINT UNSIGNED`, so an
1215/// operand is its 64-bit two's-complement pattern (a negative integer:
1216/// `-5` is `0xFFFF…FB`), rounded to the nearest integer (a float / numeric:
1217/// `2.9` is 3), its big-endian value (a `0x…` binary string), or its
1218/// leading number (a string). Anything else (an inet, a range, a
1219/// bit-string) returns None so the operator keeps its own meaning.
1220#[allow(clippy::cast_possible_truncation, clippy::cast_sign_loss)]
1221fn mysql_bit_u64(v: &Value<'_>) -> Option<u64> {
1222 match v {
1223 Value::SmallInt(n) => Some(i64::from(*n) as u64),
1224 Value::Int(n) => Some(i64::from(*n) as u64),
1225 Value::BigInt(n) => Some(*n as u64),
1226 Value::Bool(b) => Some(u64::from(*b)),
1227 Value::Float(x) => Some(x.round() as i64 as u64),
1228 Value::Real(x) => Some(f64::from(*x).round() as i64 as u64),
1229 Value::Numeric {
1230 scaled,
1231 scale,
1232 kind: spg_storage::NumericKind::Finite,
1233 } => {
1234 // Round half away from zero to an integer, then take the low
1235 // 64 bits (two's complement) — `~2.9` is `~3`.
1236 if *scale > 38 {
1237 return None;
1238 }
1239 let div = 10i128.pow(u32::from(*scale));
1240 let q = scaled / div;
1241 let rem = scaled % div;
1242 let rounded = if rem.unsigned_abs() * 2 >= div.unsigned_abs() {
1243 q + scaled.signum()
1244 } else {
1245 q
1246 };
1247 Some(rounded as u64)
1248 }
1249 Value::Bytes(b) => mysql_bit_u64(&mysql_bytes_as_number(b)),
1250 Value::Text(s) => mysql_bit_u64(&mysql_number_of(s)),
1251 _ => None,
1252 }
1253}
1254
1255/// A MySQL bitwise result — a `BIGINT UNSIGNED`. It stays a signed
1256/// `BigInt` while it fits (so an integer-context consumer — MAKE_SET
1257/// bits, ELT / SUBSTRING / REPEAT counts — still takes it); a value past
1258/// `i64::MAX` (a set bit 63, e.g. `~5`) has no signed integer type, so it
1259/// becomes a scale-0 NUMERIC, which holds the whole `0..=2^64-1` range and
1260/// renders as the plain integer MySQL prints.
1261fn u64_as_value(n: u64) -> Value<'static> {
1262 match i64::try_from(n) {
1263 Ok(v) => Value::BigInt(v),
1264 Err(_) => Value::numeric(i128::from(n), 0),
1265 }
1266}
1267
1268/// v7.39 (round 383) — the MySQL bitwise operators on UNSIGNED 64-bit
1269/// integers. Returns None when either operand is not number-like (so an
1270/// `inet << int` / `bit(n) & bit(n)` / geometric `#` keeps its own path)
1271/// or the operator is not bitwise. A shift of 64 or more is 0 (MySQL does
1272/// not mask the shift count).
1273pub(crate) fn mysql_bitwise(op: BinOp, l: &Value<'_>, r: &Value<'_>) -> Option<Value<'static>> {
1274 let out = match op {
1275 BinOp::BitAnd => mysql_bit_u64(l)? & mysql_bit_u64(r)?,
1276 BinOp::BitOr => mysql_bit_u64(l)? | mysql_bit_u64(r)?,
1277 BinOp::BitXor => mysql_bit_u64(l)? ^ mysql_bit_u64(r)?,
1278 // `<<` / `>>` share the inet-containment BinOps; a numeric pair is a
1279 // shift, anything else stays inet / range / bit-string.
1280 BinOp::InetContainedBy => {
1281 let (a, n) = (mysql_bit_u64(l)?, mysql_bit_u64(r)?);
1282 if n >= 64 { 0 } else { a << n }
1283 }
1284 BinOp::InetContains => {
1285 let (a, n) = (mysql_bit_u64(l)?, mysql_bit_u64(r)?);
1286 if n >= 64 { 0 } else { a >> n }
1287 }
1288 _ => return None,
1289 };
1290 Some(u64_as_value(out))
1291}
1292
1293/// v7.39 (round 383) — MySQL unary `~x`: the UNSIGNED 64-bit complement
1294/// (`~5` is 18446744073709551610). None for a non-number operand (a
1295/// bit-string / inet / macaddr keeps PG's typed complement).
1296pub(crate) fn mysql_bit_not(v: &Value<'_>) -> Option<Value<'static>> {
1297 Some(u64_as_value(!mysql_bit_u64(v)?))
1298}
1299
1300/// v7.39 (round 390, type-fidelity epic P5) — the inline `SET('a','b',…)`
1301/// variant list an expression's column is declared with, or None. Mirrors
1302/// `expr_enum_type_name` — a bare `Expr::Column` looked up by name.
1303pub(crate) fn expr_set_variants<'e>(
1304 e: &'e Expr,
1305 columns: &'e [ColumnSchema],
1306) -> Option<&'e [String]> {
1307 match e {
1308 Expr::Column(c) => columns
1309 .iter()
1310 .find(|col| col.name == c.name)
1311 .and_then(|col| col.inline_set_variants.as_deref()),
1312 _ => None,
1313 }
1314}
1315
1316/// v7.39 (round 402) — the inline `ENUM('a','b',…)` variant list an
1317/// expression's column is declared with, or None. Like `expr_set_variants`.
1318pub(crate) fn expr_inline_enum_variants<'e>(
1319 e: &'e Expr,
1320 columns: &'e [ColumnSchema],
1321) -> Option<&'e [String]> {
1322 match e {
1323 Expr::Column(c) => columns
1324 .iter()
1325 .find(|col| col.name == c.name)
1326 .and_then(|col| col.inline_enum_variants.as_deref()),
1327 _ => None,
1328 }
1329}
1330
1331/// The 1-based ordinal a stored inline-ENUM text carries in a numeric
1332/// context (`e + 0` is 1 for the first member); the empty string / an
1333/// unknown member is 0 (MySQL's implicit `''` enum error value).
1334pub(crate) fn enum_text_to_ordinal(text: &str, variants: &[String]) -> i64 {
1335 variants
1336 .iter()
1337 .position(|v| v == text)
1338 .map_or(0, |p| p as i64 + 1)
1339}
1340
1341/// The bitmask a stored SET text carries in a numeric context: each
1342/// comma-separated member contributes `1 << its position` in the declared
1343/// variant list (`'a,c'` over `('a','b','c','d')` is 1 | 4 = 5). An empty
1344/// string is 0; an unknown member (should not occur — the write path
1345/// validates) contributes nothing.
1346pub(crate) fn set_text_to_bitmask(text: &str, variants: &[String]) -> i64 {
1347 if text.is_empty() {
1348 return 0;
1349 }
1350 let mut bits = 0i64;
1351 for member in text.split(',') {
1352 if let Some(pos) = variants.iter().position(|v| v == member) {
1353 bits |= 1i64 << pos;
1354 }
1355 }
1356 bits
1357}
1358
1359/// Is this an arithmetic / bitwise operator MySQL evaluates a SET column
1360/// numerically under? (`s + 0`, `s & flag`, …). The comparison operators
1361/// are NOT here — `s = 'a,c'` stays a text compare.
1362pub(crate) const fn is_mysql_numeric_binop(op: BinOp) -> bool {
1363 matches!(
1364 op,
1365 BinOp::Add
1366 | BinOp::Sub
1367 | BinOp::Mul
1368 | BinOp::Div
1369 | BinOp::Mod
1370 | BinOp::BitAnd
1371 | BinOp::BitOr
1372 | BinOp::BitXor
1373 | BinOp::InetContainedBy
1374 | BinOp::InetContains
1375 )
1376}
1377
1378/// v7.39 (round 372) — is `v` a numeric zero (any width / kind)? Used to
1379/// route `x % 0` / `MOD(x, 0)` to NULL under the MySQL dialect.
1380pub(crate) fn value_is_zero(v: &Value<'_>) -> bool {
1381 match v {
1382 Value::SmallInt(n) => *n == 0,
1383 Value::Int(n) => *n == 0,
1384 Value::BigInt(n) => *n == 0,
1385 Value::Float(x) => *x == 0.0,
1386 Value::Real(x) => *x == 0.0,
1387 Value::Numeric { scaled, .. } => *scaled == 0,
1388 _ => false,
1389 }
1390}
1391
1392/// `-'5'` in the MySQL dialect. Out-of-line for the same frame reason.
1393#[inline(never)]
1394fn mysql_negate_text(
1395 op: spg_sql::ast::UnOp,
1396 v: &Value<'static>,
1397) -> Option<Result<Value<'static>, EvalError>> {
1398 match v {
1399 Value::Text(t) => Some(apply_unary(op, mysql_number_of(t))),
1400 _ => None,
1401 }
1402}
1403
1404/// The MySQL reading of a unary operator, or None to let `apply_unary` (the
1405/// PG path) run. Kept out of `eval_expr`'s recursive frame — see the
1406/// round-383 frame cliff. Covers `NOT` on any truth value (round 346),
1407/// `-'str'` numeric negation (round 351), and the unsigned `~` complement
1408/// (round 383).
1409#[inline(never)]
1410fn mysql_unary_arm(
1411 op: spg_sql::ast::UnOp,
1412 v: &Value<'static>,
1413) -> Option<Result<Value<'static>, EvalError>> {
1414 use spg_sql::ast::UnOp;
1415 match op {
1416 // `NOT 5` is 0 — read any non-bool as a truth value; a bool / NULL
1417 // keeps the PG path (still refused there for non-bool).
1418 UnOp::Not if !matches!(v, Value::Bool(_) | Value::Null) => Some(mysql_not(v)),
1419 // `-'5'` is -5, `-'abc'` is 0.
1420 UnOp::Neg => mysql_negate_text(op, v),
1421 // `+ anything` is that thing: measured on MariaDB 11, `+'x'` is
1422 // 'x', `+TRUE` is 1, `+NULL` is NULL. No type check at all, unlike
1423 // PG, which refuses every non-numeric operand.
1424 UnOp::Plus => Some(Ok(v.clone())),
1425 // `~5` is the unsigned 64-bit complement; NULL stays NULL (PG path).
1426 UnOp::BitNot if !matches!(v, Value::Null) => mysql_bit_not(v).map(Ok),
1427 _ => None,
1428 }
1429}
1430
1431/// A date / timestamp string as its temporal value, or `None` when it is
1432/// not one (in which case the operand is left exactly as it was).
1433#[inline(never)]
1434fn text_as_temporal(t: &str) -> Option<Value<'static>> {
1435 parse_timestamp_literal(t)
1436 .map(Value::Timestamp)
1437 .or_else(|| parse_date_literal(t).map(Value::Date))
1438}
1439
1440/// v7.39 (round 620) — an unadorned string literal, which is what PG calls
1441/// `unknown`: a value whose type the context gets to choose. `''::TEXT` is
1442/// not one, and neither is a text column.
1443fn is_unknown_string_literal(e: &Expr) -> bool {
1444 matches!(e, Expr::Literal(spg_sql::ast::Literal::String(_)))
1445}
1446
1447/// v7.39 (round 620) — resolve such a literal to boolean, which is what a
1448/// boolean connective asks of it. An unparseable one is PG's input-syntax
1449/// error (22P02), not a type complaint: `'a' AND true` says
1450/// `invalid input syntax for type boolean: "a"`, exactly as `'a'::BOOLEAN`
1451/// does — same failure, same words, because it is the same coercion.
1452#[inline(never)]
1453fn coerce_unknown_literal_to_bool(e: &Expr) -> Result<Value<'static>, EvalError> {
1454 let Expr::Literal(spg_sql::ast::Literal::String(s)) = e else {
1455 unreachable!("guarded by is_unknown_string_literal")
1456 };
1457 cast::cast_value_in(
1458 Value::Text(s.clone().into()),
1459 spg_sql::ast::CastTarget::Bool,
1460 false,
1461 )
1462}
1463
1464/// v7.39 (round 621) — a literal that is plainly not a boolean, and the PG
1465/// type name for it. `NULL` and a bare string literal are deliberately absent:
1466/// neither carries a type of its own, and a boolean connective is a context
1467/// that gives them one.
1468fn non_boolean_literal_type(e: &Expr) -> Option<&'static str> {
1469 use spg_sql::ast::Literal as L;
1470 match e {
1471 Expr::Literal(L::Integer(_)) => Some("integer"),
1472 Expr::Literal(L::Float(_)) => Some("double precision"),
1473 Expr::Literal(L::Numeric { .. } | L::NumericBig(_)) => Some("numeric"),
1474 _ => None,
1475 }
1476}
1477
1478/// v7.39 (round 621) — `AND` / `OR`, evaluated the way PG evaluates them.
1479///
1480/// Round 620 handled the unknown literal here; round 621 adds the part that
1481/// makes `WHERE x <> 0 AND 1/x > 0` work at all. SPG evaluated both sides
1482/// always, so the guard idiom — the whole reason that predicate is written
1483/// that way — raised on the very rows the guard exists to exclude. Measured
1484/// against PG: `false AND (1/0 = 0)` answers `f`, `true OR (1/0 = 0)` answers
1485/// `t`, and a filter guarded that way returns its rows.
1486///
1487/// PG affords that AND still refuses `false AND 1`, because the two happen at
1488/// different times: the operand types are checked during ANALYSIS, before any
1489/// evaluation, and the short circuit is a RUN-TIME decision. Both parts are
1490/// here — the right-hand operand's type is read statically (it is the side
1491/// that may go unevaluated), and only a type that is definitively known and
1492/// definitively not boolean is refused. An unknown type is left alone, so a
1493/// shape the describer cannot type keeps the old behaviour rather than
1494/// earning a spurious error.
1495///
1496/// Order is PG's too, and it is strictly left-first: `(1/0 = 0) AND false`
1497/// raises on both, because the left is evaluated before anything can decide
1498/// that it did not need to be.
1499///
1500/// Out-of-line so it costs `eval_expr` no frame (the round-305 frame cliff).
1501#[inline(never)]
1502fn eval_connective(
1503 lhs: &Expr,
1504 op: BinOp,
1505 rhs: &Expr,
1506 row: &Row<'static>,
1507 ctx: &EvalContext<'_>,
1508) -> Result<Value<'static>, EvalError> {
1509 let side = |e: &Expr| -> Result<Value<'static>, EvalError> {
1510 if is_unknown_string_literal(e) {
1511 coerce_unknown_literal_to_bool(e)
1512 } else {
1513 eval_expr(e, row, ctx)
1514 }
1515 };
1516 let l = side(lhs)?;
1517 // The analysis-time half: refuse a right-hand operand that is plainly not
1518 // boolean, whether or not the short circuit would reach it.
1519 //
1520 // Only a LITERAL is read this way. The first cut asked
1521 // `describe_expr_type` for any expression's type, and it answers
1522 // confidently and wrongly for shapes that matter here — `NULL` comes back
1523 // as text, so `true AND NULL` earned a type error; and a MATCH … AGAINST
1524 // folds internally into an OR over tsvector operands, so full-text search
1525 // stopped working. Three existing pins caught all of it. A literal cannot
1526 // be misread, and it is what PG's own refusals in this area are about.
1527 if let Some(ty) = non_boolean_literal_type(rhs) {
1528 return Err(EvalError::TypeMismatch {
1529 detail: alloc::format!(
1530 "argument of {} must be type boolean, not type {ty}",
1531 if matches!(op, BinOp::And) {
1532 "AND"
1533 } else {
1534 "OR"
1535 },
1536 ),
1537 });
1538 }
1539 // Resolving an unknown literal belongs to the same half — it is a
1540 // coercion PG performs while analysing, so `false AND 'a'` says
1541 // `invalid input syntax for type boolean: "a"` rather than answering `f`.
1542 let rhs_resolved = if is_unknown_string_literal(rhs) {
1543 Some(coerce_unknown_literal_to_bool(rhs)?)
1544 } else {
1545 None
1546 };
1547 // The run-time half.
1548 match (op, &l) {
1549 (BinOp::And, Value::Bool(false)) => return Ok(Value::Bool(false)),
1550 (BinOp::Or, Value::Bool(true)) => return Ok(Value::Bool(true)),
1551 _ => {}
1552 }
1553 let r = match rhs_resolved {
1554 Some(v) => v,
1555 None => side(rhs)?,
1556 };
1557 if matches!(op, BinOp::And) {
1558 and_3vl(l, r)
1559 } else {
1560 apply_binary(op, l, r)
1561 }
1562}
1563
1564/// v7.39 (round 346, M1) — the MySQL reading of `AND` / `OR`, out-of-line
1565/// so it costs `eval_expr` no frame (see the round-305 frame cliff).
1566#[inline(never)]
1567fn eval_mysql_connective(
1568 lhs: &Expr,
1569 op: BinOp,
1570 rhs: &Expr,
1571 row: &Row<'static>,
1572 ctx: &EvalContext<'_>,
1573) -> Result<Value<'static>, EvalError> {
1574 let l = as_mysql_truth(eval_expr(lhs, row, ctx)?)?;
1575 let r = as_mysql_truth(eval_expr(rhs, row, ctx)?)?;
1576 apply_mysql_connective(op, l, r)
1577}
1578
1579/// v7.39 (round 407) — apply a MySQL logical connective (`AND` / `OR` /
1580/// `XOR`) to two operands already reduced to truth values (`Bool` or
1581/// `Null`). AND / OR reuse the dialect-blind `apply_binary`; `XOR` is
1582/// MySQL-only (no `apply_binary` arm) and computed here: NULL on either
1583/// side yields NULL, otherwise the exclusive-or of the two truth values.
1584pub(crate) fn apply_mysql_connective(
1585 op: BinOp,
1586 l: Value<'static>,
1587 r: Value<'static>,
1588) -> Result<Value<'static>, EvalError> {
1589 if op == BinOp::LogicalXor {
1590 return Ok(match (&l, &r) {
1591 (Value::Bool(a), Value::Bool(b)) => Value::Bool(a != b),
1592 _ => Value::Null,
1593 });
1594 }
1595 apply_binary(op, l, r)
1596}
1597
1598#[inline(never)]
1599pub(crate) fn as_mysql_truth(v: Value<'static>) -> Result<Value<'static>, EvalError> {
1600 Ok(match v {
1601 Value::Null => Value::Null,
1602 other => Value::Bool(predicate_is_true(&other, "AND", true)?),
1603 })
1604}
1605
1606/// The MySQL reading of `NOT`, likewise out-of-line.
1607#[inline(never)]
1608fn mysql_not(v: &Value<'_>) -> Result<Value<'static>, EvalError> {
1609 Ok(Value::Bool(!predicate_is_true(v, "NOT", true)?))
1610}
1611
1612/// v7.39 (round 346, M1) — is this value TRUE, in a position that wants a
1613/// truth value (WHERE / CASE WHEN / NOT / AND / OR / HAVING / ON)?
1614///
1615/// The engine used to write `matches!(v, Value::Bool(true))` at every such
1616/// position, so anything that was not already a boolean silently read as
1617/// FALSE. `SELECT CASE WHEN 1 THEN 'a' END` answered NULL and — far worse —
1618/// `SELECT … WHERE 1` returned **no rows at all**. Neither dialect does
1619/// that: MariaDB 11 takes any non-zero number as true, and PG 18.4 raises
1620/// `argument of WHERE must be type boolean, not type integer`.
1621///
1622/// NULL is not true (three-valued logic) and is not an error in either.
1623pub(crate) fn predicate_is_true(v: &Value<'_>, kw: &str, mysql: bool) -> Result<bool, EvalError> {
1624 match v {
1625 Value::Bool(b) => Ok(*b),
1626 Value::Null => Ok(false),
1627 _ if mysql => Ok(mysql_truthy(v)),
1628 // PG resolves a bare literal in this position through boolean
1629 // INPUT, so `CASE WHEN 'true'` is legal and `'abc'` is not.
1630 Value::Text(t) => match crate::eval::cast::cast_value(
1631 Value::text(t.to_string()),
1632 spg_sql::ast::CastTarget::Bool,
1633 )? {
1634 Value::Bool(b) => Ok(b),
1635 _ => Ok(false),
1636 },
1637 other => Err(EvalError::TypeMismatch {
1638 detail: alloc::format!(
1639 "argument of {kw} must be type boolean, not type {}",
1640 crate::eval::strings::pg_typeof_name(other)
1641 ),
1642 }),
1643 }
1644}
1645
1646/// MariaDB 11's reading, measured: a number is true when it is not zero
1647/// (`-1` and `0.5` are both true); a string contributes its LEADING
1648/// number, so `'1abc'` is true while `'abc'` and `''` are false.
1649fn mysql_truthy(v: &Value<'_>) -> bool {
1650 match v {
1651 Value::Bool(b) => *b,
1652 Value::Null => false,
1653 Value::SmallInt(n) => *n != 0,
1654 Value::Int(n) => *n != 0,
1655 Value::BigInt(n) => *n != 0,
1656 Value::Float(f) => *f != 0.0,
1657 Value::Real(f) => *f != 0.0,
1658 Value::Numeric { scaled, .. } => *scaled != 0,
1659 Value::Text(t) => mysql_leading_number(t) != 0.0,
1660 Value::BpChar(t) => mysql_leading_number(t) != 0.0,
1661 // Everything else converts to a non-zero number in MariaDB (a
1662 // DATE reads as its YYYYMMDD digits, for one).
1663 _ => true,
1664 }
1665}
1666
1667/// The leading numeric prefix of a string, MySQL-style: `'1abc'` is 1,
1668/// `'abc'` and `''` are 0.
1669#[inline(never)]
1670pub(crate) fn mysql_leading_number(s: &str) -> f64 {
1671 let t = s.trim_start();
1672 let mut end = 0usize;
1673 let mut seen_dot = false;
1674 let mut seen_digit = false;
1675 // v7.39 (round 351, M11) — the exponent form counts: MariaDB reads
1676 // `'1e3'` as 1000 and `'1.5e2'` as 150 (measured). A trailing `e`
1677 // with no digits after it is not part of the number (`'1e'` is 1).
1678 let mut seen_exp = false;
1679 let mut exp_at = 0usize;
1680 for (i, c) in t.char_indices() {
1681 match c {
1682 '-' | '+' if i == 0 => {}
1683 '-' | '+' if seen_exp && i == exp_at + 1 => {}
1684 '0'..='9' => seen_digit = true,
1685 '.' if !seen_dot && !seen_exp => seen_dot = true,
1686 'e' | 'E' if seen_digit && !seen_exp => {
1687 seen_exp = true;
1688 exp_at = i;
1689 }
1690 _ => break,
1691 }
1692 end = i + c.len_utf8();
1693 }
1694 if !seen_digit {
1695 return 0.0;
1696 }
1697 // Trim an exponent that never got its digits.
1698 let mut text = &t[..end];
1699 while !text.is_empty() && text.parse::<f64>().is_err() {
1700 text = &text[..text.len() - 1];
1701 }
1702 text.parse::<f64>().unwrap_or(0.0)
1703}
1704
1705/// v7.39 (read01 ruleutils.c) — resolve a relation name to its synthetic
1706/// oid: user tables in the 16384+ band (table_names order), views at
1707/// 32768+, and the synthesised system catalogs at their REAL PG oids.
1708/// `None` when the name is unknown (the caller keeps the legacy text
1709/// behaviour so `'anything'::regclass::text` still round-trips).
1710pub(crate) fn regclass_name_to_oid(cat: &spg_storage::Catalog, bare: &str) -> Option<i64> {
1711 // v7.39 (round 337, V62) — an INDEX and a SEQUENCE are relations too:
1712 // both have a `pg_class` row, so both answer to `::regclass` in PG.
1713 // v7.39 (round 338, V64) — and the bands live in ONE allocator now,
1714 // shared with the catalog synths, so `pg_class.oid = 'x'::regclass`
1715 // holds for every kind rather than only for tables.
1716 if let Some(oid) = crate::system_catalog::relation_oid(cat, bare) {
1717 return Some(oid);
1718 }
1719 Some(match bare {
1720 "pg_type" => 1247,
1721 "pg_attribute" => 1249,
1722 "pg_proc" => 1255,
1723 "pg_class" => 1259,
1724 "pg_database" => 1262,
1725 "pg_constraint" => 2606,
1726 "pg_index" => 2610,
1727 "pg_namespace" => 2615,
1728 // v7.39 (round 650) — the text-search catalogs. This list is a
1729 // hand-kept subset of `CATALOG_RELATIONS`, which is why adding a
1730 // catalog there was not enough for `'pg_ts_config'::regclass`.
1731 "pg_ts_config" => 3602,
1732 "pg_ts_config_map" => 3603,
1733 "pg_ts_dict" => 3600,
1734 "pg_ts_parser" => 3601,
1735 "pg_ts_template" => 3764,
1736 // 7.38.1 S5.1 — stop hand-copying: anything CATALOG_RELATIONS
1737 // publishes resolves here too (pg_dump's dependency pass casts
1738 // 'pg_extension' / 'pg_amop' / 'pg_opfamily'::regclass).
1739 other => {
1740 return crate::system_catalog::CATALOG_RELATIONS
1741 .iter()
1742 .find(|(n, _)| other.eq_ignore_ascii_case(n))
1743 .map(|(_, oid)| *oid);
1744 }
1745 })
1746}
1747
1748/// v7.39 (round 263) — crate-visible wrapper so the write path can
1749/// relabel + coerce a value into a composite column's declared type.
1750pub(crate) fn apply_composite_cast_pub(
1751 v: Value<'static>,
1752 comp: &spg_storage::CompositeDef,
1753 cat: Option<&spg_storage::Catalog>,
1754) -> Result<Value<'static>, EvalError> {
1755 apply_composite_cast_in(v, comp, cat)
1756}
1757
1758/// v7.39 (round 264) — resolve one field's value, recursing when the
1759/// field is itself a COMPOSITE. Without this a nested field kept the
1760/// inner record's TEXT rendering, so `(x).inner.street` errored and
1761/// `row_to_json` nested a string rather than an object.
1762fn coerce_composite_field(
1763 val: Value<'static>,
1764 fname: &str,
1765 fty: spg_storage::DataType,
1766 user_ty: Option<&str>,
1767 cat: Option<&spg_storage::Catalog>,
1768) -> Result<Value<'static>, EvalError> {
1769 if matches!(val, Value::Null) {
1770 return Ok(val);
1771 }
1772 if let Some(tn) = user_ty
1773 && let Some(inner) = cat.and_then(|c| c.composite_types().get(tn))
1774 {
1775 return apply_composite_cast_in(val, inner, cat);
1776 }
1777 crate::conversions::coerce_value(val, fty, fname, 0).map_err(|e| EvalError::TypeMismatch {
1778 detail: alloc::format!("{e}"),
1779 })
1780}
1781
1782fn apply_composite_cast(
1783 v: Value<'static>,
1784 comp: &spg_storage::CompositeDef,
1785) -> Result<Value<'static>, EvalError> {
1786 apply_composite_cast_in(v, comp, None)
1787}
1788
1789fn apply_composite_cast_in(
1790 v: Value<'static>,
1791 comp: &spg_storage::CompositeDef,
1792 cat: Option<&spg_storage::Catalog>,
1793) -> Result<Value<'static>, EvalError> {
1794 match v {
1795 Value::Null => Ok(Value::Null),
1796 Value::Composite(fields) => {
1797 if fields.len() != comp.fields.len() {
1798 // PG reports the SHAPE mismatch as a plain cast refusal.
1799 return Err(EvalError::TypeMismatch {
1800 detail: alloc::format!("cannot cast type record to {}", comp.name),
1801 });
1802 }
1803 // v7.39 (round 263) — relabel AND coerce: this branch only
1804 // renamed the fields, so `ROW('x','notanint')::addr` kept the
1805 // text in an int field and PG's input error never fired.
1806 let mut out: alloc::vec::Vec<(alloc::string::String, Value<'static>)> =
1807 alloc::vec::Vec::with_capacity(comp.fields.len());
1808 for (i, ((name, fty), (_, val))) in comp.fields.iter().zip(fields).enumerate() {
1809 let ut = comp.field_user_types.get(i).and_then(Option::as_deref);
1810 let coerced = coerce_composite_field(val, name, *fty, ut, cat)?;
1811 out.push((name.clone(), coerced));
1812 }
1813 Ok(Value::Composite(out))
1814 }
1815 Value::Text(s) => {
1816 let raw = parse_record_text(s.as_ref()).ok_or_else(|| EvalError::TypeMismatch {
1817 detail: alloc::format!("malformed record literal: \"{s}\""),
1818 })?;
1819 if raw.len() != comp.fields.len() {
1820 return Err(EvalError::TypeMismatch {
1821 detail: alloc::format!("malformed record literal: \"{s}\""),
1822 });
1823 }
1824 let mut out: alloc::vec::Vec<(alloc::string::String, Value<'static>)> =
1825 alloc::vec::Vec::with_capacity(raw.len());
1826 for (i, ((fname, fty), field_text)) in comp.fields.iter().zip(raw).enumerate() {
1827 let ut = comp.field_user_types.get(i).and_then(Option::as_deref);
1828 let val = match field_text {
1829 None => Value::Null,
1830 Some(t) => coerce_composite_field(Value::text(t), fname, *fty, ut, cat)?,
1831 };
1832 out.push((fname.clone(), val));
1833 }
1834 Ok(Value::Composite(out))
1835 }
1836 other => Err(EvalError::TypeMismatch {
1837 detail: alloc::format!(
1838 "cannot cast {} to composite type \"{}\"",
1839 crate::conversions::pg_type_name_for_error_opt(other.data_type()),
1840 comp.name
1841 ),
1842 }),
1843 }
1844}
1845
1846/// Split PG's record text `(f1,f2,...)` into per-field raw strings
1847/// (None = empty field = NULL). Double quotes wrap fields containing
1848/// metacharacters; `""` inside is a literal quote; a backslash escapes
1849/// the next character.
1850fn parse_record_text(s: &str) -> Option<alloc::vec::Vec<Option<alloc::string::String>>> {
1851 let t = s.trim();
1852 let inner = t.strip_prefix('(')?.strip_suffix(')')?;
1853 let mut out: alloc::vec::Vec<Option<alloc::string::String>> = alloc::vec::Vec::new();
1854 let chars: alloc::vec::Vec<char> = inner.chars().collect();
1855 let mut field = alloc::string::String::new();
1856 let mut quoted_seen = false;
1857 let mut i = 0usize;
1858 let mut in_quotes = false;
1859 loop {
1860 if i >= chars.len() {
1861 if in_quotes {
1862 return None;
1863 }
1864 out.push(if field.is_empty() && !quoted_seen {
1865 None
1866 } else {
1867 Some(field.clone())
1868 });
1869 break;
1870 }
1871 let c = chars[i];
1872 if in_quotes {
1873 match c {
1874 '"' if chars.get(i + 1) == Some(&'"') => {
1875 field.push('"');
1876 i += 2;
1877 }
1878 '"' => {
1879 in_quotes = false;
1880 i += 1;
1881 }
1882 '\\' => {
1883 field.push(*chars.get(i + 1)?);
1884 i += 2;
1885 }
1886 _ => {
1887 field.push(c);
1888 i += 1;
1889 }
1890 }
1891 } else {
1892 match c {
1893 '"' => {
1894 in_quotes = true;
1895 quoted_seen = true;
1896 i += 1;
1897 }
1898 ',' => {
1899 out.push(if field.is_empty() && !quoted_seen {
1900 None
1901 } else {
1902 Some(core::mem::take(&mut field))
1903 });
1904 quoted_seen = false;
1905 i += 1;
1906 }
1907 '\\' => {
1908 field.push(*chars.get(i + 1)?);
1909 i += 2;
1910 }
1911 _ => {
1912 field.push(c);
1913 i += 1;
1914 }
1915 }
1916 }
1917 }
1918 Some(out)
1919}
1920
1921fn apply_enum_cast<'a>(
1922 v: Value<'a>,
1923 en: &spg_storage::EnumDef,
1924 name: &str,
1925) -> Result<Value<'a>, EvalError> {
1926 match &v {
1927 Value::Null => Ok(v),
1928 Value::Text(s) => {
1929 if en.labels.iter().any(|l| l.as_str() == s.as_ref()) {
1930 Ok(v)
1931 } else {
1932 Err(EvalError::TypeMismatch {
1933 detail: alloc::format!("invalid input value for enum {name}: {s:?}"),
1934 })
1935 }
1936 }
1937 other => Err(EvalError::TypeMismatch {
1938 detail: alloc::format!(
1939 "cannot cast {} to enum {name}",
1940 crate::conversions::pg_type_name_for_error_opt(other.data_type())
1941 ),
1942 }),
1943 }
1944}
1945
1946/// v7.39 (read01 utils/adt, enum.c) — enum_first / enum_last /
1947/// enum_range resolved from the argument's STATIC enum type (an explicit
1948/// `::enumtype` cast or a column's `ColumnSchema.user_enum_type`) over the
1949/// catalog's member order. Returns None when no argument names a known
1950/// enum, letting the generic function path produce its usual error.
1951/// Out-of-line (`inline(never)`) so the sizable locals don't land in
1952/// `eval_expr`'s recursion frame.
1953/// Enum-ness lives outside the DataType lattice: the witness for "this
1954/// expression is enum-typed" is an explicit `::enumtype` cast or a column
1955/// whose `ColumnSchema.user_enum_type` is set.
1956/// v7.39 (round 258) — crate-visible wrapper so the projection builder
1957/// can keep an expression's enum identity (see `select.rs`).
1958pub(crate) fn expr_enum_type_name_pub<'e>(
1959 e: &'e Expr,
1960 columns: &'e [ColumnSchema],
1961) -> Option<&'e str> {
1962 expr_enum_type_name(e, columns)
1963}
1964
1965/// v7.39 (round 425) — the fractional-seconds precision a projected
1966/// expression should RENDER with: the widest declared precision among the
1967/// MySQL temporal columns it reads. `MAX(d3)` and `d3 + INTERVAL 1 SECOND`
1968/// both keep `d3`'s three digits, as MariaDB does. `None` when the
1969/// expression touches no such column, which leaves PG rendering untouched.
1970///
1971/// Residual (recorded, not modelled): MariaDB also WIDENS the precision from
1972/// some operands — `DATE_ADD(d3, INTERVAL 1 MICROSECOND)` prints six digits
1973/// there. Taking the max over referenced columns covers the common shapes
1974/// and never narrows below the source column.
1975pub(crate) fn expr_mysql_fsp(e: &Expr, columns: &[ColumnSchema]) -> Option<u8> {
1976 fn walk(e: &Expr, columns: &[ColumnSchema], best: &mut Option<u8>) {
1977 match e {
1978 Expr::Column(c) => {
1979 if let Some(f) = columns
1980 .iter()
1981 .find(|col| col.name == c.name)
1982 .and_then(|col| col.mysql_fsp)
1983 {
1984 *best = Some(best.map_or(f, |b: u8| b.max(f)));
1985 }
1986 }
1987 Expr::Binary { lhs, rhs, .. } => {
1988 walk(lhs, columns, best);
1989 walk(rhs, columns, best);
1990 }
1991 Expr::Unary { expr, .. } | Expr::Cast { expr, .. } => walk(expr, columns, best),
1992 Expr::FunctionCall { args, .. } => {
1993 for a in args {
1994 walk(a, columns, best);
1995 }
1996 }
1997 Expr::Case {
1998 operand,
1999 branches,
2000 else_branch,
2001 } => {
2002 if let Some(o) = operand.as_deref() {
2003 walk(o, columns, best);
2004 }
2005 for (w, t) in branches {
2006 walk(w, columns, best);
2007 walk(t, columns, best);
2008 }
2009 if let Some(el) = else_branch.as_deref() {
2010 walk(el, columns, best);
2011 }
2012 }
2013 _ => {}
2014 }
2015 }
2016 let mut best = None;
2017 walk(e, columns, &mut best);
2018 best
2019}
2020
2021/// v7.39 (round 467) — is this expression MySQL-UNSIGNED?
2022///
2023/// MySQL decides unsignedness statically, from the expression's type, not
2024/// from the value it happens to produce. Measured on MariaDB 11: `SUM(a) -
2025/// 100` answers -99 even though `a` is `INT UNSIGNED`, because SUM's result
2026/// type is not unsigned; `a - 5` on the same column raises 1690. So this
2027/// walks the expression the way MySQL's type resolution does.
2028///
2029/// A cast names its target `unsigned` (the parser lowercases MySQL's
2030/// `CAST(x AS UNSIGNED)` into `CastTarget::Named`). Arithmetic is unsigned
2031/// when EITHER operand is — that is MySQL's rule, and it is why `1 - b`
2032/// raises while `5 - a` does not: both are unsigned expressions, but only
2033/// the first has a negative result.
2034///
2035/// Deliberately NOT unsigned: unary minus (MariaDB answers -1 for
2036/// `-CAST(1 AS UNSIGNED)`), and every function result including the
2037/// aggregates. Both measured.
2038pub(crate) fn expr_is_mysql_unsigned(e: &Expr, columns: &[ColumnSchema]) -> bool {
2039 match e {
2040 Expr::Column(c) => columns
2041 .iter()
2042 .find(|col| col.name == c.name)
2043 .is_some_and(|col| col.is_unsigned),
2044 Expr::Cast {
2045 target: CastTarget::Named(n),
2046 ..
2047 } => n.eq_ignore_ascii_case("unsigned"),
2048 Expr::Binary {
2049 lhs,
2050 op: BinOp::Add | BinOp::Sub | BinOp::Mul,
2051 rhs,
2052 } => expr_is_mysql_unsigned(lhs, columns) || expr_is_mysql_unsigned(rhs, columns),
2053 _ => false,
2054 }
2055}
2056
2057/// v7.39 (round 467) — MySQL arithmetic over an UNSIGNED operand, with
2058/// MySQL's range check.
2059///
2060/// `INT UNSIGNED` columns holding 1 and 5 made `a - b` answer **-4** in a
2061/// MySQL session. MariaDB raises `ERROR 1690 (22003): BIGINT UNSIGNED value
2062/// is out of range`. A negative answer where the server promises a
2063/// non-negative one is the kind of thing an application stores back into
2064/// the same column, so it was silent and wrong in the worst direction.
2065///
2066/// The check runs in i128 so the subtraction that underflows is observed
2067/// rather than wrapped, and it only fires when the expression is unsigned
2068/// AND both operands are integers — a NUMERIC or float operand takes the
2069/// ordinary path, as it does in MySQL.
2070///
2071/// `#[inline(never)]`: this is called from the recursive evaluator's
2072/// hottest frame, which already sits against the stack guard.
2073#[inline(never)]
2074fn apply_binary_mysql_unsigned(
2075 op: BinOp,
2076 lhs: &Expr,
2077 rhs: &Expr,
2078 l: Value<'static>,
2079 r: Value<'static>,
2080 ctx: &EvalContext,
2081) -> Result<Value<'static>, EvalError> {
2082 if ctx.mysql_dialect
2083 && matches!(op, BinOp::Add | BinOp::Sub | BinOp::Mul)
2084 && let Some(a) = mysql_int_operand(&l)
2085 && let Some(b) = mysql_int_operand(&r)
2086 && (expr_is_mysql_unsigned(lhs, ctx.columns) || expr_is_mysql_unsigned(rhs, ctx.columns))
2087 {
2088 let out = match op {
2089 BinOp::Add => a.checked_add(b),
2090 BinOp::Sub => a.checked_sub(b),
2091 _ => a.checked_mul(b),
2092 };
2093 let in_range = out.is_some_and(|v| (0..=i128::from(u64::MAX)).contains(&v));
2094 if !in_range {
2095 // MariaDB names the offending expression in the message, with
2096 // minimal parentheses — `a * 0 - 1`, not `((a * 0) - 1)`.
2097 // `pretty_expr` is the deparser that already produces that
2098 // shape. Residual, recorded rather than faked: MariaDB writes
2099 // its columns fully qualified in backticks
2100 // (`db`.`tbl`.`col`), and the database name is not something
2101 // the evaluation context carries.
2102 return Err(EvalError::TypeMismatch {
2103 detail: alloc::format!(
2104 "BIGINT UNSIGNED value is out of range in '{}'",
2105 spg_sql::ast::pretty_expr_mysql(&Expr::Binary {
2106 lhs: alloc::boxed::Box::new(lhs.clone()),
2107 op,
2108 rhs: alloc::boxed::Box::new(rhs.clone()),
2109 })
2110 ),
2111 });
2112 }
2113 }
2114 let out = apply_binary_in(op, l, r, ctx.mysql_dialect);
2115 // v7.39 (round 503) — MariaDB answers NULL for division / modulo by
2116 // zero; SPG raised.
2117 //
2118 // Measured against MariaDB 11: `SELECT 1/0`, `SELECT 5 DIV 0` and
2119 // `SELECT 5 % 0` are all NULL — and they are NULL under the DEFAULT
2120 // sql_mode too, which contains `ERROR_FOR_DIVISION_BY_ZERO`. That flag
2121 // governs WRITES, not the expression: the division evaluates to NULL,
2122 // and a strict-mode INSERT of that result is what raises 1365.
2123 //
2124 // The rule is therefore the DIALECT's, not the mode's: in a MySQL
2125 // session the expression is NULL. It is deliberately not gated on
2126 // `mysql_strict` — an earlier cut of this was, and the gate fired only
2127 // because the probe's context happens to carry no engine, which a
2128 // later round attaching one would have silently reversed.
2129 //
2130 // RESIDUAL, recorded rather than faked: MariaDB's strict-mode INSERT
2131 // of a division by zero raises 1365 and its non-strict INSERT stores
2132 // NULL. SPG's INSERT path evaluates elsewhere and still raises, so it
2133 // matches strict and diverges from non-strict. Closing that needs the
2134 // expression to know it is in a write, which nothing here carries.
2135 if ctx.mysql_dialect && matches!(out, Err(EvalError::DivisionByZero)) {
2136 return Ok(Value::Null);
2137 }
2138 out
2139}
2140
2141/// The integer an operand contributes to the unsigned range check, or
2142/// `None` when it is not an integer at all (NULL, text, NUMERIC, float).
2143fn mysql_int_operand(v: &Value<'_>) -> Option<i128> {
2144 match v {
2145 Value::SmallInt(n) => Some(i128::from(*n)),
2146 Value::Int(n) => Some(i128::from(*n)),
2147 Value::BigInt(n) => Some(i128::from(*n)),
2148 // v7.39 (round 471) — a BIGINT UNSIGNED cell is stored as Numeric
2149 // with scale 0, so the range check has to see it as the integer it
2150 // is. Without this arm the column's own type moved it out of reach
2151 // of round 467's guard and `c - 5` went back to answering -4.
2152 Value::Numeric { scaled, scale, .. } if *scale == 0 => Some(*scaled),
2153 _ => None,
2154 }
2155}
2156
2157fn expr_enum_type_name<'e>(e: &'e Expr, columns: &'e [ColumnSchema]) -> Option<&'e str> {
2158 match e {
2159 Expr::Cast {
2160 target: CastTarget::Named(n),
2161 ..
2162 } => Some(n.as_str()),
2163 Expr::Column(c) => columns
2164 .iter()
2165 .find(|col| col.name == c.name)
2166 // v7.39 (round 259) — a DOMAIN column carries its name in its
2167 // own field; both are "the user type this column is declared
2168 // as", which is what the callers (enum-order comparison,
2169 // pg_typeof) want. Callers gate on the catalog, so a name that
2170 // is one kind never resolves as the other.
2171 .and_then(|col| {
2172 col.user_enum_type
2173 .as_deref()
2174 .or(col.user_domain_type.as_deref())
2175 }),
2176 _ => None,
2177 }
2178}
2179
2180/// v7.39 (enum order knife) — the member-label list for an enum-typed
2181/// expression, or None when the expression carries no enum witness or the
2182/// name is not a known enum. The returned slice borrows the catalog.
2183pub(crate) fn expr_enum_labels<'c>(
2184 e: &Expr,
2185 columns: &[ColumnSchema],
2186 catalog: Option<&'c spg_storage::Catalog>,
2187) -> Option<&'c [String]> {
2188 let name = expr_enum_type_name(e, columns)?;
2189 catalog
2190 .and_then(|cat| cat.enum_types().get(name))
2191 .map(|en| en.labels.as_slice())
2192}
2193
2194/// v7.39 (enum order knife) — compare two enum labels by member order.
2195/// None when either side is not Text or not a member (caller falls back to
2196/// the generic comparison, so a stray value never panics or misorders
2197/// silently differently from before).
2198pub(crate) fn enum_ord_cmp(
2199 labels: &[String],
2200 a: &Value<'_>,
2201 b: &Value<'_>,
2202) -> Option<core::cmp::Ordering> {
2203 let pos = |v: &Value<'_>| -> Option<usize> {
2204 match v {
2205 Value::Text(s) => labels.iter().position(|l| l.as_str() == s.as_ref()),
2206 _ => None,
2207 }
2208 };
2209 Some(pos(a)?.cmp(&pos(b)?))
2210}
2211
2212/// v7.39 (enum order knife) — Binary-comparison hook: when either side's
2213/// static type witnesses an enum and both runtime values are member labels,
2214/// compare by member order (PG's enumsortorder semantics). Out-of-line to
2215/// keep `eval_expr`'s recursion frame small.
2216#[inline(never)]
2217fn enum_compare_hook(
2218 op: BinOp,
2219 lhs: &Expr,
2220 rhs: &Expr,
2221 l: &Value<'_>,
2222 r: &Value<'_>,
2223 ctx: &EvalContext<'_>,
2224) -> Option<Result<Value<'static>, EvalError>> {
2225 let cat = ctx.catalog?;
2226 if cat.enum_types().is_empty() {
2227 return None;
2228 }
2229 let labels = expr_enum_labels(lhs, ctx.columns, ctx.catalog)
2230 .or_else(|| expr_enum_labels(rhs, ctx.columns, ctx.catalog))?;
2231 let ord = enum_ord_cmp(labels, l, r)?;
2232 let b = match op {
2233 BinOp::Eq => ord == core::cmp::Ordering::Equal,
2234 BinOp::NotEq => ord != core::cmp::Ordering::Equal,
2235 BinOp::Lt => ord == core::cmp::Ordering::Less,
2236 BinOp::LtEq => ord != core::cmp::Ordering::Greater,
2237 BinOp::Gt => ord == core::cmp::Ordering::Greater,
2238 BinOp::GtEq => ord != core::cmp::Ordering::Less,
2239 _ => return None,
2240 };
2241 Some(Ok(Value::Bool(b)))
2242}
2243
2244/// v7.39 (round 693) — Binary-comparison hook for a declared collation, the
2245/// last shape F36 left open: `loc BETWEEN 'a' AND 'd'` returns a different
2246/// ROW SET under `en_US.utf8` than under byte order, not merely a different
2247/// order.
2248///
2249/// It sits beside [`enum_compare_hook`] because it is the same kind of fact
2250/// — something about the operand COLUMNS that `compare` cannot look up from
2251/// two values — and takes the same two protections: `#[inline(never)]`, so
2252/// `eval_expr`'s recursion frame does not grow (the comment at the call site
2253/// records a fourth `||` there tipping the 768 KiB guard on its own), and
2254/// the caller's Text/Text gate, so no integer comparison reaches it.
2255///
2256/// EQUALITY is deliberately not handled. PG18's `en_US.utf8` is
2257/// deterministic, so `=`, `<>`, `LIKE`, `IN` and `count(DISTINCT …)` give
2258/// byte-equality's answer — measured, all five. Only the ordering operators
2259/// change, and `least`/`greatest` follow them through their own comparator.
2260#[inline(never)]
2261fn collate_compare_hook(
2262 op: BinOp,
2263 lhs: &Expr,
2264 rhs: &Expr,
2265 l: &Value<'_>,
2266 r: &Value<'_>,
2267 ctx: &EvalContext<'_>,
2268) -> Option<Result<Value<'static>, EvalError>> {
2269 if !matches!(op, BinOp::Lt | BinOp::LtEq | BinOp::Gt | BinOp::GtEq) {
2270 return None;
2271 }
2272 let (Value::Text(a), Value::Text(b)) = (l, r) else {
2273 return None;
2274 };
2275 let resolve = |c: &spg_sql::ast::ColumnName| -> Option<alloc::string::String> {
2276 let pos = find_column_pos(c, ctx)?;
2277 ctx.columns.get(pos)?.collation_name.clone()
2278 };
2279 let derived = crate::collate_derive::derive(lhs, &resolve)
2280 .combine_pub(crate::collate_derive::derive(rhs, &resolve));
2281 if let Some((x, y)) = derived.conflict() {
2282 return Some(Err(EvalError::TypeMismatch {
2283 detail: alloc::format!(
2284 "collation mismatch between implicit collations \"{x}\" and \"{y}\""
2285 ),
2286 }));
2287 }
2288 // v7.38.18 (S2) — and the DATABASE's collation when neither side
2289 // declares one, which is what an undeclared text column is compared
2290 // under. `C` is byte order and resolves to `None`, so a database
2291 // that never asked for a locale takes the same path it always did.
2292 //
2293 // Ordering only. Whether a comparison PADS is decided in
2294 // `text_compare_of` from a MySQL collation NAME, and a PostgreSQL
2295 // database collating as `en_US.utf8` does not pad — inheritance
2296 // reaching that would make `'a' = 'a '` true everywhere.
2297 // v7.39.2 — an operand WRITTEN as `BINARY x`, `CAST(x AS BINARY)` or
2298 // `x COLLATE utf8mb4_bin` asks for byte order, and the database's
2299 // collation must not answer over it.
2300 //
2301 // The `_bin` spelling lowers onto the BINARY cast, so it never
2302 // reaches `derive` as a collation at all — the fallback below then
2303 // ordered by the database's locale. Measured on MySQL 9.7.2 against
2304 // an image collating `en_US.utf8`: `'B' COLLATE utf8mb4_bin < 'a'`
2305 // is 1 there and was 0 here, and `>` was wrong to match. The fold
2306 // was already suppressed, so `=` was right and only the ORDERING
2307 // comparisons were wrong — the half that works is what hid it.
2308 //
2309 // A COLUMN whose stored collation is `Binary` is NOT this: that is
2310 // the struct's default and means "nothing was said", which is why
2311 // `operand_is_binary_column` is deliberately not consulted here.
2312 if crate::eval::is_binary_coerced(lhs) || crate::eval::is_binary_coerced(rhs) {
2313 return None;
2314 }
2315 // v7.39.4 — the SESSION sits between the two, and it was missing.
2316 //
2317 // The chain here was "what the operands declare, else the DATABASE",
2318 // and a session collation declares nothing on an operand. So under
2319 // `SET NAMES utf8mb4 COLLATE utf8mb4_bin` on a database collating
2320 // `en_US.UTF-8` — which is what the server ships — `'a' < 'B'`
2321 // answered 1 where MySQL 9.7.2 answers 0: the collator was handed
2322 // the database's locale and the session was never consulted.
2323 // Measured against the published 7.39.3 image, side by side with
2324 // MySQL, and reproduced here only after the test engine was given
2325 // the SHIPPED database collation — on a byte-ordering database the
2326 // defect cannot exist, which is why every pin passed.
2327 //
2328 // A byte-wise session ends the chain rather than choosing a name:
2329 // returning `None` hands the comparison back to byte order, which is
2330 // what `utf8mb4_bin` means. The comment above explains why `=` was
2331 // right all along — the fold reads the session already, and only the
2332 // ordering operators come through here.
2333 let session = ctx
2334 .mysql_dialect
2335 .then(|| crate::eval::resolve::session_collation_name(ctx))
2336 .flatten();
2337 if derived.name().is_none() && session.as_deref().is_some_and(crate::collate::is_byte_wise) {
2338 return None;
2339 }
2340 let db = ctx
2341 .catalog
2342 .map(spg_storage::Catalog::db_collation)
2343 .filter(|d| !crate::collate::is_byte_wise(d));
2344 let ord = crate::collate::compare(derived.name().or(session.as_deref()).or(db)?, a, b)?;
2345 let b = match op {
2346 BinOp::Lt => ord == core::cmp::Ordering::Less,
2347 BinOp::LtEq => ord != core::cmp::Ordering::Greater,
2348 BinOp::Gt => ord == core::cmp::Ordering::Greater,
2349 BinOp::GtEq => ord != core::cmp::Ordering::Less,
2350 _ => return None,
2351 };
2352 Some(Ok(Value::Bool(b)))
2353}
2354
2355/// v7.39 (round 693) — the collation `least`/`greatest` should compare by,
2356/// derived across every argument the same way a comparison's two operands
2357/// are. `None` keeps byte order, which is right for arguments that declare
2358/// nothing.
2359#[inline(never)]
2360fn greatest_least_collation(args: &[Expr], ctx: &EvalContext<'_>) -> Option<alloc::string::String> {
2361 let resolve = |c: &spg_sql::ast::ColumnName| -> Option<alloc::string::String> {
2362 let pos = find_column_pos(c, ctx)?;
2363 ctx.columns.get(pos)?.collation_name.clone()
2364 };
2365 let derived = args
2366 .iter()
2367 .fold(crate::collate_derive::Derived::None, |acc, a| {
2368 acc.combine_pub(crate::collate_derive::derive(a, &resolve))
2369 });
2370 derived
2371 .name()
2372 .filter(|n| crate::collate::is_supported(n))
2373 .map(alloc::string::ToString::to_string)
2374}
2375
2376/// v7.39 (round 704) — rewrite a comparison's operator-not-found error when
2377/// the operand at fault is an UNKNOWN string literal against a numeric-family
2378/// value. PG commits such a literal to the other side's type before comparing,
2379/// so its error is the input function's — `invalid input syntax for type
2380/// integer: "abc"` — not `operator does not exist: integer = text`. An
2381/// explicit `::text` operand keeps the operator error (`1 IS DISTINCT FROM
2382/// 'a'::text`, measured on PG18), which is precisely the distinction two
2383/// `Value`s cannot carry: the first cut of this round rewrote inside
2384/// `binop::compare` and the r238 pin plus corpus 19 caught it the same day.
2385///
2386/// Error-path only — a comparison that succeeds never calls this — so the
2387/// 35.6 %-of-self-time note on `compare` is untouched.
2388#[cold]
2389#[inline(never)]
2390/// v7.39.13 — rename the operands of an ambiguous `||` refusal the way
2391/// PostgreSQL names them.
2392///
2393/// `char1_concat_is_ambiguous` builds the message from the VALUES, and
2394/// `Value::Text` is what an untyped literal and a real `text` both
2395/// arrive as. PostgreSQL calls the first `unknown`:
2396///
2397/// ```text
2398/// 'x' || 'r'::"char" unknown || "char"
2399/// 'x'::text || 'r'::"char" text || "char"
2400/// ```
2401///
2402/// Only the expression knows which was written, so the substitution
2403/// happens here, where it is in hand.
2404fn name_untyped_concat_operands(e: EvalError, lhs: &Expr, rhs: &Expr) -> EvalError {
2405 const PREFIX: &str = "operator is not unique: ";
2406 let EvalError::TypeMismatch { detail } = &e else {
2407 return e;
2408 };
2409 let Some(rest) = detail.strip_prefix(PREFIX) else {
2410 return e;
2411 };
2412 let Some((l_name, r_name)) = rest.split_once(" || ") else {
2413 return e;
2414 };
2415 let fix = |written: &str, side: &Expr| -> alloc::string::String {
2416 if written == "text" && is_unknown_string_literal(side) {
2417 return alloc::string::String::from("unknown");
2418 }
2419 alloc::string::String::from(written)
2420 };
2421 EvalError::TypeMismatch {
2422 detail: alloc::format!("{PREFIX}{} || {}", fix(l_name, lhs), fix(r_name, rhs)),
2423 }
2424}
2425
2426fn unknown_literal_cmp_error(
2427 err: EvalError,
2428 lhs: &Expr,
2429 rhs: &Expr,
2430 lv: &Value<'_>,
2431 rv: &Value<'_>,
2432) -> EvalError {
2433 let EvalError::TypeMismatch { detail } = &err else {
2434 return err;
2435 };
2436 // Two spellings of the same fall-through: `compare`'s operator error,
2437 // and the owned numeric path's conversion error (`f = 'y'` reaches
2438 // "cannot convert text to FLOAT"). Both mean the literal failed to
2439 // lift; neither is what PG says about an unknown literal.
2440 if !detail.starts_with("operator does not exist")
2441 && !detail.starts_with("cannot convert text to")
2442 {
2443 return err;
2444 }
2445 let numeric = |v: &Value<'_>| {
2446 matches!(
2447 v.data_type(),
2448 Some(
2449 spg_storage::DataType::SmallInt
2450 | spg_storage::DataType::Int
2451 | spg_storage::DataType::BigInt
2452 | spg_storage::DataType::Float
2453 | spg_storage::DataType::Real
2454 | spg_storage::DataType::Numeric { .. }
2455 )
2456 )
2457 };
2458 let rewrite = |s: &Value<'_>, other: &Value<'_>| -> Option<EvalError> {
2459 let Value::Text(text) = s else { return None };
2460 let dt = other.data_type()?;
2461 Some(EvalError::TypeMismatch {
2462 detail: alloc::format!(
2463 "invalid input syntax for type {}: \"{text}\"",
2464 crate::conversions::pg_type_name_for_error(dt)
2465 ),
2466 })
2467 };
2468 if is_unknown_string_literal(lhs)
2469 && numeric(rv)
2470 && let Some(e) = rewrite(lv, rv)
2471 {
2472 return e;
2473 }
2474 if is_unknown_string_literal(rhs)
2475 && numeric(lv)
2476 && let Some(e) = rewrite(rv, lv)
2477 {
2478 return e;
2479 }
2480 err
2481}
2482
2483fn enum_arg_type_name<'e>(args: &'e [Expr], ctx: &EvalContext<'e>) -> Option<&'e str> {
2484 args.iter()
2485 .find_map(|a| expr_enum_type_name(a, ctx.columns))
2486 .filter(|n| {
2487 ctx.catalog
2488 .is_some_and(|cat| cat.enum_types().contains_key(*n))
2489 })
2490}
2491
2492/// Cheap value-free precheck so `eval_expr`'s recursion frame carries no
2493/// binding for the enum path (stack-depth guard budget).
2494#[inline(never)]
2495fn enum_introspection_applies(args: &[Expr], ctx: &EvalContext<'_>) -> bool {
2496 enum_arg_type_name(args, ctx).is_some()
2497}
2498
2499#[inline(never)]
2500fn eval_enum_introspection(
2501 name: &str,
2502 args: &[Expr],
2503 row: &Row<'static>,
2504 ctx: &EvalContext<'_>,
2505) -> Result<Value<'static>, EvalError> {
2506 let Some(en) = enum_arg_type_name(args, ctx)
2507 .and_then(|n| ctx.catalog.and_then(|cat| cat.enum_types().get(n)))
2508 else {
2509 // The precheck guarantees this arm is unreachable; keep a typed
2510 // error rather than a panic if the two ever drift.
2511 return Err(EvalError::TypeMismatch {
2512 detail: "could not determine polymorphic type".into(),
2513 });
2514 };
2515 let labels = &en.labels;
2516 if labels.is_empty() {
2517 return Ok(Value::Null);
2518 }
2519 if name.eq_ignore_ascii_case("enum_first") {
2520 return Ok(Value::text(labels[0].clone()));
2521 }
2522 if name.eq_ignore_ascii_case("enum_last") {
2523 return Ok(Value::text(labels[labels.len() - 1].clone()));
2524 }
2525 // enum_range(NULL) = all; enum_range(lo, hi) slices inclusively,
2526 // NULL bound = open end (PG).
2527 let pos_of = |v: &Value<'_>| -> Option<usize> {
2528 match v {
2529 Value::Text(s) => labels.iter().position(|l| l == s.as_ref()),
2530 _ => None,
2531 }
2532 };
2533 let (lo, hi) = if args.len() == 2 {
2534 let a = eval_expr(&args[0], row, ctx)?;
2535 let b = eval_expr(&args[1], row, ctx)?;
2536 (
2537 pos_of(&a).unwrap_or(0),
2538 pos_of(&b).unwrap_or(labels.len() - 1),
2539 )
2540 } else {
2541 (0, labels.len() - 1)
2542 };
2543 let out: alloc::vec::Vec<Option<String>> = labels
2544 .get(lo..=hi)
2545 .unwrap_or(&[])
2546 .iter()
2547 .map(|l| Some(l.clone()))
2548 .collect();
2549 Ok(Value::TextArray(out))
2550}
2551
2552/// Apply one `[index]` subscript to a value — the single-step semantics shared
2553/// by 1-D array elements and JSON path access (`j['a']`, `j[0]`). NULL target
2554/// or index → NULL; a 1-based integer indexes a 1-D array (out of range → NULL,
2555/// non-array → error); JSON delegates to `path_get`.
2556fn apply_one_subscript(
2557 target_v: Value<'static>,
2558 index: &Expr,
2559 row: &Row<'static>,
2560 ctx: &EvalContext<'_>,
2561) -> Result<Value<'static>, EvalError> {
2562 let idx_v = eval_expr(index, row, ctx)?;
2563 if matches!(target_v, Value::Null) || matches!(idx_v, Value::Null) {
2564 return Ok(Value::Null);
2565 }
2566 // v7.38 (read01) — JSON/JSONB subscripting (`j['a']`, `j[0]`, chained
2567 // `j['a']['b']`) is object/array access, identical to the `->` operator
2568 // (text key → object field, integer → 0-based array element). PG 14+.
2569 if matches!(target_v, Value::Json(_)) {
2570 return crate::json::path_get(&target_v, &idx_v, false);
2571 }
2572 let i: i64 = match idx_v {
2573 Value::Int(n) => i64::from(n),
2574 Value::BigInt(n) => n,
2575 Value::SmallInt(n) => i64::from(n),
2576 other => {
2577 return Err(EvalError::TypeMismatch {
2578 detail: format!(
2579 "array subscript must be integer, got {}",
2580 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2581 ),
2582 });
2583 }
2584 };
2585 // v7.39.11 — PG's catalog vectors subscript from 0, where every
2586 // array type subscripts from 1. `indkey[0]` is the first key
2587 // column there, and returning NULL for it would be a wrong answer
2588 // rather than an error.
2589 let lower = i64::from(!matches!(
2590 target_v,
2591 Value::Int2Vector(_) | Value::OidVector(_)
2592 ));
2593 if i < lower {
2594 return Ok(Value::Null);
2595 }
2596 let pos = (i - lower) as usize;
2597 match array_element_at(&target_v, pos) {
2598 Some(v) => Ok(v),
2599 None if array_len(&target_v).is_some() => Ok(Value::Null),
2600 None => Err(EvalError::TypeMismatch {
2601 detail: format!(
2602 "subscript target must be an array, got {}",
2603 crate::conversions::pg_type_name_for_error_opt(target_v.data_type())
2604 ),
2605 }),
2606 }
2607}
2608
2609/// v7.38 (read01, 2D-subscript) — index a 2-D array (`arr[i][j]`). PG needs
2610/// exactly two subscripts to reach an element; a single subscript on a 2-D
2611/// array yields NULL (not the row), and any out-of-range index → NULL. Both
2612/// subscripts are 1-based.
2613fn eval_matrix_subscript(
2614 base: &Value<'static>,
2615 idx_exprs: &[&Expr],
2616 row: &Row<'static>,
2617 ctx: &EvalContext<'_>,
2618) -> Result<Value<'static>, EvalError> {
2619 if idx_exprs.len() != 2 {
2620 return Ok(Value::Null);
2621 }
2622 let mut idx = [0i64; 2];
2623 for (k, ix) in idx_exprs.iter().enumerate() {
2624 idx[k] = match eval_expr(ix, row, ctx)? {
2625 Value::Null => return Ok(Value::Null),
2626 Value::Int(n) => i64::from(n),
2627 Value::BigInt(n) => n,
2628 Value::SmallInt(n) => i64::from(n),
2629 other => {
2630 return Err(EvalError::TypeMismatch {
2631 detail: format!(
2632 "array subscript must be integer, got {}",
2633 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2634 ),
2635 });
2636 }
2637 };
2638 }
2639 let (r, c) = (idx[0], idx[1]);
2640 if r < 1 || c < 1 {
2641 return Ok(Value::Null);
2642 }
2643 let (ri, ci) = ((r - 1) as usize, (c - 1) as usize);
2644 macro_rules! elem {
2645 ($rows:expr, $map:expr) => {
2646 Ok($rows
2647 .get(ri)
2648 .and_then(|inner| inner.get(ci))
2649 .map_or(Value::Null, |cell| cell.as_ref().map_or(Value::Null, $map)))
2650 };
2651 }
2652 match base {
2653 Value::IntArray2D(rows) => elem!(rows, |n| Value::Int(*n)),
2654 Value::BigIntArray2D(rows) => elem!(rows, |n| Value::BigInt(*n)),
2655 Value::BoolArray2D(rows) => elem!(rows, |b| Value::Bool(*b)),
2656 Value::TextArray2D(rows) => {
2657 elem!(rows, |s| Value::Text(alloc::borrow::Cow::Owned(s.clone())))
2658 }
2659 _ => Ok(Value::Null),
2660 }
2661}
2662
2663/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
2664/// (stack-depth guard budget); body unchanged.
2665#[inline(never)]
2666fn eval_cast_arm(
2667 expr: &Expr,
2668 target: &CastTarget,
2669 row: &Row<'static>,
2670 ctx: &EvalContext<'_>,
2671) -> Result<Value<'static>, EvalError> {
2672 let v = eval_expr(expr, row, ctx)?;
2673 // v7.39 (round 473) — `<oid>::regclass` names the relation.
2674 //
2675 // The cast itself has no catalog, so it answered the bare number:
2676 // `indexrelid::regclass` printed `100001` where PG prints `ix1`, and
2677 // pg_class / pg_index rows are read by joining on oids and rendering
2678 // them — a tool cannot match the two up. `relation_name_for_oid`
2679 // mirrors `relation_oid`'s walks so the two directions agree; an oid
2680 // that names nothing keeps rendering as the number, which is what PG
2681 // does for a dropped relation's oid too.
2682 if matches!(target, CastTarget::RegClass)
2683 && let Some(cat) = ctx.catalog
2684 {
2685 let oid = match &v {
2686 Value::Int(n) => Some(i64::from(*n)),
2687 Value::BigInt(n) => Some(*n),
2688 _ => None,
2689 };
2690 if let Some(oid) = oid
2691 && let Some(name) = crate::system_catalog::relation_name_for_oid(cat, oid)
2692 {
2693 return Ok(Value::text(name));
2694 }
2695 }
2696 // v7.38 (read01 P6.40) — a cast to a user DOMAIN (`x::posint`)
2697 // enforces the domain's NOT NULL + CHECK constraints, matching PG.
2698 // The base-type coercion already happened when `v` was produced
2699 // (the domain is a constrained alias of its base type); here we
2700 // only run the constraints.
2701 if let CastTarget::Named(name) = target
2702 && let Some(cat) = ctx.catalog
2703 {
2704 if let Some(dom) = cat.domain_types().get(name.as_str()) {
2705 return apply_domain_constraints(v, dom, name, cat);
2706 }
2707 // v7.38 (read01 P6.67) — `'label'::<user enum>` validates the
2708 // label against the enum's members (a non-member errors like
2709 // PG). A typed NULL passes through carrying the enum type.
2710 if let Some(en) = cat.enum_types().get(name.as_str()) {
2711 return apply_enum_cast(v, en, name);
2712 }
2713 // v7.39 (read01 rowtypes.c) — `'(1,x)'::<composite>` parses PG's
2714 // record text form against the type's field list; a ROW value
2715 // re-labels its fields.
2716 if let Some(comp) = cat.composite_types().get(name.as_str()) {
2717 // v7.39 (round 264) — pass the catalog so a NESTED composite
2718 // field resolves into a record rather than staying text.
2719 return apply_composite_cast_in(v, comp, ctx.catalog);
2720 }
2721 // v7.39 (round 509) — every TABLE also names a row type, and
2722 // `jsonb_populate_record(NULL::mytable, …)` is PG's canonical
2723 // spelling for "shaped like this table". PG accepts `NULL::mytable`
2724 // and refuses `1::mytable` with "cannot cast type integer to
2725 // mytable" — the type exists, the conversion does not. This only
2726 // ever worked here because a NULL skipped cast resolution entirely;
2727 // now that it does not, the row type has to be named explicitly.
2728 if cat.get(name.as_str()).is_some() {
2729 return if matches!(v, Value::Null) {
2730 Ok(Value::Null)
2731 } else {
2732 Err(EvalError::TypeMismatch {
2733 detail: alloc::format!(
2734 "cannot cast type {} to {name}",
2735 crate::eval::strings::pg_typeof_name(&v),
2736 ),
2737 })
2738 };
2739 }
2740 // v7.39 (round 513) — `regnamespace` and `regrole` resolve against
2741 // things that live outside the type table: schemas on the catalog,
2742 // roles on the engine. They belong here for the same reason the
2743 // relation check above does — this is the arm that can see them.
2744 // v7.39 (round 526) — the NUMERIC direction, which is the one a
2745 // catalog join uses: `relnamespace::regnamespace` names the
2746 // schema a relation lives in, and it errored with "unsupported
2747 // cast target" while `'public'::regnamespace` worked. Round 513
2748 // added the name direction only, so the half that reads a
2749 // catalog was the half missing.
2750 if (name.eq_ignore_ascii_case("regnamespace") || name.eq_ignore_ascii_case("regrole"))
2751 && let Some(oid) = match &v {
2752 Value::Int(n) => Some(i64::from(*n)),
2753 Value::BigInt(n) => Some(*n),
2754 _ => None,
2755 }
2756 {
2757 let named = if name.eq_ignore_ascii_case("regnamespace") {
2758 crate::system_catalog::schema_name_for_oid(oid)
2759 } else {
2760 ctx.engine.and_then(|e| e.role_name_for_oid(oid))
2761 };
2762 // PG prints the bare number for an oid that names nothing,
2763 // exactly as `regclass` does.
2764 return Ok(Value::text(
2765 named.unwrap_or_else(|| alloc::format!("{oid}")),
2766 ));
2767 }
2768 if name.eq_ignore_ascii_case("regnamespace")
2769 && let Value::Text(t) = &v
2770 {
2771 let want = t.trim().trim_matches('"');
2772 // 7.38.1 S5.1 — the name direction answers the DUAL
2773 // (oid, name) value for the schemas with a published oid:
2774 // regnamespace IS an oid in PG, and pg_dump compares it
2775 // against numeric namespace columns (`opcnamespace =
2776 // 'pg_catalog'::regnamespace`) — while the wire render
2777 // stays the NAME, as PG's does (the round-513 contract).
2778 // The RegClass dual carries exactly that pair. A user
2779 // schema without a published oid keeps plain text.
2780 return if spg_storage::is_builtin_schema(want) || cat.schema_exists(want) {
2781 Ok(match want {
2782 "pg_catalog" => Value::RegClass(11, "pg_catalog".into()),
2783 "public" => Value::RegClass(2200, "public".into()),
2784 "information_schema" => Value::RegClass(13000, "information_schema".into()),
2785 _ => Value::text(want.to_string()),
2786 })
2787 } else {
2788 Err(EvalError::TypeMismatch {
2789 detail: alloc::format!("schema \"{want}\" does not exist"),
2790 })
2791 };
2792 }
2793 if name.eq_ignore_ascii_case("regrole")
2794 && let Value::Text(t) = &v
2795 {
2796 let want = t.trim().trim_matches('"').to_string();
2797 // PG ships predefined roles that exist whether or not anybody
2798 // created them; SPG carries the rest on the engine.
2799 const PREDEFINED: &[&str] = &[
2800 "pg_read_all_data",
2801 "pg_write_all_data",
2802 "pg_monitor",
2803 "pg_read_all_settings",
2804 "pg_read_all_stats",
2805 "pg_stat_scan_tables",
2806 "pg_signal_backend",
2807 "pg_checkpoint",
2808 "pg_maintain",
2809 "pg_use_reserved_connections",
2810 "pg_create_subscription",
2811 ];
2812 let known = PREDEFINED.iter().any(|r| r.eq_ignore_ascii_case(&want))
2813 || ctx.engine.is_some_and(|e| e.role_exists(&want));
2814 return if known {
2815 Ok(Value::text(want))
2816 } else {
2817 Err(EvalError::TypeMismatch {
2818 detail: alloc::format!("role \"{want}\" does not exist"),
2819 })
2820 };
2821 }
2822 // v7.39 (round 509) — the cast target is checked even when the
2823 // operand is NULL. `cast_value_in` short-circuits a NULL before it
2824 // looks at the target, so `NULL::nosuchtype` silently answered NULL
2825 // and `pg_typeof(NULL::nosuchtype)` answered `unknown`, while
2826 // `1::nosuchtype` errored — the gap was exactly the NULL case, in
2827 // both spellings. Everything a catalog can name has been tried by
2828 // now; what is left is the builtin table.
2829 if matches!(v, Value::Null)
2830 && !crate::eval::cast::builtin_target_resolves(name, ctx.mysql_dialect)
2831 {
2832 return Err(EvalError::TypeMismatch {
2833 detail: cast::unknown_type_error_text(name),
2834 });
2835 }
2836 }
2837 // v7.39 (round 285) — `::record`, the anonymous composite type. PG
2838 // treats it as an IDENTITY cast on anything already composite: the
2839 // value keeps its fields and their names, so `(ROW(1,2)::record).f1`
2840 // and `(r).x` still resolve. Only a non-composite is refused, with
2841 // PG's wording. `record` is not a catalog type, so this cannot live
2842 // in the lookups above.
2843 if let CastTarget::Named(name) = target
2844 && name.eq_ignore_ascii_case("record")
2845 {
2846 return match v {
2847 Value::Composite(_) | Value::Null => Ok(v),
2848 other => Err(EvalError::TypeMismatch {
2849 detail: alloc::format!(
2850 "cannot cast type {} to record",
2851 crate::eval::strings::pg_typeof_name(&other),
2852 ),
2853 }),
2854 };
2855 }
2856 // v7.38 (read01, T22) — a numeric OID cast to regclass reverse-looks
2857 // up the user relation name (PG's 16384+ band, assigned in
2858 // table_names() order). System OIDs / non-matches fall through to
2859 // the integer-rendering path in cast_value.
2860 if matches!(target, CastTarget::RegClass) {
2861 // v7.39 (read01 ruleutils.c) — regclass is DUAL-shape: oid for
2862 // catalog joins (conrelid = 't'::regclass), name for display.
2863 let oid_in = match &v {
2864 Value::SmallInt(n) => Some(i64::from(*n)),
2865 Value::Int(n) => Some(i64::from(*n)),
2866 Value::BigInt(n) => Some(*n),
2867 _ => None,
2868 };
2869 if let (Some(oid), Some(cat)) = (oid_in, ctx.catalog) {
2870 if oid >= 16384 {
2871 if let Some(name) = cat.table_names().into_iter().nth((oid - 16384) as usize) {
2872 return Ok(Value::RegClass(oid, name.into()));
2873 }
2874 }
2875 }
2876 if let (Value::Text(s), Some(cat)) = (&v, ctx.catalog) {
2877 let bare = s
2878 .rsplit('.')
2879 .next()
2880 .unwrap_or(s)
2881 .trim_matches('"')
2882 .to_string();
2883 if let Some(oid) = regclass_name_to_oid(cat, &bare) {
2884 return Ok(Value::RegClass(oid, bare.into()));
2885 }
2886 // v7.39 (round 337, V62) — a name that is no relation at all is
2887 // PG's error, not a silent pass-through. `'nope'::regclass`
2888 // used to answer the TEXT `nope`, so a downstream
2889 // `pg_get_viewdef('nope'::regclass)` reported "no such view"
2890 // when the truth is there is no such relation — and a
2891 // catalog join on it quietly matched nothing. PG 18.4:
2892 // `ERROR: relation "nope" does not exist`. (`to_regclass` is
2893 // the spelling that answers NULL instead, and still does.)
2894 //
2895 // The system views SPG synthesises have no oid space, so they
2896 // keep the textual form rather than erroring.
2897 const SYSTEM_RELS: &[&str] = &[
2898 "pg_roles",
2899 "pg_user",
2900 "pg_tables",
2901 "pg_views",
2902 "pg_settings",
2903 "pg_stat_activity",
2904 "pg_stat_database",
2905 "pg_stat_user_tables",
2906 "pg_class",
2907 "pg_attribute",
2908 "pg_type",
2909 "pg_proc",
2910 "pg_namespace",
2911 "pg_constraint",
2912 "pg_index",
2913 "pg_rewrite",
2914 ];
2915 if !SYSTEM_RELS.contains(&bare.as_str()) {
2916 return Err(EvalError::TypeMismatch {
2917 detail: alloc::format!("relation \"{bare}\" does not exist"),
2918 });
2919 }
2920 }
2921 }
2922 // v7.39 (round 339, V63) — `::regproc` / `::regprocedure` resolve
2923 // against the USER function catalog too. Name resolution ran against
2924 // the static pg_proc table alone, so `'my_fn'::regproc` — the form
2925 // every catalog query and pg_dump uses to name a function — raised
2926 // `function "my_fn" does not exist` for a function that plainly did.
2927 // The cast layer has no catalog handle; this is the same interception
2928 // point the `::regclass` block above uses.
2929 if let (CastTarget::Named(tname), Some(cat), Value::Text(s)) = (target, ctx.catalog, &v) {
2930 let lower = tname.to_ascii_lowercase();
2931 if matches!(lower.as_str(), "regproc" | "regprocedure") {
2932 let raw = s.trim();
2933 // regprocedure carries the argument list: `f(int,text)`.
2934 let (name_part, args_part) = match raw.split_once('(') {
2935 Some((n, rest)) => (n.trim(), Some(rest.trim_end_matches(')'))),
2936 None => (raw, None),
2937 };
2938 let bare = name_part
2939 .strip_prefix("public.")
2940 .unwrap_or(name_part)
2941 .trim_matches('"');
2942 let cands = cat.functions_named(bare);
2943 if let Some(args_txt) = args_part {
2944 // An overload IS distinguishable here — the argument list
2945 // is what regprocedure exists to carry.
2946 let want =
2947 crate::system_catalog::canonical_arg_types(&alloc::format!("({args_txt})"));
2948 if let Some(f) = cands
2949 .iter()
2950 .find(|f| crate::system_catalog::canonical_arg_types(&f.args_repr) == want)
2951 {
2952 let rendered = alloc::format!(
2953 "{bare}({})",
2954 crate::system_catalog::canonical_arg_types(&f.args_repr)
2955 );
2956 // v7.39 (round 342, V65) — dual shape: the oid for
2957 // catalog joins, the rendering for display.
2958 let oid = crate::system_catalog::function_oid_by_signature(cat, bare, &want)
2959 .unwrap_or(0);
2960 return Ok(Value::RegProc(oid, rendered.into()));
2961 }
2962 } else {
2963 match cands.len() {
2964 0 => {}
2965 1 => {
2966 let oid = crate::system_catalog::function_oid(cat, bare).unwrap_or(0);
2967 return Ok(Value::RegProc(oid, bare.into()));
2968 }
2969 _ => {
2970 return Err(EvalError::TypeMismatch {
2971 detail: alloc::format!("more than one function named \"{bare}\""),
2972 });
2973 }
2974 }
2975 }
2976 }
2977 }
2978 // v7.38 (T-tstz Phase 1) — `<timestamptz>::text` renders the offset
2979 // (`2024-01-15 10:30:00+00`); plain timestamp does not. The runtime
2980 // value is the same tz-less `Value::Timestamp`, so consult the
2981 // inner expression's static type. Falls through to the ordinary
2982 // cast on any shape the static typer can't resolve — worst case is
2983 // today's no-offset rendering, never a wrong instant.
2984 if matches!(target, CastTarget::Text)
2985 && let Value::Timestamp(t) = &v
2986 && crate::describe::describe_expr(expr, ctx.columns)
2987 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
2988 {
2989 // v7.39 (tz epic) — per-VALUE session offset (DST zones
2990 // vary within a statement); non-ISO DateStyles carry the
2991 // zone designation instead of the numeric offset.
2992 let off = ctx.session_tz_offset_at(*t);
2993 let abbr = ctx.session_tz_abbrev_at(*t);
2994 return Ok(Value::text(format::format_timestamptz_tz(
2995 *t,
2996 &ctx.render_style,
2997 off,
2998 abbr.as_deref(),
2999 )));
3000 }
3001 // v7.39 (read01 utils/adt, datetime.c) — the relative
3002 // reserved words resolve against the transaction clock:
3003 // 'today'/'tomorrow'/'yesterday' are midnight dates, 'now'
3004 // is the current instant ('now'::date = today). Clockless
3005 // engines fall through to the parser (which rejects them).
3006 if matches!(
3007 target,
3008 CastTarget::Date | CastTarget::Timestamp | CastTarget::Timestamptz
3009 ) && let Value::Text(word) = &v
3010 && let Some(clock) = ctx.clock
3011 {
3012 let w = word.trim().to_ascii_lowercase();
3013 if matches!(w.as_str(), "today" | "tomorrow" | "yesterday" | "now") {
3014 let now_us = clock();
3015 let today = i32::try_from(now_us.div_euclid(86_400_000_000)).ok();
3016 if let Some(today) = today {
3017 let day = match w.as_str() {
3018 "tomorrow" => today + 1,
3019 "yesterday" => today - 1,
3020 _ => today,
3021 };
3022 return Ok(match (&target, w.as_str()) {
3023 (CastTarget::Date, _) => Value::Date(day),
3024 (_, "now") => Value::Timestamp(now_us),
3025 _ => Value::Timestamp(crate::conversions::date_days_to_micros(day)),
3026 });
3027 }
3028 }
3029 }
3030 // v7.39 (GUC knife 5) — text INPUT to date/timestamp under a
3031 // non-MDY DateOrder disambiguates by the session order
3032 // (`'01/02/2024'::date` is Feb 1 under DMY). The default MDY
3033 // order flows through cast_value's parse_date_literal.
3034 if ctx.render_style.date_order != format::DateOrder::Mdy {
3035 match (&target, &v) {
3036 (CastTarget::Date, Value::Text(s)) => {
3037 if let Some(d) = format::parse_date_literal_ordered(s, ctx.render_style.date_order)
3038 {
3039 return Ok(Value::Date(d));
3040 }
3041 }
3042 (CastTarget::Timestamp, Value::Text(s)) => {
3043 if let Some(t) =
3044 format::parse_timestamp_literal_ordered(s, ctx.render_style.date_order)
3045 {
3046 return Ok(Value::Timestamp(t));
3047 }
3048 }
3049 _ => {}
3050 }
3051 }
3052 // v7.39 (round 309, V30) — the mirror of the timestamptz arm
3053 // below. A literal carrying a zone NAME is legal input to the
3054 // zone-less types, and PG throws the zone away rather than
3055 // converting: `'2020-01-01 10:00:00 America/New_York'::timestamp`
3056 // is 10:00, not 15:00. Round 289 did this for a numeric `+02`
3057 // offset; a named zone still failed to parse at all.
3058 //
3059 // The name is not simply stripped — PG validates it, and says so
3060 // (`time zone "bogus/zone" not recognized`, lowercased) rather than
3061 // reporting a malformed literal. That check is why this belongs
3062 // here and not in `cast_value`: resolving a zone needs the host
3063 // functions, which only the context carries.
3064 let zoneless_target = match &target {
3065 CastTarget::Timestamp => Some("timestamp"),
3066 CastTarget::Date => Some("date"),
3067 // `::time` has no CastTarget of its own; it arrives named.
3068 CastTarget::Named(n) if n.eq_ignore_ascii_case("time") => Some("time"),
3069 _ => None,
3070 };
3071 if let Some(kind) = zoneless_target
3072 && let Value::Text(txt) = &v
3073 && let Some((wall, zone)) = split_trailing_zone_name(txt, ctx.render_style.date_order)
3074 {
3075 if ctx.zone_local_to_utc(zone, wall).is_none() {
3076 // Measured boundary: PG calls the token a ZONE NAME — and
3077 // so reports a misspelling as such — only when it is
3078 // path-shaped. A bare word it does not know (`ABCD`, `QQQ`,
3079 // `UTC_X`) makes the whole literal invalid syntax instead,
3080 // because nothing marks it as having meant a zone at all.
3081 if zone.contains('/') {
3082 return Err(EvalError::TypeMismatch {
3083 detail: alloc::format!(
3084 "time zone \"{}\" not recognized",
3085 zone.to_ascii_lowercase()
3086 ),
3087 });
3088 }
3089 } else {
3090 return Ok(match kind {
3091 "date" => {
3092 Value::Date(i32::try_from(wall.div_euclid(86_400_000_000)).map_err(|_| {
3093 EvalError::TypeMismatch {
3094 detail: "timestamp out of DATE range".into(),
3095 }
3096 })?)
3097 }
3098 "time" => Value::Time(wall.rem_euclid(86_400_000_000)),
3099 _ => Value::Timestamp(wall),
3100 });
3101 }
3102 }
3103 // v7.39 (tz epic) — timestamptz INPUT: an offset-less
3104 // literal is a wall-clock reading in the session zone
3105 // (PG); a trailing IANA zone name localises there. Both
3106 // fall through to cast_value when nothing matches (its
3107 // parse treats naive input as UTC — correct for a UTC
3108 // session).
3109 if matches!(target, CastTarget::Timestamptz)
3110 && let Value::Text(txt) = &v
3111 {
3112 let order = ctx.render_style.date_order;
3113 let sess_zone = ctx
3114 .session_gucs
3115 .and_then(|g| g.get("timezone"))
3116 .map(String::as_str);
3117 // Trailing zone name: the last space-separated token,
3118 // when it names a resolvable zone (contains a letter
3119 // and isn't consumed by the plain parse).
3120 if let Some(idx) = txt.trim_end().rfind(' ') {
3121 let (head, tail) = (txt[..idx].trim(), txt[idx + 1..].trim());
3122 let tail_is_zoneish = tail.len() > 1
3123 && tail.bytes().any(|b| b.is_ascii_alphabetic())
3124 && !tail.eq_ignore_ascii_case("bc")
3125 && !tail.eq_ignore_ascii_case("ad");
3126 if tail_is_zoneish
3127 && format::parse_timestamp_literal_tz_ordered(txt, order).is_none()
3128 && let Some((wall, false)) = format::parse_timestamp_literal_tz_ordered(head, order)
3129 && let Some(utc) = ctx.zone_local_to_utc(tail, wall)
3130 {
3131 return Ok(Value::Timestamp(utc));
3132 }
3133 }
3134 if let Some((wall, had_tz)) = format::parse_timestamp_literal_tz_ordered(txt, order) {
3135 if had_tz {
3136 return Ok(Value::Timestamp(wall));
3137 }
3138 if let Some(zone) = sess_zone
3139 && !zone.eq_ignore_ascii_case("utc")
3140 && !zone.eq_ignore_ascii_case("gmt")
3141 && let Some(utc) = ctx.zone_local_to_utc(zone, wall)
3142 {
3143 return Ok(Value::Timestamp(utc));
3144 }
3145 return Ok(Value::Timestamp(wall));
3146 }
3147 }
3148 // v7.39 (round 523) — a NAIVE timestamp cast to timestamptz is a
3149 // wall-clock reading in the session zone, exactly as the text form
3150 // above already is. This was a no-op, so under `SET TimeZone =
3151 // 'Asia/Tokyo'` a `TIMESTAMP '2020-01-01 00:00:00'::timestamptz`
3152 // named 09:00 JST — a different INSTANT, nine hours from the one PG
3153 // stores, not a different rendering of the same one.
3154 //
3155 // The source's static type is the witness: SPG keeps timestamptz in
3156 // the same `Value::Timestamp`, so only an expression that is not
3157 // ALREADY timestamptz may be shifted, or a tstz-to-tstz cast would
3158 // move the instant twice.
3159 if matches!(target, CastTarget::Timestamptz)
3160 && let Value::Timestamp(wall) = &v
3161 && !matches!(
3162 crate::describe::describe_expr(expr, ctx.columns).map(|s| s.ty),
3163 Some(spg_storage::DataType::Timestamptz)
3164 )
3165 && let Some(zone) = ctx.session_gucs.and_then(|g| g.get("timezone"))
3166 && !zone.eq_ignore_ascii_case("utc")
3167 && !zone.eq_ignore_ascii_case("gmt")
3168 && let Some(utc) = ctx.zone_local_to_utc(zone, *wall)
3169 {
3170 return Ok(Value::Timestamp(utc));
3171 }
3172 // v7.39 (round 523) — and the other direction: a timestamptz cast
3173 // DOWN to a zone-free type reads the local clock in the session
3174 // zone. `(TIMESTAMPTZ '2020-01-01 15:00:00Z')::date` answered
3175 // 2020-01-01 in Tokyo where PG answers 2020-01-02 — a whole day out
3176 // for every instant in the last nine hours of a UTC day, which is
3177 // exactly the shape a daily report groups on. `now()::timestamp`
3178 // likewise disagreed with `now() AT TIME ZONE <session zone>`, which
3179 // PG defines to be the same value.
3180 if matches!(target, CastTarget::Timestamp | CastTarget::Date)
3181 && let Value::Timestamp(t) = &v
3182 && crate::describe::describe_expr(expr, ctx.columns)
3183 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
3184 {
3185 let local = t.saturating_add(ctx.session_tz_offset_at(*t));
3186 return Ok(match target {
3187 CastTarget::Date => i32::try_from(local.div_euclid(86_400_000_000))
3188 .map_or(Value::Timestamp(local), Value::Date),
3189 _ => Value::Timestamp(local),
3190 });
3191 }
3192 // v7.39 (GUC knife 3) — the out-function casts honour the
3193 // session render style, like PG's date_out/interval_out/
3194 // float8out under DateStyle/IntervalStyle/extra_float_digits.
3195 if matches!(target, CastTarget::Text) {
3196 match &v {
3197 // v7.39 (round 524) — `bytea_out` is a render GUC too.
3198 Value::Bytes(b) if ctx.render_style.bytea_escape => {
3199 return Ok(Value::text(format::format_bytea_escape(b)));
3200 }
3201 Value::Date(d) => {
3202 return Ok(Value::text(format::format_date_styled(
3203 *d,
3204 &ctx.render_style,
3205 )));
3206 }
3207 Value::Timestamp(t) => {
3208 return Ok(Value::text(format::format_timestamp_styled(
3209 *t,
3210 &ctx.render_style,
3211 )));
3212 }
3213 Value::Interval {
3214 months,
3215 days,
3216 micros,
3217 kind,
3218 } => {
3219 // v7.38.19 — an infinity is the word, in every style.
3220 if !kind.is_finite() {
3221 return Ok(Value::text(format::format_interval_kinded(0, 0, 0, *kind)));
3222 }
3223 return Ok(Value::text(format::format_interval_styled(
3224 *months,
3225 *days,
3226 *micros,
3227 &ctx.render_style,
3228 )));
3229 }
3230 Value::Float(x) => {
3231 return Ok(Value::text(format::format_float_styled(
3232 *x,
3233 &ctx.render_style,
3234 )));
3235 }
3236 Value::Real(x) => {
3237 return Ok(Value::text(format::format_real_styled(
3238 *x,
3239 &ctx.render_style,
3240 )));
3241 }
3242 _ => {}
3243 }
3244 }
3245 crate::eval::cast::cast_value_ref_in(v, target, ctx.mysql_dialect)
3246}
3247
3248/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
3249/// (stack-depth guard budget); body unchanged.
3250#[inline(never)]
3251fn eval_array_arm(
3252 items: &[Expr],
3253 row: &Row<'static>,
3254 ctx: &EvalContext<'_>,
3255) -> Result<Value<'static>, EvalError> {
3256 let mut materialised: Vec<Value<'static>> = Vec::with_capacity(items.len());
3257 for elem in items {
3258 materialised.push(eval_expr(elem, row, ctx)?);
3259 }
3260 // v7.38 (read01, T10) — a constructor whose elements are all 1-D
3261 // arrays builds a 2-D array (`ARRAY[[1,2],[3,4]]`). All rows must
3262 // share a length (PG: "multidimensional arrays must have array
3263 // expressions with matching dimensions"). Int rows promote to
3264 // bigint if any row is bigint; a text row makes the whole thing text.
3265 // v7.39 (read01 round 73) — a row of ANY array kind counts. Round 72 gave
3266 // `ARRAY[true,false]` its real `bool[]` type, and this detector only knew
3267 // Int / BigInt / Text rows — so `ARRAY[ARRAY[true,false]]` stopped being a
3268 // 2-D array at all and collapsed into a 1-D text[] of rendered rows, with
3269 // `[1][2]` failing outright. A regression THIS campaign introduced, caught by
3270 // the very sweep that was chasing its own residual. Rows that are not
3271 // int/bigint arrays render into the text 2-D form below (SPG has no bool 2-D
3272 // storage variant — a recorded residual), but they stay 2-D.
3273 let all_arrays = !materialised.is_empty()
3274 && materialised.iter().all(|v| {
3275 values::array_len(v).is_some()
3276 && !matches!(
3277 v,
3278 Value::TextArray2D(_)
3279 | Value::IntArray2D(_)
3280 | Value::BigIntArray2D(_)
3281 | Value::BoolArray2D(_)
3282 )
3283 });
3284 if all_arrays {
3285 let row_len = values::array_len(&materialised[0]).unwrap_or(0);
3286 let same_len = materialised
3287 .iter()
3288 .all(|v| values::array_len(v) == Some(row_len));
3289 if !same_len {
3290 return Err(EvalError::TypeMismatch {
3291 detail: "multidimensional arrays must have array expressions \
3292 with matching dimensions"
3293 .into(),
3294 });
3295 }
3296 // v7.39 (read01 round 75) — all-BOOL rows build a real `bool[][]`. BOOL is
3297 // the one element type whose ARRAY rendering (`t`) differs from its
3298 // scalar one (`true`), so the text-backed 2-D could not be right for it:
3299 // `ARRAY[ARRAY[true,false]]::text` wants `{{t,f}}` while `[1][2]::text`
3300 // wants `false`. Every other type renders the same either way — which is
3301 // why this is the only typed 2-D SPG needs.
3302 if materialised
3303 .iter()
3304 .all(|v| matches!(v, Value::BoolArray(_)))
3305 {
3306 let rows: Vec<Vec<Option<bool>>> = materialised
3307 .into_iter()
3308 .map(|v| match v {
3309 Value::BoolArray(r) => r,
3310 _ => unreachable!("checked above"),
3311 })
3312 .collect();
3313 return Ok(Value::BoolArray2D(rows));
3314 }
3315 let any_text = materialised
3316 .iter()
3317 .any(|v| !matches!(v, Value::IntArray(_) | Value::BigIntArray(_)));
3318 let any_big = materialised
3319 .iter()
3320 .any(|v| matches!(v, Value::BigIntArray(_)));
3321 if any_text {
3322 let rows: Vec<Vec<Option<String>>> = materialised
3323 .into_iter()
3324 .map(|v| match v {
3325 Value::TextArray(r) => r,
3326 // Any other element type renders into the text 2-D form,
3327 // element by element. SPG has no typed 2-D storage beyond
3328 // int / bigint / text, so a bool 2-D array IS text — and the
3329 // SCALAR rendering is the one to use: `(arr)[1][2]::text`
3330 // must read `false`, as in PG. (`pg_typeof` reporting
3331 // `text[]` rather than `boolean[]` is the recorded residual;
3332 // a typed 2-D needs new storage variants.)
3333 other => {
3334 let n = values::array_len(&other).unwrap_or(0);
3335 (0..n)
3336 .map(|i| match values::array_element_at(&other, i) {
3337 None | Some(Value::Null) => None,
3338 Some(v) => Some(value_to_text(&v)),
3339 })
3340 .collect()
3341 }
3342 })
3343 .collect();
3344 return Ok(Value::TextArray2D(rows));
3345 }
3346 if any_big {
3347 let rows: Vec<Vec<Option<i64>>> = materialised
3348 .into_iter()
3349 .map(|v| match v {
3350 Value::BigIntArray(r) => r,
3351 Value::IntArray(r) => r.into_iter().map(|c| c.map(i64::from)).collect(),
3352 _ => unreachable!(),
3353 })
3354 .collect();
3355 return Ok(Value::BigIntArray2D(rows));
3356 }
3357 let rows: Vec<Vec<Option<i32>>> = materialised
3358 .into_iter()
3359 .map(|v| match v {
3360 Value::IntArray(r) => r,
3361 _ => unreachable!(),
3362 })
3363 .collect();
3364 return Ok(Value::IntArray2D(rows));
3365 }
3366 // v7.39 (read01 round 72) — a HOMOGENEOUS array of a non-numeric, non-text
3367 // type keeps that type, and is unambiguous, so it is decided BEFORE the
3368 // numeric/text unification below. Everything outside the numeric ladder and
3369 // text used to fall into that loop's `_ => has_text = true` — a silent
3370 // degradation, not a decision: `ARRAY[true, false]` came back as `text[]`.
3371 // It usually LOOKED right (array_to_string renders `t` either way), which is
3372 // exactly what let it sit; the array FUNCTIONS are what tripped over it.
3373 if let Some(v) = values::homogeneous_typed_array(&materialised) {
3374 return Ok(crate::describe::upgrade_timestamptz_array(
3375 v,
3376 items,
3377 ctx.columns,
3378 ));
3379 }
3380 // v7.39 (round 236) — PG resolves an ARRAY constructor's elements to ONE
3381 // element type and refuses the constructor when they have no common one:
3382 // `ARRAY[1, 'a'::text]` is "ARRAY types integer and text cannot be
3383 // matched". SPG degraded to `text[]` instead, so `ARRAY[1, true]` came
3384 // back as `{1,t}` — a column of rendered strings that then behaved like
3385 // text everywhere downstream. Same rule (and the same untyped-literal
3386 // subtlety) as the set-operation resolution in round 233: a bare string
3387 // literal is PG's `unknown` and takes the other elements' type, so it is
3388 // identified from the SYNTAX, not from the value's runtime type.
3389 unify_array_elements(items, &mut materialised)?;
3390 // Coercing the untyped elements can make the array homogeneous
3391 // (`ARRAY[true,'t']` becomes two booleans), so re-try the typed-array
3392 // path before falling into the numeric/text ladder below — otherwise
3393 // the now-uniform boolean array would still degrade to text[].
3394 if let Some(v) = values::homogeneous_typed_array(&materialised) {
3395 return Ok(crate::describe::upgrade_timestamptz_array(
3396 v,
3397 items,
3398 ctx.columns,
3399 ));
3400 }
3401 let mut has_text = false;
3402 let mut has_float = false;
3403 // v7.40.0 — `Value::Real` used to land in this loop's `_ =>
3404 // has_text` arm, so `ARRAY[1.5::real, 2]` came back `text[]` where
3405 // PostgreSQL 18.6 answers `real[]` (measured). Only a `float8`
3406 // widens the array; `real` beside an int or a numeric stays real.
3407 let mut has_real = false;
3408 let mut has_numeric = false;
3409 let mut has_bigint = false;
3410 let mut has_int = false;
3411 // A NumericBig or non-finite (NaN/Inf) numeric can't be held in
3412 // NumericArray's `(i128, scale)` cells, so it forces the text[]
3413 // fallback rather than a lossy/panicking conversion.
3414 let mut numeric_representable = true;
3415 for v in &materialised {
3416 match v {
3417 Value::Null => {}
3418 Value::Int(_) | Value::SmallInt(_) => has_int = true,
3419 Value::BigInt(_) => has_bigint = true,
3420 Value::Numeric {
3421 kind: spg_storage::NumericKind::Finite,
3422 ..
3423 } => {
3424 has_numeric = true;
3425 }
3426 Value::Numeric { .. } => {
3427 has_numeric = true;
3428 numeric_representable = false;
3429 }
3430 Value::NumericBig(_) => {
3431 has_numeric = true;
3432 numeric_representable = false;
3433 }
3434 Value::Float(_) => has_float = true,
3435 Value::Real(_) => has_real = true,
3436 Value::Text(_) | Value::Json(_) => has_text = true,
3437 // v7.39 (round 652) — a reg value belongs to the array by
3438 // its OID half. Falling into the catch-all made
3439 // `ARRAY['pg_class'::regclass]` a text array, so
3440 // `oid = ANY(…)` compared bigint against text and was
3441 // refused — while the identical `oid = 'pg_class'::regclass`
3442 // worked. Same defect the IN-list gate had, one layer down.
3443 Value::RegClass(..) | Value::RegProc(..) | Value::RegType(..) => has_bigint = true,
3444 _ => has_text = true,
3445 }
3446 }
3447 let any_numlike = has_int || has_bigint || has_numeric || has_float || has_real;
3448 if has_text || !any_numlike || (has_numeric && !numeric_representable) {
3449 let out: Vec<Option<String>> = materialised
3450 .into_iter()
3451 .map(|v| match v {
3452 Value::Null => None,
3453 Value::Text(s) | Value::Json(s) => Some(s.into_owned()),
3454 other => Some(value_to_text_for_array(&other, &ctx.render_style)),
3455 })
3456 .collect();
3457 return Ok(Value::TextArray(out));
3458 }
3459 // v7.38 (read01) — PG array-element unification across the numeric
3460 // ladder: any float → double precision[]; else any numeric →
3461 // numeric[] (each element keeps its own scale, PG's behaviour);
3462 // else the integer widths. Matches `pg_typeof(ARRAY[1, 2.5])` =
3463 // numeric[] and keeps downstream `[i]` arithmetic numeric.
3464 if has_real && !has_float {
3465 #[allow(clippy::cast_possible_truncation)]
3466 let out: Vec<Option<f32>> = materialised
3467 .into_iter()
3468 .map(|v| match v {
3469 Value::Null => None,
3470 Value::Real(f) => Some(f),
3471 #[allow(clippy::cast_precision_loss)]
3472 Value::Int(n) => Some(n as f32),
3473 Value::SmallInt(n) => Some(f32::from(n)),
3474 #[allow(clippy::cast_precision_loss)]
3475 Value::BigInt(n) => Some(n as f32),
3476 #[allow(clippy::cast_precision_loss)]
3477 Value::Numeric { scaled, scale, .. } => {
3478 Some((scaled as f64 / libm::pow(10.0, f64::from(scale))) as f32)
3479 }
3480 _ => None,
3481 })
3482 .collect();
3483 return Ok(Value::RealArray(out));
3484 }
3485 if has_float || has_real {
3486 let out: Vec<Option<f64>> = materialised
3487 .into_iter()
3488 .map(|v| match v {
3489 Value::Null => None,
3490 Value::Float(f) => Some(f),
3491 Value::Real(f) => Some(f64::from(f)),
3492 Value::Int(n) => Some(f64::from(n)),
3493 Value::SmallInt(n) => Some(f64::from(n)),
3494 #[allow(clippy::cast_precision_loss)]
3495 Value::BigInt(n) => Some(n as f64),
3496 #[allow(clippy::cast_precision_loss)]
3497 Value::Numeric { scaled, scale, .. } => {
3498 Some(scaled as f64 / libm::pow(10.0, f64::from(scale)))
3499 }
3500 _ => None,
3501 })
3502 .collect();
3503 return Ok(Value::FloatArray(out));
3504 }
3505 if has_numeric {
3506 let out: Vec<Option<(i128, u16)>> = materialised
3507 .into_iter()
3508 .map(|v| match v {
3509 Value::Null => None,
3510 Value::SmallInt(n) => Some((i128::from(n), 0)),
3511 Value::Int(n) => Some((i128::from(n), 0)),
3512 Value::BigInt(n) => Some((i128::from(n), 0)),
3513 Value::Numeric { scaled, scale, .. } => Some((scaled, scale)),
3514 _ => None,
3515 })
3516 .collect();
3517 return Ok(Value::NumericArray(out));
3518 }
3519 if has_bigint {
3520 let out: Vec<Option<i64>> = materialised
3521 .into_iter()
3522 .map(|v| match v {
3523 Value::Null => None,
3524 Value::Int(n) => Some(i64::from(n)),
3525 Value::SmallInt(n) => Some(i64::from(n)),
3526 Value::BigInt(n) => Some(n),
3527 // Keep in step with the `has_bigint` classification above:
3528 // whatever is counted there has to be convertible here, and
3529 // the arm below panics rather than errors. Round 652 added
3530 // the reg family to the classifier and this materialiser
3531 // took a wire-visible panic until it learned them too.
3532 Value::RegClass(oid, _) | Value::RegProc(oid, _) | Value::RegType(oid, _) => {
3533 Some(oid)
3534 }
3535 _ => unreachable!(),
3536 })
3537 .collect();
3538 return Ok(Value::BigIntArray(out));
3539 }
3540 let out: Vec<Option<i32>> = materialised
3541 .into_iter()
3542 .map(|v| match v {
3543 Value::Null => None,
3544 Value::Int(n) => Some(n),
3545 Value::SmallInt(n) => Some(i32::from(n)),
3546 _ => unreachable!(),
3547 })
3548 .collect();
3549 Ok(Value::IntArray(out))
3550}
3551
3552/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
3553/// (stack-depth guard budget); body unchanged.
3554#[inline(never)]
3555/// MySQL's collation-derivation rank for an expression. See the call
3556/// site in `eval_function_call_arm` for the measured table.
3557///
3558/// A function over several arguments takes the STRONGEST derivation
3559/// among them — the lowest rank — which is what `CONCAT('a', s)`
3560/// answering 2 shows. One with no arguments is a system constant only
3561/// if it is one of the three named below; every other zero-argument
3562/// builtin returns a value, and a value is a literal here.
3563fn mysql_coercibility(e: &Expr) -> i32 {
3564 match e {
3565 Expr::Collate { .. } => 0,
3566 Expr::Column(_) => 2,
3567 Expr::Literal(spg_sql::ast::Literal::Null) => 6,
3568 Expr::Literal(
3569 spg_sql::ast::Literal::Integer(_)
3570 | spg_sql::ast::Literal::Float(_)
3571 | spg_sql::ast::Literal::Numeric { .. },
3572 ) => 5,
3573 Expr::Literal(_) => 4,
3574 Expr::FunctionCall { name, args } => {
3575 if matches!(
3576 name.to_ascii_lowercase().as_str(),
3577 "user"
3578 | "current_user"
3579 | "session_user"
3580 | "system_user"
3581 | "database"
3582 | "schema"
3583 | "version"
3584 ) {
3585 return 3;
3586 }
3587 args.iter().map(mysql_coercibility).min().unwrap_or(4)
3588 }
3589 Expr::Cast { expr, .. } => mysql_coercibility(expr),
3590 Expr::Binary { lhs, rhs, .. } => mysql_coercibility(lhs).min(mysql_coercibility(rhs)),
3591 Expr::Unary { expr, .. } => mysql_coercibility(expr),
3592 // Anything else carries a value that came from somewhere in the
3593 // row, which is a column's derivation.
3594 _ => 2,
3595 }
3596}
3597
3598fn eval_function_call_arm(
3599 name: &str,
3600 args: &[Expr],
3601 row: &Row<'static>,
3602 ctx: &EvalContext<'_>,
3603) -> Result<Value<'static>, EvalError> {
3604 // v7.39 (read01 round 77) — named arguments (`f(x := 1)` / `f(x => 1)`).
3605 // The parser leaves them in the tree because only the catalog knows a user
3606 // function's parameter names; here they become positional, once, for
3607 // builtins and user functions alike.
3608 // v7.39 (read01 round 100) — `VARIADIC <array>` splices the array's
3609 // elements in as individual trailing arguments before dispatch, so a
3610 // variadic builtin (concat / concat_ws / format / …) sees them exactly as
3611 // if they had been written out. Done before the named-arg pass and the
3612 // positional dispatch.
3613 if args.iter().any(|a| matches!(a, Expr::Variadic(_))) {
3614 let expanded = expand_variadic_args(args, row, ctx)?;
3615 return eval_function_call_arm(name, &expanded, row, ctx);
3616 }
3617 // v7.39 (round 276) — `date_part('timezone'|…, <timestamp>)` must be
3618 // REJECTED the way EXTRACT already rejects it, and the judgement
3619 // needs the argument's STATIC declared type: SPG stores timestamptz
3620 // in the same `Value::Timestamp`, so a timestamptz legitimately
3621 // answers 0 while a plain timestamp must error. The dispatch below
3622 // receives values, not expressions, so the check belongs here —
3623 // the same place, and the same r237 trust rule (only a cast or a
3624 // column is believed), that the EXTRACT arm uses.
3625 if name.eq_ignore_ascii_case("date_part")
3626 && args.len() == 2
3627 && let Expr::Literal(spg_sql::ast::Literal::String(unit)) = &args[0]
3628 && matches!(
3629 unit.to_ascii_lowercase().as_str(),
3630 "timezone" | "timezone_hour" | "timezone_minute"
3631 )
3632 && matches!(&args[1], Expr::Cast { .. } | Expr::Column(_))
3633 && let Some(sch) = crate::describe::describe_expr(&args[1], ctx.columns)
3634 && matches!(sch.ty, spg_storage::DataType::Timestamp)
3635 {
3636 return Err(EvalError::TypeMismatch {
3637 detail: alloc::format!(
3638 "unit \"{}\" not supported for type timestamp without time zone",
3639 unit.to_ascii_lowercase()
3640 ),
3641 });
3642 }
3643 // v7.40.0 — MySQL's `COERCIBILITY(expr)`.
3644 //
3645 // It reports the collation DERIVATION of the expression, which is a
3646 // property of the SYNTAX and not of the value: `COERCIBILITY(s)` and
3647 // `COERCIBILITY(s COLLATE utf8mb4_bin)` are 2 and 0 for the same
3648 // column, and no evaluated value can tell them apart. So it is
3649 // answered here, where the argument is still an expression.
3650 //
3651 // The ranks are MySQL 9.7.2's own, measured:
3652 //
3653 // ```text
3654 // COLLATE … 0 a column 2
3655 // USER() DATABASE() 3 a string literal 4
3656 // VERSION() a number 5
3657 // NULL 6
3658 // ```
3659 if name.eq_ignore_ascii_case("coercibility")
3660 && let [arg] = args
3661 {
3662 return Ok(Value::Int(mysql_coercibility(arg)));
3663 }
3664 // v7.39 (round 258) — `pg_typeof` over an ENUM. An enum value travels
3665 // as `Value::Text` (its label), so the value-driven namer answered
3666 // `text`; the type lives in the EXPRESSION, which this arm still has.
3667 // `expr_enum_type_name` resolves a column or a cast statically — the
3668 // same static-only discipline round 253 used for EXTRACT's type name.
3669 if name.eq_ignore_ascii_case("pg_typeof")
3670 && let [arg] = args
3671 {
3672 // The name must be a REAL enum in the catalog. `expr_enum_type_name`
3673 // returns any named cast's target verbatim — it is only a
3674 // pre-filter for `expr_enum_labels`, which does the catalog
3675 // lookup — so using it alone hijacked every `x::float8` /
3676 // `x::int2` and reported SPG's internal spelling instead of PG's.
3677 // v7.39 (round 259) — domains report their own name too; like an
3678 // enum, a domain value travels as its BASE type's value, so the
3679 // name has to come from the expression. Both lookups are gated on
3680 // the catalog: `expr_enum_type_name` returns ANY named cast's
3681 // target verbatim, so an ungated use hijacks `x::float8`.
3682 let is_user_type = |e: &Expr| {
3683 expr_enum_type_name(e, ctx.columns)
3684 .filter(|n| {
3685 // v7.39 (round 330, V48) — the information_schema
3686 // domains are built into the server rather than
3687 // catalog objects (a catalog domain is user data and
3688 // would be dumped), so they are recognised here too.
3689 crate::system_catalog::is_information_schema_domain(n)
3690 || ctx.catalog.is_some_and(|cat| {
3691 cat.enum_types().contains_key(*n)
3692 || cat.domain_types().contains_key(*n)
3693 // v7.39 (round 263) — composites too: a cast
3694 // to one reported the generic `record`.
3695 || cat.composite_types().contains_key(*n)
3696 })
3697 })
3698 .map(alloc::string::String::from)
3699 };
3700 // v7.38.19 — a cast to a PSEUDO-type reports what PostgreSQL
3701 // reports, which is not always the name written.
3702 //
3703 // Measured on 18.4 rather than reasoned about: `cstring` and
3704 // `void` report themselves, while `anyelement`, `anynonarray`
3705 // and `unknown` all report `unknown` -- a polymorphic
3706 // placeholder resolves against the argument, and a bare literal
3707 // gives it nothing to resolve to. The value travels as text
3708 // either way, which is why the name has to come from the
3709 // expression: `'x'::cstring` renders `x` on both engines and
3710 // answered `text` here.
3711 //
3712 // The list here is SHORTER than `pseudo_type`'s on purpose: it
3713 // is the names measured on 18.4 for this call, no more. `record`
3714 // is the reason it has to be. `pg_typeof(ROW(1,'x')::r285::record)`
3715 // answers `r285` -- the composite's own name, not `record` --
3716 // and a first draft that reported every pseudo-type here turned
3717 // that into `unknown`, which `e2e_record_type_round285` caught.
3718 if let Some(named) = expr_enum_type_name(arg, ctx.columns)
3719 && let Some(pseudo) = crate::conversions::pseudo_type(named)
3720 && matches!(
3721 pseudo,
3722 "cstring" | "void" | "anyelement" | "anynonarray" | "unknown"
3723 )
3724 {
3725 return Ok(Value::text(match pseudo {
3726 "cstring" | "void" => pseudo,
3727 _ => "unknown",
3728 }));
3729 }
3730 let is_enum = is_user_type;
3731 if let Some(en) = is_enum(arg) {
3732 return Ok(Value::text(en));
3733 }
3734 // `ARRAY[<enum>, …]` reports the array form.
3735 if let Expr::Array(items) = arg
3736 && let Some(first) = items.first()
3737 && let Some(en) = is_enum(first)
3738 {
3739 return Ok(Value::text(alloc::format!("{en}[]")));
3740 }
3741 }
3742 if args.iter().any(|a| matches!(a, Expr::NamedArg { .. })) {
3743 let positional = resolve_named_args(name, args, ctx)?;
3744 return eval_function_call_arm(name, &positional, row, ctx);
3745 }
3746 eval_function_call_positional(name, args, row, ctx)
3747}
3748
3749/// v7.39 (read01 round 100) — rewrite a call's argument list, replacing each
3750/// `VARIADIC <array>` with the array's elements as literal arguments. A NULL
3751/// array contributes no elements (PG treats `VARIADIC NULL` as empty). Regular
3752/// arguments are carried through untouched so they still evaluate against the
3753/// row in the recursive call.
3754fn expand_variadic_args(
3755 args: &[Expr],
3756 row: &Row<'static>,
3757 ctx: &EvalContext<'_>,
3758) -> Result<alloc::vec::Vec<Expr>, EvalError> {
3759 let mut out = alloc::vec::Vec::with_capacity(args.len());
3760 for a in args {
3761 if let Expr::Variadic(inner) = a {
3762 let v = eval_expr(inner, row, ctx)?;
3763 let elems = crate::select::array_value_to_elements(&v).map_err(|_| {
3764 EvalError::TypeMismatch {
3765 detail: "VARIADIC argument must be an array".into(),
3766 }
3767 })?;
3768 for e in elems {
3769 out.push(Expr::Literal(crate::value_to_literal(e)));
3770 }
3771 } else {
3772 out.push(a.clone());
3773 }
3774 }
3775 Ok(out)
3776}
3777
3778/// The declared parameter names of `fname`, or `None` when it takes none.
3779/// Builtins whose parameters PG names live in the table; everything else asks
3780/// the catalog, where a user function's `args_repr` has carried its parameter
3781/// names since the day CREATE FUNCTION stored them.
3782fn declared_param_names(fname: &str, ctx: &EvalContext<'_>) -> Option<alloc::vec::Vec<String>> {
3783 let lower = fname.to_ascii_lowercase();
3784 let builtin: &[&str] = match lower.as_str() {
3785 "make_date" => &["year", "month", "day"],
3786 "make_time" => &["hour", "min", "sec"],
3787 "make_timestamp" | "make_timestamptz" => &["year", "month", "mday", "hour", "min", "sec"],
3788 "make_interval" => &["years", "months", "weeks", "days", "hours", "mins", "secs"],
3789 _ => &[],
3790 };
3791 if !builtin.is_empty() {
3792 return Some(builtin.iter().map(|s| (*s).to_string()).collect());
3793 }
3794 let cat = ctx.catalog?;
3795 let def = cat
3796 .functions()
3797 .values()
3798 .find(|f| f.name.eq_ignore_ascii_case(&lower))?;
3799 let names = spg_storage::function_arg_names(&def.args_repr);
3800 if names.iter().all(alloc::string::String::is_empty) {
3801 return None;
3802 }
3803 Some(names)
3804}
3805
3806/// Rewrite a call's arguments into positional order. Positional arguments fill
3807/// slots left to right; a named one goes to its declared slot. Slots nobody
3808/// filled stay absent for a user function (arity is checked at the call) and
3809/// become integer 0 for the `make_*` builtins, whose trailing fields PG
3810/// defaults that way.
3811fn resolve_named_args(
3812 fname: &str,
3813 args: &[Expr],
3814 ctx: &EvalContext<'_>,
3815) -> Result<alloc::vec::Vec<Expr>, EvalError> {
3816 let Some(params) = declared_param_names(fname, ctx) else {
3817 return Err(EvalError::TypeMismatch {
3818 detail: alloc::format!("function {fname}(...) does not support named arguments"),
3819 });
3820 };
3821 let mut slots: alloc::vec::Vec<Option<Expr>> = (0..params.len()).map(|_| None).collect();
3822 let mut next_positional = 0usize;
3823 for a in args {
3824 let (idx, val) = match a {
3825 Expr::NamedArg { name, expr } => {
3826 let i = params
3827 .iter()
3828 .position(|p| p.eq_ignore_ascii_case(name))
3829 .ok_or_else(|| EvalError::TypeMismatch {
3830 detail: alloc::format!("{fname}(...) has no argument named \"{name}\""),
3831 })?;
3832 (i, (**expr).clone())
3833 }
3834 other => {
3835 let i = next_positional;
3836 next_positional += 1;
3837 (i, other.clone())
3838 }
3839 };
3840 if idx >= slots.len() {
3841 return Err(EvalError::TypeMismatch {
3842 detail: alloc::format!("{fname}(...) got too many arguments"),
3843 });
3844 }
3845 if slots[idx].is_some() {
3846 return Err(EvalError::TypeMismatch {
3847 detail: alloc::format!("{fname}(...) got multiple values for one argument"),
3848 });
3849 }
3850 slots[idx] = Some(val);
3851 }
3852 let make_family = fname.to_ascii_lowercase().starts_with("make_");
3853 let mut out = alloc::vec::Vec::with_capacity(slots.len());
3854 for slot in slots {
3855 match slot {
3856 Some(e) => out.push(e),
3857 None if make_family => {
3858 out.push(Expr::Literal(spg_sql::ast::Literal::Integer(0)));
3859 }
3860 // A user function's unfilled slot is simply not passed; the call's
3861 // own arity check phrases the error.
3862 None => {}
3863 }
3864 }
3865 Ok(out)
3866}
3867
3868fn eval_function_call_positional(
3869 name: &str,
3870 args: &[Expr],
3871 row: &Row<'static>,
3872 ctx: &EvalContext<'_>,
3873) -> Result<Value<'static>, EvalError> {
3874 // v7.39 (round 237) — COALESCE / GREATEST / LEAST resolve their
3875 // arguments to one type the way CASE and ARRAY do. Checked statically:
3876 // an argument may have side effects (`COALESCE(nextval('s'), 1)`), so
3877 // its declared type is read rather than its value.
3878 if matches!(args.len(), 2..) {
3879 let construct = if name.eq_ignore_ascii_case("coalesce") {
3880 Some("COALESCE")
3881 } else if name.eq_ignore_ascii_case("greatest") {
3882 Some("GREATEST")
3883 } else if name.eq_ignore_ascii_case("least") {
3884 Some("LEAST")
3885 } else {
3886 None
3887 };
3888 if let Some(construct) = construct {
3889 unify_branch_types_static(construct, args.iter(), ctx)?;
3890 }
3891 }
3892 // v7.39 (read01 utils/adt, enum.c) — the enum introspection
3893 // v7.39 (read01 utils/adt, enum.c) — the enum introspection
3894 // family needs the ARGUMENT'S STATIC TYPE (the value is
3895 // usually NULL::enumtype): first/last/range over the
3896 // catalog's member order. Out-of-line so eval_expr's
3897 // recursion frame stays small (stack-depth guard budget).
3898 if (name.eq_ignore_ascii_case("enum_first")
3899 || name.eq_ignore_ascii_case("enum_last")
3900 || name.eq_ignore_ascii_case("enum_range"))
3901 && enum_introspection_applies(args, ctx)
3902 {
3903 return eval_enum_introspection(name, args, row, ctx);
3904 }
3905 // v7.39 (tz epic) — AT TIME ZONE (fn form: timezone(zone, ts))
3906 // with a NAMED zone needs the host tzdb and the argument's
3907 // static type for its two directions:
3908 // naive AT ZONE -> that zone's wall clock -> UTC instant
3909 // tstz AT ZONE -> UTC instant -> that zone's wall clock
3910 // Fixed offsets / abbreviations keep the legacy path below.
3911 // v7.39 (round 523) — `to_char(tstz, fmt)` renders the LOCAL clock
3912 // in the session zone, and its zone tokens name that zone. It was
3913 // rendering the UTC reading and spelling it `UTC`, so a formatted
3914 // stamp disagreed with the same value's own `::text`.
3915 if args.len() == 2
3916 && name.eq_ignore_ascii_case("to_char")
3917 && let Some(zone) = ctx.session_gucs.and_then(|g| g.get("timezone"))
3918 && !zone.eq_ignore_ascii_case("utc")
3919 && !zone.eq_ignore_ascii_case("gmt")
3920 && crate::describe::describe_expr(&args[0], ctx.columns)
3921 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
3922 && let Value::Timestamp(t) = eval_expr(&args[0], row, ctx)?
3923 {
3924 let off = ctx.session_tz_offset_at(t);
3925 let abbrev = ctx
3926 .session_tz_abbrev_at(t)
3927 .unwrap_or_else(|| zone.to_uppercase());
3928 let vals = [
3929 Value::Timestamp(t.saturating_add(off)),
3930 eval_expr(&args[1], row, ctx)?,
3931 ];
3932 return crate::eval::strings::to_char_in_zone(&vals, Some((&abbrev, off)));
3933 }
3934 // v7.39 (round 523) — `date_trunc(unit, tstz)` truncates on the
3935 // LOCAL calendar in the session zone. It was truncating in UTC and
3936 // rendering the result in the session zone, so under `SET TimeZone =
3937 // 'Asia/Tokyo'` a day truncation answered `2020-01-01 09:00:00+09` —
3938 // not a day boundary at all, and the wrong day for anything before
3939 // 09:00. Every report grouped by day was cut nine hours late.
3940 //
3941 // The three-argument form already does exactly this, DST reverse
3942 // lookup and all, so the session zone is passed to THAT rather than
3943 // written a second time. Only a statically-known timestamptz shifts:
3944 // a naive timestamp has no zone to be read in.
3945 if args.len() == 2
3946 && (name.eq_ignore_ascii_case("date_trunc") || name.eq_ignore_ascii_case("date_bin"))
3947 && let Some(zone) = ctx.session_gucs.and_then(|g| g.get("timezone"))
3948 && !zone.eq_ignore_ascii_case("utc")
3949 && !zone.eq_ignore_ascii_case("gmt")
3950 && crate::describe::describe_expr(&args[1], ctx.columns)
3951 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
3952 {
3953 let vals = [
3954 eval_expr(&args[0], row, ctx)?,
3955 eval_expr(&args[1], row, ctx)?,
3956 Value::text(zone.clone()),
3957 ];
3958 return datetime::date_trunc(&vals, ctx);
3959 }
3960 if args.len() == 2
3961 && name.eq_ignore_ascii_case("timezone")
3962 && let zone_v = eval_expr(&args[0], row, ctx)?
3963 && let Value::Text(zone) = &zone_v
3964 && datetime::resolve_zone_offset(zone.as_ref()).is_none()
3965 && !zone.trim().eq_ignore_ascii_case("utc")
3966 && !zone.trim().eq_ignore_ascii_case("gmt")
3967 && zone.parse::<i64>().is_err()
3968 && ctx.tz_offset_fn.is_some()
3969 {
3970 let zone = zone.trim();
3971 let src_is_tstz = crate::describe::describe_expr(&args[1], ctx.columns)
3972 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz));
3973 let inner = eval_expr(&args[1], row, ctx)?;
3974 if let Value::Timestamp(t) = inner {
3975 if src_is_tstz {
3976 if let Some(off) = ctx.zone_offset_at(zone, t) {
3977 return Ok(Value::Timestamp(t + off));
3978 }
3979 } else if let Some(utc) = ctx.zone_local_to_utc(zone, t) {
3980 return Ok(Value::Timestamp(utc));
3981 }
3982 return Err(EvalError::TypeMismatch {
3983 detail: alloc::format!("time zone \"{zone}\" not recognized"),
3984 });
3985 }
3986 }
3987 // v7.29 (round-22 phase 3) - prefix fast path: LEFT(col, n)
3988 // on a TEXT column borrows the cell and clones only the
3989 // prefix. The generic path clones the WHOLE cell first -
3990 // a LEFT(body, 120) over 24k x 30 KB rows spent 383 ms
3991 // copying bytes it then threw away (7 ms without LEFT).
3992 if args.len() == 2
3993 && name.eq_ignore_ascii_case("left")
3994 && let Expr::Column(c) = &args[0]
3995 && let Some(cell) = resolve_column_borrowed(c, row, ctx)?
3996 {
3997 {
3998 match cell {
3999 Value::Null => return Ok(Value::Null),
4000 Value::Text(t) => {
4001 let n_v = eval_expr(&args[1], row, ctx)?;
4002 if let Value::SmallInt(_) | Value::Int(_) | Value::BigInt(_) = n_v {
4003 let n = match n_v {
4004 Value::SmallInt(x) => i64::from(x),
4005 Value::Int(x) => i64::from(x),
4006 Value::BigInt(x) => x,
4007 _ => 0,
4008 };
4009 return Ok(Value::text(text_prefix_chars(t, n)));
4010 }
4011 }
4012 _ => {}
4013 }
4014 }
4015 }
4016 // v7.38 (T-tstz Phase 1) — the ONE case where pg_typeof needs the
4017 // static type: timestamptz. The runtime value is a tz-less
4018 // Value::Timestamp, so the value-driven answer below can only ever
4019 // say "without time zone". For every other type the value-driven
4020 // path is strictly better (it distinguishes json vs jsonb, keeps
4021 // NULL as "unknown", and is not fooled by describe_expr's lossy
4022 // heuristics), so we consult the static typer ONLY when it says
4023 // Timestamptz and otherwise fall through untouched.
4024 if args.len() == 1
4025 && name.eq_ignore_ascii_case("pg_typeof")
4026 && crate::describe::describe_expr(&args[0], ctx.columns)
4027 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
4028 {
4029 return Ok(Value::text::<alloc::string::String>(
4030 "timestamp with time zone".into(),
4031 ));
4032 }
4033 // v7.39 (round 694) — `oid[]`. Its VALUE is a BigIntArray, so the
4034 // value-driven namer answers `bigint[]`; the declared type lives in the
4035 // expression, exactly as it does for the Timestamptz arm above and for
4036 // the enum / domain / composite arms further down. The scalar `oid`
4037 // needed the same treatment in round 667.
4038 if args.len() == 1
4039 && name.eq_ignore_ascii_case("pg_typeof")
4040 && crate::describe::describe_expr(&args[0], ctx.columns)
4041 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::OidArray))
4042 {
4043 return Ok(Value::text::<alloc::string::String>("oid[]".into()));
4044 }
4045 // v7.39 (read01 round 56) — a COMPOSITE column reports its type NAME, not
4046 // the generic `record` the runtime value would give. Composite-ness lives
4047 // outside the DataType lattice (the stored form is JSON), so the witness is
4048 // the column's `user_composite_type` — the same shape as the enum witness.
4049 if args.len() == 1
4050 && name.eq_ignore_ascii_case("pg_typeof")
4051 && let Expr::Column(c) = &args[0]
4052 && let Some(cname) = ctx
4053 .columns
4054 .iter()
4055 .find(|sc| sc.name.eq_ignore_ascii_case(&c.name))
4056 .and_then(|sc| sc.user_composite_type.as_deref())
4057 {
4058 return Ok(Value::text::<alloc::string::String>(cname.into()));
4059 }
4060 // v7.39 (round 291) — `name` is a DECLARED type over a Value::Text,
4061 // so the value can never witness it; the schema is the only witness,
4062 // exactly as for a composite column above.
4063 if args.len() == 1
4064 && name.eq_ignore_ascii_case("pg_typeof")
4065 && let Expr::Column(c) = &args[0]
4066 && ctx
4067 .columns
4068 .iter()
4069 .any(|sc| sc.name.eq_ignore_ascii_case(&c.name) && sc.ty == spg_storage::DataType::Name)
4070 {
4071 return Ok(Value::text::<alloc::string::String>("name".into()));
4072 }
4073 // …and the same for a bare `'abc'::name`, where the cast TARGET is
4074 // the witness. PG computes pg_typeof statically; SPG reads the
4075 // value, which by then is an ordinary text.
4076 if args.len() == 1
4077 && name.eq_ignore_ascii_case("pg_typeof")
4078 && let Expr::Cast {
4079 target: spg_sql::ast::CastTarget::Named(n),
4080 ..
4081 } = &args[0]
4082 && n.eq_ignore_ascii_case("name")
4083 {
4084 return Ok(Value::text::<alloc::string::String>("name".into()));
4085 }
4086 // v7.39 (read01 round 116) — a bare, uncoerced string literal is PG's
4087 // `unknown` type, not text: `pg_typeof('x')` / `pg_typeof('123')` /
4088 // `pg_typeof('2024-01-01')` all report `unknown`. The literal only becomes
4089 // text once context coerces it — a cast (`'x'::text`), a concatenation, or
4090 // a function argument — each of which is a different Expr node that falls
4091 // through to the value-driven path below (which correctly says text).
4092 if args.len() == 1
4093 && name.eq_ignore_ascii_case("pg_typeof")
4094 && matches!(&args[0], Expr::Literal(spg_sql::ast::Literal::String(_)))
4095 {
4096 return Ok(Value::text::<alloc::string::String>("unknown".into()));
4097 }
4098 // v7.37.16 — pg_typeof of a NULL cell reports the COLUMN's
4099 // static type when it has one (PG: `VALUES (NULL),(1.5)`
4100 // types the column numeric and its NULL row's pg_typeof says
4101 // numeric, not unknown).
4102 //
4103 // A BARE `NULL` stays "unknown", as PG has it. That used to fall
4104 // out of TEXT being absent from the name table — a NULL literal
4105 // describes as TEXT, so the lookup returned None and the caller
4106 // reported unknown. Round 871 filled that table in, which silently
4107 // took the bare-NULL behaviour with it and broke
4108 // `pg_typeof_null_returns_unknown`. The rule is now stated rather
4109 // than emergent: an untyped NULL literal is unknown, a NULL that
4110 // was cast reports what it was cast to.
4111 if args.len() == 1 && name.eq_ignore_ascii_case("pg_typeof") {
4112 if matches!(&args[0], Expr::Literal(spg_sql::ast::Literal::Null)) {
4113 return Ok(Value::text::<alloc::string::String>("unknown".into()));
4114 }
4115 let v = eval_expr(&args[0], row, ctx)?;
4116 if matches!(v, Value::Null)
4117 && let Some(shape) = crate::describe::describe_expr(&args[0], ctx.columns)
4118 && let Some(n) = pg_typeof_name_for_datatype(shape.ty)
4119 {
4120 return Ok(Value::text(n));
4121 }
4122 // v7.39 (round 640) — a NON-null cell normally answers from the
4123 // value, which is right for every type whose identity the value
4124 // carries. `xid8` has no value of its own — a cell is a
4125 // `Value::BigInt` — so it can only ever say "bigint" unless the
4126 // schema is asked. `xid` is listed with it because a cell that
4127 // reached here as a plain integer (a synthesised catalog row
4128 // that has not been converted) should still name its column's
4129 // type rather than the storage it arrived in.
4130 //
4131 // v7.39 (round 667) — `oid` joins them for the same reason and no
4132 // other: its cell is a `Value::BigInt` too, so `pg_typeof(1::oid)`
4133 // answered `bigint`. Still not a general switch — where the value
4134 // knows its own identity it is the better witness, because an
4135 // expression's static shape is an approximation and its result is
4136 // the fact.
4137 if !matches!(v, Value::Null)
4138 && let Some(shape) = crate::describe::describe_expr(&args[0], ctx.columns)
4139 && matches!(
4140 shape.ty,
4141 spg_storage::DataType::Xid
4142 | spg_storage::DataType::Xid8
4143 | spg_storage::DataType::Oid
4144 // v7.39.13 — `jsonb` for the same reason and no
4145 // other: SPG carries JSON and JSONB in one
4146 // `Value::Json`, so the value-driven namer can only
4147 // ever say `json`. Reported by sentori against
4148 // 7.39.12 — `pg_typeof` over a `jsonb` scalar
4149 // subquery answered `json`, and an ORM branching on
4150 // the type reads that.
4151 | spg_storage::DataType::Jsonb
4152 )
4153 && let Some(n) = pg_typeof_name_for_datatype(shape.ty)
4154 {
4155 return Ok(Value::text(n));
4156 }
4157 return apply_function(name, &[v], ctx);
4158 }
4159 // v7.37 D.1 — COALESCE result-type coercion. PG gives COALESCE the
4160 // common type of its branches, so a typed sibling (`NULL::time`,
4161 // `col::time`) makes the whole expression that type and an untyped
4162 // string-literal branch is coerced to it. Without this,
4163 // `COALESCE(NULL::time, '12:00')::text` rendered the raw `12:00`
4164 // instead of `12:00:00`. Only kicks in when the picked value is a
4165 // bare Text and a non-text cast-target sibling exists.
4166 if name.eq_ignore_ascii_case("coalesce") && !args.is_empty() {
4167 // v7.39 (round 609) — PG's COALESCE does not evaluate a branch past
4168 // the first non-NULL one. This evaluated every branch into a `Vec`
4169 // and so RAISED errors PG never raises: `coalesce(1, 1/0)`,
4170 // `coalesce(NULL, 2, 1/0)` and `coalesce(1, NULL, 1/0)` all failed
4171 // with "division by zero" where PG answers 1, 2 and 1.
4172 //
4173 // A branch after the pick is still READ for its type — that is what
4174 // decides the result's, and `COALESCE(1, 2.5)` is numeric in both
4175 // engines — but its error is discarded, because PG never runs it and
4176 // so never reports it. A branch that fails contributes no type,
4177 // which is the same as SPG having no declared type to widen to.
4178 //
4179 // The two `Vec`s this replaces cost two allocations a row even for
4180 // `coalesce(id, 0)` over a plain INTEGER column, where the answer
4181 // needs none.
4182 let mut result: Option<Value<'static>> = None;
4183 let mut tbuf = [spg_storage::DataType::Int; 8];
4184 let mut ntypes = 0usize;
4185 let mut spill: Vec<spg_storage::DataType> = Vec::new();
4186 for a in args {
4187 let v = if result.is_none() {
4188 eval_expr(a, row, ctx)?
4189 } else {
4190 match eval_expr(a, row, ctx) {
4191 Ok(v) => v,
4192 Err(_) => continue,
4193 }
4194 };
4195 // v7.39 (round 649) — a NULL branch still has a TYPE, and PG
4196 // resolves COALESCE's result from the branches' declared
4197 // types, not from the values that survive. `Value::Null` has
4198 // no `data_type()`, so `coalesce(1::int, NULL::float8)`
4199 // collected only `integer` and answered integer where PG
4200 // answers double precision. Ask the expression when the
4201 // value cannot say — inside the arm that already runs, and
4202 // only on the NULL that would otherwise contribute nothing.
4203 let branch_ty = match v.data_type() {
4204 Some(t) => Some(t),
4205 None => crate::describe::describe_expr(a, ctx.columns).map(|sh| sh.ty),
4206 };
4207 if let Some(t) = branch_ty {
4208 if ntypes < tbuf.len() {
4209 tbuf[ntypes] = t;
4210 ntypes += 1;
4211 } else {
4212 spill.push(t);
4213 }
4214 }
4215 if result.is_none() && !matches!(v, Value::Null) {
4216 result = Some(v);
4217 }
4218 }
4219 let result = result.unwrap_or(Value::Null);
4220 if matches!(result, Value::Text(_)) {
4221 if let Some(target) = args.iter().find_map(coalesce_type_hint) {
4222 return crate::eval::cast::cast_value(result, target);
4223 }
4224 }
4225 // v7.38 (read01) — otherwise widen the picked value to the PG
4226 // common type of all branches (COALESCE(1, 2.5) → numeric).
4227 if spill.is_empty() {
4228 return Ok(widen_to_common(result, &tbuf[..ntypes]));
4229 }
4230 let mut types: Vec<spg_storage::DataType> = tbuf[..ntypes].to_vec();
4231 types.append(&mut spill);
4232 return Ok(widen_to_common(result, &types));
4233 }
4234 let evaluated: Result<Vec<Value<'static>>, _> =
4235 args.iter().map(|a| eval_expr(a, row, ctx)).collect();
4236 let evaluated = evaluated?;
4237 // v7.39 (read01 json.c) — to_json(timestamptz) spells the instant in
4238 // ISO 8601 WITH the session-zone offset ("2024-03-09T14:05:06+00:00"),
4239 // unlike plain timestamp. The runtime value carries no tz tag, so the
4240 // argument's static type is the witness.
4241 if (name.eq_ignore_ascii_case("to_json") || name.eq_ignore_ascii_case("to_jsonb"))
4242 && evaluated.len() == 1
4243 && let Some(Value::Timestamp(t)) = evaluated.first()
4244 && args.first().is_some_and(|a| {
4245 crate::describe::describe_expr(a, ctx.columns)
4246 .is_some_and(|sh| matches!(sh.ty, spg_storage::DataType::Timestamptz))
4247 })
4248 {
4249 let off = ctx.session_tz_offset_at(*t);
4250 let local = t + off;
4251 let days = local.div_euclid(86_400_000_000);
4252 let day_us = local.rem_euclid(86_400_000_000);
4253 let (y, mo, d) = civil_from_days(i32::try_from(days).unwrap_or(0));
4254 let secs = day_us / 1_000_000;
4255 let frac = day_us % 1_000_000;
4256 let (hh, mi, ss) = (secs / 3600, (secs / 60) % 60, secs % 60);
4257 let mut txt = alloc::format!("{y:04}-{mo:02}-{d:02}T{hh:02}:{mi:02}:{ss:02}");
4258 if frac != 0 {
4259 let f = alloc::format!("{frac:06}");
4260 txt.push('.');
4261 txt.push_str(f.trim_end_matches('0'));
4262 }
4263 let (sign, omag) = if off < 0 { ('-', -off) } else { ('+', off) };
4264 let (oh, om) = (omag / 3_600_000_000, (omag / 60_000_000) % 60);
4265 let _ = core::fmt::Write::write_fmt(&mut txt, format_args!("{sign}{oh:02}:{om:02}"));
4266 return Ok(Value::json(alloc::format!("\"{txt}\"")));
4267 }
4268 // v7.39 (enum order knife) — greatest/least over enum-typed arguments
4269 // pick by member order, not label text (PG). The witness needs the arg
4270 // ASTs, so this can't live in the value-level function dispatch.
4271 if (name.eq_ignore_ascii_case("greatest") || name.eq_ignore_ascii_case("least"))
4272 && let Some(labels) = args
4273 .iter()
4274 .find_map(|a| expr_enum_labels(a, ctx.columns, ctx.catalog))
4275 && evaluated
4276 .iter()
4277 .all(|v| matches!(v, Value::Text(_) | Value::Null))
4278 {
4279 let is_greatest = name.eq_ignore_ascii_case("greatest");
4280 let mut best: Option<&Value<'static>> = None;
4281 for v in evaluated.iter().filter(|v| !matches!(v, Value::Null)) {
4282 best = Some(match best {
4283 None => v,
4284 Some(b) => match enum_ord_cmp(labels, v, b) {
4285 Some(core::cmp::Ordering::Greater) if is_greatest => v,
4286 Some(core::cmp::Ordering::Less) if !is_greatest => v,
4287 Some(_) => b,
4288 // A non-member snuck in — fall out to the generic path.
4289 None => return apply_function(name, &evaluated, ctx),
4290 },
4291 });
4292 }
4293 return Ok(best.cloned().unwrap_or(Value::Null));
4294 }
4295 // v7.39 (round 693) — and the same for a declared COLLATION, which
4296 // `least`/`greatest` need for the same structural reason: the witness
4297 // is the argument's column, so it cannot live in the value-level
4298 // dispatch either. Measured on PG18 over a column declaring
4299 // en_US.utf8: `least(a,'d')` is `d` and `greatest(a,'d')` is `Zebra`,
4300 // where byte order gives the pair reversed.
4301 if (name.eq_ignore_ascii_case("greatest") || name.eq_ignore_ascii_case("least"))
4302 && evaluated
4303 .iter()
4304 .all(|v| matches!(v, Value::Text(_) | Value::Null))
4305 && let Some(coll) = greatest_least_collation(args, ctx)
4306 {
4307 let is_greatest = name.eq_ignore_ascii_case("greatest");
4308 let mut best: Option<&Value<'static>> = None;
4309 for v in evaluated.iter().filter(|v| !matches!(v, Value::Null)) {
4310 best = Some(match (best, v) {
4311 (None, _) => v,
4312 (Some(Value::Text(y)), Value::Text(x)) => {
4313 match crate::collate::compare(&coll, x, y) {
4314 Some(core::cmp::Ordering::Greater) if is_greatest => v,
4315 Some(core::cmp::Ordering::Less) if !is_greatest => v,
4316 Some(_) => best.unwrap_or(v),
4317 // Not a collation this build performs after all —
4318 // one answer, from the generic path.
4319 None => return apply_function(name, &evaluated, ctx),
4320 }
4321 }
4322 (Some(b), _) => b,
4323 });
4324 }
4325 return Ok(best.cloned().unwrap_or(Value::Null));
4326 }
4327 // v7.39 (round 621) — an unadorned string literal takes the type the
4328 // function's parameter asks for. `justify_interval('36 hours')` is answered
4329 // by PG and was refused here, and so were `justify_days('35 days')` and
4330 // `justify_hours('27 hours')` — the spelling everyone writes, since typing
4331 // `INTERVAL` in front of the literal is exactly what PG saves you from.
4332 //
4333 // Only a LITERAL is resolved, which is the same boundary round 620 drew for
4334 // the boolean connectives: `justify_interval(t)` over a TEXT column stays
4335 // refused, because PG refuses it too (no such overload). The arg ASTs are
4336 // needed to tell those apart, so this cannot live in the value-level
4337 // dispatch — the same reason the enum witness above sits here.
4338 if let Some(want) = unknown_literal_param_type(name) {
4339 let mut coerced = evaluated;
4340 for (i, a) in args.iter().enumerate() {
4341 if is_unknown_string_literal(a)
4342 && let Some(slot) = coerced.get_mut(i)
4343 {
4344 *slot = cast::cast_value_in(
4345 core::mem::replace(slot, Value::Null),
4346 want.clone(),
4347 false,
4348 )?;
4349 }
4350 }
4351 return apply_function(name, &coerced, ctx);
4352 }
4353 apply_function(name, &evaluated, ctx)
4354}
4355
4356/// v7.39 (round 621) — the parameter type a bare string literal resolves to.
4357///
4358/// PG resolves an `unknown` literal to whatever the chosen overload declares.
4359/// SPG has no overload resolution to hang that on, so the functions whose only
4360/// parameter is unambiguous are listed. `None` leaves the argument alone.
4361fn unknown_literal_param_type(name: &str) -> Option<spg_sql::ast::CastTarget> {
4362 match name.to_ascii_lowercase().as_str() {
4363 "justify_days" | "justify_hours" | "justify_interval" => {
4364 Some(spg_sql::ast::CastTarget::Interval)
4365 }
4366 _ => None,
4367 }
4368}
4369
4370/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4371/// (stack-depth guard budget); body unchanged.
4372#[inline(never)]
4373fn eval_any_all_arm(
4374 expr: &Expr,
4375 op: &BinOp,
4376 array: &Expr,
4377 is_any: bool,
4378 row: &Row<'static>,
4379 ctx: &EvalContext<'_>,
4380) -> Result<Value<'static>, EvalError> {
4381 let lhs = eval_expr(expr, row, ctx)?;
4382 let arr = eval_expr(array, row, ctx)?;
4383 any_all_over(lhs, arr, op, is_any)
4384}
4385
4386/// v7.39 (round 597) — the ANY/ALL comparison with both sides already in
4387/// hand. Split out so a CONSTANT right-hand array can be built once at
4388/// compile time instead of once per row: `WHERE id = ANY (ARRAY[1..10])`
4389/// rebuilt the array for all 500k rows and cost 268 ms against PG18's 8.3,
4390/// rising to 494 ms at twenty elements, while the equivalent
4391/// `id IN (1..10)` took 2.3. The body below is unchanged; it never touched
4392/// `row`.
4393pub(crate) fn any_all_over(
4394 lhs: Value<'static>,
4395 arr: Value<'static>,
4396 op: &BinOp,
4397 is_any: bool,
4398) -> Result<Value<'static>, EvalError> {
4399 if matches!(arr, Value::Null) {
4400 return Ok(Value::Null);
4401 }
4402 // v7.38 (read01) — an unknown-string RHS (`x = ANY('{1,2,3}')`)
4403 // takes the LHS's type: coerce the external array text to the array
4404 // type matching the LHS's element type, like PG.
4405 let arr = match &arr {
4406 Value::Text(_) => {
4407 // The LHS's element type, or TEXT when the LHS is an
4408 // untyped NULL (PG's unknown → text default).
4409 let arr_ty = match lhs.data_type() {
4410 Some(spg_storage::DataType::SmallInt) => spg_storage::DataType::SmallIntArray,
4411 Some(spg_storage::DataType::Int) => spg_storage::DataType::IntArray,
4412 Some(spg_storage::DataType::BigInt) => spg_storage::DataType::BigIntArray,
4413 Some(spg_storage::DataType::Numeric { .. }) => spg_storage::DataType::NumericArray,
4414 Some(spg_storage::DataType::Float) => spg_storage::DataType::FloatArray,
4415 Some(spg_storage::DataType::Bool) => spg_storage::DataType::BoolArray,
4416 Some(spg_storage::DataType::Date) => spg_storage::DataType::DateArray,
4417 _ => spg_storage::DataType::TextArray,
4418 };
4419 crate::conversions::coerce_value(arr.clone(), arr_ty, "", 0).unwrap_or(arr)
4420 }
4421 _ => arr,
4422 };
4423 // Build the element list generically so every scalar array type
4424 // (numeric[], float8[], bool[], date[], …) is accepted, not just
4425 // int/bigint/text.
4426 let Some(len) = array_len(&arr) else {
4427 return Err(EvalError::TypeMismatch {
4428 detail: format!(
4429 "ANY/ALL right-hand side must be an array, got {}",
4430 crate::conversions::pg_type_name_for_error_opt(arr.data_type())
4431 ),
4432 });
4433 };
4434 let elems: Vec<Option<Value>> = (0..len)
4435 .map(|i| match array_element_at(&arr, i) {
4436 Some(Value::Null) | None => None,
4437 Some(v) => Some(v),
4438 })
4439 .collect();
4440 // PG: `x op ANY (empty)` → false and `x op ALL (empty)` →
4441 // true, decided purely by emptiness — the comparison is
4442 // never evaluated, so a NULL LHS is irrelevant. This must
4443 // short-circuit before `saw_null` is seeded from the LHS,
4444 // otherwise `NULL op ANY/ALL (empty)` wrongly yields NULL.
4445 if elems.is_empty() {
4446 return Ok(Value::Bool(!is_any));
4447 }
4448 let mut saw_null = matches!(lhs, Value::Null);
4449 let mut saw_match = false;
4450 let mut saw_mismatch = false;
4451 for elem in elems {
4452 let elem_v = match elem {
4453 Some(v) => v,
4454 None => {
4455 saw_null = true;
4456 continue;
4457 }
4458 };
4459 if matches!(lhs, Value::Null) {
4460 saw_null = true;
4461 continue;
4462 }
4463 match apply_binary(*op, lhs.clone(), elem_v) {
4464 Ok(Value::Bool(true)) => saw_match = true,
4465 Ok(Value::Bool(false)) => saw_mismatch = true,
4466 Ok(Value::Null) => saw_null = true,
4467 Ok(other) => {
4468 return Err(EvalError::TypeMismatch {
4469 detail: format!(
4470 "ANY/ALL comparison didn't return Bool: {}",
4471 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4472 ),
4473 });
4474 }
4475 Err(e) => return Err(e),
4476 }
4477 }
4478 let result = if is_any {
4479 if saw_match {
4480 Value::Bool(true)
4481 } else if saw_null {
4482 Value::Null
4483 } else {
4484 Value::Bool(false)
4485 }
4486 } else if saw_mismatch {
4487 Value::Bool(false)
4488 } else if saw_null {
4489 Value::Null
4490 } else {
4491 Value::Bool(true)
4492 };
4493 Ok(result)
4494}
4495
4496/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4497/// (stack-depth guard budget); body unchanged.
4498#[inline(never)]
4499fn eval_case_arm(
4500 operand: &Option<alloc::boxed::Box<Expr>>,
4501 branches: &[(Expr, Expr)],
4502 else_branch: &Option<alloc::boxed::Box<Expr>>,
4503 row: &Row<'static>,
4504 ctx: &EvalContext<'_>,
4505) -> Result<Value<'static>, EvalError> {
4506 // v7.39 (round 237) — PG resolves the RESULT branches to one type and
4507 // refuses the CASE when they have no common one, before running
4508 // anything. SPG returned whichever branch fired, so
4509 // `CASE WHEN true THEN 1 ELSE 'a'::text END` answered `1` and the same
4510 // expression answered text on another row.
4511 {
4512 // PG resolves the ELSE branch FIRST and then the WHEN results, and
4513 // its message names the running type before the conflicting one —
4514 // which is why `THEN 1 ELSE 'a'::text` reports "text and integer"
4515 // while a two-WHEN `THEN 1 ... THEN true` reports "integer and
4516 // boolean". Probed against 18.4; the order is observable.
4517 let mut results: Vec<&Expr> = Vec::with_capacity(branches.len() + 1);
4518 if let Some(e) = else_branch {
4519 results.push(e);
4520 }
4521 results.extend(branches.iter().map(|(_, r)| r));
4522 unify_branch_types_static("CASE", results, ctx)?;
4523 }
4524 let operand_value = match operand {
4525 Some(o) => Some(eval_expr(o, row, ctx)?),
4526 None => None,
4527 };
4528 // v7.37 D.1 — CASE result-type coercion (same rule as COALESCE): a
4529 // typed result branch (`... THEN '10:00'::time`) makes the whole
4530 // CASE that type, so an untyped string-literal branch is coerced to
4531 // it. Compute the hint once from every THEN/ELSE branch.
4532 let case_hint = branches
4533 .iter()
4534 .map(|(_, t)| t)
4535 .chain(else_branch.iter().map(|b| b.as_ref()))
4536 .find_map(coalesce_type_hint);
4537 // v7.38 (read01) — the CASE result is PG's common type of every
4538 // THEN/ELSE branch, so a taken integer branch is widened to
4539 // numeric when a sibling branch is numeric (and `pg_typeof` /
4540 // downstream division match PG). Only one branch is evaluated, so
4541 // the type must come from the branch expressions statically.
4542 let branch_types: Vec<spg_storage::DataType> = branches
4543 .iter()
4544 .map(|(_, t)| t)
4545 .chain(else_branch.iter().map(|b| b.as_ref()))
4546 .filter_map(|e| crate::describe::describe_expr(e, ctx.columns).map(|s| s.ty))
4547 .collect();
4548 let coerce = |v: Value<'static>| -> Result<Value<'static>, EvalError> {
4549 let v = match (&v, &case_hint) {
4550 (Value::Text(_), Some(target)) => cast::cast_value(v, target.clone())?,
4551 _ => v,
4552 };
4553 Ok(widen_to_common(v, &branch_types))
4554 };
4555 for (when_expr, then_expr) in branches {
4556 let when_value = eval_expr(when_expr, row, ctx)?;
4557 let matched = match &operand_value {
4558 // v7.39 (round 346, M1) — the WHEN condition is a truth value,
4559 // not a boolean-shaped one: `CASE WHEN 1 THEN 'a' END` used to
4560 // answer NULL in BOTH dialects, where MariaDB answers `a` and
4561 // PG raises `argument of CASE/WHEN must be type boolean`.
4562 None => predicate_is_true(&when_value, "CASE/WHEN", ctx.mysql_dialect)?,
4563 // v7.39 (round 412) — under the MySQL default collation the
4564 // `CASE op WHEN v` equality folds Text/BpChar operands (CI +
4565 // accent + PAD SPACE), matching `op = v` outside CASE.
4566 Some(op_v) => {
4567 let (l, r) = if ctx.mysql_dialect {
4568 match (op_v, &when_value) {
4569 // v7.38.18 — fold each side on its OWN type. This
4570 // pair match missed `CASE <char col> WHEN
4571 // '<literal>'`: a BpChar against a Text is neither
4572 // arm, so it compared bytes with the CHAR still
4573 // padded and answered ELSE.
4574 (x, y)
4575 if spg_storage::mysql_fold_value(x).is_some()
4576 && spg_storage::mysql_fold_value(y).is_some() =>
4577 {
4578 (
4579 Value::text(spg_storage::mysql_fold_value(x).unwrap()),
4580 Value::text(spg_storage::mysql_fold_value(y).unwrap()),
4581 )
4582 }
4583 _ => (op_v.clone(), when_value),
4584 }
4585 } else {
4586 (op_v.clone(), when_value)
4587 };
4588 matches!(
4589 apply_binary(spg_sql::ast::BinOp::Eq, l, r)?,
4590 Value::Bool(true)
4591 )
4592 }
4593 };
4594 if matched {
4595 return coerce(eval_expr(then_expr, row, ctx)?);
4596 }
4597 }
4598 match else_branch {
4599 Some(e) => coerce(eval_expr(e, row, ctx)?),
4600 None => Ok(Value::Null),
4601 }
4602}
4603
4604/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4605/// (stack-depth guard budget); body unchanged.
4606#[inline(never)]
4607fn eval_array_slice_arm(
4608 target: &Expr,
4609 lo: &Option<alloc::boxed::Box<Expr>>,
4610 hi: &Option<alloc::boxed::Box<Expr>>,
4611 row: &Row<'static>,
4612 ctx: &EvalContext<'_>,
4613) -> Result<Value<'static>, EvalError> {
4614 let target_v = eval_expr(target, row, ctx)?;
4615 if matches!(target_v, Value::Null) {
4616 return Ok(Value::Null);
4617 }
4618 let bound = |e: Option<&Expr>| -> Result<Option<i64>, EvalError> {
4619 match e {
4620 None => Ok(None),
4621 Some(b) => match eval_expr(b, row, ctx)? {
4622 Value::Null => Ok(None),
4623 Value::Int(n) => Ok(Some(i64::from(n))),
4624 Value::BigInt(n) => Ok(Some(n)),
4625 Value::SmallInt(n) => Ok(Some(i64::from(n))),
4626 other => Err(EvalError::TypeMismatch {
4627 detail: format!(
4628 "array slice bound must be integer, got {}",
4629 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4630 ),
4631 }),
4632 },
4633 }
4634 };
4635 let lo_b = bound(lo.as_deref())?;
4636 let hi_b = bound(hi.as_deref())?;
4637 fn window(len: usize, lo: Option<i64>, hi: Option<i64>) -> (usize, usize) {
4638 let start = lo.map_or(0, |l| (l.max(1) - 1) as usize).min(len);
4639 let end = hi.map_or(len, |h| h.max(0) as usize).min(len);
4640 (start, end.max(start))
4641 }
4642 match target_v {
4643 Value::TextArray(items) => {
4644 let (s, e) = window(items.len(), lo_b, hi_b);
4645 Ok(Value::TextArray(items[s..e].to_vec()))
4646 }
4647 Value::IntArray(items) => {
4648 let (s, e) = window(items.len(), lo_b, hi_b);
4649 Ok(Value::IntArray(items[s..e].to_vec()))
4650 }
4651 Value::BigIntArray(items) => {
4652 let (s, e) = window(items.len(), lo_b, hi_b);
4653 Ok(Value::BigIntArray(items[s..e].to_vec()))
4654 }
4655 other => Err(EvalError::TypeMismatch {
4656 detail: format!(
4657 "slice target must be an array, got {}",
4658 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4659 ),
4660 }),
4661 }
4662}
4663
4664/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4665/// (stack-depth guard budget); body unchanged.
4666#[inline(never)]
4667fn eval_in_list_arm(
4668 expr: &Expr,
4669 list: &[Expr],
4670 negated: bool,
4671 row: &Row<'static>,
4672 ctx: &EvalContext<'_>,
4673) -> Result<Value<'static>, EvalError> {
4674 // v7.39 (round 238) — PG resolves the whole list's type BEFORE comparing
4675 // anything, so `1 IN (1, 'a'::text)` is refused. SPG compared item by
4676 // item and broke on the first match, so the offending element was never
4677 // reached and the predicate quietly answered true. Checked statically,
4678 // like round 237: evaluating the rest of the list to inspect it would
4679 // change when side effects fire.
4680 require_in_list_comparable(expr, list, ctx)?;
4681 // v7.39 (round 364, M4 P2) — a MySQL session folds text before the
4682 // membership test, so `t IN ('FOO')` matches 'Foo'. `BINARY` is not
4683 // reachable through a bare column needle here; the fold is text-only.
4684 // v7.39 (round 370, M4 P4a) — an explicit `COLLATE utf8mb4_bin` needle
4685 // column is byte-wise, so it does not fold. v7.39 (round 371, M4 P4b) —
4686 // a per-expression `… COLLATE utf8mb4_bin` / `BINARY …` on the needle
4687 // OR any list item forces the whole membership test byte-wise.
4688 let in_fold = ctx.mysql_dialect
4689 && !resolve::operand_is_binary_column(expr, ctx)
4690 && !resolve::is_binary_coerced(expr)
4691 && !list.iter().any(|i| resolve::is_binary_coerced(i));
4692 // v7.38.18 — the membership test has ONE collation, the needle's,
4693 // and its NAME decides whether trailing spaces count.
4694 // v7.39 — same as `text_compare`: a literal needle has no column, so
4695 // this used to fall to `None` = NO PAD. `'a ' IN ('a')` answered 1 on
4696 // MySQL 9.7.2 and 0 here under a matched `utf8mb4_general_ci`.
4697 let in_pads = crate::collate::pads_space(
4698 resolve::column_collation_name(expr, ctx)
4699 .or_else(|| resolve::session_collation_name(ctx))
4700 .as_deref(),
4701 );
4702 let needle = mysql_collation_key(eval_expr(expr, row, ctx)?, in_fold, in_pads);
4703 let needle_null = matches!(needle, Value::Null);
4704 let mut saw_null = needle_null && !list.is_empty();
4705 let mut matched = false;
4706 if !needle_null {
4707 for item in list {
4708 let v = mysql_collation_key(eval_expr(item, row, ctx)?, in_fold, in_pads);
4709 if matches!(v, Value::Null) {
4710 saw_null = true;
4711 continue;
4712 }
4713 match apply_binary(BinOp::Eq, needle.clone(), v)? {
4714 Value::Bool(true) => {
4715 matched = true;
4716 break;
4717 }
4718 Value::Bool(false) => {}
4719 Value::Null => saw_null = true,
4720 other => {
4721 return Err(EvalError::TypeMismatch {
4722 detail: format!(
4723 "IN comparison didn't return Bool: {}",
4724 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4725 ),
4726 });
4727 }
4728 }
4729 }
4730 }
4731 let inner = if matched {
4732 Value::Bool(true)
4733 } else if saw_null {
4734 Value::Null
4735 } else {
4736 Value::Bool(false)
4737 };
4738 Ok(match (negated, inner) {
4739 (true, Value::Bool(b)) => Value::Bool(!b),
4740 (_, v) => v,
4741 })
4742}
4743
4744/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4745/// (stack-depth guard budget); body unchanged.
4746#[inline(never)]
4747fn eval_like_arm(
4748 expr: &Expr,
4749 pattern: &Expr,
4750 negated: bool,
4751 case_insensitive: bool,
4752 row: &Row<'static>,
4753 ctx: &EvalContext<'_>,
4754) -> Result<Value<'static>, EvalError> {
4755 let v = eval_expr(expr, row, ctx)?;
4756 let p = eval_expr(pattern, row, ctx)?;
4757 // NULL on either side propagates to NULL — same as PG.
4758 // v7.39 (bpchar epic) — LIKE matches bpchar on its PADDED
4759 // stored form, per PG's bpchar pattern operators.
4760 let (text, pat) = match (v, p) {
4761 (Value::Null, _) | (_, Value::Null) => return Ok(Value::Null),
4762 (Value::Text(a) | Value::BpChar(a), Value::Text(b) | Value::BpChar(b)) => (a, b),
4763 (Value::Text(_) | Value::BpChar(_), other) | (other, _) => {
4764 return Err(EvalError::TypeMismatch {
4765 detail: format!(
4766 "LIKE requires text operands, got {}",
4767 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4768 ),
4769 });
4770 }
4771 };
4772 // v7.25 (round-17) — ILIKE folds both operands (PG
4773 // lowercases per the default collation).
4774 // v7.39 (round 364, M4 P2) — a MySQL session's default collation is
4775 // accent- and case-insensitive, so `LIKE` folds both sides the way
4776 // `ILIKE` does; the wildcards `%` / `_` are not Latin letters so the
4777 // fold leaves them alone.
4778 // v7.39 (round 370, M4 P4a) — an explicit `COLLATE utf8mb4_bin` column
4779 // matches byte-wise, so it does not fold. v7.39 (round 371, M4 P4b) —
4780 // a per-expression `… COLLATE utf8mb4_bin` / `BINARY …` on either the
4781 // value or the pattern forces byte-wise too.
4782 let mysql = ctx.mysql_dialect
4783 && !resolve::operand_is_binary_column(expr, ctx)
4784 && !resolve::operand_is_binary_column(pattern, ctx)
4785 && !resolve::is_binary_coerced(expr)
4786 && !resolve::is_binary_coerced(pattern);
4787 let m = if case_insensitive {
4788 like_match(&text.to_lowercase(), &pat.to_lowercase())?
4789 } else if mysql {
4790 like_match(
4791 &spg_storage::mysql_ci_fold(&text),
4792 &spg_storage::mysql_ci_fold(&pat),
4793 )?
4794 } else {
4795 like_match(&text, &pat)?
4796 };
4797 Ok(Value::Bool(if negated { !m } else { m }))
4798}
4799
4800/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4801/// (stack-depth guard budget); body unchanged.
4802#[inline(never)]
4803fn eval_extract_arm(
4804 field: &spg_sql::ast::ExtractField,
4805 source: &Expr,
4806 row: &Row<'static>,
4807 ctx: &EvalContext<'_>,
4808) -> Result<Value<'static>, EvalError> {
4809 let v = eval_expr(source, row, ctx)?;
4810 extract_from_value(field, v, source, ctx)
4811}
4812
4813/// v7.39 (round 595) — the field extraction, with the source value already
4814/// in hand. Split out so the compiled-predicate program can pop the source
4815/// off its stack instead of handing the whole node back to the interpreter:
4816/// one non-compilable node used to disqualify the entire WHERE, and
4817/// `WHERE extract(year FROM t) = 2020` was interpreting the column read and
4818/// the comparison too. The body below is unchanged; it never touched `row`.
4819pub(crate) fn extract_from_value(
4820 field: &spg_sql::ast::ExtractField,
4821 v: Value<'static>,
4822 source: &Expr,
4823 ctx: &EvalContext<'_>,
4824) -> Result<Value<'static>, EvalError> {
4825 // v7.39 (round 382) — MySQL coerces a date/time STRING to its temporal
4826 // value for EXTRACT (`EXTRACT(YEAR FROM '2020-05-15')` is 2020, and the
4827 // time fields read a `'... HH:MM:SS'` string); PG needs a typed source.
4828 let v = match &v {
4829 Value::Text(s) if ctx.mysql_dialect => text_as_temporal(s).unwrap_or(v),
4830 _ => v,
4831 };
4832 // v7.39 (tz epic) — timezone[_hour|_minute] of a timestamptz
4833 // reports the SESSION offset at that instant (PG: 32400 for
4834 // Tokyo; -14400 for New York in July).
4835 if matches!(
4836 field,
4837 spg_sql::ast::ExtractField::Timezone
4838 | spg_sql::ast::ExtractField::TimezoneHour
4839 | spg_sql::ast::ExtractField::TimezoneMinute
4840 ) && let Value::Timestamp(t) = &v
4841 && crate::describe::describe_expr(source, ctx.columns)
4842 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
4843 {
4844 let off_secs = ctx.session_tz_offset_at(*t) / 1_000_000;
4845 let n = match field {
4846 spg_sql::ast::ExtractField::Timezone => off_secs,
4847 spg_sql::ast::ExtractField::TimezoneHour => off_secs / 3600,
4848 _ => (off_secs / 60) % 60,
4849 };
4850 // v7.39 (round 253) — numeric, like every other EXTRACT result.
4851 return Ok(Value::Numeric {
4852 scaled: i128::from(n),
4853 scale: 0,
4854 kind: spg_storage::NumericKind::Finite,
4855 });
4856 }
4857 // v7.39 (round 523) — and every OTHER field of a timestamptz reads
4858 // the local clock in the session zone, which is the whole reason PG
4859 // has the type. `extract(hour from …)` answered the UTC hour under
4860 // `SET TimeZone = 'Asia/Tokyo'` — 0 where PG says 9 — and
4861 // `extract(dow …)` therefore named the wrong DAY, so a report
4862 // grouped by weekday put nine hours of every Sunday under Saturday.
4863 // Only fields of the local clock shift; epoch and julian are
4864 // absolute, and the timezone fields answered above.
4865 let v = match &v {
4866 Value::Timestamp(t)
4867 if !matches!(
4868 field,
4869 spg_sql::ast::ExtractField::Epoch
4870 | spg_sql::ast::ExtractField::Julian
4871 | spg_sql::ast::ExtractField::Timezone
4872 | spg_sql::ast::ExtractField::TimezoneHour
4873 | spg_sql::ast::ExtractField::TimezoneMinute
4874 ) && crate::describe::describe_expr(source, ctx.columns)
4875 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz)) =>
4876 {
4877 Value::Timestamp(t.saturating_add(ctx.session_tz_offset_at(*t)))
4878 }
4879 _ => v,
4880 };
4881 // v7.39 (round 253) — the source's PG type name for error wording,
4882 // upgraded from the declared type when statically known (a tstz
4883 // VALUE is indistinguishable from a timestamp). Only a cast /
4884 // column is trusted (the r237 lesson: describe_expr reports a
4885 // binary operator as its left operand's type).
4886 let static_declared = matches!(source, Expr::Cast { .. } | Expr::Column(_))
4887 .then(|| crate::describe::describe_expr(source, ctx.columns))
4888 .flatten()
4889 .map(|sch| sch.ty);
4890 let src_name = match static_declared {
4891 Some(spg_storage::DataType::Timestamptz) => "timestamp with time zone",
4892 _ => datetime::value_src_type_name(&v),
4893 };
4894 // PG rejects the timezone family on a plain timestamp (0A000);
4895 // only reject when the declared type is STATICALLY timestamp — a
4896 // dynamic value stays lenient (the pre-r253 zero answer).
4897 if matches!(
4898 field,
4899 spg_sql::ast::ExtractField::Timezone
4900 | spg_sql::ast::ExtractField::TimezoneHour
4901 | spg_sql::ast::ExtractField::TimezoneMinute
4902 ) && matches!(static_declared, Some(spg_storage::DataType::Timestamp))
4903 {
4904 return Err(EvalError::TypeMismatch {
4905 detail: alloc::format!(
4906 "unit \"{}\" not supported for type timestamp without time zone",
4907 alloc::format!("{field}").to_lowercase()
4908 ),
4909 });
4910 }
4911 // v7.39 (round 418) — MySQL's compound units (`DAY_SECOND`, `YEAR_MONTH`,
4912 // …) reach here as `ExtractField::Other`, which PG rejects. Under the
4913 // MySQL dialect they pack several components into one integer instead.
4914 if ctx.mysql_dialect
4915 && let spg_sql::ast::ExtractField::Other(name) = field
4916 && let Some(packed) = crate::eval::datetime::mysql_compound_extract(name, &v)
4917 {
4918 return Ok(packed);
4919 }
4920 extract_field(field, &v, src_name)
4921}
4922
4923/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4924/// (stack-depth guard budget); body unchanged.
4925#[inline(never)]
4926fn eval_array_subscript_arm(
4927 expr: &Expr,
4928 row: &Row<'static>,
4929 ctx: &EvalContext<'_>,
4930) -> Result<Value<'static>, EvalError> {
4931 // Collect the whole subscript chain so PG's multi-dimensional
4932 // access (`arr[i][j]` is ONE N-subscript op) is distinguishable
4933 // from chained 1-D indexing. `arr[1][2]` parses as
4934 // `(arr[1])[2]`; PG indexes the matrix directly and returns NULL
4935 // for a partial subscript (`arr[1]` on a 2-D array is NULL).
4936 let mut idx_exprs: Vec<&Expr> = Vec::new();
4937 let mut base = expr;
4938 while let Expr::ArraySubscript { target, index } = base {
4939 idx_exprs.push(index);
4940 base = target;
4941 }
4942 idx_exprs.reverse();
4943 let base_v = eval_expr(base, row, ctx)?;
4944 if matches!(
4945 base_v,
4946 Value::IntArray2D(_)
4947 | Value::BigIntArray2D(_)
4948 | Value::TextArray2D(_)
4949 | Value::BoolArray2D(_)
4950 ) {
4951 return eval_matrix_subscript(&base_v, &idx_exprs, row, ctx);
4952 }
4953 // 1-D array / JSON: apply each subscript left-to-right. This
4954 // reproduces the prior single-subscript semantics exactly, and
4955 // chained JSON (`j['a']['b']`) still resolves step by step.
4956 let mut cur = base_v;
4957 for ix in idx_exprs {
4958 cur = apply_one_subscript(cur, ix, row, ctx)?;
4959 }
4960 Ok(cur)
4961}
4962
4963/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4964/// (stack-depth guard budget); body unchanged.
4965#[inline(never)]
4966fn eval_field_access_arm(
4967 base: &Expr,
4968 field: &str,
4969 row: &Row<'static>,
4970 ctx: &EvalContext<'_>,
4971) -> Result<Value<'static>, EvalError> {
4972 // v7.38 (read01, T9) — composite field access `(expr).field`.
4973 // The base evaluates to a record; look the member up by name
4974 // (`f1`..`fN` for an anonymous ROW, base column names for a
4975 // whole-row). A NULL record yields NULL (PG semantics).
4976 let v = eval_expr(base, row, ctx)?;
4977 match v {
4978 Value::Null => Ok(Value::Null),
4979 Value::Composite(fields) => fields
4980 .into_iter()
4981 .find(|(name, _)| name == field)
4982 .map(|(_, val)| val)
4983 .ok_or_else(|| missing_field_error(base, field, ctx)),
4984 _ => Err(not_a_composite_error(base, field, ctx)),
4985 }
4986}
4987
4988/// v7.39 (round 285) — PG words a missing composite field three ways, and
4989/// which one you get depends on the base expression's STATIC type, not on
4990/// the value:
4991///
4992/// * a named composite — `column "nosuch" not found in data type rc9`
4993/// * a whole-row table reference — `column rt8.nosuch does not exist`
4994/// (unquoted, and qualified — the odd one out)
4995/// * an anonymous ROW or `::record` — `could not identify column
4996/// "nosuch" in record data type`
4997///
4998/// All three read off live PG 18.4. A `Value::Composite` carries its field
4999/// names but not its type name, so the base expression is what decides.
5000fn missing_field_error(base: &Expr, field: &str, ctx: &EvalContext<'_>) -> EvalError {
5001 // v7.39 (round 307, V25) — the named type may arrive by cast OR from
5002 // the schema of a column that a projection produced, so ask once and
5003 // let the catalog say whether it is a composite. Before this only
5004 // the cast spelling was recognised, which is why a composite that
5005 // came through a derived table or a CTE — where the base is a plain
5006 // column — fell through to the anonymous-record wording.
5007 if let Some(name) = base_named_type(base, ctx)
5008 && ctx
5009 .catalog
5010 .is_some_and(|c| c.composite_types().contains_key(name))
5011 {
5012 return EvalError::TypeMismatch {
5013 detail: alloc::format!("column \"{field}\" not found in data type {name}"),
5014 };
5015 }
5016 if let Expr::Column(c) = base
5017 && c.qualifier.is_none()
5018 {
5019 // A column DECLARED as a named composite reports that type — the
5020 // schema records it in `user_composite_type`, which is the only
5021 // place the name survives (a `Value::Composite` does not carry it).
5022 if let Some(name) = ctx
5023 .columns
5024 .iter()
5025 .find(|col| col.name.eq_ignore_ascii_case(&c.name))
5026 .and_then(|col| col.user_composite_type.as_ref())
5027 {
5028 return EvalError::TypeMismatch {
5029 detail: alloc::format!("column \"{field}\" not found in data type {name}"),
5030 };
5031 }
5032 // A whole-row reference to a real table is the odd wording out:
5033 // qualified, and unquoted.
5034 if ctx.catalog.is_some_and(|cat| cat.get(&c.name).is_some()) {
5035 return EvalError::TypeMismatch {
5036 detail: alloc::format!("column {}.{field} does not exist", c.name),
5037 };
5038 }
5039 }
5040 EvalError::TypeMismatch {
5041 detail: alloc::format!("could not identify column \"{field}\" in record data type"),
5042 }
5043}
5044
5045/// v7.39 (round 307, V25) — the user-declared type name behind a field
5046/// access, if any: either written as a cast (`ROW(…)::rc9`) or carried on
5047/// the column's schema.
5048///
5049/// All three schema slots are consulted rather than just the composite
5050/// one. A projection currently files the name of a `::rc9` cast under
5051/// `user_enum_type` — `expr_enum_type_name` answers for ANY named cast,
5052/// without asking whether the name is an enum — so keying off one slot
5053/// would answer for some shapes and not others. The caller decides what
5054/// the name MEANS by asking the catalog, which is the only thing that
5055/// actually knows; this function just finds it.
5056fn base_named_type<'c>(base: &'c Expr, ctx: &'c EvalContext<'_>) -> Option<&'c str> {
5057 match base {
5058 Expr::Cast {
5059 target: CastTarget::Named(name),
5060 ..
5061 } => Some(name.as_str()),
5062 Expr::Column(c) => ctx
5063 .columns
5064 .iter()
5065 .find(|sc| sc.name.eq_ignore_ascii_case(&c.name))
5066 .and_then(|sc| {
5067 sc.user_composite_type
5068 .as_deref()
5069 .or(sc.user_domain_type.as_deref())
5070 .or(sc.user_enum_type.as_deref())
5071 }),
5072 _ => None,
5073 }
5074}
5075
5076/// v7.39 (round 307, V25) — PG's wording when field notation is applied
5077/// to something that is not a composite at all. It names the type:
5078/// `column notation .f applied to type pos9, which is not a composite
5079/// type` — for a domain and an enum alike. Only when the base has no
5080/// user-declared type at all does the generic message stand.
5081fn not_a_composite_error(base: &Expr, field: &str, ctx: &EvalContext<'_>) -> EvalError {
5082 if let Some(name) = base_named_type(base, ctx)
5083 && ctx.catalog.is_some_and(|c| {
5084 c.enum_types().contains_key(name) || c.domain_types().contains_key(name)
5085 })
5086 {
5087 return EvalError::TypeMismatch {
5088 detail: alloc::format!(
5089 "column notation .{field} applied to type {name}, which is not a composite type"
5090 ),
5091 };
5092 }
5093 EvalError::TypeMismatch {
5094 detail: alloc::format!("field access `.{field}` requires a composite (record) value"),
5095 }
5096}
5097
5098/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
5099/// (stack-depth guard budget); body unchanged.
5100#[inline(never)]
5101/// v7.39 (round 328, V45) — the three-valued boolean tests. None of them
5102/// ever answers NULL: a NULL input is "not true" and "not false", and IS
5103/// UNKNOWN is precisely the NULL case. Verified against PG 18.4 —
5104/// `NULL::bool IS TRUE` is false, `IS NOT TRUE` true, `IS UNKNOWN` true,
5105/// and `false IS NOT FALSE` false.
5106fn eval_bool_test_arm(
5107 expr: &Expr,
5108 value: Option<bool>,
5109 negated: bool,
5110 row: &Row<'static>,
5111 ctx: &EvalContext<'_>,
5112) -> Result<Value<'static>, EvalError> {
5113 let v = eval_expr(expr, row, ctx)?;
5114 let hit = match (value, &v) {
5115 // IS UNKNOWN — the input is NULL.
5116 (None, Value::Null) => true,
5117 // v7.39 (round 625) — and PG rejects a non-boolean here too:
5118 // `argument of IS UNKNOWN must be type boolean`. MySQL has no
5119 // IS UNKNOWN, so there is no dialect branch.
5120 (None, Value::Bool(_)) => false,
5121 (None, other) => {
5122 return Err(EvalError::TypeMismatch {
5123 detail: alloc::format!(
5124 "argument of IS {}UNKNOWN must be type boolean, not type {}",
5125 if negated { "NOT " } else { "" },
5126 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5127 ),
5128 });
5129 }
5130 (Some(_), Value::Null) => false,
5131 (Some(want), Value::Bool(b)) => *b == want,
5132 // v7.39 (round 397) — MySQL reads a non-boolean as a truth value
5133 // for `IS TRUE` / `IS FALSE` (`5 IS TRUE` is 1, `0 IS FALSE` is 1,
5134 // `'abc' IS TRUE` is 0). PG rejects a non-boolean at parse time, so
5135 // this only fires under the dialect; a NULL is already handled.
5136 (Some(want), other) if ctx.mysql_dialect => mysql_truthy(other) == want,
5137 // v7.39 (round 625, S05b/F29) — on PG a non-boolean is REJECTED, and
5138 // the comment above said so while the arm below answered `false`
5139 // anyway. `1 IS TRUE` came back false, which reads as "the test was
5140 // run and did not hold" rather than "you cannot ask this of an
5141 // integer" — the wrong answer for every non-boolean type, eight of
5142 // them measured. PG's own sentence, which names the operator and the
5143 // type it got.
5144 (Some(_), other) => {
5145 return Err(EvalError::TypeMismatch {
5146 detail: alloc::format!(
5147 "argument of IS {}{} must be type boolean, not type {}",
5148 if negated { "NOT " } else { "" },
5149 if value == Some(true) { "TRUE" } else { "FALSE" },
5150 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5151 ),
5152 });
5153 }
5154 };
5155 Ok(Value::Bool(hit != negated))
5156}
5157
5158fn eval_is_null_arm(
5159 expr: &Expr,
5160 negated: bool,
5161 row: &Row<'static>,
5162 ctx: &EvalContext<'_>,
5163) -> Result<Value<'static>, EvalError> {
5164 // v7.38 (read01 P4.11) — `ROW(...) IS [NOT] NULL` is evaluated
5165 // field-wise, not as a whole-value null test: a row IS NULL when
5166 // every field is null, and IS NOT NULL when every field is
5167 // non-null — so the two are NOT simple negations (ROW(1,NULL) is
5168 // neither). A field that is itself a row is a non-null value, so
5169 // the check does not recurse. The `(a, b) IS NULL` tuple spelling
5170 // is already desugared to `a IS NULL AND b IS NULL` in the parser;
5171 // this covers the explicit `ROW(...)` constructor.
5172 if let Expr::FunctionCall { name, args } = expr
5173 && name.eq_ignore_ascii_case("row")
5174 {
5175 let mut all_null = true;
5176 let mut all_non_null = true;
5177 for a in args {
5178 if matches!(eval_expr(a, row, ctx)?, Value::Null) {
5179 all_non_null = false;
5180 } else {
5181 all_null = false;
5182 }
5183 }
5184 return Ok(Value::Bool(if negated { all_non_null } else { all_null }));
5185 }
5186 // v7.39 (round 962) — the same field-wise rule for a row-valued
5187 // EXPRESSION, not just the `ROW(...)` spelling. P4.11 keyed on the
5188 // syntax, so every other way to hold a row got the whole-value test:
5189 // measured against PG18.4, `SELECT an IS NULL FROM an` on a row whose
5190 // every column is NULL answered `t` there and `f` here, and a column
5191 // declared with a composite type behaved the same way. Round 961 made
5192 // whole-row references reachable through a projection, which is what
5193 // surfaced it.
5194 //
5195 // Fields are tested exactly as the `ROW(...)` arm tests its
5196 // arguments, without recursing — a field that is itself a row is a
5197 // non-null value.
5198 let v = eval_expr(expr, row, ctx)?;
5199 if let Value::Composite(fields) = &v {
5200 let mut all_null = true;
5201 let mut all_non_null = true;
5202 for (_, f) in fields {
5203 if matches!(f, Value::Null) {
5204 all_non_null = false;
5205 } else {
5206 all_null = false;
5207 }
5208 }
5209 return Ok(Value::Bool(if negated { all_non_null } else { all_null }));
5210 }
5211 let is_null = matches!(v, Value::Null);
5212 Ok(Value::Bool(if negated { !is_null } else { is_null }))
5213}
5214
5215pub fn eval_expr(
5216 expr: &Expr,
5217 row: &Row<'static>,
5218 ctx: &EvalContext<'_>,
5219) -> Result<Value<'static>, EvalError> {
5220 // v7.38 (read01 P3.25) — guard against a native stack overflow on a
5221 // pathologically nested expression (`a AND a AND … ` × thousands): the
5222 // recursion base is seeded on the outermost call, and once a deeper
5223 // call has consumed more than the budget we return an error the way
5224 // PG's check_stack_depth() does, rather than aborting the process.
5225 let sp = eval_stack_ptr();
5226 let base = ctx.recursion_base.get();
5227 if base == 0 {
5228 ctx.recursion_base.set(sp);
5229 } else if base.saturating_sub(sp) > MAX_EVAL_STACK_BYTES {
5230 return Err(EvalError::StackDepthExceeded);
5231 }
5232 match expr {
5233 // v7.39.2 — a collation does not change the VALUE, only what
5234 // comparing it means. So evaluation is the inner expression's,
5235 // and the collation is read by the comparison sites rather than
5236 // here. An arm rather than a traversal join, because that is a
5237 // decision and not a walk.
5238 Expr::Collate { expr, collation } => {
5239 // The name is checked HERE rather than in a walk over every
5240 // statement: a `COLLATE` that is never evaluated is never
5241 // reached, and one that is reached is reached exactly once
5242 // per evaluation. PG rejects an unknown name whether or not
5243 // anything compares it — `SELECT 'a' COLLATE "nosuch"` is an
5244 // error there — and evaluating the projection is that.
5245 // The wire question is asked by the PARSER, which knows
5246 // the session's dialect; this context does not always (the
5247 // INSERT path builds one with `mysql_dialect: false`). Here
5248 // it is only "does SPG have a collation by this name".
5249 if !crate::collate::is_known(collation) {
5250 return Err(EvalError::TypeMismatch {
5251 detail: crate::collate::unknown_collation_text(collation, ctx.mysql_dialect),
5252 });
5253 }
5254 eval_expr(expr, row, ctx)
5255 }
5256 Expr::AggregateOrdered { .. } => Err(EvalError::TypeMismatch {
5257 detail: "aggregate ORDER BY is only valid inside an aggregating SELECT".into(),
5258 }),
5259 // A named argument is only meaningful inside a call, where the callee's
5260 // parameter names give it a slot. Anywhere else it is a syntax error,
5261 // and saying so beats silently evaluating it as if the name were absent.
5262 Expr::NamedArg { name, .. } => Err(EvalError::TypeMismatch {
5263 detail: alloc::format!("named argument \"{name}\" is only valid in a function call"),
5264 }),
5265 Expr::Literal(l) => Ok(literal_to_value(l)),
5266 Expr::Column(c) => resolve_column(c, row, ctx),
5267 Expr::Placeholder(n) => {
5268 let idx = usize::from(*n).saturating_sub(1);
5269 ctx.params
5270 .get(idx)
5271 .cloned()
5272 .ok_or_else(|| EvalError::PlaceholderOutOfRange {
5273 n: *n,
5274 bound: u16::try_from(ctx.params.len()).unwrap_or(u16::MAX),
5275 })
5276 }
5277 // v7.39 (round 620) — an unadorned string literal carries PG's
5278 // `unknown` type, and a boolean connective is a context that resolves
5279 // it TO boolean. `'true' AND true`, `'f' OR false` and `NOT 'a'` are
5280 // answered by PG (`t`, `f`, and the input-syntax error respectively)
5281 // and were all refused here with `argument of … must be type boolean,
5282 // not type text` — the message PG reserves for an operand that really
5283 // IS text (`''::TEXT AND true`), which stays refused. Out-of-line and
5284 // behind a literal-shaped guard: this is the recursive frame the
5285 // 768 KiB stack budget is tuned against.
5286 Expr::Unary {
5287 op: spg_sql::ast::UnOp::Not,
5288 expr,
5289 } if !ctx.mysql_dialect && is_unknown_string_literal(expr) => apply_unary(
5290 spg_sql::ast::UnOp::Not,
5291 coerce_unknown_literal_to_bool(expr)?,
5292 ),
5293 Expr::Unary { op, expr } => {
5294 let v = eval_expr(expr, row, ctx)?;
5295 // The MySQL-specific unary readings (NOT any truth value, `-`/`~`
5296 // on a string, `~` unsigned) live out-of-line: `eval_expr` is the
5297 // recursive frame the 768 KiB stack-depth budget is tuned
5298 // against, and locals added here cost one nesting level each (the
5299 // round-305 / round-383 frame cliff).
5300 if ctx.mysql_dialect {
5301 if let Some(r) = mysql_unary_arm(*op, &v) {
5302 return r;
5303 }
5304 }
5305 apply_unary(*op, v)
5306 }
5307 // v7.39 (round 346, M1) — MariaDB reads both sides of AND / OR as
5308 // truth values (`1 AND 2` is 1, measured). apply_binary has no
5309 // dialect, so the coercion happens here, where it does. The body
5310 // is out-of-line: `eval_expr` is the recursive frame the 768 KiB
5311 // stack-depth budget is tuned against, and locals added here cost
5312 // one nesting level each (the round-305 frame cliff).
5313 Expr::Binary { lhs, op, rhs }
5314 if ctx.mysql_dialect && matches!(op, BinOp::And | BinOp::Or | BinOp::LogicalXor) =>
5315 {
5316 eval_mysql_connective(lhs, *op, rhs, row, ctx)
5317 }
5318 // v7.39 (round 620/621) — the unknown-literal resolution and the
5319 // short circuit, both out of line. Placed AFTER the MySQL arm so the
5320 // dialect keeps its own reading of these connectives.
5321 Expr::Binary { lhs, op, rhs } if matches!(op, BinOp::And | BinOp::Or) => {
5322 eval_connective(lhs, *op, rhs, row, ctx)
5323 }
5324 Expr::Binary { lhs, op, rhs } => {
5325 // v7.32 (P4 borrow channel) — comparison fast path. A pure
5326 // comparison op only reads its operands and returns Bool,
5327 // and for non-NUMERIC / non-INTERVAL / non-CI-collation
5328 // operands `apply_binary` IS just the NULL-3VL check plus
5329 // the ref-based `compare` (NUMERIC routes through fixed-
5330 // point `apply_binary_numeric`; INTERVAL through
5331 // `apply_binary_interval`; CI columns fold). So read the
5332 // operands borrowed — a column cell is no longer cloned
5333 // just to compare it (`WHERE thread_id != ''` alone cloned
5334 // one Text cell per scanned row). Anything that needs the
5335 // owned path falls through unchanged.
5336 if matches!(
5337 op,
5338 BinOp::Eq | BinOp::NotEq | BinOp::Lt | BinOp::LtEq | BinOp::Gt | BinOp::GtEq
5339 ) {
5340 let lc = eval_expr_cow(lhs, row, ctx)?;
5341 let rc = eval_expr_cow(rhs, row, ctx)?;
5342 // v7.39 (enum order knife) — enum-typed operands compare by
5343 // member order, not label text. Cold unless both sides are
5344 // Text and the catalog has enum types at all.
5345 if matches!(lc.as_ref(), Value::Text(_)) && matches!(rc.as_ref(), Value::Text(_)) {
5346 if let Some(r) = enum_compare_hook(*op, lhs, rhs, lc.as_ref(), rc.as_ref(), ctx)
5347 {
5348 return r;
5349 }
5350 // v7.39 (round 693) — and the collation hook, under the
5351 // same Text/Text gate for the same reason.
5352 if let Some(r) =
5353 collate_compare_hook(*op, lhs, rhs, lc.as_ref(), rc.as_ref(), ctx)
5354 {
5355 return r;
5356 }
5357 }
5358 // v7.39 (round 351, M11) — the three conditions fold into
5359 // ONE call. Adding a fourth as another `||` here tipped the
5360 // 768 KiB stack guard on its own (measured — this is the
5361 // hottest recursive frame there is); one call is cheaper
5362 // than the three it replaces.
5363 let owned_path = needs_owned_compare(lc.as_ref(), rc.as_ref(), lhs, rhs, ctx);
5364 if !owned_path {
5365 if lc.as_ref().is_null() || rc.as_ref().is_null() {
5366 return Ok(Value::Null);
5367 }
5368 return compare(*op, lc.as_ref(), rc.as_ref()).map_err(|e| {
5369 unknown_literal_cmp_error(e, lhs, rhs, lc.as_ref(), rc.as_ref())
5370 });
5371 }
5372 let (l, r) = collation_fold_for_compare(
5373 *op,
5374 lhs,
5375 rhs,
5376 lc.into_owned(),
5377 rc.into_owned(),
5378 ctx,
5379 );
5380 // The owned call consumes the values; the rewrite needs the
5381 // literal's text and the other side's type only on the ERROR
5382 // path, so capture those two up front — the capture itself is
5383 // gated on the cheap expr test, so a comparison with no
5384 // unknown literal pays one branch.
5385 let probe = (is_unknown_string_literal(lhs) || is_unknown_string_literal(rhs))
5386 .then(|| (l.clone(), r.clone()));
5387 return apply_binary_in(*op, l, r, ctx.mysql_dialect).map_err(|e| match &probe {
5388 Some((pl, pr)) => unknown_literal_cmp_error(e, lhs, rhs, pl, pr),
5389 None => e,
5390 });
5391 }
5392 let l = eval_expr(lhs, row, ctx)?;
5393 let r = eval_expr(rhs, row, ctx)?;
5394 // v7.17.0 Phase 2.5 — collation-aware text comparison.
5395 // When either operand of a comparison op references a
5396 // column declared `COLLATE "case_insensitive"` (or any
5397 // MySQL `_ci` collation), case-fold both sides before
5398 // the byte-wise compare so `WHERE name = 'foo'` matches
5399 // stored `'Foo'`. Non-Text values fall straight through
5400 // — the helper is a no-op outside Text-Text equality
5401 // and inequality.
5402 let (l, r) = collation_fold_for_compare(*op, lhs, rhs, l, r, ctx);
5403 // v7.39 (GUC knife 4) — `date/interval/float || text` textifies
5404 // through the out-functions, which honour the session render
5405 // style. Pre-render the style-sensitive operand here (the
5406 // orthodox home is an implicit-cast node at type resolution;
5407 // until then this keeps apply_binary style-free). Default
5408 // style short-circuits — text_concat's own value_to_text
5409 // produces the identical bytes.
5410 if matches!(op, spg_sql::ast::BinOp::Concat)
5411 && ctx.render_style != format::RenderStyle::default()
5412 {
5413 let styled = |v: Value<'static>| -> Value<'static> {
5414 match &v {
5415 Value::Date(_)
5416 | Value::Timestamp(_)
5417 | Value::Interval { .. }
5418 | Value::Float(_)
5419 | Value::Real(_) => {
5420 Value::text(values::value_to_text_styled(&v, &ctx.render_style))
5421 }
5422 _ => v,
5423 }
5424 };
5425 let (sl, sr) = (styled(l), styled(r));
5426 return apply_binary(*op, sl, sr);
5427 }
5428 // v7.38.13 — in PG mode `apply_binary_mysql_unsigned` checks a
5429 // dialect flag and forwards, and `apply_binary_in` does the
5430 // same; both take two 48-byte `Value`s by value. Skip them.
5431 if ctx.mysql_dialect {
5432 apply_binary_mysql_unsigned(*op, lhs, rhs, l, r, ctx)
5433 } else {
5434 // v7.39.13 — `||` against a `"char"` refuses, and PG names
5435 // an untyped literal `unknown` in the refusal. SPG holds
5436 // both `'x'` and `'x'::text` as `Value::Text`, so only the
5437 // EXPRESSION can tell them apart — the same knowledge
5438 // `unknown_literal_cmp_error` already uses one arm above,
5439 // for comparisons. Reported by sentori: the refusal is
5440 // right, the left-hand type named in it was not.
5441 //
5442 // On the error path only; a successful concat pays one
5443 // branch on the op.
5444 let untyped = matches!(op, spg_sql::ast::BinOp::Concat)
5445 && (is_unknown_string_literal(lhs) || is_unknown_string_literal(rhs));
5446 binop::apply_binary(*op, l, r).map_err(|e| {
5447 if untyped {
5448 name_untyped_concat_operands(e, lhs, rhs)
5449 } else {
5450 e
5451 }
5452 })
5453 }
5454 }
5455 Expr::Cast { expr, target } => eval_cast_arm(expr, target, row, ctx),
5456 Expr::FieldAccess { base, field } => eval_field_access_arm(base, field, row, ctx),
5457 Expr::IsNull { expr, negated } => eval_is_null_arm(expr, *negated, row, ctx),
5458 // v7.39 (round 328, V45) — `x IS [NOT] TRUE | FALSE | UNKNOWN`.
5459 // Out-of-line like its IS NULL neighbour: an inline body here
5460 // grows every frame of the recursive evaluator, which is what
5461 // tipped the 512KB depth guard in round 305.
5462 Expr::BoolTest {
5463 expr,
5464 value,
5465 negated,
5466 } => eval_bool_test_arm(expr, *value, *negated, row, ctx),
5467 Expr::FunctionCall { name, args } => eval_function_call_arm(name, args, row, ctx),
5468 // v7.39 (read01 round 100) — VARIADIC is spliced into its enclosing
5469 // call before the args are evaluated (see eval_function_call_arm); a
5470 // bare one reaching here was written outside a function call.
5471 Expr::Variadic(_) => Err(EvalError::TypeMismatch {
5472 detail: "VARIADIC is only valid as a function-call argument".into(),
5473 }),
5474 Expr::Like {
5475 expr,
5476 pattern,
5477 negated,
5478 case_insensitive,
5479 } => eval_like_arm(expr, pattern, *negated, *case_insensitive, row, ctx),
5480 Expr::Extract { field, source } => eval_extract_arm(field, source, row, ctx),
5481 // v4.10: subquery nodes should have been resolved into
5482 // Literal / InList nodes by Engine::resolve_select_subqueries
5483 // before the row loop. Anything reaching here is a bug.
5484 Expr::ScalarSubquery(_)
5485 | Expr::Exists { .. }
5486 | Expr::InSubquery { .. }
5487 | Expr::RowInSubquery { .. }
5488 | Expr::RowCmpSubquery { .. } => Err(EvalError::TypeMismatch {
5489 detail: "subquery reached row eval — engine resolver bug".into(),
5490 }),
5491 // v7.30.2 (mailrs round-25) — flat `expr [NOT] IN (a, b, …)`.
5492 // Iterative scan with PG three-valued logic: TRUE on the first
5493 // Eq match; if nothing matched, NULL when the needle is NULL or
5494 // any comparison was NULL; FALSE otherwise. Empty list (only
5495 // reachable via an empty subquery result) is FALSE / TRUE even
5496 // for a NULL needle — no comparison ever happens.
5497 Expr::InList {
5498 expr,
5499 list,
5500 negated,
5501 } => eval_in_list_arm(expr, list, *negated, row, ctx),
5502 // v4.12: window functions should have been rewritten into
5503 // synthetic __win_N column references by
5504 // exec_select_with_window before row eval. Anything
5505 // reaching here is similarly a bug.
5506 Expr::WindowFunction { .. } => Err(EvalError::TypeMismatch {
5507 detail: "window function reached row eval — engine rewrite bug".into(),
5508 }),
5509 // v7.10.10 — `ARRAY[expr, expr, …]` constructor.
5510 // v7.11.13 — element-type detection: all integers →
5511 // IntArray (or BigIntArray when widening), any Text →
5512 // TextArray. Non-TEXT non-integer elements (Bool, Float)
5513 // stringify into TextArray as the safe default.
5514 Expr::Array(items) => eval_array_arm(items, row, ctx),
5515 // v7.10.12 — `arr[i]` PG-style 1-based indexing.
5516 // Out-of-range indices (including i ≤ 0) return NULL.
5517 Expr::ArraySubscript { .. } => eval_array_subscript_arm(expr, row, ctx),
5518 // Array slice `arr[lo:hi]` — PG 1-based, both ends
5519 // inclusive, out-of-range bounds clamp, missing bounds
5520 // extend to the array's ends. Result keeps the element
5521 // type; an empty window yields an empty array.
5522 Expr::ArraySlice { target, lo, hi } => eval_array_slice_arm(target, lo, hi, row, ctx),
5523 // v7.10.12 — `x op ANY(arr)` / `x op ALL(arr)`. PG
5524 // 3VL: ANY → true if any element compares-true; NULL if
5525 // no true but some NULL; false otherwise. ALL: false if
5526 // any compares-false; NULL if no false but some NULL;
5527 // true otherwise.
5528 Expr::AnyAll {
5529 expr,
5530 op,
5531 array,
5532 is_any,
5533 } => eval_any_all_arm(expr, op, array, *is_any, row, ctx),
5534 // v7.13.0 — CASE WHEN … END (mailrs round-5 G9).
5535 // Short-circuit on the first matching branch. Searched form
5536 // (operand=None) treats each branch's WHEN as a Bool
5537 // predicate. Simple form (operand=Some) compares with =.
5538 // ELSE on no match; NULL if no ELSE.
5539 Expr::Case {
5540 operand,
5541 branches,
5542 else_branch,
5543 } => eval_case_arm(operand, branches, else_branch, row, ctx),
5544 }
5545}
5546
5547/// v7.10.10 — best-effort text rendering for non-TEXT array
5548/// elements (numbers, bools, etc.). The PG rule is that
5549/// `ARRAY[1, 2]` is `int[]`, but SPG's v7.10 only models TEXT[],
5550/// so we widen by stringifying. NUMERIC formatting goes through
5551/// the existing canonical helpers to stay consistent with
5552/// `format_numeric` / `format_date` etc.
5553/// v7.37 D.1 — the COALESCE result-type hint: a sibling branch's explicit
5554/// cast target (`NULL::time`, `col::time`), unless it is Text (Text carries no
5555/// coercion). Returns the first non-Text `CastTarget` found, mirroring PG's
5556/// left-to-right common-type resolution for the common single-typed-branch case.
5557fn coalesce_type_hint(e: &Expr) -> Option<CastTarget> {
5558 match e {
5559 Expr::Cast { target, .. } if !matches!(target, CastTarget::Text) => Some(target.clone()),
5560 _ => None,
5561 }
5562}
5563
5564/// v7.38 (read01) — widen `v` to the already-resolved common type `common`
5565/// of a `CASE`/`COALESCE`/`GREATEST`/`LEAST`/`NULLIF` result, so the value's
5566/// type matches the one PG reports and downstream operators (e.g. `/`) see
5567/// the widened type (integer division vs numeric division). Only widens;
5568/// anything already at the common type, or that fails to coerce, is returned
5569/// untouched (this must never turn a working expression into an error).
5570/// NUMERIC is scale-preserving: an existing exact-numeric keeps its own scale
5571/// (PG renders `COALESCE(1.50, 2)` as `1.50`); only integers promote, to
5572/// scale 0.
5573pub(crate) fn widen_value_to(v: Value<'static>, common: spg_storage::DataType) -> Value<'static> {
5574 use spg_storage::DataType as DT;
5575 // Only widen numeric- and temporal-category results: these are the ones
5576 // whose type actually changes a downstream value (integer vs numeric
5577 // division, date vs timestamp). Widening a string result (varchar ∪ text)
5578 // would only relabel the type while risking a spurious length-limit error
5579 // when coercing into a modelled-length varchar/char, so leave it as-is.
5580 if !matches!(
5581 common,
5582 DT::SmallInt
5583 | DT::Int
5584 | DT::BigInt
5585 | DT::Numeric { .. }
5586 // v7.39 (round 649) — `real` was absent here too, so even once
5587 // `common_type` learned to rank it, the value was handed back
5588 // unwidened: `coalesce(1::int, 1::real)` stayed integer where
5589 // PG says real. Two lists, one ladder — the gap had to be
5590 // closed in both.
5591 | DT::Real
5592 | DT::Float
5593 | DT::Date
5594 | DT::Time
5595 | DT::Timestamp
5596 | DT::Timestamptz
5597 ) {
5598 return v;
5599 }
5600 if matches!(v, Value::Null) {
5601 return v;
5602 }
5603 if v.data_type() == Some(common) {
5604 return v;
5605 }
5606 if matches!(common, spg_storage::DataType::Numeric { .. })
5607 && matches!(v, Value::Numeric { .. } | Value::NumericBig(_))
5608 {
5609 return v;
5610 }
5611 let target = if matches!(common, spg_storage::DataType::Numeric { .. }) {
5612 spg_storage::DataType::Numeric {
5613 precision: 0,
5614 scale: 0,
5615 }
5616 } else {
5617 common
5618 };
5619 match crate::conversions::coerce_value(v.clone(), target, "", 0) {
5620 Ok(cv) => cv,
5621 Err(_) => v,
5622 }
5623}
5624
5625/// Widen `v` to the PG common type of the sibling `types`, or leave it as-is
5626/// when the types don't resolve to a single widening type. See
5627/// [`widen_value_to`] and [`crate::describe::common_type`].
5628pub(crate) fn widen_to_common(
5629 v: Value<'static>,
5630 types: &[spg_storage::DataType],
5631) -> Value<'static> {
5632 match crate::describe::common_type(types) {
5633 Some(common) => widen_value_to(v, common),
5634 None => v,
5635 }
5636}
5637
5638pub(crate) fn value_to_text_for_array(v: &Value, style: &format::RenderStyle) -> String {
5639 match v {
5640 Value::Text(s) | Value::Json(s) => s.to_string(),
5641 Value::Int(n) => n.to_string(),
5642 Value::BigInt(n) => n.to_string(),
5643 Value::SmallInt(n) => n.to_string(),
5644 // PG renders booleans in array external form as `t` / `f`
5645 // (the bool type's output function), not `true` / `false`.
5646 Value::Bool(b) => {
5647 if *b {
5648 "t".into()
5649 } else {
5650 "f".into()
5651 }
5652 }
5653 Value::Float(x) => format::format_float_styled(*x, style),
5654 Value::Real(x) => format::format_real_styled(*x, style),
5655 Value::Date(d) => format::format_date_styled(*d, style),
5656 Value::Timestamp(t) => format::format_timestamp_styled(*t, style),
5657 Value::Numeric {
5658 scaled,
5659 scale,
5660 kind,
5661 } => format_numeric_kind(*kind, *scaled, *scale),
5662 // v7.39 — everything else renders its canonical PG text (this
5663 // Debug fallback is how `ARRAY['\xff'::bytea]` printed
5664 // `{Bytes([255])}` on the wire).
5665 _ => values::value_to_text_styled(v, style),
5666 }
5667}
5668
5669/// SQL `LIKE` matcher. Wildcards are `%` (any run, possibly empty) and `_`
5670/// (exactly one char). `\` escapes the next pattern char so `\%` matches a
5671/// literal `%`. Matches the whole input — no implicit anchoring needed
5672/// since SQL `LIKE` is always full-string. Errs on a trailing unpaired
5673/// escape the matcher actually reaches with text left (PG's lazy 22025).
5674fn like_match(text: &str, pattern: &str) -> Result<bool, EvalError> {
5675 let pat: Vec<char> = pattern.chars().collect();
5676 like_match_str(text, &pat, 0)
5677}
5678
5679/// v7.37.16 — pg_typeof spelling for a STATIC column type (the
5680/// NULL-cell fallback; the value-level table is `pg_typeof_name`).
5681/// TEXT maps to None because a NULL literal describes as TEXT — the
5682/// unknown stand-in — and pg_typeof(NULL) must stay "unknown"; every
5683/// type not listed also returns None (caller keeps the value answer).
5684pub(crate) fn pg_typeof_name_for_datatype(t: spg_storage::DataType) -> Option<&'static str> {
5685 use spg_storage::DataType as D;
5686 Some(match t {
5687 D::SmallInt => "smallint",
5688 D::Int => "integer",
5689 D::BigInt => "bigint",
5690 D::Float => "double precision",
5691 D::Real => "real",
5692 D::Numeric { .. } => "numeric",
5693 D::Bool => "boolean",
5694 D::Date => "date",
5695 D::Time => "time without time zone",
5696 D::Timestamp => "timestamp without time zone",
5697 D::Timestamptz => "timestamp with time zone",
5698 D::Name => "name",
5699 D::Xid => "xid",
5700 D::Xid8 => "xid8",
5701 D::Oid => "oid",
5702 // v7.39 (round 694) — and its array, for the same reason.
5703 D::OidArray => "oid[]",
5704 D::Uuid => "uuid",
5705 D::Interval => "interval",
5706 // v7.39 (round 871) — the rest of what a NULL cast can be
5707 // annotated with. This table decided which types survived
5708 // `pg_typeof(NULL::t)`: the twenty above answered, everything
5709 // else fell to `_ => None` and reported `unknown`. That reads
5710 // as a NULL problem and is not one — `NULL::uuid` was right all
5711 // along while `NULL::text` was wrong, because one was listed
5712 // and the other was not.
5713 //
5714 // Names are PG18's own, taken from running `pg_typeof` there
5715 // rather than from memory: `bit varying` not `varbit`, `bit`
5716 // for any width, `character varying`, `"char"` quoted.
5717 D::Text => "text",
5718 D::Multirange(k) => match k {
5719 spg_storage::RangeKind::Int4 => "int4multirange",
5720 spg_storage::RangeKind::Int8 => "int8multirange",
5721 spg_storage::RangeKind::Num => "nummultirange",
5722 spg_storage::RangeKind::Ts => "tsmultirange",
5723 spg_storage::RangeKind::TsTz => "tstzmultirange",
5724 spg_storage::RangeKind::Date => "datemultirange",
5725 },
5726 D::Varchar(_) => "character varying",
5727 // PG names `char(n)` "character"; the one-byte internal type
5728 // spelled `"char"` is a DIFFERENT type there, and SPG maps both
5729 // onto `Char(u32)` — so this arm must answer for the declared
5730 // one. Round 871's first attempt said `"char"` here and would
5731 // have reported `char(5)` as the internal type.
5732 D::Char(_) => "character",
5733 D::Json => "json",
5734 D::Jsonb => "jsonb",
5735 D::Bytes => "bytea",
5736 D::Inet => "inet",
5737 D::Cidr => "cidr",
5738 D::Macaddr => "macaddr",
5739 D::Macaddr8 => "macaddr8",
5740 D::Bit(_) => "bit",
5741 D::BitVarying(_) => "bit varying",
5742 D::Xml => "xml",
5743 D::Money => "money",
5744 D::Point => "point",
5745 D::Lseg => "lseg",
5746 D::Path => "path",
5747 D::PgBox => "box",
5748 D::Polygon => "polygon",
5749 D::Line => "line",
5750 D::Circle => "circle",
5751 D::TextArray => "text[]",
5752 D::IntArray => "integer[]",
5753 D::BigIntArray => "bigint[]",
5754 D::SmallIntArray => "smallint[]",
5755 D::FloatArray => "double precision[]",
5756 D::NumericArray => "numeric[]",
5757 D::BoolArray => "boolean[]",
5758 D::DateArray => "date[]",
5759 D::TimestampArray => "timestamp without time zone[]",
5760 D::TimestamptzArray => "timestamp with time zone[]",
5761 D::RealArray => "real[]",
5762 D::TimeArray => "time without time zone[]",
5763 D::TimeTzArray => "time with time zone[]",
5764 D::InetArray => "inet[]",
5765 D::XmlArray => "xml[]",
5766 D::UuidArray => "uuid[]",
5767 D::JsonArray => "json[]",
5768 D::JsonbArray => "jsonb[]",
5769 D::BytesArray => "bytea[]",
5770 D::VarcharArray => "character varying[]",
5771 D::CharArray => "\"char\"[]",
5772 D::IntervalArray => "interval[]",
5773 // v7.39.13 — the two temporal types this table never named.
5774 // `timetz` answered correctly through the VALUE path and
5775 // `unknown` through this one, so `pg_typeof(NULL::timetz)` and
5776 // `pg_typeof(k)` disagreed about one column.
5777 D::TimeTz => "time with time zone",
5778 D::Year => "year",
5779 _ => return None,
5780 })
5781}
5782
5783/// v7.37.16 — zero-allocation LIKE core: the text side walks a `&str`
5784/// cursor (char-semantic — `_` consumes one CHARACTER, `%` backtracks
5785/// only at char boundaries) instead of collecting a `Vec<char>` per
5786/// call. The old per-row collect was ~50 ns/row of pure allocator
5787/// traffic on a 50 k-row `WHERE s LIKE '%…%'` scan (the heavy.rs
5788/// like_filter 2.8× loss); the pattern side stays a compile-once
5789/// `&[char]` (see `Step::Like`).
5790pub(crate) fn like_match_str(text: &str, pat: &[char], mut pi: usize) -> Result<bool, EvalError> {
5791 let mut t = text;
5792 while pi < pat.len() {
5793 match pat[pi] {
5794 '%' => {
5795 // Collapse consecutive `%` and try every possible split.
5796 while pi < pat.len() && pat[pi] == '%' {
5797 pi += 1;
5798 }
5799 if pi == pat.len() {
5800 return Ok(true);
5801 }
5802 let mut rest = t;
5803 loop {
5804 if like_match_str(rest, pat, pi)? {
5805 return Ok(true);
5806 }
5807 match rest.chars().next() {
5808 Some(c) => rest = &rest[c.len_utf8()..],
5809 None => return Ok(false),
5810 }
5811 }
5812 }
5813 '_' => match t.chars().next() {
5814 Some(c) => {
5815 t = &t[c.len_utf8()..];
5816 pi += 1;
5817 }
5818 None => return Ok(false),
5819 },
5820 // v7.39 (round 144, like_match.c) — a trailing unpaired escape is
5821 // PG's 22025 error, but LAZILY: only when the matcher reaches it
5822 // with text left. A branch where the text is already exhausted
5823 // returns false without ever "seeing" the trailing escape
5824 // ('x' LIKE 'x\' is false; 'xy' LIKE 'x\' errors).
5825 '\\' if pi + 1 >= pat.len() => {
5826 if t.is_empty() {
5827 return Ok(false);
5828 }
5829 return Err(EvalError::TypeMismatch {
5830 detail: "LIKE pattern must not end with escape character".into(),
5831 });
5832 }
5833 '\\' => {
5834 let want = pat[pi + 1];
5835 match t.chars().next() {
5836 Some(c) if c == want => {
5837 t = &t[c.len_utf8()..];
5838 pi += 2;
5839 }
5840 _ => return Ok(false),
5841 }
5842 }
5843 c => match t.chars().next() {
5844 Some(tc) if tc == c => {
5845 t = &t[c.len_utf8()..];
5846 pi += 1;
5847 }
5848 _ => return Ok(false),
5849 },
5850 }
5851 }
5852 Ok(t.is_empty())
5853}
5854
5855/// v7.24 (round-15) — `string_to_array(text, delimiter)`.
5856fn fn_string_to_array(args: &[Value<'_>]) -> Result<Value<'static>, EvalError> {
5857 if args.len() != 2 && args.len() != 3 {
5858 return Err(crate::eval::functions::wrong_arity("string_to_array", args));
5859 }
5860 // v7.37.17 (17.6 siblings) — the 3-arg PG form adds
5861 // `null_string`: elements equal to it become SQL NULL.
5862 let (text_arg, delim_arg, null_arg) = match args {
5863 [t, d] => (t, d, None),
5864 [t, d, n] => (t, d, Some(n)),
5865 _ => {
5866 return Err(EvalError::TypeMismatch {
5867 detail: alloc::format!(
5868 "string_to_array expects 2 or 3 arguments, got {}",
5869 args.len()
5870 ),
5871 });
5872 }
5873 };
5874 let null_string: Option<&str> = match null_arg {
5875 None | Some(Value::Null) => None,
5876 Some(Value::Text(s)) => Some(s.as_ref()),
5877 Some(other) => {
5878 return Err(EvalError::TypeMismatch {
5879 detail: alloc::format!(
5880 "string_to_array null_string must be text, got {}",
5881 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5882 ),
5883 });
5884 }
5885 };
5886 let text = match text_arg {
5887 Value::Null => return Ok(Value::Null),
5888 Value::Text(t) => t,
5889 other => {
5890 return Err(EvalError::TypeMismatch {
5891 detail: alloc::format!(
5892 "string_to_array expects text, got {}",
5893 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5894 ),
5895 });
5896 }
5897 };
5898 // PG (9.1+): empty input → empty array, regardless of delimiter.
5899 if text.is_empty() {
5900 return Ok(Value::TextArray(Vec::new()));
5901 }
5902 let nullify = |p: String| -> Option<String> {
5903 if null_string == Some(p.as_str()) {
5904 None
5905 } else {
5906 Some(p)
5907 }
5908 };
5909 let parts: Vec<Option<String>> = match delim_arg {
5910 // NULL delimiter → one element per character.
5911 Value::Null => text.chars().map(|c| nullify(c.to_string())).collect(),
5912 Value::Text(d) if d.is_empty() => alloc::vec![nullify(text.to_string())],
5913 Value::Text(d) => text
5914 .split(d.as_ref())
5915 .map(|p| nullify(p.to_string()))
5916 .collect(),
5917 other => {
5918 return Err(EvalError::TypeMismatch {
5919 detail: alloc::format!(
5920 "string_to_array delimiter must be text, got {}",
5921 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5922 ),
5923 });
5924 }
5925 };
5926 Ok(Value::TextArray(parts))
5927}
5928
5929/// v6.4.3 — `error_on_null(v)`. Returns `v` unchanged if non-NULL;
5930/// errors otherwise. Convenience to assert NOT NULL inside an
5931/// expression without wrapping it in COALESCE + raise hacks.
5932fn error_on_null(args: &[Value<'_>]) -> Result<Value<'static>, EvalError> {
5933 if args.len() != 1 {
5934 return Err(EvalError::WrongArity {
5935 name: alloc::string::String::from("error_on_null"),
5936 types: crate::eval::functions::arg_type_list(args),
5937 });
5938 }
5939 if matches!(args[0], Value::Null) {
5940 return Err(EvalError::TypeMismatch {
5941 detail: "error_on_null(): argument is NULL".into(),
5942 });
5943 }
5944 Ok(args[0].clone().into_owned())
5945}
5946
5947/// Helper: coerce a Value to an Option<String> for regex args. NULL
5948/// propagates as None (caller short-circuits to Value::Null).
5949fn text_arg(v: &Value) -> Result<Option<String>, EvalError> {
5950 match v {
5951 Value::Text(s) => Ok(Some(s.to_string())),
5952 Value::Null => Ok(None),
5953 other => Err(EvalError::TypeMismatch {
5954 detail: alloc::format!(
5955 "regex function expects TEXT arg, got {}",
5956 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5957 ),
5958 }),
5959 }
5960}
5961
5962// Month-name tables shared by the date formatters in `eval::strings`
5963// (`date_format_mysql`) and `eval::datetime` via `use super::`. Kept in
5964// `eval.rs` alongside `civil_from_days` so the calendar primitives live
5965// in one place.
5966const MONTH_FULL: [&str; 12] = [
5967 "January",
5968 "February",
5969 "March",
5970 "April",
5971 "May",
5972 "June",
5973 "July",
5974 "August",
5975 "September",
5976 "October",
5977 "November",
5978 "December",
5979];
5980const MONTH_ABBR: [&str; 12] = [
5981 "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec",
5982];
5983
5984/// Howard Hinnant's `civil_from_days` — converts days since the Unix
5985/// epoch back to a proleptic-Gregorian (year, month, day) triple. Stays
5986/// in `eval.rs` (shared with the date SQL functions here and with
5987/// `eval::strings`); the inverse `days_from_civil` lives in
5988/// `eval::format`. Both keep the engine off `std` time facilities.
5989#[allow(clippy::cast_possible_truncation, clippy::cast_sign_loss)]
5990fn civil_from_days(days: i32) -> (i32, u32, u32) {
5991 let z = i64::from(days) + 719_468;
5992 let era = z.div_euclid(146_097);
5993 // doe ∈ [0, 146_097); fits in u32 with room to spare. Same for
5994 // every other quantity below — `as u32` truncations are safe by
5995 // construction.
5996 let doe = (z - era * 146_097) as u32;
5997 let yoe = (doe.saturating_sub(doe / 1460) + doe / 36524 - doe / 146_096) / 365;
5998 let y_base = i64::from(yoe) + era * 400;
5999 let doy = doe.saturating_sub(365 * yoe + yoe / 4 - yoe / 100);
6000 let mp = (5 * doy + 2) / 153;
6001 let d = doy.saturating_sub((153 * mp + 2) / 5) + 1;
6002 let m = if mp < 10 { mp + 3 } else { mp - 9 };
6003 let y = if m <= 2 { y_base + 1 } else { y_base };
6004 (y as i32, m, d)
6005}
6006
6007/// Add `months` (signed) to a `(year, month, day)` triple using PG's
6008/// clamp-to-last-day rule (so `'2024-01-31' + 1 month` → `'2024-02-29'`).
6009fn add_months_to_civil(y: i32, m: u32, d: u32, months: i32) -> (i32, u32, u32) {
6010 let total_months = i64::from(y) * 12 + i64::from(m) - 1 + i64::from(months);
6011 let new_year = i32::try_from(total_months.div_euclid(12)).unwrap_or(i32::MAX);
6012 let new_month_zero = total_months.rem_euclid(12);
6013 #[allow(clippy::cast_possible_truncation, clippy::cast_sign_loss)]
6014 let new_month = (new_month_zero as u32) + 1;
6015 let max_day = days_in_month(new_year, new_month);
6016 (new_year, new_month, d.min(max_day))
6017}
6018
6019const fn days_in_month(y: i32, m: u32) -> u32 {
6020 match m {
6021 1 | 3 | 5 | 7 | 8 | 10 | 12 => 31,
6022 2 => {
6023 // Proleptic Gregorian leap rule.
6024 if y.rem_euclid(4) == 0 && (y.rem_euclid(100) != 0 || y.rem_euclid(400) == 0) {
6025 29
6026 } else {
6027 28
6028 }
6029 }
6030 // 4 / 6 / 9 / 11 plus any out-of-range month (callers normalise
6031 // first, but be defensive) get the 30-day fallback.
6032 _ => 30,
6033 }
6034}
6035
6036pub(crate) fn literal_to_value(l: &Literal) -> Value<'static> {
6037 match l {
6038 Literal::Integer(n) => {
6039 if let Ok(small) = i32::try_from(*n) {
6040 Value::Int(small)
6041 } else {
6042 Value::BigInt(*n)
6043 }
6044 }
6045 Literal::Float(x) => Value::Float(*x),
6046 Literal::Numeric { unscaled, scale } => Value::Numeric {
6047 scaled: *unscaled,
6048 scale: *scale,
6049 kind: spg_storage::NumericKind::Finite,
6050 },
6051 Literal::NumericBig(s) => crate::conversions::big_literal_to_value(s),
6052 // v7.38.8 — already decoded, so the row loop neither clones a
6053 // string nor coerces one back into a timestamp.
6054 Literal::Timestamp { micros, .. } => Value::Timestamp(*micros),
6055 Literal::Date { days, .. } => Value::Date(*days),
6056 Literal::String(s) => Value::text(s.clone()),
6057 Literal::Vector(v) => Value::vector(v.clone()),
6058 Literal::TextArray(items) => Value::TextArray(items.clone()),
6059 Literal::IntArray(items) => Value::IntArray(items.clone()),
6060 Literal::BigIntArray(items) => Value::BigIntArray(items.clone()),
6061 Literal::Bool(b) => Value::Bool(*b),
6062 Literal::Null => Value::Null,
6063 Literal::Interval {
6064 months,
6065 days,
6066 micros,
6067 ..
6068 } => Value::Interval {
6069 months: *months,
6070 days: *days,
6071 micros: *micros,
6072 kind: spg_storage::IntervalKind::Finite,
6073 },
6074 }
6075}
6076
6077impl crate::Engine {
6078 /// v7.39 (read01 round 63) — run a user function whose body has its own
6079 /// FROM. The arguments are substituted into the body as literals and the
6080 /// SELECT goes through the REAL executor, so it sees exactly the rows a
6081 /// hand-written query would: the row-header visibility filter applies, and
6082 /// under in-place MVCC a dead row stays dead.
6083 ///
6084 /// PG returns the FIRST row of a scalar SQL function's body (and NULL when
6085 /// it returns none).
6086 pub(crate) fn run_user_fn_query(
6087 &self,
6088 def: &spg_storage::FunctionDef,
6089 stmt: &spg_sql::ast::SelectStatement,
6090 arg_names: &[alloc::string::String],
6091 args: &spg_storage::Row<'static>,
6092 fn_depth: u16,
6093 ) -> Result<Value<'static>, EvalError> {
6094 const MAX_QUERY_FN_DEPTH: u16 = 8;
6095 if fn_depth >= MAX_QUERY_FN_DEPTH {
6096 return Err(EvalError::TypeMismatch {
6097 detail: alloc::format!(
6098 "function {:?}: a body with its own FROM may nest at most {MAX_QUERY_FN_DEPTH} deep",
6099 def.name
6100 ),
6101 });
6102 }
6103 // Bind the arguments — the same helper the set-returning path uses, so
6104 // both resolve an argument identically (a COLUMN of the body's own FROM
6105 // shadows a same-named argument, as in PG).
6106 let owned: alloc::vec::Vec<Value<'static>> =
6107 args.values.iter().map(|v| v.clone().into_owned()).collect();
6108 let bound =
6109 bind_user_fn_args(self.active_catalog(), stmt, arg_names, &owned).map_err(|e| {
6110 EvalError::TypeMismatch {
6111 detail: alloc::format!("function {:?}: {e}", def.name),
6112 }
6113 })?;
6114
6115 // v7.39 (round 334, V55) — a SECURITY DEFINER body is authorised as
6116 // the function's OWNER. Measured on PG 18.4: a definer function
6117 // owned by `owner55` counts rows of a table `caller55` cannot read,
6118 // while the SECURITY INVOKER sibling is refused.
6119 let as_role = def.security_definer.then(|| def.owner.as_deref()).flatten();
6120 let out = self
6121 .exec_select_cancel_as(&bound, crate::CancelToken::none(), as_role)
6122 .map_err(|e| EvalError::TypeMismatch {
6123 detail: alloc::format!("function {:?}: {e}", def.name),
6124 })?;
6125 let crate::QueryResult::Rows { rows, .. } = out else {
6126 return Ok(Value::Null);
6127 };
6128 let Some(first) = rows.first() else {
6129 // No row: PG's scalar SQL function returns NULL.
6130 return Ok(Value::Null);
6131 };
6132 let v = first.values.first().cloned().unwrap_or(Value::Null);
6133 let declared = def.returns.trim();
6134 if declared.eq_ignore_ascii_case("VOID") {
6135 return Ok(Value::Null);
6136 }
6137 crate::eval::cast::cast_value(v.into_owned(), declared_return_cast_target(declared))
6138 .or_else(|_| Ok(Value::Null))
6139 }
6140}
6141
6142/// v7.39 (read01 round 65) — bind a call's arguments into a function body's
6143/// SELECT, as literals. Shared by the scalar path (round 63) and the
6144/// set-returning one, so both resolve an argument the same way — including the
6145/// rule that a COLUMN of the body's own FROM shadows a same-named argument.
6146pub(crate) fn bind_user_fn_args(
6147 cat: &spg_storage::Catalog,
6148 stmt: &spg_sql::ast::SelectStatement,
6149 arg_names: &[alloc::string::String],
6150 args: &[Value<'static>],
6151) -> Result<spg_sql::ast::SelectStatement, EvalError> {
6152 let mut bound = stmt.clone();
6153 let mut binds: alloc::collections::BTreeMap<alloc::string::String, spg_sql::ast::Expr> =
6154 alloc::collections::BTreeMap::new();
6155 for (i, name) in arg_names.iter().enumerate() {
6156 if name.is_empty() {
6157 continue;
6158 }
6159 let shadowed = body_from_tables(stmt).iter().any(|t| {
6160 cat.get(t).is_some_and(|tb| {
6161 tb.schema()
6162 .columns
6163 .iter()
6164 .any(|c| c.name.eq_ignore_ascii_case(name))
6165 })
6166 });
6167 if shadowed {
6168 continue;
6169 }
6170 let v = args.get(i).cloned().unwrap_or(Value::Null);
6171 let lit =
6172 crate::substitute::value_to_literal_expr(v).map_err(|e| EvalError::TypeMismatch {
6173 detail: alloc::format!("argument {name} cannot be bound into the body: {e}"),
6174 })?;
6175 binds.insert(name.to_ascii_lowercase(), lit);
6176 }
6177 substitute_arg_refs_in_select(&mut bound, &binds);
6178 Ok(bound)
6179}
6180
6181/// The base tables a function body's FROM names — used to decide whether an
6182/// argument name is shadowed by a column of the same name.
6183fn body_from_tables(
6184 stmt: &spg_sql::ast::SelectStatement,
6185) -> alloc::vec::Vec<alloc::string::String> {
6186 let mut out = alloc::vec::Vec::new();
6187 if let Some(from) = &stmt.from {
6188 out.push(from.primary.name.clone());
6189 for j in &from.joins {
6190 out.push(j.table.name.clone());
6191 }
6192 }
6193 out
6194}
6195
6196/// Replace every bare reference to an argument name with its literal value.
6197fn substitute_arg_refs_in_select(
6198 stmt: &mut spg_sql::ast::SelectStatement,
6199 binds: &alloc::collections::BTreeMap<alloc::string::String, spg_sql::ast::Expr>,
6200) {
6201 use spg_sql::ast::{Expr, SelectItem};
6202 fn walk(e: &mut Expr, binds: &alloc::collections::BTreeMap<alloc::string::String, Expr>) {
6203 match e {
6204 Expr::Column(c) => {
6205 if c.qualifier.is_none()
6206 && let Some(lit) = binds.get(&c.name.to_ascii_lowercase())
6207 {
6208 *e = lit.clone();
6209 }
6210 }
6211 Expr::Binary { lhs, rhs, .. } => {
6212 walk(lhs, binds);
6213 walk(rhs, binds);
6214 }
6215 Expr::Unary { expr, .. } | Expr::Cast { expr, .. } => walk(expr, binds),
6216 Expr::FunctionCall { args, .. } => args.iter_mut().for_each(|a| walk(a, binds)),
6217 Expr::Case {
6218 operand,
6219 branches,
6220 else_branch,
6221 } => {
6222 if let Some(o) = operand {
6223 walk(o, binds);
6224 }
6225 for (c, v) in branches.iter_mut() {
6226 walk(c, binds);
6227 walk(v, binds);
6228 }
6229 if let Some(x) = else_branch {
6230 walk(x, binds);
6231 }
6232 }
6233 Expr::InList { expr, list, .. } => {
6234 walk(expr, binds);
6235 list.iter_mut().for_each(|it| walk(it, binds));
6236 }
6237 Expr::AnyAll { expr, array, .. } => {
6238 walk(expr, binds);
6239 walk(array, binds);
6240 }
6241 Expr::Array(items) => items.iter_mut().for_each(|it| walk(it, binds)),
6242 Expr::ArraySubscript { target, index } => {
6243 walk(target, binds);
6244 walk(index, binds);
6245 }
6246 _ => {}
6247 }
6248 }
6249 for item in &mut stmt.items {
6250 if let SelectItem::Expr { expr, .. } = item {
6251 walk(expr, binds);
6252 }
6253 }
6254 if let Some(w) = &mut stmt.where_ {
6255 walk(w, binds);
6256 }
6257 if let Some(h) = &mut stmt.having {
6258 walk(h, binds);
6259 }
6260 if let Some(gs) = &mut stmt.group_by {
6261 gs.iter_mut().for_each(|g| walk(g, binds));
6262 }
6263 for o in &mut stmt.order_by {
6264 walk(&mut o.expr, binds);
6265 }
6266 if let Some(from) = &mut stmt.from {
6267 for j in &mut from.joins {
6268 if let Some(on) = &mut j.on {
6269 walk(on, binds);
6270 }
6271 }
6272 }
6273 // v7.39 (read01 round 69) — the UNION peers. A body like
6274 // `SELECT k UNION ALL SELECT k * 10` has its second half in `unions`, and
6275 // leaving it unsubstituted made the argument look like a missing column.
6276 for (_, peer) in &mut stmt.unions {
6277 substitute_arg_refs_in_select(peer, binds);
6278 }
6279 for cte in &mut stmt.ctes {
6280 if let Some(s) = cte.body.as_select_mut() {
6281 substitute_arg_refs_in_select(s, binds);
6282 }
6283 }
6284}
6285
6286/// v7.38.4 (sentori step 54) — the cast target for a function's DECLARED
6287/// return type.
6288///
6289/// `def.returns` holds the type as the user wrote it, so an array is
6290/// `bigint[]`; a `CastTarget::Named` spells the same type `bigint_array`.
6291/// The coercion therefore could not resolve `RETURNS bigint[]`, and the
6292/// `or_else(NULL)` under every call site turned "I could not coerce this"
6293/// into a NULL answer: the body computed `{1,2}` and the caller got
6294/// nothing, with no error anywhere. Their version keys compare through one
6295/// of these, so every version-targeted push reached zero devices while
6296/// reporting success.
6297///
6298/// Shared by both coercion sites — the pure-expression body and the one
6299/// with its own FROM — because they had the same line written twice and
6300/// fixing one would have left the other.
6301pub(crate) fn declared_return_cast_target(declared: &str) -> spg_sql::ast::CastTarget {
6302 let name = declared.trim().strip_suffix("[]").map_or_else(
6303 || alloc::string::String::from(declared.trim()),
6304 |base| alloc::format!("{}_array", base.trim()),
6305 );
6306 spg_sql::ast::CastTarget::Named(name)
6307}
6308
6309impl crate::Engine {
6310 /// v7.39 (read01 round 64) — call a plpgsql function as a scalar. The body
6311 /// runs on the interpreter the DO block and triggers already use, with the
6312 /// arguments bound as locals; its `SELECT … INTO` and `FOR … IN SELECT`
6313 /// resolvers go through the READ path, so what the body sees is what a
6314 /// hand-written query would see (visibility filter and all).
6315 ///
6316 /// A body that writes is refused — the call arrives through expression
6317 /// evaluation, which holds the engine immutably. Refusing is the honest
6318 /// answer; silently dropping the write would be the worst one.
6319 pub(crate) fn call_plpgsql_scalar_fn(
6320 &self,
6321 def: &spg_storage::FunctionDef,
6322 arg_names: &[alloc::string::String],
6323 args: &spg_storage::Row<'static>,
6324 ) -> Result<Value<'static>, EvalError> {
6325 let block =
6326 spg_sql::parse_function_body(def.body.trim()).map_err(|e| EvalError::TypeMismatch {
6327 detail: alloc::format!("function {:?} body does not parse: {e}", def.name),
6328 })?;
6329 let mut locals: alloc::collections::BTreeMap<alloc::string::String, Value<'static>> =
6330 alloc::collections::BTreeMap::new();
6331 for (i, n) in arg_names.iter().enumerate() {
6332 if n.is_empty() {
6333 continue;
6334 }
6335 locals.insert(
6336 n.to_ascii_lowercase(),
6337 args.values.get(i).cloned().unwrap_or(Value::Null),
6338 );
6339 }
6340 let dts = self
6341 .session_param("default_text_search_config")
6342 .map(alloc::string::String::from);
6343
6344 let select_into = |stmt: &spg_sql::ast::Statement| -> Result<
6345 Value<'static>,
6346 crate::triggers::TriggerError,
6347 > {
6348 let spg_sql::ast::Statement::Select(s) = stmt else {
6349 return Err(crate::triggers::TriggerError::EvalFailed {
6350 function: def.name.clone(),
6351 cause: EvalError::TypeMismatch {
6352 detail: "SELECT … INTO body must be a SELECT".into(),
6353 },
6354 });
6355 };
6356 let r = self
6357 .exec_select_cancel(s, crate::CancelToken::none())
6358 .map_err(|e| crate::triggers::TriggerError::EvalFailed {
6359 function: def.name.clone(),
6360 cause: EvalError::TypeMismatch {
6361 detail: alloc::format!("SELECT … INTO failed: {e}"),
6362 },
6363 })?;
6364 match r {
6365 crate::QueryResult::Rows { rows, .. } => Ok(rows
6366 .into_iter()
6367 .next()
6368 .and_then(|row| row.values.into_iter().next())
6369 .unwrap_or(Value::Null)),
6370 _ => Ok(Value::Null),
6371 }
6372 };
6373 let for_query = |stmt: &spg_sql::ast::Statement| -> Result<
6374 (
6375 alloc::vec::Vec<alloc::string::String>,
6376 alloc::vec::Vec<alloc::vec::Vec<Value<'static>>>,
6377 ),
6378 crate::triggers::TriggerError,
6379 > {
6380 let spg_sql::ast::Statement::Select(s) = stmt else {
6381 return Err(crate::triggers::TriggerError::EvalFailed {
6382 function: def.name.clone(),
6383 cause: EvalError::TypeMismatch {
6384 detail: "FOR … IN body must be a SELECT".into(),
6385 },
6386 });
6387 };
6388 let r = self
6389 .exec_select_cancel(s, crate::CancelToken::none())
6390 .map_err(|e| crate::triggers::TriggerError::EvalFailed {
6391 function: def.name.clone(),
6392 cause: EvalError::TypeMismatch {
6393 detail: alloc::format!("FOR … IN SELECT failed: {e}"),
6394 },
6395 })?;
6396 match r {
6397 crate::QueryResult::Rows { columns, rows } => Ok((
6398 columns.iter().map(|c| c.name.clone()).collect(),
6399 rows.into_iter().map(|row| row.values).collect(),
6400 )),
6401 _ => Ok((alloc::vec::Vec::new(), alloc::vec::Vec::new())),
6402 }
6403 };
6404
6405 let out = crate::triggers::call_plpgsql_scalar(
6406 &def.name,
6407 &block,
6408 locals,
6409 dts.as_deref(),
6410 Some(&select_into),
6411 Some(&for_query),
6412 // A scalar call has no set to build; RETURN NEXT / RETURN QUERY are
6413 // errors here, as in PG.
6414 // (second None below: the read path holds the engine immutably, so
6415 // RAISE messages have nowhere session-bound to go — B3 residual.)
6416 None,
6417 None,
6418 )
6419 .map_err(|e| EvalError::TypeMismatch {
6420 detail: alloc::format!("{e}"),
6421 })?;
6422 let declared = def.returns.trim();
6423 let Some(v) = out else {
6424 if declared.eq_ignore_ascii_case("VOID") {
6425 return Ok(Value::Null);
6426 }
6427 // PG: a non-void function that falls out of the bottom.
6428 return Err(EvalError::TypeMismatch {
6429 detail: alloc::format!(
6430 "control reached end of function {:?} without RETURN",
6431 def.name
6432 ),
6433 });
6434 };
6435 if declared.eq_ignore_ascii_case("VOID") {
6436 return Ok(Value::Null);
6437 }
6438 crate::eval::cast::cast_value(
6439 v.into_owned(),
6440 spg_sql::ast::CastTarget::Named(alloc::string::String::from(declared)),
6441 )
6442 .or_else(|_| Ok(Value::Null))
6443 }
6444}
6445
6446impl crate::Engine {
6447 /// v7.39 (read01 round 66) — run a `RETURNS SETOF` plpgsql function and
6448 /// collect the rows `RETURN NEXT` / `RETURN QUERY` appended. Same
6449 /// interpreter, same read-path resolvers as the scalar call — the only
6450 /// difference is that a SINK is provided, which is what makes those two
6451 /// statements legal.
6452 pub(crate) fn call_plpgsql_setof_fn(
6453 &self,
6454 def: &spg_storage::FunctionDef,
6455 arg_names: &[alloc::string::String],
6456 args: &[Value<'static>],
6457 ) -> Result<alloc::vec::Vec<alloc::vec::Vec<Value<'static>>>, EvalError> {
6458 let block =
6459 spg_sql::parse_function_body(def.body.trim()).map_err(|e| EvalError::TypeMismatch {
6460 detail: alloc::format!("function {:?} body does not parse: {e}", def.name),
6461 })?;
6462 let mut locals: alloc::collections::BTreeMap<alloc::string::String, Value<'static>> =
6463 alloc::collections::BTreeMap::new();
6464 for (i, n) in arg_names.iter().enumerate() {
6465 if n.is_empty() {
6466 continue;
6467 }
6468 locals.insert(
6469 n.to_ascii_lowercase(),
6470 args.get(i).cloned().unwrap_or(Value::Null),
6471 );
6472 }
6473 let dts = self
6474 .session_param("default_text_search_config")
6475 .map(alloc::string::String::from);
6476 let run_select = |stmt: &spg_sql::ast::Statement,
6477 what: &str|
6478 -> Result<crate::QueryResult, crate::triggers::TriggerError> {
6479 let spg_sql::ast::Statement::Select(s) = stmt else {
6480 return Err(crate::triggers::TriggerError::EvalFailed {
6481 function: def.name.clone(),
6482 cause: EvalError::TypeMismatch {
6483 detail: alloc::format!("{what} body must be a SELECT"),
6484 },
6485 });
6486 };
6487 self.exec_select_cancel(s, crate::CancelToken::none())
6488 .map_err(|e| crate::triggers::TriggerError::EvalFailed {
6489 function: def.name.clone(),
6490 cause: EvalError::TypeMismatch {
6491 detail: alloc::format!("{what} failed: {e}"),
6492 },
6493 })
6494 };
6495 let select_into = |stmt: &spg_sql::ast::Statement| -> Result<
6496 Value<'static>,
6497 crate::triggers::TriggerError,
6498 > {
6499 match run_select(stmt, "SELECT … INTO")? {
6500 crate::QueryResult::Rows { rows, .. } => Ok(rows
6501 .into_iter()
6502 .next()
6503 .and_then(|r| r.values.into_iter().next())
6504 .unwrap_or(Value::Null)),
6505 _ => Ok(Value::Null),
6506 }
6507 };
6508 let for_query = |stmt: &spg_sql::ast::Statement| -> Result<
6509 (
6510 alloc::vec::Vec<alloc::string::String>,
6511 alloc::vec::Vec<alloc::vec::Vec<Value<'static>>>,
6512 ),
6513 crate::triggers::TriggerError,
6514 > {
6515 match run_select(stmt, "FOR … IN / RETURN QUERY")? {
6516 crate::QueryResult::Rows { columns, rows } => Ok((
6517 columns.iter().map(|c| c.name.clone()).collect(),
6518 rows.into_iter().map(|r| r.values).collect(),
6519 )),
6520 _ => Ok((alloc::vec::Vec::new(), alloc::vec::Vec::new())),
6521 }
6522 };
6523 let sink: core::cell::RefCell<alloc::vec::Vec<alloc::vec::Vec<Value<'static>>>> =
6524 core::cell::RefCell::new(alloc::vec::Vec::new());
6525 crate::triggers::call_plpgsql_scalar(
6526 &def.name,
6527 &block,
6528 locals,
6529 dts.as_deref(),
6530 Some(&select_into),
6531 Some(&for_query),
6532 Some(&sink),
6533 // Read path — immutable engine borrow; B3 residual.
6534 None,
6535 )
6536 .map_err(|e| EvalError::TypeMismatch {
6537 detail: alloc::format!("{e}"),
6538 })?;
6539 Ok(sink.into_inner())
6540 }
6541}
6542
6543/// v7.39 (round 236) — resolve an ARRAY constructor's element types the way
6544/// PG does. Typed elements must share a type category; a bare string or NULL
6545/// literal is untyped and converts to whatever the typed elements resolved
6546/// to, reporting the value (not a type mismatch) when it will not convert.
6547fn unify_array_elements(
6548 items: &[Expr],
6549 materialised: &mut [Value<'static>],
6550) -> Result<(), EvalError> {
6551 unify_construct_values("ARRAY", items, materialised)
6552}
6553
6554/// v7.39 (round 238) — an `IN (...)` list must be comparable with its
6555/// needle. Reports the operator the way PG does
6556/// ("operator does not exist: integer = text"), and — like round 237 —
6557/// judges only the operands whose type is genuinely known.
6558fn require_in_list_comparable(
6559 needle: &Expr,
6560 list: &[Expr],
6561 ctx: &EvalContext<'_>,
6562) -> Result<(), EvalError> {
6563 let known_ty = |e: &Expr| {
6564 matches!(e, Expr::Cast { .. } | Expr::Literal(_) | Expr::Column(_))
6565 .then(|| crate::describe::describe_expr(e, ctx.columns).map(|s| s.ty))
6566 .flatten()
6567 };
6568 // An untyped literal adopts the needle's type, so it never conflicts.
6569 //
6570 // v7.39 (round 652) — and so does a cast to one of the reg types.
6571 // `'pg_class'::regclass` carries an oid AND a name, which is why
6572 // `compare` has arms for both, but it DESCRIBES as text — SPG has no
6573 // `DataType` for it. This check ran before any comparison and
6574 // refused `WHERE oid IN ('pg_class'::regclass, …)` as `bigint =
6575 // text`, while the identical `oid = 'pg_class'::regclass` worked.
6576 // That shape is how pg_dump, ORMs and monitoring queries name a
6577 // handful of relations at once, so it is not a corner.
6578 let reg_cast = |e: &Expr| {
6579 match e {
6580 Expr::Cast { target, .. } => match target {
6581 spg_sql::ast::CastTarget::RegClass | spg_sql::ast::CastTarget::RegType => true,
6582 // `regproc` and `regnamespace` have no variant of their
6583 // own; they arrive through the generic named path.
6584 spg_sql::ast::CastTarget::Named(n) => {
6585 n.eq_ignore_ascii_case("regproc")
6586 || n.eq_ignore_ascii_case("regnamespace")
6587 || n.eq_ignore_ascii_case("regtype")
6588 || n.eq_ignore_ascii_case("regclass")
6589 }
6590 _ => false,
6591 },
6592 _ => false,
6593 }
6594 };
6595 let untyped = |e: &Expr| {
6596 reg_cast(e)
6597 || matches!(
6598 e,
6599 Expr::Literal(spg_sql::ast::Literal::String(_))
6600 | Expr::Literal(spg_sql::ast::Literal::Null)
6601 )
6602 };
6603 // The needle can be untyped too (`NULL IN (1,2)`): SPG has no `Unknown`
6604 // DataType, so a bare NULL describes as TEXT and would look like a text
6605 // needle conflicting with integer list items.
6606 if untyped(needle) {
6607 return Ok(());
6608 }
6609 let Some(nt) = known_ty(needle) else {
6610 return Ok(());
6611 };
6612 for item in list {
6613 if untyped(item) {
6614 continue;
6615 }
6616 let Some(it) = known_ty(item) else { continue };
6617 if !crate::conversions::types_unify(nt, it) {
6618 return Err(EvalError::TypeMismatch {
6619 detail: alloc::format!(
6620 "operator does not exist: {} = {}",
6621 crate::conversions::pg_type_name_for_error(nt),
6622 crate::conversions::pg_type_name_for_error(it),
6623 ),
6624 });
6625 }
6626 }
6627 Ok(())
6628}
6629
6630/// v7.39 (round 237) — STATIC branch-type resolution for the constructs
6631/// whose branches must not all be evaluated: CASE runs only the branch it
6632/// takes, and a COALESCE / GREATEST argument may have side effects
6633/// (`COALESCE(nextval('s'), 1)`), so the check reads each branch's declared
6634/// type instead of its value. Same rule and wording as the value-driven
6635/// ARRAY path below; an untyped literal is converted here (a literal has no
6636/// side effects) so a value that will not convert is reported as PG does.
6637/// v7.39 (round 609) — takes anything that yields the branches, so the
6638/// COALESCE caller no longer builds a `Vec<&Expr>` of them for every row.
6639pub(crate) fn unify_branch_types_static<'e>(
6640 construct: &str,
6641 branches: impl IntoIterator<Item = &'e Expr> + Clone,
6642 ctx: &EvalContext<'_>,
6643) -> Result<(), EvalError> {
6644 use spg_storage::DataType;
6645 let untyped = |e: &Expr| {
6646 matches!(
6647 e,
6648 Expr::Literal(spg_sql::ast::Literal::String(_))
6649 | Expr::Literal(spg_sql::ast::Literal::Null)
6650 )
6651 };
6652 let mut resolved: Option<DataType> = None;
6653 for e in branches.clone() {
6654 if untyped(e) {
6655 continue;
6656 }
6657 // Only branches whose type is GENUINELY known take part. A general
6658 // `describe_expr` is a best-effort hint for wire type tags, not a
6659 // type checker: it reports a binary operator as its left operand's
6660 // type, so `payload->'a'` (jsonb in PG) came back as text and this
6661 // check refused a working `COALESCE(payload->'a', '{}'::jsonb)`.
6662 // Refusing a valid query is worse than missing an invalid one, so
6663 // the check confines itself to an explicit cast, a typed literal and
6664 // a plain column reference.
6665 let known = matches!(e, Expr::Cast { .. } | Expr::Literal(_) | Expr::Column(_));
6666 if !known {
6667 continue;
6668 }
6669 // 7.38.1 S5.1 — a reg* cast is an OID wearing a name: describe
6670 // says Text (the wire render), but it compares and unions with
6671 // numeric catalog columns (pg_dump: `SELECT classid … UNION
6672 // ALL SELECT 'pg_opfamily'::regclass …`). Its static claim is
6673 // not genuinely known here, so it sits the check out — the
6674 // dual RegClass value reconciles at runtime.
6675 if matches!(
6676 e,
6677 Expr::Cast {
6678 target: spg_sql::ast::CastTarget::RegType | spg_sql::ast::CastTarget::RegClass,
6679 ..
6680 }
6681 ) {
6682 continue;
6683 }
6684 let Some(ty) = crate::describe::describe_expr_type(e, ctx.columns) else {
6685 continue;
6686 };
6687 match resolved {
6688 None => resolved = Some(ty),
6689 Some(prev) if crate::conversions::types_unify(prev, ty) => {
6690 if matches!(prev, DataType::Int | DataType::SmallInt) {
6691 resolved = Some(ty);
6692 }
6693 }
6694 Some(prev) => {
6695 return Err(EvalError::TypeMismatch {
6696 detail: alloc::format!(
6697 "{construct} types {} and {} cannot be matched",
6698 crate::conversions::pg_type_name_for_error(prev),
6699 crate::conversions::pg_type_name_for_error(ty),
6700 ),
6701 });
6702 }
6703 }
6704 }
6705 let Some(target) = resolved else {
6706 return Ok(());
6707 };
6708 if matches!(target, DataType::Text) {
6709 return Ok(());
6710 }
6711 // v7.39 (round 398) — MySQL aggregates a mixed int/string CASE /
6712 // COALESCE to a string (`CASE WHEN 1 THEN 1 ELSE 'x' END` is '1', not an
6713 // error); PG requires the untyped string literals to coerce to the
6714 // resolved numeric type, so it refuses. Under the dialect, skip that
6715 // coercion check — the value is returned as-is / widened by the caller.
6716 if ctx.mysql_dialect {
6717 return Ok(());
6718 }
6719 for e in branches {
6720 if !untyped(e) {
6721 continue;
6722 }
6723 if let Expr::Literal(spg_sql::ast::Literal::String(lit)) = e {
6724 crate::conversions::coerce_value(Value::text(lit.clone()), target, "", 0).map_err(
6725 |err| match err {
6726 crate::EngineError::Eval(ev) => ev,
6727 other => EvalError::TypeMismatch {
6728 detail: alloc::format!("{other}"),
6729 },
6730 },
6731 )?;
6732 }
6733 }
6734 Ok(())
6735}
6736
6737/// v7.39 (round 237) — the same resolution for every construct that builds
6738/// one value out of several branches: ARRAY, CASE, COALESCE, GREATEST and
6739/// LEAST. PG names the construct in the message ("CASE types text and
6740/// integer cannot be matched"), which is why the caller passes it in.
6741pub(crate) fn unify_construct_values(
6742 construct: &str,
6743 items: &[Expr],
6744 materialised: &mut [Value<'static>],
6745) -> Result<(), EvalError> {
6746 use spg_storage::DataType;
6747 let untyped = |e: &Expr| {
6748 matches!(
6749 e,
6750 Expr::Literal(spg_sql::ast::Literal::String(_))
6751 | Expr::Literal(spg_sql::ast::Literal::Null)
6752 )
6753 };
6754 // The type the typed elements agree on, if any.
6755 let mut resolved: Option<DataType> = None;
6756 for (i, v) in materialised.iter().enumerate() {
6757 if items.get(i).is_some_and(untyped) {
6758 continue;
6759 }
6760 let Some(ty) = v.data_type() else { continue };
6761 match resolved {
6762 None => resolved = Some(ty),
6763 Some(prev) if crate::conversions::types_unify(prev, ty) => {
6764 // Keep the wider of the two so the coercion below targets it.
6765 if matches!(prev, DataType::Int | DataType::SmallInt) {
6766 resolved = Some(ty);
6767 }
6768 }
6769 Some(prev) => {
6770 return Err(EvalError::TypeMismatch {
6771 detail: alloc::format!(
6772 "{construct} types {} and {} cannot be matched",
6773 crate::conversions::pg_type_name_for_error(prev),
6774 crate::conversions::pg_type_name_for_error(ty),
6775 ),
6776 });
6777 }
6778 }
6779 }
6780 // Untyped literals adopt that type; a failure names the value, as PG does.
6781 let Some(target) = resolved else {
6782 return Ok(());
6783 };
6784 if matches!(target, DataType::Text) {
6785 return Ok(());
6786 }
6787 for (i, v) in materialised.iter_mut().enumerate() {
6788 if !items.get(i).is_some_and(untyped) || matches!(v, Value::Null) {
6789 continue;
6790 }
6791 *v = crate::conversions::coerce_value(v.clone(), target, "", i).map_err(|e| match e {
6792 crate::EngineError::Eval(ev) => ev,
6793 other => EvalError::TypeMismatch {
6794 detail: alloc::format!("{other}"),
6795 },
6796 })?;
6797 }
6798 Ok(())
6799}
6800
6801/// v7.39 (round 309, V30) — split `'<timestamp> <zone name>'` into the
6802/// wall-clock reading and the zone token, for the zone-less target types.
6803///
6804/// Returns `None` when the literal parses on its own (nothing to strip)
6805/// or when the trailing token is not zone-SHAPED — those keep the
6806/// ordinary "invalid input syntax" path, which is what PG answers for
6807/// `'2020-01-01 10:00:00 xyz'`. Whether a zone-shaped token is a REAL
6808/// zone is the caller's question; getting that wrong is a different
6809/// error in PG, and conflating the two would report a malformed literal
6810/// for a merely-misspelled zone.
6811///
6812/// Deliberately does not accept a bare time (`'10:00:00 America/New_York'`):
6813/// PG refuses a named zone there, and only reaches this spelling through
6814/// a full timestamp literal.
6815fn split_trailing_zone_name(txt: &str, order: format::DateOrder) -> Option<(i64, &str)> {
6816 // Already valid without help — leave it alone.
6817 if format::parse_timestamp_literal_wall_ordered(txt, order).is_some() {
6818 return None;
6819 }
6820 let trimmed = txt.trim_end();
6821 let idx = trimmed.rfind(' ')?;
6822 let (head, tail) = (trimmed[..idx].trim(), trimmed[idx + 1..].trim());
6823 // An era marker is part of the timestamp, not a zone.
6824 let zone_shaped = tail.len() > 1
6825 && tail.bytes().any(|b| b.is_ascii_alphabetic())
6826 && !tail.eq_ignore_ascii_case("bc")
6827 && !tail.eq_ignore_ascii_case("ad");
6828 if !zone_shaped {
6829 return None;
6830 }
6831 let wall = format::parse_timestamp_literal_wall_ordered(head, order)?;
6832 Some((wall, tail))
6833}
6834
6835#[cfg(test)]
6836mod tests {
6837 use super::*;
6838 use alloc::vec;
6839 use spg_sql::ast::UnOp;
6840 use spg_storage::{ColumnSchema, DataType, Row};
6841
6842 fn col(name: &str, ty: DataType) -> ColumnSchema {
6843 ColumnSchema::new(name, ty, true)
6844 }
6845
6846 fn ctx<'a>(cols: &'a [ColumnSchema], alias: Option<&'a str>) -> EvalContext<'a> {
6847 EvalContext::new(cols, alias)
6848 }
6849
6850 /// v7.32 (P4 borrow channel) differential: the borrowed comparison
6851 /// fast path in `eval_expr`'s Binary arm must be byte-for-byte the
6852 /// pre-P4 owned path (`apply_binary` on cloned operands) across a
6853 /// cross-type value matrix and every comparison operator — covering
6854 /// the fast-path types (Text/Int/Float/Date/Timestamp/Bool/Null) and
6855 /// the owned-fallback types (Numeric/Interval).
6856 #[test]
6857 fn borrowed_compare_equals_owned_apply_binary() {
6858 let vals = vec![
6859 Value::Null,
6860 Value::Bool(true),
6861 Value::Bool(false),
6862 Value::SmallInt(3),
6863 Value::Int(3),
6864 Value::Int(-1),
6865 Value::BigInt(3),
6866 Value::BigInt(100),
6867 Value::Float(3.0),
6868 Value::Float(2.5),
6869 Value::text(String::new()),
6870 Value::text("a"),
6871 Value::text("b"),
6872 Value::Date(10),
6873 Value::Timestamp(1000),
6874 Value::Numeric {
6875 scaled: 30,
6876 scale: 1,
6877 kind: spg_storage::NumericKind::Finite,
6878 },
6879 Value::Interval {
6880 months: 0,
6881 days: 0,
6882 micros: 5,
6883 kind: spg_storage::IntervalKind::Finite,
6884 },
6885 ];
6886 let ops = [
6887 BinOp::Eq,
6888 BinOp::NotEq,
6889 BinOp::Lt,
6890 BinOp::LtEq,
6891 BinOp::Gt,
6892 BinOp::GtEq,
6893 ];
6894 let cs = vec![col("x", DataType::Int), col("y", DataType::Int)];
6895 let c = ctx(&cs, None);
6896 let lhs = Expr::Column(ColumnName {
6897 qualifier: None,
6898 name: "x".into(),
6899 });
6900 let rhs = Expr::Column(ColumnName {
6901 qualifier: None,
6902 name: "y".into(),
6903 });
6904 for l in &vals {
6905 for r in &vals {
6906 let row = Row::new(vec![l.clone(), r.clone()]);
6907 for op in ops {
6908 let got = eval_expr(
6909 &Expr::Binary {
6910 lhs: alloc::boxed::Box::new(lhs.clone()),
6911 op,
6912 rhs: alloc::boxed::Box::new(rhs.clone()),
6913 },
6914 &row,
6915 &c,
6916 );
6917 // Pre-P4 reference: owned operands through apply_binary
6918 // (collation fold is a no-op for non-CI columns).
6919 let want = apply_binary(op, l.clone(), r.clone());
6920 assert_eq!(
6921 format!("{got:?}"),
6922 format!("{want:?}"),
6923 "op={op:?} l={l:?} r={r:?}"
6924 );
6925 }
6926 }
6927 }
6928 }
6929
6930 fn lit(n: i64) -> Expr {
6931 Expr::Literal(Literal::Integer(n))
6932 }
6933
6934 fn null() -> Expr {
6935 Expr::Literal(Literal::Null)
6936 }
6937
6938 fn col_ref(name: &str) -> Expr {
6939 Expr::Column(ColumnName {
6940 qualifier: None,
6941 name: name.into(),
6942 })
6943 }
6944
6945 #[test]
6946 fn literal_evaluates_to_value() {
6947 let r = Row::new(vec![]);
6948 let cs: [ColumnSchema; 0] = [];
6949 let c = ctx(&cs, None);
6950 assert_eq!(eval_expr(&lit(42), &r, &c).unwrap(), Value::Int(42));
6951 assert_eq!(
6952 eval_expr(&Expr::Literal(Literal::Float(1.5)), &r, &c).unwrap(),
6953 Value::Float(1.5)
6954 );
6955 assert_eq!(eval_expr(&null(), &r, &c).unwrap(), Value::Null);
6956 }
6957
6958 #[test]
6959 fn column_lookup_unqualified() {
6960 let cs = vec![col("a", DataType::Int), col("b", DataType::Text)];
6961 let r = Row::new(vec![Value::Int(7), Value::text("hi")]);
6962 let c = ctx(&cs, None);
6963 assert_eq!(eval_expr(&col_ref("a"), &r, &c).unwrap(), Value::Int(7));
6964 assert_eq!(eval_expr(&col_ref("b"), &r, &c).unwrap(), Value::text("hi"));
6965 }
6966
6967 #[test]
6968 fn column_not_found_errors() {
6969 let cs = vec![col("a", DataType::Int)];
6970 let r = Row::new(vec![Value::Int(0)]);
6971 let c = ctx(&cs, None);
6972 let err = eval_expr(&col_ref("ghost"), &r, &c).unwrap_err();
6973 assert!(matches!(err, EvalError::ColumnNotFound { ref name } if name == "ghost"));
6974 }
6975
6976 #[test]
6977 fn qualified_column_matches_alias() {
6978 let cs = vec![col("a", DataType::Int)];
6979 let r = Row::new(vec![Value::Int(5)]);
6980 let c = ctx(&cs, Some("u"));
6981 let qualified = Expr::Column(ColumnName {
6982 qualifier: Some("u".into()),
6983 name: "a".into(),
6984 });
6985 assert_eq!(eval_expr(&qualified, &r, &c).unwrap(), Value::Int(5));
6986 }
6987
6988 #[test]
6989 fn qualified_column_unknown_alias_errors() {
6990 let cs = vec![col("a", DataType::Int)];
6991 let r = Row::new(vec![Value::Int(5)]);
6992 let c = ctx(&cs, Some("u"));
6993 let wrong = Expr::Column(ColumnName {
6994 qualifier: Some("x".into()),
6995 name: "a".into(),
6996 });
6997 assert!(matches!(
6998 eval_expr(&wrong, &r, &c).unwrap_err(),
6999 EvalError::UnknownQualifier { .. }
7000 ));
7001 }
7002
7003 #[test]
7004 fn arithmetic_with_widening() {
7005 let r = Row::new(vec![]);
7006 let cs: [ColumnSchema; 0] = [];
7007 let c = ctx(&cs, None);
7008 let e = Expr::Binary {
7009 lhs: alloc::boxed::Box::new(lit(2)),
7010 op: BinOp::Add,
7011 rhs: alloc::boxed::Box::new(Expr::Literal(Literal::Float(0.5))),
7012 };
7013 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Float(2.5));
7014 }
7015
7016 #[test]
7017 fn division_by_zero_errors() {
7018 let r = Row::new(vec![]);
7019 let cs: [ColumnSchema; 0] = [];
7020 let c = ctx(&cs, None);
7021 let e = Expr::Binary {
7022 lhs: alloc::boxed::Box::new(lit(1)),
7023 op: BinOp::Div,
7024 rhs: alloc::boxed::Box::new(lit(0)),
7025 };
7026 assert_eq!(
7027 eval_expr(&e, &r, &c).unwrap_err(),
7028 EvalError::DivisionByZero
7029 );
7030 }
7031
7032 #[test]
7033 fn comparison_returns_bool() {
7034 let r = Row::new(vec![]);
7035 let cs: [ColumnSchema; 0] = [];
7036 let c = ctx(&cs, None);
7037 let e = Expr::Binary {
7038 lhs: alloc::boxed::Box::new(lit(1)),
7039 op: BinOp::Lt,
7040 rhs: alloc::boxed::Box::new(lit(2)),
7041 };
7042 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Bool(true));
7043 }
7044
7045 #[test]
7046 fn null_propagates_through_arithmetic() {
7047 let r = Row::new(vec![]);
7048 let cs: [ColumnSchema; 0] = [];
7049 let c = ctx(&cs, None);
7050 let e = Expr::Binary {
7051 lhs: alloc::boxed::Box::new(lit(1)),
7052 op: BinOp::Add,
7053 rhs: alloc::boxed::Box::new(null()),
7054 };
7055 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Null);
7056 }
7057
7058 #[test]
7059 fn stack_depth_guard_trips_on_pathological_nesting() {
7060 // Built directly as an AST — the parser's own budgets (256
7061 // chained binary operators, 64 nesting levels) reject such SQL
7062 // long before eval sees it, so this exercises the eval-side
7063 // guard on its own. 30 000 frames overshoot the 768 KiB budget
7064 // at any conceivable frame size; the guard errors at the byte
7065 // budget, far below the worker stack, so deeper is safer.
7066 let mut e = Expr::Literal(Literal::Bool(true));
7067 for _ in 0..30_000 {
7068 e = Expr::Binary {
7069 lhs: alloc::boxed::Box::new(e),
7070 op: BinOp::And,
7071 rhs: alloc::boxed::Box::new(Expr::Literal(Literal::Bool(true))),
7072 };
7073 }
7074 let r = Row::new(vec![]);
7075 let cs: [ColumnSchema; 0] = [];
7076 let c = ctx(&cs, None);
7077 let err = eval_expr(&e, &r, &c).unwrap_err();
7078 assert!(matches!(err, EvalError::StackDepthExceeded), "{err:?}");
7079 // Dropping a 30 000-deep Box chain recurses in the drop glue —
7080 // deeper than the eval guard allows the EVAL side to go — so
7081 // leak it rather than gamble on the test thread's stack.
7082 core::mem::forget(e);
7083 }
7084
7085 #[test]
7086 fn and_three_valued_logic() {
7087 let r = Row::new(vec![]);
7088 let cs: [ColumnSchema; 0] = [];
7089 let c = ctx(&cs, None);
7090 let tt = |a: bool, b_null: bool| Expr::Binary {
7091 lhs: alloc::boxed::Box::new(Expr::Literal(Literal::Bool(a))),
7092 op: BinOp::And,
7093 rhs: alloc::boxed::Box::new(if b_null {
7094 null()
7095 } else {
7096 Expr::Literal(Literal::Bool(true))
7097 }),
7098 };
7099 // FALSE AND NULL → FALSE
7100 assert_eq!(
7101 eval_expr(&tt(false, true), &r, &c).unwrap(),
7102 Value::Bool(false)
7103 );
7104 // TRUE AND NULL → NULL
7105 assert_eq!(eval_expr(&tt(true, true), &r, &c).unwrap(), Value::Null);
7106 // TRUE AND TRUE → TRUE
7107 assert_eq!(
7108 eval_expr(&tt(true, false), &r, &c).unwrap(),
7109 Value::Bool(true)
7110 );
7111 }
7112
7113 #[test]
7114 fn or_three_valued_logic() {
7115 let r = Row::new(vec![]);
7116 let cs: [ColumnSchema; 0] = [];
7117 let c = ctx(&cs, None);
7118 let or_with_null = |a: bool| Expr::Binary {
7119 lhs: alloc::boxed::Box::new(Expr::Literal(Literal::Bool(a))),
7120 op: BinOp::Or,
7121 rhs: alloc::boxed::Box::new(null()),
7122 };
7123 // TRUE OR NULL → TRUE
7124 assert_eq!(
7125 eval_expr(&or_with_null(true), &r, &c).unwrap(),
7126 Value::Bool(true)
7127 );
7128 // FALSE OR NULL → NULL
7129 assert_eq!(
7130 eval_expr(&or_with_null(false), &r, &c).unwrap(),
7131 Value::Null
7132 );
7133 }
7134
7135 #[test]
7136 fn not_on_null_is_null() {
7137 let r = Row::new(vec![]);
7138 let cs: [ColumnSchema; 0] = [];
7139 let c = ctx(&cs, None);
7140 let e = Expr::Unary {
7141 op: UnOp::Not,
7142 expr: alloc::boxed::Box::new(null()),
7143 };
7144 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Null);
7145 }
7146
7147 #[test]
7148 fn text_comparison_lexicographic() {
7149 let r = Row::new(vec![]);
7150 let cs: [ColumnSchema; 0] = [];
7151 let c = ctx(&cs, None);
7152 let e = Expr::Binary {
7153 lhs: alloc::boxed::Box::new(Expr::Literal(Literal::String("apple".into()))),
7154 op: BinOp::Lt,
7155 rhs: alloc::boxed::Box::new(Expr::Literal(Literal::String("banana".into()))),
7156 };
7157 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Bool(true));
7158 }
7159
7160 #[test]
7161 fn interval_format_basics() {
7162 // v7.37.5 β — three-arg signature. PG byte-equal:
7163 // `'1 day'` ≠ `'24 hours'` now, the format reflects it.
7164 assert_eq!(format_interval(0, 0, 0), "00:00:00");
7165 assert_eq!(format_interval(0, 1, 0), "1 day");
7166 assert_eq!(format_interval(0, -1, 0), "-1 days");
7167 assert_eq!(format_interval(0, 0, 86_400_000_000), "24:00:00");
7168 assert_eq!(format_interval(0, 0, 3_600_000_000), "01:00:00");
7169 assert_eq!(format_interval(0, 1, 9_000_000), "1 day 00:00:09");
7170 assert_eq!(format_interval(14, 0, 0), "1 year 2 mons");
7171 assert_eq!(format_interval(-1, 0, 0), "-1 mons");
7172 }
7173
7174 #[test]
7175 fn interval_format_pg_byte_equal_day_vs_24h() {
7176 // v7.37.5 β — the PG-canonical distinction `'1 day'` ≠ `'24 hours'`
7177 // is preserved in the formatter, not just the parser.
7178 assert_eq!(format_interval(0, 1, 0), "1 day");
7179 assert_eq!(format_interval(0, 0, 86_400_000_000), "24:00:00");
7180 assert_ne!(
7181 format_interval(0, 1, 0),
7182 format_interval(0, 0, 86_400_000_000),
7183 );
7184 }
7185}