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