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