spg_engine/eval/cast.rs
1//! `expr::TYPE` CAST evaluation (cut 29 — extracted from `eval.rs`).
2//!
3//! Implements PG-style runtime coercion: the giant `cast_value`
4//! dispatcher plus its per-target helpers (numeric / bool / array /
5//! date / timestamp / interval / vector). Date and timestamp casts
6//! defer to the calendar parsers (`parse_date_literal` /
7//! `parse_timestamp_literal`) that stay in `eval.rs`; tsvector /
8//! tsquery casts defer to the FTS codecs re-exported from
9//! `eval::textsearch`.
10
11use alloc::format;
12use alloc::string::{String, ToString};
13use alloc::vec::Vec;
14
15use spg_sql::ast::CastTarget;
16use spg_storage::Value;
17
18use super::math::{f64_powi, f64_round_half_even};
19use super::{
20 EvalError, decode_tsquery_external, decode_tsvector_external, parse_date_literal,
21 parse_timestamp_literal, value_to_text,
22};
23
24/// v7.39 (read01 regproc.c) — the reg* input types SPG carries as text:
25/// resolve the name and return its canonical rendering, with PG's
26/// distinct not-found / ambiguous errors.
27/// v7.39 (round 515) — the type names each family below accepts, declared
28/// ONCE so the value path and the NULL-target check cannot drift apart.
29///
30/// They already had, three times. Round 509 added a check that a cast target
31/// names something, with its own hand-kept list; round 512 then added
32/// `Value::Cid` without registering `cid` there, round 514 found that and
33/// `xid`, and round 515 found six more — `'english'::regconfig` resolved
34/// while `NULL::regconfig` did not. The duplication was the defect, so it
35/// is gone rather than patched a third time.
36pub(crate) const REG_MISC_TYPES: &[&str] = &[
37 "regproc",
38 "regprocedure",
39 "regoper",
40 "regoperator",
41 "regconfig",
42 "regdictionary",
43 "regcollation",
44];
45
46/// The catalog-shaped scalars — see `cast_catalog_scalar`.
47pub(crate) const CATALOG_SCALAR_TYPES: &[&str] = &[
48 "cid",
49 "xid",
50 "oidvector",
51 "int2vector",
52 "aclitem",
53 "refcursor",
54 "pg_snapshot",
55 "txid_snapshot",
56 "jsonpath",
57];
58
59/// The pseudotypes and the statistics / GiST internals: a NULL is NULL, a
60/// value is "cannot accept a value of type X".
61pub(crate) const OPAQUE_TYPES: &[&str] = &[
62 "anyarray",
63 "anyelement",
64 "anyenum",
65 "anyrange",
66 "anymultirange",
67 "anynonarray",
68 "anycompatible",
69 "anycompatiblearray",
70 "anycompatiblenonarray",
71 "anycompatiblerange",
72 "anycompatiblemultirange",
73 "any",
74 "trigger",
75 "event_trigger",
76 "internal",
77 "language_handler",
78 "fdw_handler",
79 "pg_ddl_command",
80 "pg_node_tree",
81 "pg_ndistinct",
82 "pg_mcv_list",
83 "pg_dependencies",
84 "pg_brin_minmax_multi_summary",
85 "pg_brin_bloom_summary",
86 "gtsvector",
87];
88
89fn cast_reg_misc(kind: &str, s: &str) -> Result<Value<'static>, EvalError> {
90 let bare = s
91 .strip_prefix("pg_catalog.")
92 .unwrap_or(s)
93 .trim()
94 .to_string();
95 match kind {
96 "regconfig" => {
97 const CONFIGS: &[&str] = &[
98 "simple",
99 "arabic",
100 "armenian",
101 "basque",
102 "catalan",
103 "danish",
104 "dutch",
105 "english",
106 "finnish",
107 "french",
108 "german",
109 "greek",
110 "hindi",
111 "hungarian",
112 "indonesian",
113 "irish",
114 "italian",
115 "lithuanian",
116 "nepali",
117 "norwegian",
118 "portuguese",
119 "romanian",
120 "russian",
121 "serbian",
122 "spanish",
123 "swedish",
124 "tamil",
125 "turkish",
126 "yiddish",
127 ];
128 if CONFIGS.contains(&bare.as_str()) {
129 Ok(Value::text(bare))
130 } else {
131 Err(EvalError::TypeMismatch {
132 detail: alloc::format!("text search configuration \"{s}\" does not exist"),
133 })
134 }
135 }
136 // v7.39 (round 513) — `regcollation`. PG lowercases an UNQUOTED
137 // identifier before it looks, which is why `'C'::regcollation` is
138 // "collation \"c\" for encoding \"UTF8\" does not exist" there while
139 // `'\"C\"'::regcollation` resolves — measured, and the reason the
140 // quoted form is the one anybody writes. The rendering keeps the
141 // quotes PG puts back on a name that needs them.
142 "regcollation" => {
143 let quoted = bare.starts_with('"') && bare.ends_with('"') && bare.len() >= 2;
144 let name = if quoted {
145 bare[1..bare.len() - 1].to_string()
146 } else {
147 bare.to_ascii_lowercase()
148 };
149 const COLLATIONS: &[&str] = &["C", "POSIX", "default", "ucs_basic"];
150 match COLLATIONS.iter().find(|c| **c == name) {
151 // PG re-quotes anything that is not a plain lowercase word.
152 Some(c) => Ok(Value::text(
153 if c.chars().all(|ch| ch.is_ascii_lowercase() || ch == '_') && *c != "default" {
154 (*c).to_string()
155 } else {
156 alloc::format!("\"{c}\"")
157 },
158 )),
159 None => Err(EvalError::TypeMismatch {
160 detail: alloc::format!(
161 "collation \"{name}\" for encoding \"UTF8\" does not exist"
162 ),
163 }),
164 }
165 }
166 "regdictionary" => {
167 if bare == "simple" || bare.ends_with("_stem") {
168 Ok(Value::text(bare))
169 } else {
170 Err(EvalError::TypeMismatch {
171 detail: alloc::format!("text search dictionary \"{s}\" does not exist"),
172 })
173 }
174 }
175 "regproc" => {
176 let hits = crate::system_catalog::PG_PROC_FUNCS
177 .iter()
178 .filter(|(_, n, ..)| *n == bare)
179 .count();
180 match hits {
181 0 => Err(EvalError::TypeMismatch {
182 detail: alloc::format!("function \"{s}\" does not exist"),
183 }),
184 1 => Ok(Value::text(bare)),
185 _ => Err(EvalError::TypeMismatch {
186 detail: alloc::format!("more than one function named \"{bare}\""),
187 }),
188 }
189 }
190 "regprocedure" => {
191 // `name(argtype, ...)` — resolve the name, canonicalize each
192 // argument type, and re-render.
193 let Some((fname, rest)) = bare.split_once('(') else {
194 return Err(EvalError::TypeMismatch {
195 detail: alloc::format!("expected a left parenthesis in \"{s}\""),
196 });
197 };
198 let Some(args_txt) = rest.strip_suffix(')') else {
199 return Err(EvalError::TypeMismatch {
200 detail: alloc::format!("expected a right parenthesis in \"{s}\""),
201 });
202 };
203 let fname = fname.trim().to_ascii_lowercase();
204 let args: Vec<String> = if args_txt.trim().is_empty() {
205 Vec::new()
206 } else {
207 args_txt
208 .split(',')
209 .map(|a| {
210 crate::conversions::regtype_canonical_name(a.trim()).ok_or_else(|| {
211 EvalError::TypeMismatch {
212 detail: alloc::format!("type \"{}\" does not exist", a.trim()),
213 }
214 })
215 })
216 .collect::<Result<_, _>>()?
217 };
218 let nargs = args.len() as i32;
219 let known = crate::system_catalog::PG_PROC_FUNCS
220 .iter()
221 .any(|(_, n, _, na, _)| *n == fname && *na == nargs);
222 if !known {
223 return Err(EvalError::TypeMismatch {
224 detail: alloc::format!("function \"{s}\" does not exist"),
225 });
226 }
227 Ok(Value::text(alloc::format!("{fname}({})", args.join(","))))
228 }
229 // regoper / regoperator: SPG has no operator catalog; every core
230 // operator symbol is multiply overloaded in PG, so a known symbol
231 // reports PG's ambiguity and anything else does not exist.
232 _ => {
233 let sym: String = bare.chars().filter(|c| !c.is_whitespace()).collect();
234 let core_op = !sym.is_empty() && sym.chars().all(|c| "+-*/<>=~!@#%^&|`?".contains(c));
235 if core_op {
236 Err(EvalError::TypeMismatch {
237 detail: alloc::format!("more than one operator named {sym}"),
238 })
239 } else {
240 Err(EvalError::TypeMismatch {
241 detail: alloc::format!("operator does not exist: {s}"),
242 })
243 }
244 }
245 }
246}
247
248/// v7.39 (round 355, M13) — `BINARY expr` / `CAST(expr AS BINARY[(n)])`.
249fn cast_mysql_binary(v: Value<'static>, name: &str) -> Result<Value<'static>, EvalError> {
250 let limit: Option<usize> = name
251 .split_once('(')
252 .and_then(|(_, rest)| rest.trim_end_matches(')').trim().parse().ok());
253 let text = match &v {
254 Value::Null => return Ok(Value::Null),
255 Value::Text(t) => t.to_string(),
256 Value::BpChar(t) => t.to_string(),
257 other => crate::eval::values::value_to_text(other),
258 };
259 Ok(Value::text(match limit {
260 // Byte-wise, which is the point of the type.
261 Some(n) if text.len() > n => {
262 let mut cut = n;
263 while cut > 0 && !text.is_char_boundary(cut) {
264 cut -= 1;
265 }
266 text[..cut].to_string()
267 }
268 _ => text,
269 }))
270}
271
272/// v7.39 (round 352, M8) — `CAST(x AS SIGNED)` / `CAST(x AS UNSIGNED)`.
273fn cast_mysql_integer(v: Value<'static>, unsigned: bool) -> Result<Value<'static>, EvalError> {
274 // v7.39 (round 527) — an EXACT integer source must not round-trip
275 // through f64. It loses precision above 2^53, and the float→int cast
276 // SATURATES, so `CAST(18446744073709551615 AS UNSIGNED)` answered
277 // 9223372036854775807 — a different number, with nothing to say so.
278 // The value stores, compares and sums correctly at full width
279 // (measured against MariaDB 11); only the cast reduced it.
280 let exact: Option<i128> = match &v {
281 Value::Bool(b) => Some(i128::from(u8::from(*b))),
282 Value::SmallInt(x) => Some(i128::from(*x)),
283 Value::Int(x) => Some(i128::from(*x)),
284 Value::BigInt(x) => Some(i128::from(*x)),
285 Value::Numeric {
286 scaled,
287 scale: 0,
288 kind: spg_storage::NumericKind::Finite,
289 } => Some(*scaled),
290 _ => None,
291 };
292 let rounded: i128 = match exact {
293 Some(n) => n,
294 None => {
295 let n: f64 = match &v {
296 Value::Null => return Ok(Value::Null),
297 Value::Float(x) => *x,
298 Value::Real(x) => f64::from(*x),
299 #[allow(clippy::cast_precision_loss)]
300 Value::Numeric { scaled, scale, .. } => {
301 *scaled as f64 / 10_f64.powi(i32::from(*scale))
302 }
303 Value::Text(t) | Value::BpChar(t) => crate::eval::mysql_leading_number(t),
304 other => {
305 return Err(EvalError::TypeMismatch {
306 detail: alloc::format!(
307 "cannot cast {} to integer",
308 crate::conversions::pg_type_name_for_error_opt(other.data_type())
309 ),
310 });
311 }
312 };
313 // Half away from zero, which is what MariaDB does
314 // (2.5 → 3, -2.5 → -3).
315 let r = if n >= 0.0 {
316 (n + 0.5).floor()
317 } else {
318 (n - 0.5).ceil()
319 };
320 #[allow(clippy::cast_possible_truncation)]
321 let as_i64 = r as i64;
322 i128::from(as_i64)
323 }
324 };
325 if unsigned {
326 // MariaDB wraps a negative through the full u64 range.
327 let wrapped: u64 = if rounded < 0 {
328 #[allow(clippy::cast_sign_loss, clippy::cast_possible_truncation)]
329 {
330 rounded as i64 as u64
331 }
332 } else {
333 u64::try_from(rounded).unwrap_or(u64::MAX)
334 };
335 // Above i64::MAX the value only fits the numeric carrier, which
336 // is the same one a BIGINT UNSIGNED column already uses.
337 return Ok(if wrapped > i64::MAX as u64 {
338 Value::Numeric {
339 scaled: i128::from(wrapped),
340 scale: 0,
341 kind: spg_storage::NumericKind::Finite,
342 }
343 } else {
344 #[allow(clippy::cast_possible_wrap)]
345 Value::BigInt(wrapped as i64)
346 });
347 }
348 Ok(Value::BigInt(
349 i64::try_from(rounded).unwrap_or(if rounded < 0 { i64::MIN } else { i64::MAX }),
350 ))
351}
352
353/// v7.39 (round 544) — the integer a bytea's bytes spell: big-endian,
354/// right-aligned into `width` bytes, sign-extended from the leading
355/// byte when the value already fills the width. Measured on PG18:
356/// `'\x05'::bytea::int8` is 5, `'\x'::bytea::int4` is 0,
357/// `'\xffffffff'::bytea::int4` is -1.
358#[inline(never)]
359fn bytea_to_integer(v: &Value<'static>, width: usize) -> Result<Value<'static>, EvalError> {
360 let Value::Bytes(b) = v else {
361 return Err(EvalError::TypeMismatch {
362 detail: alloc::string::String::from("expected bytea"),
363 });
364 };
365 if b.len() > width {
366 return Err(EvalError::TypeMismatch {
367 detail: alloc::format!("bytea of {} bytes is too wide for the target", b.len()),
368 });
369 }
370 let negative = b.len() == width && b.first().is_some_and(|f| *f & 0x80 != 0);
371 let mut acc: i64 = if negative { -1 } else { 0 };
372 for byte in b.iter() {
373 acc = (acc << 8) | i64::from(*byte);
374 }
375 Ok(if width == 4 {
376 Value::Int(i32::try_from(acc).unwrap_or(0))
377 } else {
378 Value::BigInt(acc)
379 })
380}
381
382/// Round a numeric operand (`scaled` × 10^-`scale`) to the nearest
383/// integer, half-away-from-zero — PG's `numeric → int` coercion rule.
384fn numeric_round_to_i128(scaled: i128, scale: u16) -> i128 {
385 let factor = 10_i128.pow(u32::from(scale));
386 let neg = scaled < 0;
387 let abs = scaled.unsigned_abs() as i128;
388 let q = abs / factor;
389 let r = abs % factor;
390 let mag = if 2 * r >= factor { q + 1 } else { q };
391 if neg { -mag } else { mag }
392}
393
394/// PG-style `expr::TYPE` coercion. NULL always casts as NULL.
395pub fn cast_value(v: Value<'static>, target: CastTarget) -> Result<Value<'static>, EvalError> {
396 cast_value_in(v, target, false)
397}
398
399/// v7.39 (round 352, M8) — `cast_value` with the session dialect, for the
400/// targets the two disagree about (`SIGNED` / `UNSIGNED` exist only in
401/// MySQL: PG says `type "signed" does not exist`, measured).
402pub fn cast_value_in(
403 v: Value<'static>,
404 target: CastTarget,
405 mysql: bool,
406) -> Result<Value<'static>, EvalError> {
407 cast_value_ref_in(v, &target, mysql)
408}
409
410/// v7.39 (round 607) — the same dispatch, taking the target by REFERENCE.
411///
412/// `eval_cast_arm` cloned the target for every row. For the settled variants
413/// that clone is free, which is why `id::FLOAT` allocated nothing a row while
414/// `id::REAL` — the same conversion under a name the parser leaves as
415/// `Named(String)` — allocated one just to hand the name over, and seven more
416/// re-deriving its lowercase form inside.
417pub fn cast_value_ref_in(
418 v: Value<'static>,
419 target: &CastTarget,
420 mysql: bool,
421) -> Result<Value<'static>, EvalError> {
422 // v7.37 (round 896) — the quoted spelling of these two arrives as
423 // `Named`, the bare one as its own variant, and only the variant had an
424 // arm. `::regclass` worked and `::"regclass"` answered `type "regclass"
425 // does not exist` — and quoted identifiers are what an ORM or pg_dump
426 // writes. Folding here rather than adding a second arm keeps one
427 // implementation: whatever the variant does, the quoted form now does.
428 // Round 894 fixed `tsvector` / `tsquery` the same way round and left
429 // these open because this path had not been read; it has now.
430 if let CastTarget::Named(n) = target {
431 if n.eq_ignore_ascii_case("regclass") {
432 return cast_value_ref_in(v, &CastTarget::RegClass, mysql);
433 }
434 if n.eq_ignore_ascii_case("regtype") {
435 return cast_value_ref_in(v, &CastTarget::RegType, mysql);
436 }
437 }
438 // v7.39 (round 509) — PG validates the cast TARGET whatever the operand
439 // is: `NULL::nosuchtype` is an error there, not NULL. This returned early
440 // before ever looking at the target, so a misspelt type name silently
441 // produced NULL and `pg_typeof(NULL::nosuchtype)` answered `unknown`. A
442 // value operand DID error, so the gap was exactly the NULL case, in both
443 // spellings (`::t` and `CAST(… AS t)`).
444 //
445 // Only `Named` can fail to resolve; every other CastTarget is a variant
446 // the parser already settled. So a NULL keeps its short-circuit
447 // everywhere else and a Named target runs the real path, which is the
448 // only thing that knows every name that resolves. Writing a second
449 // resolver to check the name against looked simpler and was wrong: it
450 // missed `::binary` (the MySQL prefix's desugar), a table's row type,
451 // and the pseudotypes, all of which resolve further down this arm.
452 if matches!(v, Value::Null) {
453 return Ok(Value::Null);
454 }
455 match target {
456 CastTarget::Vector => cast_to_vector(v),
457 // v7.38 (read01) — the inet/cidr ::text cast shows the mask even for
458 // /32 and /128 (PG's cast-path form, unlike the display default).
459 CastTarget::Text => Ok(Value::text(match &v {
460 Value::Inet { family, bits, addr } | Value::Cidr { family, bits, addr } => {
461 crate::conversions::format_inet_full(*family, *bits, addr)
462 }
463 // v7.38 (read01, T11) — bpchar → text strips the trailing blanks
464 // (unlike the padded wire display).
465 Value::BpChar(s) => s.trim_end_matches(' ').to_string(),
466 // v7.39 (read01 ruleutils.c) — regclass::text is the name.
467 Value::RegClass(_, name) | Value::RegProc(_, name) | Value::RegType(_, name) => {
468 name.to_string()
469 }
470 _ => value_to_text(&v),
471 })),
472 // v7.39 (round 254) — the integer targets refuse a NUMERIC special
473 // outright (PG: `cannot convert NaN to integer`); without this the
474 // arms below read the special's canonical mantissa and answered 0.
475 // The float / numeric targets pass it through instead — handled in
476 // their own arms, which now consult `kind`.
477 // v7.39 (round 343) — an OID-typed reference casts to an integer
478 // the way PG's do (`'t'::regclass::bigint` is 27830 there). SPG
479 // reported `cannot cast None to bigint`: the integer path read the
480 // value's storage DataType, which these two deliberately do not
481 // have, and the message leaked that `None` to the client.
482 CastTarget::BigInt | CastTarget::Int
483 if matches!(
484 v,
485 Value::RegClass(..) | Value::RegProc(..) | Value::RegType(..)
486 ) =>
487 {
488 let (Value::RegClass(oid, _) | Value::RegProc(oid, _) | Value::RegType(oid, _)) = v
489 else {
490 unreachable!("guarded above")
491 };
492 Ok(if matches!(target, CastTarget::BigInt) {
493 Value::BigInt(oid)
494 } else {
495 Value::Int(i32::try_from(oid).unwrap_or(i32::MAX))
496 })
497 }
498 // v7.39 (round 544) — bytea reads back as the integer its bytes
499 // spell, big-endian and right-aligned. Measured on PG18:
500 // '\x05'::bytea::int8 is 5, '\x'::bytea::int4 is 0.
501 CastTarget::Int if matches!(v, Value::Bytes(_)) => bytea_to_integer(&v, 4),
502 CastTarget::BigInt if matches!(v, Value::Bytes(_)) => bytea_to_integer(&v, 8),
503 CastTarget::Int => {
504 cast_numeric_special_reject(&v, "integer").unwrap_or_else(|| cast_numeric_to_int(v))
505 }
506 CastTarget::BigInt => {
507 cast_numeric_special_reject(&v, "bigint").unwrap_or_else(|| cast_numeric_to_bigint(v))
508 }
509 CastTarget::Float => cast_numeric_to_float(v),
510 CastTarget::Bool => cast_to_bool(v),
511 CastTarget::Date => cast_to_date(v),
512 // TIMESTAMP and TIMESTAMPTZ share a runtime representation
513 // (i64 microseconds UTC) but NOT an input rule, and conflating
514 // the two silently stored the wrong instant. `::timestamp`
515 // keeps the wall clock a literal's offset was written against
516 // (round 289); `::timestamptz` CONVERTS by it. The evaluator's
517 // context-aware arm intercepts timestamptz before this point,
518 // so the difference only showed on the paths that reach here
519 // directly — INSERT VALUES folds its literals through
520 // `literal_expr_to_value_in`, and stored 10:00 for
521 // `'2020-01-01 10:00:00+02'::timestamptz` where PG stores
522 // 08:00 (round 310).
523 // v7.39 (round 423) — `CAST(x AS DATETIME)` / `AS TIMESTAMP` reach
524 // here as the dedicated variant rather than `Named`, so the MySQL
525 // "bare temporal type has fractional precision 0" rule has to be
526 // applied here too: MariaDB drops the fraction, PG keeps every
527 // microsecond.
528 CastTarget::Timestamp => {
529 let out = cast_to_timestamp(v)?;
530 Ok(if mysql {
531 round_temporal_to_precision(out, 0, true)
532 } else {
533 out
534 })
535 }
536 CastTarget::Timestamptz => cast_to_timestamptz(v),
537 // v7.9.25 — `expr::INTERVAL`. Currently only TEXT → Interval
538 // is supported (the mailrs idiom: `$1::INTERVAL` where the
539 // bound param is a string like `'7 days'`).
540 // v7.39 (round 544) — a time-of-day IS an interval of that
541 // length. Measured: '10:20:30.123456'::time::interval reads
542 // 10:20:30.123456 on PG18, fractional seconds and all.
543 CastTarget::Interval => match v {
544 Value::Time(us) => Ok(Value::Interval {
545 months: 0,
546 days: 0,
547 micros: us,
548 }),
549 other => cast_to_interval(other),
550 },
551 // v7.9.25 — `::json` keeps the input text verbatim (PG's json
552 // type preserves whitespace / key order / duplicates).
553 CastTarget::Json => match v {
554 Value::Json(s) => Ok(Value::json(s)),
555 Value::Text(s) => Ok(Value::json(s)),
556 other => Err(EvalError::TypeMismatch {
557 detail: alloc::format!(
558 "::json only accepts TEXT-shape inputs, got {}",
559 crate::conversions::pg_type_name_for_error_opt(other.data_type())
560 ),
561 }),
562 },
563 // v7.38 (read01) — `::jsonb` canonicalises like PG: object keys
564 // sorted (length, then bytes) + duplicates collapsed last-wins,
565 // `, ` / `: ` whitespace, and numbers normalised. Invalid JSON
566 // falls back to the verbatim text (validation stays a separate
567 // concern from this representation fix).
568 CastTarget::Jsonb => match v {
569 // v7.39 (read01 jsonb) — the explicit ::jsonb cast validates:
570 // invalid tokens (NaN / Infinity / malformed) error like PG
571 // instead of passing the raw text through.
572 Value::Json(s) | Value::Text(s) => match crate::json::canonicalize_jsonb(s.as_ref()) {
573 Ok(c) => Ok(Value::json(c)),
574 Err(_) => Err(EvalError::TypeMismatch {
575 detail: alloc::string::String::from("invalid input syntax for type json"),
576 }),
577 },
578 other => Err(EvalError::TypeMismatch {
579 detail: alloc::format!(
580 "::jsonb only accepts TEXT-shape inputs, got {}",
581 crate::conversions::pg_type_name_for_error_opt(other.data_type())
582 ),
583 }),
584 },
585 // v7.17.0 Phase 5.3 — `::regtype` / `::regclass`. PG
586 // semantics: each is a textual catalog-name surfacing as
587 // a numeric OID at the wire layer that renders back as
588 // the original name. SPG has no OID space, but pg_dump /
589 // mailrs / Django code uses the cast purely for textual
590 // round-trip — feeding `'public.t'::regclass::text` into
591 // a downstream `format(…)` or string concat. We map to
592 // that textual contract: Text in → Text out (the schema-
593 // qualifier `public.` is stripped to match PG's default
594 // search_path-aware rendering); numeric in → re-cast to
595 // Text as best-effort; anything else errors.
596 //
597 // Pre-3.3 / pre-5.3 (v7.9.26) the cast surfaced a clean
598 // error; this lifts to accept-and-textify so the dominant
599 // dump-loader pattern unblocks. SPG-shaped queries that
600 // genuinely need an OID for runtime joins are still
601 // documented as unsupported.
602 // v7.39 (round 694) — `'{text,int4}'::regtype[]`. PG canonicalises
603 // every ELEMENT (`int4` → `integer`) and rejects an unknown one, so
604 // the array runs the scalar's own name resolution per member rather
605 // than keeping the literal. `regclass[]` keeps its names — a
606 // relation name is already what PG prints.
607 CastTarget::Named(n) if n.eq_ignore_ascii_case("regtype_array") => {
608 let Value::Text(s) = &v else {
609 return Ok(v);
610 };
611 let body = s.trim();
612 let inner = body
613 .strip_prefix('{')
614 .and_then(|b| b.strip_suffix('}'))
615 .unwrap_or(body);
616 let mut out: Vec<Option<alloc::string::String>> = Vec::new();
617 for part in inner.split(',') {
618 let t = part.trim();
619 if t.is_empty() {
620 continue;
621 }
622 if t.eq_ignore_ascii_case("NULL") {
623 out.push(None);
624 continue;
625 }
626 let bare = t.rsplit('.').next().unwrap_or(t);
627 match crate::conversions::regtype_canonical_name(bare) {
628 Some(c) => out.push(Some(c)),
629 None => {
630 return Err(EvalError::TypeMismatch {
631 detail: alloc::format!("type \"{t}\" does not exist"),
632 });
633 }
634 }
635 }
636 Ok(Value::TextArray(out))
637 }
638 CastTarget::RegType | CastTarget::RegClass => match v {
639 Value::Text(s) => {
640 // Strip an optional `<schema>.` prefix — PG's
641 // regclass render drops it when the schema is on
642 // the search_path; SPG is single-schema so
643 // dropping is always safe.
644 let bare = s.rsplit('.').next().unwrap_or(&s).to_string();
645 // v7.39 (read01 regproc.c) — regtype canonicalizes the
646 // name ('int4' → 'integer') and rejects unknown types
647 // (PG 42704).
648 if matches!(target, CastTarget::RegType) {
649 // v7.39 (round 648) — carry the OID as well as the
650 // name, the way `::regclass` and `::regproc` already
651 // do. As a plain Text this rendered correctly and
652 // then failed everything downstream: `'text'::regtype
653 // ::oid` parsed the NAME as a number and answered
654 // `invalid input syntax for type oid: "text"` where
655 // PG answers 25, and `pg_typeof` said `text`.
656 return match crate::conversions::regtype_canonical_name(&bare) {
657 Some(c) => {
658 let oid =
659 crate::conversions::regtype_name_to_oid(&c.to_ascii_lowercase())
660 .unwrap_or(0);
661 Ok(Value::RegType(oid, c.into_boxed_str()))
662 }
663 None => Err(EvalError::TypeMismatch {
664 detail: alloc::format!("type \"{s}\" does not exist"),
665 }),
666 };
667 }
668 Ok(Value::text(bare))
669 }
670 // A numeric OID → its type name for `::regtype` (the common
671 // `atttypid::regtype` column-type-name shape). `::regclass`
672 // needs a catalog reverse-lookup for user relations, which
673 // this cast has no access to, so it keeps rendering the OID.
674 Value::Int(_) | Value::BigInt(_) => {
675 let n = match v {
676 Value::Int(n) => i64::from(n),
677 Value::BigInt(n) => n,
678 _ => unreachable!(),
679 };
680 if matches!(target, CastTarget::RegType)
681 && let Some(name) = crate::conversions::regtype_oid_to_name_owned(n)
682 {
683 Ok(Value::RegType(n, name.into_boxed_str()))
684 } else {
685 Ok(Value::text(alloc::format!("{n}")))
686 }
687 }
688 other => Err(EvalError::TypeMismatch {
689 detail: alloc::format!(
690 "::regtype / ::regclass accepts TEXT (name) or integer (oid), got {}",
691 crate::conversions::pg_type_name_for_error_opt(other.data_type())
692 ),
693 }),
694 },
695 // v7.10.11 — `::TEXT[]`. Decode PG external array form
696 // when input is Text; pass through unchanged when it is
697 // already TextArray. Anything else is a type mismatch.
698 CastTarget::TextArray => match v {
699 Value::TextArray(items) => Ok(Value::TextArray(items)),
700 Value::Text(s) => {
701 if let Some(r) = try_cast_2d_array(&s, |row| {
702 decode_text_array_external(row).map(Value::TextArray)
703 }) {
704 return r;
705 }
706 decode_text_array_external(&s).map(Value::TextArray)
707 }
708 // Other scalar arrays cast element-wise, each element
709 // rendered as its own text (NULLs preserved). PG allows
710 // `ARRAY[1,2,3]::text[]`.
711 Value::IntArray(items) => Ok(Value::TextArray(
712 items
713 .into_iter()
714 .map(|o| o.map(|n| alloc::format!("{n}")))
715 .collect(),
716 )),
717 Value::BigIntArray(items) => Ok(Value::TextArray(
718 items
719 .into_iter()
720 .map(|o| o.map(|n| alloc::format!("{n}")))
721 .collect(),
722 )),
723 Value::SmallIntArray(items) => Ok(Value::TextArray(
724 items
725 .into_iter()
726 .map(|o| o.map(|n| alloc::format!("{n}")))
727 .collect(),
728 )),
729 Value::BoolArray(items) => Ok(Value::TextArray(
730 items
731 .into_iter()
732 .map(|o| o.map(|b| String::from(if b { "t" } else { "f" })))
733 .collect(),
734 )),
735 Value::FloatArray(items) => Ok(Value::TextArray(
736 items
737 .into_iter()
738 .map(|o| o.map(|x| value_to_text(&Value::Float(x))))
739 .collect(),
740 )),
741 other => Err(EvalError::TypeMismatch {
742 detail: alloc::format!(
743 "::TEXT[] only accepts TEXT / array inputs, got {}",
744 crate::conversions::pg_type_name_for_error_opt(other.data_type())
745 ),
746 }),
747 },
748 // v7.11.13 — `::INT[]` / `::BIGINT[]`. Decode PG external
749 // form `{1,2,3}` when input is Text; widen TextArray /
750 // IntArray as appropriate.
751 CastTarget::IntArray => cast_to_int_array(v),
752 CastTarget::BigIntArray => cast_to_bigint_array(v),
753 // v7.12.0 — `::tsvector` / `::tsquery`. Decodes PG external
754 // form when input is Text; passes through unchanged when the
755 // input is already the target type. Other inputs are a type
756 // mismatch. Lexer / Porter stemmer arrive in v7.12.1; the
757 // external-form cast at v7.12.0 is the path pg_dump and
758 // direct-literal callers use.
759 CastTarget::TsVector => match v {
760 Value::TsVector(items) => Ok(Value::TsVector(items)),
761 Value::Text(s) => decode_tsvector_external(&s).map(Value::TsVector),
762 other => Err(EvalError::TypeMismatch {
763 detail: alloc::format!(
764 "::tsvector only accepts TEXT / tsvector inputs, got {}",
765 crate::conversions::pg_type_name_for_error_opt(other.data_type())
766 ),
767 }),
768 },
769 CastTarget::TsQuery => match v {
770 Value::TsQuery(ast) => Ok(Value::TsQuery(ast)),
771 Value::Text(s) => decode_tsquery_external(&s).map(Value::TsQuery),
772 other => Err(EvalError::TypeMismatch {
773 detail: alloc::format!(
774 "::tsquery only accepts TEXT / tsquery inputs, got {}",
775 crate::conversions::pg_type_name_for_error_opt(other.data_type())
776 ),
777 }),
778 },
779 // v7.17.0 — `::uuid`. Identity for `uuid → uuid`; parse
780 // text via the shared `parse_uuid_str`. Anything else is a
781 // type mismatch — PG also rejects e.g. INT → UUID without
782 // an explicit text bridge.
783 CastTarget::Uuid => match v {
784 Value::Uuid(b) => Ok(Value::Uuid(b)),
785 Value::Text(s) => match spg_storage::parse_uuid_str(&s) {
786 Some(b) => Ok(Value::Uuid(b)),
787 None => Err(EvalError::TypeMismatch {
788 detail: alloc::format!("invalid input syntax for type uuid: {s:?}"),
789 }),
790 },
791 other => Err(EvalError::TypeMismatch {
792 detail: alloc::format!(
793 "::uuid only accepts TEXT / uuid inputs, got {}",
794 crate::conversions::pg_type_name_for_error_opt(other.data_type())
795 ),
796 }),
797 },
798 // v7.18 — `::bytea`. Identity for `Bytes → Bytes`; decode
799 // Text via the engine's PG-format bytea decoder (`\x`
800 // hex form + `\NNN` escape form). Anything else is a type
801 // mismatch — same shape as PG's contract. Closes the
802 // mailrs D-pre #3 reverse-acceptance gap.
803 CastTarget::Bytea => match v {
804 Value::Bytes(b) => Ok(Value::bytes(b)),
805 Value::Text(s) => match crate::conversions::decode_bytea_literal(&s) {
806 Ok(b) => Ok(Value::bytes(b)),
807 Err(msg) => Err(EvalError::TypeMismatch {
808 detail: alloc::format!("invalid input syntax for type bytea: {msg}"),
809 }),
810 },
811 // v7.39 (round 544) — an integer's two's-complement bytes,
812 // big-endian, at the source type's width. Measured on PG18:
813 // 5::int2 -> \x0005, 5::int4 -> \x00000005,
814 // 5::int8 -> \x0000000000000005, (-1)::int4 -> \xffffffff.
815 Value::SmallInt(n) => Ok(Value::bytes(n.to_be_bytes().to_vec())),
816 Value::Int(n) => Ok(Value::bytes(n.to_be_bytes().to_vec())),
817 Value::BigInt(n) => Ok(Value::bytes(n.to_be_bytes().to_vec())),
818 other => Err(EvalError::TypeMismatch {
819 detail: alloc::format!(
820 "::bytea only accepts TEXT / bytea / integer inputs, got {}",
821 crate::conversions::pg_type_name_for_error_opt(other.data_type())
822 ),
823 }),
824 },
825 CastTarget::Named(name) => {
826 // v7.39 (round 777, F31-E1) — a typmod'd ARRAY cast:
827 // `::numeric(3,1)[]` arrives as Named("numeric(3,1)_array")
828 // and fell through to the user-type lookup ('type
829 // "numeric(3,1)_array" does not exist'). PG applies the
830 // modifier per element ({1.5, 2.3}, measured). Cast to the
831 // bare base array first, then run every element through the
832 // scalar typmod cast.
833 if let Some(base_paren) = name.strip_suffix("_array")
834 && base_paren.ends_with(')')
835 && let Some(popen) = base_paren.find('(')
836 {
837 let base = &base_paren[..popen];
838 let arr = cast_value_ref_in(
839 v,
840 &CastTarget::Named(alloc::format!("{base}_array")),
841 mysql,
842 )?;
843 let scalar = CastTarget::Named(alloc::string::String::from(base_paren));
844 return match arr {
845 Value::NumericArray(items) => {
846 let mut out = alloc::vec::Vec::with_capacity(items.len());
847 for it in items {
848 out.push(match it {
849 None => None,
850 Some((scaled, scale)) => {
851 match cast_value_ref_in(
852 Value::Numeric { scaled, scale, kind: spg_storage::NumericKind::Finite },
853 &scalar,
854 mysql,
855 )? {
856 Value::Numeric { scaled, scale, .. } => {
857 Some((scaled, scale))
858 }
859 Value::NumericBig(b) => {
860 return Err(EvalError::TypeMismatch {
861 detail: alloc::format!(
862 "numeric value too large for {base_paren}[]: {b:?}"
863 ),
864 });
865 }
866 Value::Null => None,
867 other => {
868 return Err(EvalError::TypeMismatch {
869 detail: alloc::format!(
870 "unexpected element cast result {other:?}"
871 ),
872 });
873 }
874 }
875 }
876 });
877 }
878 Ok(Value::NumericArray(out))
879 }
880 other => Ok(other),
881 };
882 }
883 // v7.39 (round 613) — a plain scalar spelling goes straight to
884 // the tail. See `PLAIN_NAMED_TARGETS` for why that is the same
885 // thing as walking the arm, and the pin for the check that says
886 // so mechanically.
887 if let Some(dt) = plain_named_target(name) {
888 return finish_named_cast(v, dt, name, None, mysql);
889 }
890 // v7.38 (read01) — a temporal type with a fractional-seconds
891 // precision (`time(3)`, `timestamp(0)`, `timestamptz(2)`) rounds the
892 // sub-second field to that many digits, like PG. Resolve against the
893 // base type (`type_name_to_data_type` does not know the `(N)` form)
894 // and round the coerced result below.
895 // v7.38 (read01, T20) — an integer casts to `bit(n)` as the low n
896 // bits of its two's-complement representation (PG; int→varbit is
897 // rejected there, so only fixed-length `bit` is handled here).
898 if matches!(v, Value::Int(_) | Value::BigInt(_) | Value::SmallInt(_)) {
899 if let Some(width) = bit_cast_width(name) {
900 return int_to_bit_string(v, width.0);
901 }
902 }
903 // v7.39 (read01 varbit.c) — internal exact-length form for
904 // B'...' literals (an explicit ::bit means bit(1) below).
905 if name == "__bit_literal" {
906 return match &v {
907 Value::Null => Ok(Value::Null),
908 Value::Text(s) => match crate::conversions::parse_bit_string_text(s) {
909 Some((nb, by)) => Ok(Value::bit_string(nb, by)),
910 None => Err(EvalError::TypeMismatch {
911 detail: alloc::format!("invalid input syntax for type bit: \"{s}\""),
912 }),
913 },
914 Value::BitString { .. } => Ok(v),
915 other => Err(EvalError::TypeMismatch {
916 detail: alloc::format!(
917 "cannot cast {} to bit",
918 crate::conversions::pg_type_name_for_error_opt(other.data_type())
919 ),
920 }),
921 };
922 }
923 // v7.39 (read01 varbit.c) — `bit(n)` over a bit string (or a
924 // '0101' text form) zero-extends on the RIGHT or truncates to
925 // n (PG's bit() cast, unlike the input-time exact-length rule).
926 let bit_src: Option<Value<'static>> = match &v {
927 Value::BitString { .. } => Some(v.clone()),
928 Value::Text(s) if bit_cast_width(name).is_some() => {
929 match crate::conversions::parse_bit_string_text(s) {
930 Some((nb, by)) => Some(Value::bit_string(nb, by)),
931 None => {
932 let bad = s.chars().find(|c| *c != '0' && *c != '1');
933 return Err(EvalError::TypeMismatch {
934 detail: match bad {
935 Some(c) => {
936 alloc::format!("\"{c}\" is not a valid binary digit")
937 }
938 None => {
939 alloc::format!("invalid input syntax for type bit: \"{s}\"")
940 }
941 },
942 });
943 }
944 }
945 }
946 _ => None,
947 };
948 if let Some(Value::BitString { nbits, bytes }) = &bit_src {
949 if let Some((width, pads)) = bit_cast_width(name) {
950 // varbit truncates but never pads.
951 if !pads && *nbits <= width {
952 return Ok(Value::BitString {
953 nbits: *nbits,
954 bytes: alloc::borrow::Cow::Owned(bytes.to_vec()),
955 });
956 }
957 let mut bits: alloc::vec::Vec<bool> = (0..*nbits as usize)
958 .map(|i| bytes[i / 8] & (0x80 >> (i % 8)) != 0)
959 .collect();
960 bits.resize(width as usize, false);
961 let mut out = alloc::vec![0u8; width.div_ceil(8) as usize];
962 for (i, b) in bits.iter().enumerate() {
963 if *b {
964 out[i / 8] |= 0x80 >> (i % 8);
965 }
966 }
967 return Ok(Value::BitString {
968 nbits: width,
969 bytes: alloc::borrow::Cow::Owned(out),
970 });
971 }
972 }
973 // v7.39 (read01 oid.c) — OID is unsigned 32-bit: a negative
974 // integer wraps (PG's (Oid) cast semantics: -1 -> 4294967295),
975 // beyond u32 errors "OID out of range", bad text is 22P02.
976 // v7.39 (read01 oid.c) — OID is unsigned 32-bit: a negative
977 // integer wraps (PG's (Oid) cast semantics: -1 -> 4294967295),
978 // beyond u32 errors "OID out of range", bad text is 22P02.
979 //
980 // Round 667 moved the rules to `conversions::coerce_to_oid` so
981 // the column-assignment path shares them instead of growing a
982 // second copy.
983 if name.eq_ignore_ascii_case("oid")
984 && let Some(out) = crate::conversions::coerce_to_oid(&v)?
985 {
986 return Ok(out);
987 }
988 // v7.39 (read01 mac8.c) — macaddr8 -> macaddr requires the
989 // EUI-64 ff:fe infix; anything else is PG's dedicated error.
990 if name.eq_ignore_ascii_case("macaddr") {
991 if let Value::Macaddr8(b) = &v {
992 if b[3] == 0xff && b[4] == 0xfe {
993 return Ok(Value::Macaddr([b[0], b[1], b[2], b[5], b[6], b[7]]));
994 }
995 return Err(EvalError::TypeMismatch {
996 detail: "macaddr8 data out of range to convert to macaddr".into(),
997 });
998 }
999 }
1000 // v7.39 (read01 regproc.c) — the remaining reg* input types.
1001 // SPG carries them as their canonical text rendering; name
1002 // resolution runs against the static pg_proc table / the FTS
1003 // configuration list.
1004 // v7.39 (round 607) — matched against the static list rather than
1005 // through an owned lowercase copy. The copy was built for every
1006 // row and thrown away on every row that is not one of these.
1007 if let Some(lower_name) = REG_MISC_TYPES
1008 .iter()
1009 .copied()
1010 .find(|k| name.eq_ignore_ascii_case(k))
1011 {
1012 let s = match &v {
1013 Value::Null => return Ok(Value::Null),
1014 Value::Text(s) => s.as_ref().trim().to_string(),
1015 // v7.39 (round 634) — an OID reaches these types too.
1016 // PG registers int2/int4/int8/oid -> regproc as IMPLICIT
1017 // casts and renders an oid with no matching entry as the
1018 // number itself: `1::INT::REGPROC` is `1`, and
1019 // `1247::OID::REGPROC` is `1247`. SPG refused the whole
1020 // integer family with "accepts TEXT".
1021 Value::SmallInt(n) => return Ok(Value::text(n.to_string())),
1022 Value::Int(n) => return Ok(Value::text(n.to_string())),
1023 Value::BigInt(n) => return Ok(Value::text(n.to_string())),
1024 other => {
1025 return Err(EvalError::TypeMismatch {
1026 detail: alloc::format!(
1027 "::{lower_name} accepts TEXT, got {}",
1028 crate::conversions::pg_type_name_for_error_opt(other.data_type())
1029 ),
1030 });
1031 }
1032 };
1033 return cast_reg_misc(lower_name, &s);
1034 }
1035 // v7.39 (round 514) — the remaining catalog-shaped types. Each
1036 // validates its own text form and keeps it, which is what PG's
1037 // input functions do; the wordings below are PG18 readings.
1038 if let Some(out) = cast_catalog_scalar(name, &v)? {
1039 return Ok(out);
1040 }
1041 // v7.39 (round 511) — `'(0,1)'::tid`, so a caller can name a row
1042 // it read a ctid from earlier. PG's text form is the only input
1043 // shape it has.
1044 if name.eq_ignore_ascii_case("tid") {
1045 return match &v {
1046 Value::Tid(..) => Ok(v),
1047 Value::Text(t) => parse_tid_text(t).ok_or_else(|| EvalError::TypeMismatch {
1048 detail: alloc::format!("invalid input syntax for type tid: \"{t}\""),
1049 }),
1050 other => Err(EvalError::TypeMismatch {
1051 detail: alloc::format!(
1052 "cannot cast type {} to tid",
1053 crate::eval::strings::pg_typeof_name(other)
1054 ),
1055 }),
1056 };
1057 }
1058 // v7.39 (read01 pseudotypes.c) — casting a value INTO a
1059 // pseudotype hits PG's dummy input functions (0A000).
1060 if let Some(lower) = OPAQUE_TYPES
1061 .iter()
1062 .copied()
1063 .find(|k| name.eq_ignore_ascii_case(k))
1064 {
1065 // v7.39 (round 509) — a pseudotype is a REAL type name,
1066 // so `NULL::anyarray` is NULL on PG, not an error. Only a
1067 // VALUE hits the dummy input function. Before this the
1068 // NULL case fell through to the type table below, which
1069 // does not carry the pseudotypes, and once NULL stopped
1070 // short-circuiting the whole cast it started reporting
1071 // them as unknown types.
1072 return if matches!(v, Value::Null) {
1073 Ok(Value::Null)
1074 } else {
1075 Err(EvalError::TypeMismatch {
1076 detail: alloc::format!("cannot accept a value of type {lower}"),
1077 })
1078 };
1079 }
1080 // v7.39 (read01 pseudotypes.c) — `::cstring` is PG's I/O-form
1081 // pseudotype: text in, text out (cstring_in/out are identity).
1082 // SPG carries it as text; pg_typeof(cstring) reading "text" is
1083 // a recorded delta alongside the literal projection OIDs.
1084 // v7.39 (read01 xid8funcs.c) — `::xid` (32-bit, wrapping) and
1085 // `::xid8` (64-bit, full) parse an integer text and render it
1086 // back verbatim. SPG carries them as BigInt.
1087 if name.eq_ignore_ascii_case("xid") || name.eq_ignore_ascii_case("xid8") {
1088 return Ok(match v {
1089 Value::Null => Value::Null,
1090 Value::SmallInt(n) => Value::BigInt(i64::from(n)),
1091 Value::Int(n) => Value::BigInt(i64::from(n)),
1092 Value::BigInt(n) => Value::BigInt(n),
1093 Value::Text(s) => {
1094 let t = s.trim();
1095 match t.parse::<u64>() {
1096 Ok(n) => Value::BigInt(n as i64),
1097 Err(_) => {
1098 return Err(EvalError::TypeMismatch {
1099 detail: alloc::format!(
1100 "invalid input syntax for type {}: \"{s}\"",
1101 name.to_ascii_lowercase()
1102 ),
1103 });
1104 }
1105 }
1106 }
1107 other => {
1108 return Err(EvalError::TypeMismatch {
1109 detail: alloc::format!(
1110 "cannot cast {} to {name}",
1111 crate::conversions::pg_type_name_for_error_opt(other.data_type())
1112 ),
1113 });
1114 }
1115 });
1116 }
1117 // v7.39 (read01 varchar.c) — `::name` is text truncated to
1118 // NAMEDATALEN-1 (63) bytes.
1119 if name.eq_ignore_ascii_case("name") {
1120 return Ok(match v {
1121 Value::Null => Value::Null,
1122 other => {
1123 let t = match other {
1124 Value::Text(s) => s.into_owned(),
1125 o => value_to_text(&o),
1126 };
1127 let mut cut = t;
1128 if cut.len() > 63 {
1129 let mut idx = 63;
1130 while !cut.is_char_boundary(idx) {
1131 idx -= 1;
1132 }
1133 cut.truncate(idx);
1134 }
1135 Value::text(cut)
1136 }
1137 });
1138 }
1139 if name.eq_ignore_ascii_case("cstring") {
1140 return Ok(match v {
1141 Value::Null => Value::Null,
1142 Value::Text(s) => Value::Text(s),
1143 other => Value::text(value_to_text(&other)),
1144 });
1145 }
1146 // v7.39 (read01 jsonpath.c) — `::jsonpath` parses and prints
1147 // the canonical form (PG's jsonpath type; SPG carries it as
1148 // text — the wire OID is a recorded residual with the other
1149 // literal projection OIDs).
1150 if name.eq_ignore_ascii_case("jsonpath") {
1151 return match v {
1152 Value::Null => Ok(Value::Null),
1153 Value::Text(s) => Ok(Value::text(crate::json::jsonpath_canonical(s.as_ref())?)),
1154 other => Err(EvalError::TypeMismatch {
1155 detail: alloc::format!(
1156 "cannot cast {} to jsonpath",
1157 crate::conversions::pg_type_name_for_error_opt(other.data_type())
1158 ),
1159 }),
1160 };
1161 }
1162 // v7.39 (round 355, M13) — MySQL's `BINARY` / `BINARY(n)`.
1163 // It is a COLLATION coercion, not a type change: MariaDB
1164 // renders `BINARY 'abc'` as `abc` (and `HEX()` of it as
1165 // 616263), so the value passes through unchanged; `(n)`
1166 // truncates to n bytes (`CAST('abc' AS BINARY(2))` is `ab`,
1167 // measured). What it really buys is byte-wise comparison,
1168 // which `compare_is_case_insensitive` now refuses to fold.
1169 if mysql
1170 && (name.eq_ignore_ascii_case("binary")
1171 || name.to_ascii_lowercase().starts_with("binary("))
1172 {
1173 return cast_mysql_binary(v, name);
1174 }
1175 // v7.39 (round 352, M8) — MySQL's SIGNED / UNSIGNED targets.
1176 // Measured on MariaDB 11: a string gives its LEADING number
1177 // (`'12abc'` → 12, `'abc'` → 0); a fractional value ROUNDS
1178 // half-away-from-zero (1.5 → 2, 2.5 → 3, -2.5 → -3) rather
1179 // than truncating; and UNSIGNED wraps a negative through u64
1180 // (`-1` → 18446744073709551615).
1181 // PG has no such type — `type "signed" does not exist` — so the
1182 // reading is gated on the dialect, not just on the spelling.
1183 if mysql
1184 && (name.eq_ignore_ascii_case("signed") || name.eq_ignore_ascii_case("unsigned"))
1185 {
1186 return cast_mysql_integer(v, name.eq_ignore_ascii_case("unsigned"));
1187 }
1188 // v7.39 (round 423) — a bare MySQL temporal type carries
1189 // fractional precision 0, so `CAST(x AS DATETIME)` drops the
1190 // fraction (measured on MariaDB 11). PG's `::timestamp` keeps
1191 // every microsecond, so the default is dialect-gated.
1192 let temporal_prec = temporal_typmod(name)
1193 .or_else(|| (mysql && is_bare_temporal_type(name)).then_some(0));
1194 let resolve_name: alloc::borrow::Cow<'_, str> = if temporal_prec.is_some() {
1195 alloc::borrow::Cow::Owned(name.split('(').next().unwrap_or(name).trim().to_string())
1196 } else {
1197 alloc::borrow::Cow::Borrowed(name.as_str())
1198 };
1199 // v7.37.5 ship triage — generic typed-cast dispatch.
1200 // Resolve the ident to a `DataType` and route the value
1201 // through the existing `coerce_value` text-decoder for
1202 // every v7.37.5 γ/δ/ε/ζ-A type that already speaks
1203 // Text→typed via codec.
1204 let dt =
1205 crate::conversions::type_name_to_data_type(&resolve_name).ok_or_else(|| {
1206 // v7.39 (round 272) — a numeric typmod outside PG's
1207 // bounds gets PG's own wording rather than being
1208 // reported as an unknown type.
1209 // v7.39 (round 620) — and an unknown one is PG's
1210 // wording, which also earns it PG's SQLSTATE (42704
1211 // UNDEFINED_OBJECT; `unsupported cast target` fell
1212 // through to the generic 42000).
1213 EvalError::TypeMismatch {
1214 detail: crate::conversions::numeric_typmod_error(&resolve_name)
1215 .unwrap_or_else(|| unknown_type_error_text(name)),
1216 }
1217 })?;
1218 finish_named_cast(v, dt, &resolve_name, temporal_prec, mysql)
1219 }
1220 }
1221}
1222
1223/// v7.39 (round 613) — the tail of the `Named` arm: stringify for the text
1224/// targets, coerce, and round a temporal precision. Split out so the fast
1225/// path below reaches exactly this code rather than a copy of it.
1226/// v7.39 (round 722) — the compiled `Step::CastPlain` entry: same tail,
1227/// name pre-resolved at compile time.
1228pub(crate) fn finish_named_cast_plain(
1229 v: Value<'static>,
1230 dt: spg_storage::DataType,
1231 resolve_name: &str,
1232 mysql: bool,
1233) -> Result<Value<'static>, EvalError> {
1234 finish_named_cast(v, dt, resolve_name, None, mysql)
1235}
1236
1237fn finish_named_cast(
1238 v: Value<'static>,
1239 dt: spg_storage::DataType,
1240 resolve_name: &str,
1241 temporal_prec: Option<u8>,
1242 mysql: bool,
1243) -> Result<Value<'static>, EvalError> {
1244 // PG semantics: any value casts to varchar(n) / char(n) through its text
1245 // representation (`99::char(2)` → '99'), and an EXPLICIT cast truncates
1246 // to n characters — only column assignment errors on overflow. Stringify
1247 // a non-text source first, then truncate up front so the coerce path's
1248 // length contract never fires here.
1249 let v = match (&dt, v) {
1250 // v7.38 (read01) — an explicit cast to TEXT stringifies any
1251 // value (`text(42)` → '42'), matching `42::text`. (coerce_value
1252 // deliberately rejects a bare INT→TEXT so INSERT stays strict.)
1253 (spg_storage::DataType::Text, v) => match v {
1254 Value::Text(s) => Value::Text(s),
1255 // v7.39 (read01 inet family) — inet/cidr ::text carries
1256 // the mask even at full length (cast-path form).
1257 Value::Inet { family, bits, addr } | Value::Cidr { family, bits, addr } => {
1258 Value::text(crate::conversions::format_inet_full(family, bits, &addr))
1259 }
1260 other => Value::text(value_to_text(&other)),
1261 },
1262 (spg_storage::DataType::Varchar(n) | spg_storage::DataType::Char(n), v) => {
1263 // v7.37 D.36 — previously only `Value::Text` was handled, so
1264 // `99::char(2)` reached coerce_value as an INT and hit a
1265 // CHAR/INT storage type-mismatch.
1266 // v7.39 (bpchar epic) — a bpchar source enters through its
1267 // text cast (trailing blanks stripped): `::varchar` keeps
1268 // the stripped form, `::char(m)` re-pads in coerce_value.
1269 let s = match v {
1270 Value::Text(s) => s.into_owned(),
1271 Value::BpChar(s) => s.trim_end_matches(' ').to_string(),
1272 other => value_to_text(&other),
1273 };
1274 let s = if *n > 0 && s.chars().count() > *n as usize {
1275 s.chars()
1276 .take(*n as usize)
1277 .collect::<alloc::string::String>()
1278 } else {
1279 s
1280 };
1281 Value::text(s)
1282 }
1283 (_, v) => v,
1284 };
1285 let coerced =
1286 crate::conversions::coerce_value(v, dt, resolve_name, 0).map_err(|e| match e {
1287 // v7.39 (read01 round 113) — pass an already-classed engine
1288 // error through unchanged. Re-stringifying via Display would
1289 // double the "eval: type mismatch: " class prefix (the wire
1290 // strips only the outermost one), leaking it into the message
1291 // — visible now that jsonb → numeric casts error with PG's
1292 // exact "cannot cast jsonb string to type numeric" wording.
1293 crate::EngineError::Eval(ev) => ev,
1294 // v7.39 (round 622, S05a) — `coerce_value` is the INSERT-time
1295 // COLUMN coercion, and a cast borrows it. Its rejection is
1296 // phrased for a column, so `SELECT 1::INET` answered
1297 //
1298 // type mismatch in column "inet" (position 0): expected INET,
1299 // got INT
1300 //
1301 // naming a column that does not exist, at a position that means
1302 // nothing, in the storage layer's own vocabulary. PG says
1303 // `cannot cast type integer to inet`. The column phrasing stays
1304 // where it belongs — an INSERT still says which column — and a
1305 // failed cast now says what it failed to cast, like every other
1306 // arm in this file already did.
1307 //
1308 // The two type names come off the error itself, which already
1309 // carries them as `DataType`. Naming them BEFORE the call — the
1310 // obvious way to write this, since the value and the target both
1311 // move into it — costs two `String`s on every SUCCESSFUL cast,
1312 // and the panel caught exactly that: `id::NUMERIC` 23.75 ->
1313 // 55.48 ms, `id::REAL` 21.55 -> 50.48. This is the same eager
1314 // error construction round 614 removed from 28 call sites,
1315 // rebuilt by hand a round later.
1316 crate::EngineError::Storage(spg_storage::StorageError::TypeMismatch {
1317 expected,
1318 actual,
1319 ..
1320 }) => EvalError::TypeMismatch {
1321 detail: alloc::format!(
1322 "cannot cast {} to {}",
1323 crate::conversions::pg_type_name_for_error(actual),
1324 crate::conversions::pg_type_name_for_error(expected)
1325 ),
1326 },
1327 other => EvalError::TypeMismatch {
1328 detail: alloc::format!("{other}"),
1329 },
1330 })?;
1331 Ok(match temporal_prec {
1332 Some(prec) => round_temporal_to_precision(coerced, prec, mysql),
1333 None => coerced,
1334 })
1335}
1336
1337/// v7.39 (round 613) — the plain scalar spellings, with the type each one
1338/// resolves to.
1339///
1340/// Round 612 measured the `Named` arm re-deriving everything for every row:
1341/// `s::VARCHAR` cost 30.6 ms over 200k rows where `s::TEXT` — the identical
1342/// conversion, under a spelling the parser settles into a `CastTarget`
1343/// variant — cost 11.2, and probes split the difference across the whole arm
1344/// rather than any one place in it. These names reach the tail directly.
1345///
1346/// Both halves of that shortcut are checked mechanically by the pin, not by
1347/// eye: every entry's type is asserted to equal `type_name_to_data_type`'s
1348/// answer, and every entry is asserted absent from each arm above the
1349/// resolve (the reg-misc / catalog-scalar / opaque lists, `tid`, `xid`,
1350/// `xid8`, `jsonpath`, the MySQL `binary` / `signed` / `unsigned` names, and
1351/// the bit and temporal spellings). A name that grows a special case has to
1352/// leave this table, and the pin says so.
1353const PLAIN_NAMED_TARGETS: &[(&str, spg_storage::DataType)] = &[
1354 ("text", spg_storage::DataType::Text),
1355 ("varchar", spg_storage::DataType::Varchar(0)),
1356 ("character varying", spg_storage::DataType::Varchar(0)),
1357 (
1358 "numeric",
1359 spg_storage::DataType::Numeric {
1360 precision: 0,
1361 scale: 0,
1362 },
1363 ),
1364 (
1365 "decimal",
1366 spg_storage::DataType::Numeric {
1367 precision: 0,
1368 scale: 0,
1369 },
1370 ),
1371 ("real", spg_storage::DataType::Real),
1372 ("float4", spg_storage::DataType::Real),
1373 ("float8", spg_storage::DataType::Float),
1374 ("double precision", spg_storage::DataType::Float),
1375 ("int2", spg_storage::DataType::SmallInt),
1376 ("smallint", spg_storage::DataType::SmallInt),
1377 ("int4", spg_storage::DataType::Int),
1378 ("integer", spg_storage::DataType::Int),
1379 ("int8", spg_storage::DataType::BigInt),
1380 ("bool", spg_storage::DataType::Bool),
1381 ("boolean", spg_storage::DataType::Bool),
1382 ("date", spg_storage::DataType::Date),
1383 ("bytea", spg_storage::DataType::Bytes),
1384 ("uuid", spg_storage::DataType::Uuid),
1385];
1386
1387/// v7.39 (round 613) — the heads that may carry a typmod and are still
1388/// plain: `varchar(20)`, `char(4)`, `numeric(10,2)`. The type comes from
1389/// `type_name_to_data_type` over the WHOLE name, so the typmod is parsed
1390/// exactly where it always was; only the walk down the arm is skipped. The
1391/// pin checks each head against every arm above the resolve, and that none
1392/// of them is a bit or temporal spelling.
1393pub(crate) const PLAIN_NAMED_HEADS: &[&str] = &[
1394 "varchar",
1395 "character varying",
1396 "char",
1397 "character",
1398 "bpchar",
1399 "numeric",
1400 "decimal",
1401];
1402
1403/// The type a plain scalar spelling resolves to, or `None` when the name
1404/// needs the whole arm.
1405pub(crate) fn plain_named_target(name: &str) -> Option<spg_storage::DataType> {
1406 if let Some(dt) = PLAIN_NAMED_TARGETS
1407 .iter()
1408 .find(|(k, _)| name.eq_ignore_ascii_case(k))
1409 .map(|(_, dt)| *dt)
1410 {
1411 return Some(dt);
1412 }
1413 let head = name.split('(').next()?.trim();
1414 if name.len() == head.len()
1415 || !PLAIN_NAMED_HEADS
1416 .iter()
1417 .any(|k| head.eq_ignore_ascii_case(k))
1418 {
1419 return None;
1420 }
1421 crate::conversions::type_name_to_data_type(name)
1422}
1423
1424/// The scalar type names the `Named` arm resolves without a catalog.
1425fn is_known_scalar_name(lower: &str) -> bool {
1426 REG_MISC_TYPES.contains(&lower)
1427 || CATALOG_SCALAR_TYPES.contains(&lower)
1428 || OPAQUE_TYPES.contains(&lower)
1429 || matches!(
1430 lower,
1431 "tid"
1432 | "record"
1433 | "cstring"
1434 | "regnamespace"
1435 | "regrole"
1436 // Round 896 — the target validator runs before the arm, so
1437 // the quoted spelling has to be a known name here too or it
1438 // is rejected before the fold above ever sees it.
1439 | "regclass"
1440 | "regtype"
1441 )
1442}
1443
1444/// v7.39 (round 509) — does this name a type at all?
1445///
1446/// PG validates the cast TARGET whatever the operand is: `NULL::nosuchtype`
1447/// is an error there, not NULL. `cast_value_in` short-circuits a NULL before
1448/// it ever looks at the target, so the check has to happen in the caller —
1449/// and `eval_cast_arm` is the caller that has a catalog, which is what
1450/// enums, domains, composites and table row types need.
1451///
1452/// This lists what the `Named` arm below resolves WITHOUT a catalog. Keeping
1453/// the two in step is a real hazard: a first cut of this check missed three
1454/// live spellings — `::binary` (the MySQL prefix's desugar), a table's row
1455/// type, and the pseudotypes — and the e2e suite caught every one. It is the
1456/// check on this function.
1457/// v7.39 (round 620) — PG's wording for a cast target that names no type.
1458///
1459/// SPG said ``unsupported cast target `::nosuchtype` ``, which reads as "SPG
1460/// has not got round to that one" when what happened is that no such type
1461/// exists anywhere. PG says `type "nosuchtype" does not exist`, and because
1462/// the wire classifies by message text, saying it also moves the code off the
1463/// generic 42000 onto 42704 UNDEFINED_OBJECT.
1464pub(crate) fn unknown_type_error_text(name: &str) -> alloc::string::String {
1465 alloc::format!("type \"{name}\" does not exist")
1466}
1467
1468pub(crate) fn builtin_target_resolves(name: &str, mysql: bool) -> bool {
1469 if name == "__bit_literal" || bit_cast_width(name).is_some() {
1470 return true;
1471 }
1472 crate::conversions::with_lower_name(name, |lower| {
1473 builtin_target_resolves_lower(name, lower, mysql)
1474 })
1475}
1476
1477fn builtin_target_resolves_lower(name: &str, lower: &str, mysql: bool) -> bool {
1478 // The three families, read from the same declarations the value path
1479 // dispatches on — see their doc comment for why that matters.
1480 if is_known_scalar_name(lower) {
1481 return true;
1482 }
1483 // v7.39 (round 515) — `<element>[]`, which this parser names
1484 // `<element>_array`. PG has an array type for every scalar, so the rule
1485 // is the stem's: `NULL::cstring[]`, `NULL::aclitem[]` and
1486 // `NULL::"char"[]` all resolve there. A general rule rather than three
1487 // entries, because the next scalar added would otherwise need a fourth.
1488 if let Some(stem) = lower.strip_suffix("_array")
1489 && (is_known_scalar_name(stem)
1490 || crate::conversions::type_name_to_data_type(stem).is_some())
1491 {
1492 return true;
1493 }
1494 if mysql && matches!(lower, "binary" | "signed" | "unsigned") {
1495 return true;
1496 }
1497 let base = if temporal_typmod(name).is_some() || (mysql && is_bare_temporal_type(name)) {
1498 name.split('(').next().unwrap_or(name).trim()
1499 } else {
1500 name
1501 };
1502 crate::conversions::type_name_to_data_type(base).is_some()
1503 || crate::conversions::numeric_typmod_error(base).is_some()
1504}
1505
1506/// v7.39 (round 511) — PG's `(block,offset)` text form for a tid.
1507fn parse_tid_text(t: &str) -> Option<Value<'static>> {
1508 let inner = t.trim().strip_prefix('(')?.strip_suffix(')')?;
1509 let (b, o) = inner.split_once(',')?;
1510 Some(Value::Tid(
1511 b.trim().parse::<u32>().ok()?,
1512 o.trim().parse::<u32>().ok()?,
1513 ))
1514}
1515
1516/// v7.39 (round 514) — the catalog-shaped scalar types: the ids, the oid
1517/// vectors, an ACL item, a cursor name and a transaction snapshot.
1518///
1519/// `Some` when `name` is one of them, so the caller can fall through to
1520/// everything else. Every error wording is a PG18 reading — they differ per
1521/// type and per ELEMENT (`::oidvector` complains about `oid`,
1522/// `::int2vector` about `smallint`), which is why they are spelled out
1523/// rather than shared.
1524fn cast_catalog_scalar(name: &str, v: &Value<'_>) -> Result<Option<Value<'static>>, EvalError> {
1525 crate::conversions::with_lower_name(name, |lower| cast_catalog_scalar_lower(lower, v))
1526}
1527
1528fn cast_catalog_scalar_lower(
1529 lower: &str,
1530 v: &Value<'_>,
1531) -> Result<Option<Value<'static>>, EvalError> {
1532 // v7.39 (round 515) — `<element>[]` runs the element's own check over
1533 // each member and keeps the literal, which is what PG does: measured,
1534 // `'{a,b}'::aclitem[]` is "unrecognized key word: \"a\"".
1535 if let Some(stem) = lower.strip_suffix("_array")
1536 && (CATALOG_SCALAR_TYPES.contains(&stem) || OPAQUE_TYPES.contains(&stem))
1537 {
1538 let Value::Text(t) = v else {
1539 return Ok(None);
1540 };
1541 let body = t.trim();
1542 let inner = body
1543 .strip_prefix('{')
1544 .and_then(|b| b.strip_suffix('}'))
1545 .unwrap_or(body);
1546 for part in inner.split(',').filter(|p| !p.trim().is_empty()) {
1547 cast_catalog_scalar(stem, &Value::text(part.trim().to_string()))?;
1548 }
1549 return Ok(Some(Value::text(body.to_string())));
1550 }
1551 if !CATALOG_SCALAR_TYPES.contains(&lower) {
1552 return Ok(None);
1553 }
1554 let text = match v {
1555 Value::Text(t) => t.to_string(),
1556 Value::Cid(c) if lower == "cid" => return Ok(Some(Value::Cid(*c))),
1557 Value::Xid(x) if lower == "xid" => return Ok(Some(Value::Xid(*x))),
1558 // v7.39 (round 641) — PG has no cast between an integer and a
1559 // transaction id in either direction: `5::xid` is "cannot cast
1560 // type integer to xid" and `'5'::xid::int` is the mirror of it,
1561 // measured. The unknown-literal spelling `'5'::xid` is a
1562 // different thing — that is the type's input function, and it is
1563 // the Text arm above. Only `xid` is carved out here; `cid`,
1564 // `oid` and the vector types keep taking an integer.
1565 Value::SmallInt(_) | Value::Int(_) | Value::BigInt(_) if lower == "xid" => {
1566 return Err(EvalError::TypeMismatch {
1567 detail: alloc::format!(
1568 "cannot cast type {} to xid",
1569 crate::eval::strings::pg_typeof_name(v)
1570 ),
1571 });
1572 }
1573 Value::SmallInt(n) => alloc::format!("{n}"),
1574 Value::Int(n) => alloc::format!("{n}"),
1575 Value::BigInt(n) => alloc::format!("{n}"),
1576 other => {
1577 return Err(EvalError::TypeMismatch {
1578 detail: alloc::format!(
1579 "cannot cast type {} to {lower}",
1580 crate::eval::strings::pg_typeof_name(other)
1581 ),
1582 });
1583 }
1584 };
1585 let t = text.trim();
1586 let bad = |ty: &str, what: &str| EvalError::TypeMismatch {
1587 detail: alloc::format!("invalid input syntax for type {ty}: \"{what}\""),
1588 };
1589 let out = match lower {
1590 "cid" => Value::Cid(t.parse::<u32>().map_err(|_| bad("cid", t))?),
1591 "xid" => Value::Xid(t.parse::<u32>().map_err(|_| bad("xid", t))?),
1592 // Space-separated element lists, validated element by element and
1593 // kept in their own spelling.
1594 "oidvector" | "int2vector" => {
1595 let elem_ty = if lower == "oidvector" {
1596 "oid"
1597 } else {
1598 "smallint"
1599 };
1600 for part in t.split_whitespace() {
1601 let ok = if elem_ty == "oid" {
1602 part.parse::<u32>().is_ok()
1603 } else {
1604 part.parse::<i16>().is_ok()
1605 };
1606 if !ok {
1607 return Err(bad(elem_ty, part));
1608 }
1609 }
1610 Value::text(t.to_string())
1611 }
1612 // `grantee=privileges/grantor`, and PG checks the key word first:
1613 // anything before the `=` that is not a role name, `group` or
1614 // `user` is "unrecognized key word".
1615 "aclitem" => {
1616 let Some((who, rest)) = t.split_once('=') else {
1617 return Err(EvalError::TypeMismatch {
1618 detail: alloc::format!("unrecognized key word: \"{t}\""),
1619 });
1620 };
1621 if !rest.contains('/') {
1622 return Err(EvalError::TypeMismatch {
1623 detail: alloc::format!("a name must follow the \"/\" sign"),
1624 });
1625 }
1626 let _ = who;
1627 Value::text(t.to_string())
1628 }
1629 // A cursor name is just a name.
1630 "refcursor" => Value::text(t.to_string()),
1631 // `xmin:xmax:xip_list` — two numbers and a comma-separated tail.
1632 "pg_snapshot" | "txid_snapshot" => {
1633 let parts: alloc::vec::Vec<&str> = t.splitn(3, ':').collect();
1634 let shaped = parts.len() == 3
1635 && parts[0].parse::<u64>().is_ok()
1636 && parts[1].parse::<u64>().is_ok()
1637 && (parts[2].is_empty() || parts[2].split(',').all(|x| x.parse::<u64>().is_ok()));
1638 if !shaped {
1639 return Err(bad(lower, t));
1640 }
1641 Value::text(t.to_string())
1642 }
1643 // PG normalises a path on input: `$.a` reads back `$."a"`. The
1644 // engine already has the parser its operators use.
1645 "jsonpath" => Value::text(crate::json::jsonpath_canonical(t)?),
1646 _ => unreachable!("guarded above"),
1647 };
1648 Ok(Some(out))
1649}
1650
1651/// v7.38 (read01, T20) — width of a `bit` cast target: bare `bit` is `bit(1)`,
1652/// `bit(N)` is N. `None` for `varbit` / `bit varying` (PG rejects int→varbit) and
1653/// any non-bit name.
1654fn bit_cast_width(name: &str) -> Option<(u32, bool)> {
1655 crate::conversions::with_lower_name(name, bit_cast_width_lower)
1656}
1657
1658fn bit_cast_width_lower(lower: &str) -> Option<(u32, bool)> {
1659 let trimmed = lower.trim();
1660 if trimmed == "bit" {
1661 return Some((1, true));
1662 }
1663 // v7.39 (round 281) — `varbit(n)` / `bit varying(n)` adjust on an
1664 // explicit cast too, but only DOWN: PG truncates a too-long value
1665 // and leaves a shorter one alone, where `bit(n)` also pads.
1666 for (prefix, pads) in [("varbit", false), ("bit varying", false), ("bit", true)] {
1667 if let Some(rest) = trimmed.strip_prefix(prefix) {
1668 let rest = rest.trim_start();
1669 if let Some(inner) = rest.strip_prefix('(').and_then(|r| r.strip_suffix(')'))
1670 && let Ok(n) = inner.trim().parse::<u32>()
1671 {
1672 return Some((n, pads));
1673 }
1674 }
1675 }
1676 None
1677}
1678
1679/// v7.38 (read01, T20) — build a `bit(width)` value from an integer: the low
1680/// `width` bits of the two's-complement, packed MSB-first / left-aligned (the
1681/// on-wire bit layout). Widths past 64 sign-extend.
1682fn int_to_bit_string(v: Value<'static>, width: u32) -> Result<Value<'static>, EvalError> {
1683 let n: i64 = match v {
1684 Value::Int(x) => i64::from(x),
1685 Value::BigInt(x) => x,
1686 Value::SmallInt(x) => i64::from(x),
1687 _ => {
1688 return Err(EvalError::TypeMismatch {
1689 detail: "int_to_bit_string: non-integer source".into(),
1690 });
1691 }
1692 };
1693 let w = width as usize;
1694 let mut bytes = alloc::vec![0u8; w.div_ceil(8)];
1695 for i in 0..w {
1696 let p = w - 1 - i; // bit position counted from the LSB
1697 let bit = if p >= 64 {
1698 u8::from(n < 0) // sign-extend beyond the integer's width
1699 } else {
1700 ((n >> p) & 1) as u8
1701 };
1702 if bit != 0 {
1703 bytes[i / 8] |= 1 << (7 - (i % 8));
1704 }
1705 }
1706 Ok(Value::bit_string(width, bytes))
1707}
1708
1709/// Extract the fractional-seconds precision from a temporal cast name like
1710/// `time(3)` / `timestamp(0)` / `timestamptz(2)`; `None` for any non-temporal
1711/// type or a bare temporal type with no `(N)`.
1712fn temporal_typmod(name: &str) -> Option<u8> {
1713 crate::conversions::with_lower_name(name, |lower| {
1714 let (base, rest) = lower.split_once('(')?;
1715 if !matches!(
1716 base.trim(),
1717 "time" | "timetz" | "timestamp" | "timestamptz" | "datetime"
1718 ) {
1719 return None;
1720 }
1721 let digits = rest.trim_start();
1722 let end = digits
1723 .find(|c: char| !c.is_ascii_digit())
1724 .unwrap_or(digits.len());
1725 digits[..end].parse::<u8>().ok()
1726 })
1727}
1728
1729/// Round a TIME / TIMESTAMP value's microsecond field to `prec` fractional-
1730/// second digits (`prec` 0..=6), half-away-from-zero as PG's AdjustTimestamp.
1731/// v7.39 (round 423) — `truncate` selects MySQL's reduction mode. PG's
1732/// AdjustTimestamp ROUNDS half-away-from-zero (`::timestamp(1)` of `.256` is
1733/// `.3`); MariaDB TRUNCATES toward zero (`.2`, measured). Same function, one
1734/// flag, because everything else about the reduction is identical.
1735fn round_temporal_to_precision(v: Value<'static>, prec: u8, truncate: bool) -> Value<'static> {
1736 if prec >= 6 {
1737 return v;
1738 }
1739 let scale = 10i64.pow(u32::from(6 - prec));
1740 let reduce = |micros: i64| -> i64 {
1741 if truncate {
1742 // Toward zero, so a negative time-of-day loses the same digits.
1743 (micros / scale) * scale
1744 } else {
1745 let half = scale / 2;
1746 if micros >= 0 {
1747 ((micros + half) / scale) * scale
1748 } else {
1749 -(((-micros + half) / scale) * scale)
1750 }
1751 }
1752 };
1753 match v {
1754 Value::Timestamp(m) => Value::Timestamp(reduce(m)),
1755 Value::Time(m) => Value::Time(reduce(m)),
1756 other => other,
1757 }
1758}
1759
1760/// v7.39 (round 423) — is `name` a bare temporal type (no `(N)` modifier)?
1761/// MySQL gives those fractional precision ZERO — `CAST(x AS DATETIME)` drops
1762/// the fraction entirely — where PG's `::timestamp` keeps full microseconds.
1763fn is_bare_temporal_type(name: &str) -> bool {
1764 let t = name.trim();
1765 ["time", "timestamp", "datetime"]
1766 .iter()
1767 .any(|k| t.eq_ignore_ascii_case(k))
1768}
1769
1770fn cast_to_int_array(v: Value) -> Result<Value, EvalError> {
1771 match v {
1772 Value::IntArray(items) => Ok(Value::IntArray(items)),
1773 Value::BigIntArray(items) => {
1774 let mut out: Vec<Option<i32>> = Vec::with_capacity(items.len());
1775 for item in items {
1776 match item {
1777 None => out.push(None),
1778 Some(n) => match i32::try_from(n) {
1779 Ok(x) => out.push(Some(x)),
1780 Err(_) => {
1781 return Err(EvalError::TypeMismatch {
1782 detail: alloc::format!("::INT[] element {n} overflows i32"),
1783 });
1784 }
1785 },
1786 }
1787 }
1788 Ok(Value::IntArray(out))
1789 }
1790 Value::Text(s) => {
1791 if let Some(r) = try_cast_2d_array(&s, |row| {
1792 decode_int_array_external(row).map(Value::IntArray)
1793 }) {
1794 return r;
1795 }
1796 decode_int_array_external(&s).map(Value::IntArray)
1797 }
1798 Value::TextArray(items) => {
1799 let mut out: Vec<Option<i32>> = Vec::with_capacity(items.len());
1800 for item in items {
1801 match item {
1802 None => out.push(None),
1803 Some(s) => match s.parse::<i32>() {
1804 Ok(n) => out.push(Some(n)),
1805 Err(_) => {
1806 return Err(EvalError::TypeMismatch {
1807 detail: alloc::format!("::INT[] cannot parse {s:?}"),
1808 });
1809 }
1810 },
1811 }
1812 }
1813 Ok(Value::IntArray(out))
1814 }
1815 other => Err(EvalError::TypeMismatch {
1816 detail: alloc::format!(
1817 "::INT[] does not accept {}",
1818 crate::conversions::pg_type_name_for_error_opt(other.data_type())
1819 ),
1820 }),
1821 }
1822}
1823
1824fn cast_to_bigint_array(v: Value) -> Result<Value, EvalError> {
1825 match v {
1826 Value::BigIntArray(items) => Ok(Value::BigIntArray(items)),
1827 Value::IntArray(items) => Ok(Value::BigIntArray(
1828 items.into_iter().map(|x| x.map(i64::from)).collect(),
1829 )),
1830 Value::Text(s) => {
1831 if let Some(r) = try_cast_2d_array(&s, |row| {
1832 decode_bigint_array_external(row).map(Value::BigIntArray)
1833 }) {
1834 return r;
1835 }
1836 decode_bigint_array_external(&s).map(Value::BigIntArray)
1837 }
1838 Value::TextArray(items) => {
1839 let mut out: Vec<Option<i64>> = Vec::with_capacity(items.len());
1840 for item in items {
1841 match item {
1842 None => out.push(None),
1843 Some(s) => match s.parse::<i64>() {
1844 Ok(n) => out.push(Some(n)),
1845 Err(_) => {
1846 return Err(EvalError::TypeMismatch {
1847 detail: alloc::format!("::BIGINT[] cannot parse {s:?}"),
1848 });
1849 }
1850 },
1851 }
1852 }
1853 Ok(Value::BigIntArray(out))
1854 }
1855 other => Err(EvalError::TypeMismatch {
1856 detail: alloc::format!(
1857 "::BIGINT[] does not accept {}",
1858 crate::conversions::pg_type_name_for_error_opt(other.data_type())
1859 ),
1860 }),
1861 }
1862}
1863
1864/// Cast a possibly-2-D array literal: parse each top-level row with `elem` (the
1865/// 1-D element decoder) and fold into a 2-D value; `None` when the literal is 1-D.
1866fn try_cast_2d_array(
1867 s: &str,
1868 elem: impl Fn(&str) -> Result<Value<'static>, EvalError>,
1869) -> Option<Result<Value<'static>, EvalError>> {
1870 let rows = crate::eval::values::split_2d_rows(s)?;
1871 let mut row_vals: alloc::vec::Vec<Value<'static>> = alloc::vec::Vec::with_capacity(rows.len());
1872 for r in &rows {
1873 match elem(r) {
1874 Ok(v) => row_vals.push(v),
1875 Err(e) => return Some(Err(e)),
1876 }
1877 }
1878 Some(
1879 crate::eval::values::build_2d_from_rows(&row_vals).ok_or_else(|| EvalError::TypeMismatch {
1880 detail: crate::conversions::malformed_array_literal(s),
1881 }),
1882 )
1883}
1884
1885fn decode_int_array_external(s: &str) -> Result<Vec<Option<i32>>, EvalError> {
1886 let trimmed = s.trim();
1887 // v7.39 (read01 jsonfuncs.c) — the json_to_record/populate desugar
1888 // routes JSON array text ("[1,2]") through this cast; accept the
1889 // bracket form alongside PG's brace form.
1890 let inner = trimmed
1891 .strip_prefix('{')
1892 .and_then(|x| x.strip_suffix('}'))
1893 .or_else(|| trimmed.strip_prefix('[').and_then(|x| x.strip_suffix(']')))
1894 .ok_or_else(|| EvalError::TypeMismatch {
1895 detail: crate::conversions::malformed_array_literal(s),
1896 })?;
1897 if inner.trim().is_empty() {
1898 return Ok(Vec::new());
1899 }
1900 inner
1901 .split(',')
1902 .map(|part| {
1903 let p = part.trim();
1904 if p.eq_ignore_ascii_case("NULL") {
1905 Ok(None)
1906 } else {
1907 p.parse::<i32>()
1908 .map(Some)
1909 .map_err(|_| EvalError::TypeMismatch {
1910 detail: alloc::format!("invalid input syntax for type integer: {p:?}"),
1911 })
1912 }
1913 })
1914 .collect()
1915}
1916
1917fn decode_bigint_array_external(s: &str) -> Result<Vec<Option<i64>>, EvalError> {
1918 let trimmed = s.trim();
1919 let inner = trimmed
1920 .strip_prefix('{')
1921 .and_then(|x| x.strip_suffix('}'))
1922 .or_else(|| trimmed.strip_prefix('[').and_then(|x| x.strip_suffix(']')))
1923 .ok_or_else(|| EvalError::TypeMismatch {
1924 // v7.39 (round 325) — was "BIGmalformed array literal", a
1925 // stray edit that shipped: the message a client saw for
1926 // `'abc'::bigint[]` began with three letters of BIGINT.
1927 detail: crate::conversions::malformed_array_literal(s),
1928 })?;
1929 if inner.trim().is_empty() {
1930 return Ok(Vec::new());
1931 }
1932 inner
1933 .split(',')
1934 .map(|part| {
1935 let p = part.trim();
1936 if p.eq_ignore_ascii_case("NULL") {
1937 Ok(None)
1938 } else {
1939 p.parse::<i64>()
1940 .map(Some)
1941 .map_err(|_| EvalError::TypeMismatch {
1942 detail: alloc::format!("invalid input syntax for type bigint: {p:?}"),
1943 })
1944 }
1945 })
1946 .collect()
1947}
1948
1949/// v7.10.11 — same decoder as `decode_text_array_literal` in
1950/// `lib.rs`, but lives here so the eval-time cast path stays
1951/// inside `spg-engine::eval`. Kept in lock-step with the engine
1952/// `coerce_value` decoder by tests.
1953fn decode_text_array_external(s: &str) -> Result<Vec<Option<String>>, EvalError> {
1954 let trimmed = s.trim();
1955 let inner = trimmed
1956 .strip_prefix('{')
1957 .and_then(|x| x.strip_suffix('}'))
1958 .or_else(|| trimmed.strip_prefix('[').and_then(|x| x.strip_suffix(']')))
1959 .ok_or_else(|| EvalError::TypeMismatch {
1960 detail: alloc::format!("TEXT[] literal {s:?} must be enclosed in '{{...}}'"),
1961 })?;
1962 let mut out: Vec<Option<String>> = Vec::new();
1963 if inner.trim().is_empty() {
1964 return Ok(out);
1965 }
1966 let bytes = inner.as_bytes();
1967 let mut i = 0;
1968 while i <= bytes.len() {
1969 while i < bytes.len() && (bytes[i] == b' ' || bytes[i] == b'\t') {
1970 i += 1;
1971 }
1972 if i < bytes.len() && bytes[i] == b'"' {
1973 i += 1;
1974 let mut buf = String::new();
1975 while i < bytes.len() && bytes[i] != b'"' {
1976 if bytes[i] == b'\\' && i + 1 < bytes.len() {
1977 buf.push(bytes[i + 1] as char);
1978 i += 2;
1979 } else {
1980 buf.push(bytes[i] as char);
1981 i += 1;
1982 }
1983 }
1984 if i >= bytes.len() {
1985 return Err(EvalError::TypeMismatch {
1986 detail: "unterminated quoted element in TEXT[] literal".into(),
1987 });
1988 }
1989 i += 1;
1990 out.push(Some(buf));
1991 } else {
1992 let start = i;
1993 while i < bytes.len() && bytes[i] != b',' {
1994 i += 1;
1995 }
1996 let raw = inner[start..i].trim();
1997 if raw.eq_ignore_ascii_case("NULL") {
1998 out.push(None);
1999 } else {
2000 out.push(Some(raw.to_string()));
2001 }
2002 }
2003 while i < bytes.len() && (bytes[i] == b' ' || bytes[i] == b'\t') {
2004 i += 1;
2005 }
2006 if i >= bytes.len() {
2007 break;
2008 }
2009 if bytes[i] != b',' {
2010 return Err(EvalError::TypeMismatch {
2011 detail: "expected ',' between TEXT[] elements".into(),
2012 });
2013 }
2014 i += 1;
2015 }
2016 Ok(out)
2017}
2018
2019fn cast_to_interval(v: Value) -> Result<Value, EvalError> {
2020 match v {
2021 Value::Interval {
2022 months,
2023 days,
2024 micros,
2025 } => Ok(Value::Interval {
2026 months,
2027 days,
2028 micros,
2029 }),
2030 Value::Text(s) => {
2031 let (months, days, micros) =
2032 spg_sql::parser::parse_interval_text(&s).ok_or_else(|| {
2033 EvalError::TypeMismatch {
2034 // v7.39 (round 324, V42) — PG's wording.
2035 detail: alloc::format!("invalid input syntax for type interval: \"{s}\""),
2036 }
2037 })?;
2038 Ok(Value::Interval {
2039 months,
2040 days,
2041 micros,
2042 })
2043 }
2044 other => Err(EvalError::TypeMismatch {
2045 detail: alloc::format!(
2046 "::INTERVAL only accepts TEXT-shape inputs, got {}",
2047 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2048 ),
2049 }),
2050 }
2051}
2052
2053fn cast_to_date(v: Value) -> Result<Value, EvalError> {
2054 match v {
2055 Value::Date(d) => Ok(Value::Date(d)),
2056 // Integer literals carry days since the Unix epoch — used by
2057 // the `CURRENT_DATE` AST rewrite to inject the wall clock.
2058 Value::Int(n) => Ok(Value::Date(n)),
2059 Value::BigInt(n) => {
2060 i32::try_from(n)
2061 .map(Value::Date)
2062 .map_err(|_| EvalError::TypeMismatch {
2063 detail: "bigint days-since-epoch out of DATE range".into(),
2064 })
2065 }
2066 // Timestamp truncates to its day boundary.
2067 Value::Timestamp(t) => {
2068 let days = t.div_euclid(86_400_000_000);
2069 i32::try_from(days)
2070 .map(Value::Date)
2071 .map_err(|_| EvalError::TypeMismatch {
2072 detail: "timestamp out of DATE range".into(),
2073 })
2074 }
2075 Value::Text(s) => {
2076 if let Some(d) = parse_date_literal(&s) {
2077 return Ok(Value::Date(d));
2078 }
2079 // PG accepts a full timestamp string in a DATE cast and
2080 // truncates to the day (verified vs live PG18.4:
2081 // `'2020-01-01 12:00:00'::date` → 2020-01-01; a bad time
2082 // like `'... 25:00:00'` still raises). Reuse the timestamp
2083 // parser — it validates the time-of-day + optional TZ — then
2084 // floor to the date via the same path as the Timestamp arm.
2085 if let Some(t) = parse_timestamp_literal(&s) {
2086 let days = t.div_euclid(86_400_000_000);
2087 return i32::try_from(days)
2088 .map(Value::Date)
2089 .map_err(|_| EvalError::TypeMismatch {
2090 detail: "timestamp out of DATE range".into(),
2091 });
2092 }
2093 // PG error split: numeric-shaped input whose field values
2094 // fail the calendar checks is "out of range" (plus PG's
2095 // DateStyle hint); anything else is an input-syntax error.
2096 if super::format::date_text_is_field_shaped(&s) {
2097 return Err(EvalError::TypeMismatch {
2098 detail: format!(
2099 "date/time field value out of range: {s:?}\n\
2100 HINT: Perhaps you need a different \"DateStyle\" setting."
2101 ),
2102 });
2103 }
2104 Err(EvalError::TypeMismatch {
2105 detail: format!("invalid input syntax for type date: {s:?}"),
2106 })
2107 }
2108 other => Err(EvalError::TypeMismatch {
2109 detail: format!(
2110 "cannot cast {} to DATE",
2111 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2112 ),
2113 }),
2114 }
2115}
2116
2117fn cast_to_timestamp(v: Value) -> Result<Value, EvalError> {
2118 match v {
2119 Value::Timestamp(t) => Ok(Value::Timestamp(t)),
2120 // Int / BigInt carry microseconds since the Unix epoch — used
2121 // by the `NOW()` / `CURRENT_TIMESTAMP` AST rewrite to inject
2122 // the wall clock as a plain integer literal.
2123 Value::Int(n) => Ok(Value::Timestamp(i64::from(n))),
2124 Value::BigInt(n) => Ok(Value::Timestamp(n)),
2125 // DATE → TIMESTAMP picks midnight on the date.
2126 // v7.39 (read01 timestamp.c) — sentinel-aware (the plain multiply
2127 // overflowed on ±infinity dates).
2128 Value::Date(d) => Ok(Value::Timestamp(crate::conversions::date_days_to_micros(d))),
2129 Value::Text(s) => {
2130 // v7.39 (round 289) — the target has no zone, so PG ignores
2131 // any the literal carries: `'…+02'::timestamp` keeps the
2132 // wall clock rather than converting to UTC.
2133 crate::eval::format::parse_timestamp_literal_wall_ordered(
2134 &s,
2135 crate::eval::format::DateOrder::Mdy,
2136 )
2137 .map(Value::Timestamp)
2138 .ok_or_else(|| EvalError::TypeMismatch {
2139 // v7.39 (round 324, V42) — PG's wording, and PG's split
2140 // between "invalid input syntax" and "date/time field
2141 // value out of range".
2142 detail: crate::eval::format::datetime_input_error_text(&s, "timestamp"),
2143 })
2144 }
2145 other => Err(EvalError::TypeMismatch {
2146 detail: format!(
2147 "cannot cast {} to TIMESTAMP",
2148 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2149 ),
2150 }),
2151 }
2152}
2153
2154/// v7.39 (round 310) — `::timestamptz` from text: an offset in the
2155/// literal is APPLIED, unlike the zone-less sibling which discards it.
2156/// Naive input (no offset) is read as UTC, which is what the
2157/// context-aware arm already assumed when it fell through to here.
2158fn cast_to_timestamptz(v: Value) -> Result<Value, EvalError> {
2159 let Value::Text(s) = &v else {
2160 return cast_to_timestamp(v);
2161 };
2162 crate::eval::format::parse_timestamp_literal_tz_ordered(s, crate::eval::format::DateOrder::Mdy)
2163 .map(|(micros, _had_tz)| Value::Timestamp(micros))
2164 .ok_or_else(|| EvalError::TypeMismatch {
2165 // v7.39 (round 324, V42) — and with the RIGHT type name: this arm
2166 // used to report `TIMESTAMP` for a `::timestamptz` cast.
2167 detail: crate::eval::format::datetime_input_error_text(s, "timestamp with time zone"),
2168 })
2169}
2170
2171/// v7.39 (round 254) — PG refuses to cast a NUMERIC special into any
2172/// integer type: `cannot convert NaN to integer` / `cannot convert
2173/// infinity to bigint` (an infinity is named without its sign, probed
2174/// live). Returns `None` for an ordinary value so the caller runs its
2175/// normal conversion.
2176fn cast_numeric_special_reject(
2177 v: &Value,
2178 target: &str,
2179) -> Option<Result<Value<'static>, EvalError>> {
2180 let Value::Numeric { kind, .. } = v else {
2181 return None;
2182 };
2183 if *kind == spg_storage::NumericKind::Finite {
2184 return None;
2185 }
2186 let what = if *kind == spg_storage::NumericKind::NaN {
2187 "NaN"
2188 } else {
2189 "infinity"
2190 };
2191 Some(Err(EvalError::TypeMismatch {
2192 detail: alloc::format!("cannot convert {what} to {target}"),
2193 }))
2194}
2195
2196fn cast_numeric_to_int(v: Value) -> Result<Value, EvalError> {
2197 match v {
2198 // v7.39 (round 633) — SMALLINT. `1::SMALLINT::INT` answered
2199 // "cannot cast smallint to int": the arm was simply absent, next to
2200 // the Int and BigInt ones. Widening a smallint is about as ordinary
2201 // as a cast gets, and PG has it registered as an IMPLICIT cast.
2202 // Same omission shape as the sum accumulator missing SmallInt in
2203 // round 626 — a variant list written out by hand, one entry short.
2204 Value::SmallInt(n) => Ok(Value::Int(i32::from(n))),
2205 Value::Int(n) => Ok(Value::Int(n)),
2206 Value::BigInt(n) => i32::try_from(n)
2207 .map(Value::Int)
2208 // v7.39 (read01 round 79) — PG's wording, which the Float arm two
2209 // arms down was already using: "integer out of range". Drivers match
2210 // on it. Three arms of one function had two different messages.
2211 .map_err(|_| EvalError::TypeMismatch {
2212 detail: "integer out of range".into(),
2213 }),
2214 // PG rounds (half-to-even) coercing a real number to an integer, and
2215 // errors on a non-finite or out-of-range value (`'inf'::int`,
2216 // `1e20::int`) rather than saturating.
2217 #[allow(clippy::cast_possible_truncation)]
2218 Value::Float(x) => {
2219 let r = f64_round_half_even(x);
2220 if !r.is_finite() || !(-2_147_483_648.0..=2_147_483_647.0).contains(&r) {
2221 return Err(EvalError::TypeMismatch {
2222 detail: "integer out of range".into(),
2223 });
2224 }
2225 Ok(Value::Int(r as i32))
2226 }
2227 // v7.39 (read01 round 112) — `real` (float4) rounds/range-checks the
2228 // same way float8 does; only the float8 arm existed.
2229 #[allow(clippy::cast_possible_truncation)]
2230 Value::Real(x) => {
2231 let r = f64_round_half_even(f64::from(x));
2232 if !r.is_finite() || !(-2_147_483_648.0..=2_147_483_647.0).contains(&r) {
2233 return Err(EvalError::TypeMismatch {
2234 detail: "integer out of range".into(),
2235 });
2236 }
2237 Ok(Value::Int(r as i32))
2238 }
2239 Value::Numeric { scaled, scale, .. } => {
2240 let rounded = numeric_round_to_i128(scaled, scale);
2241 i32::try_from(rounded)
2242 .map(Value::Int)
2243 .map_err(|_| EvalError::TypeMismatch {
2244 detail: "integer out of range".into(),
2245 })
2246 }
2247 Value::Text(s) => crate::conversions::parse_pg_int(&s)
2248 .and_then(|n| i32::try_from(n).ok())
2249 .map(Value::Int)
2250 .ok_or_else(|| EvalError::TypeMismatch {
2251 detail: format!("invalid input syntax for type integer: {s:?}"),
2252 }),
2253 Value::Bool(b) => Ok(Value::Int(i32::from(b))),
2254 // v7.39 (read01 char.c) — ("char")::int is the byte value.
2255 Value::Char1(b) => Ok(Value::Int(i32::from(b))),
2256 // PG `bit`/`varbit` → int is the MSB-first bit value.
2257 #[allow(clippy::cast_possible_truncation)]
2258 Value::BitString { nbits, bytes } => Ok(Value::Int(crate::conversions::bit_string_to_i64(
2259 nbits, &bytes,
2260 ) as i32)),
2261 // v7.39 (read01 round 113) — jsonb → int: decode the JSON scalar, then
2262 // round via the numeric arm above. String/array/object/boolean error.
2263 Value::Json(s) => match crate::conversions::jsonb_scalar_for_cast(&s, "integer")? {
2264 crate::conversions::JsonbScalar::Numeric(n) => cast_numeric_to_int(n),
2265 crate::conversions::JsonbScalar::Bool(_) => Err(
2266 crate::conversions::jsonb_cast_type_error("boolean", "integer"),
2267 ),
2268 crate::conversions::JsonbScalar::Null => Ok(Value::Null),
2269 },
2270 other => Err(EvalError::TypeMismatch {
2271 detail: format!(
2272 "cannot cast {} to int",
2273 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2274 ),
2275 }),
2276 }
2277}
2278
2279fn cast_numeric_to_bigint(v: Value) -> Result<Value, EvalError> {
2280 match v {
2281 Value::Int(n) => Ok(Value::BigInt(i64::from(n))),
2282 // v7.39 (round 633) — SMALLINT, missing here for the same reason.
2283 Value::SmallInt(n) => Ok(Value::BigInt(i64::from(n))),
2284 Value::BigInt(n) => Ok(Value::BigInt(n)),
2285 // PG rounds (half-to-even) coercing a real number to bigint, and errors
2286 // on a non-finite or out-of-range value rather than saturating.
2287 #[allow(clippy::cast_possible_truncation)]
2288 Value::Float(x) => {
2289 let r = f64_round_half_even(x);
2290 if !r.is_finite()
2291 || !(-9_223_372_036_854_775_808.0..9_223_372_036_854_775_808.0).contains(&r)
2292 {
2293 return Err(EvalError::TypeMismatch {
2294 detail: "bigint out of range".into(),
2295 });
2296 }
2297 Ok(Value::BigInt(r as i64))
2298 }
2299 // v7.39 (read01 round 112) — `real` (float4) → bigint, matching float8.
2300 #[allow(clippy::cast_possible_truncation)]
2301 Value::Real(x) => {
2302 let r = f64_round_half_even(f64::from(x));
2303 if !r.is_finite()
2304 || !(-9_223_372_036_854_775_808.0..9_223_372_036_854_775_808.0).contains(&r)
2305 {
2306 return Err(EvalError::TypeMismatch {
2307 detail: "bigint out of range".into(),
2308 });
2309 }
2310 Ok(Value::BigInt(r as i64))
2311 }
2312 Value::Numeric { scaled, scale, .. } => {
2313 let rounded = numeric_round_to_i128(scaled, scale);
2314 i64::try_from(rounded)
2315 .map(Value::BigInt)
2316 .map_err(|_| EvalError::TypeMismatch {
2317 detail: format!("numeric {rounded} does not fit in bigint"),
2318 })
2319 }
2320 Value::Text(s) => crate::conversions::parse_pg_int(&s)
2321 .map(Value::BigInt)
2322 .ok_or_else(|| EvalError::TypeMismatch {
2323 // v7.39 (round 324, V42) — PG's wording.
2324 detail: format!("invalid input syntax for type bigint: \"{s}\""),
2325 }),
2326 Value::Bool(b) => Ok(Value::BigInt(i64::from(b))),
2327 // PG `bit`/`varbit` → bigint is the MSB-first bit value.
2328 Value::BitString { nbits, bytes } => Ok(Value::BigInt(
2329 crate::conversions::bit_string_to_i64(nbits, &bytes),
2330 )),
2331 // v7.39 (read01 round 113) — jsonb → bigint.
2332 Value::Json(s) => match crate::conversions::jsonb_scalar_for_cast(&s, "bigint")? {
2333 crate::conversions::JsonbScalar::Numeric(n) => cast_numeric_to_bigint(n),
2334 crate::conversions::JsonbScalar::Bool(_) => Err(
2335 crate::conversions::jsonb_cast_type_error("boolean", "bigint"),
2336 ),
2337 crate::conversions::JsonbScalar::Null => Ok(Value::Null),
2338 },
2339 other => Err(EvalError::TypeMismatch {
2340 detail: format!(
2341 "cannot cast {} to bigint",
2342 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2343 ),
2344 }),
2345 }
2346}
2347
2348fn cast_numeric_to_float(v: Value) -> Result<Value, EvalError> {
2349 match v {
2350 Value::Int(n) => Ok(Value::Float(f64::from(n))),
2351 #[allow(clippy::cast_precision_loss)]
2352 Value::BigInt(n) => Ok(Value::Float(n as f64)),
2353 Value::Float(x) => Ok(Value::Float(x)),
2354 // PG's numeric→double precision is an implicit cast; a
2355 // `Value::Numeric` (from `::numeric`, a numeric column, or
2356 // numeric arithmetic) must convert to f64, not error.
2357 #[allow(clippy::cast_precision_loss)]
2358 // v7.39 (round 254) — a special crosses to its IEEE twin.
2359 Value::Numeric { kind, .. } if kind != spg_storage::NumericKind::Finite => {
2360 Ok(Value::Float(match kind {
2361 spg_storage::NumericKind::NaN => f64::NAN,
2362 spg_storage::NumericKind::PosInf => f64::INFINITY,
2363 _ => f64::NEG_INFINITY,
2364 }))
2365 }
2366 Value::Numeric { scaled, scale, .. } => Ok(Value::Float(
2367 (scaled as f64) / f64_powi(10.0, i32::from(scale)),
2368 )),
2369 Value::Text(s) => {
2370 let t = s.trim();
2371 // Unparseable → invalid syntax; parseable-but-out-of-range (overflow
2372 // to ±∞ / nonzero underflow to 0) → out of range, the way PG's
2373 // float8in does, rather than silently yielding Infinity/0. Shared
2374 // with the Named-cast coerce path so `::float` and `::float8` agree.
2375 if t.parse::<f64>().is_err() {
2376 return Err(EvalError::TypeMismatch {
2377 detail: format!("cannot parse {s:?} as float"),
2378 });
2379 }
2380 crate::conversions::parse_float8(t)
2381 .map(Value::Float)
2382 .ok_or_else(|| EvalError::TypeMismatch {
2383 detail: format!("\"{t}\" is out of range for type double precision"),
2384 })
2385 }
2386 // v7.39 (read01 round 113) — jsonb → double precision.
2387 Value::Json(s) => {
2388 match crate::conversions::jsonb_scalar_for_cast(&s, "double precision")? {
2389 crate::conversions::JsonbScalar::Numeric(n) => cast_numeric_to_float(n),
2390 crate::conversions::JsonbScalar::Bool(_) => Err(
2391 crate::conversions::jsonb_cast_type_error("boolean", "double precision"),
2392 ),
2393 crate::conversions::JsonbScalar::Null => Ok(Value::Null),
2394 }
2395 }
2396 other => Err(EvalError::TypeMismatch {
2397 detail: format!(
2398 "cannot cast {} to float",
2399 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2400 ),
2401 }),
2402 }
2403}
2404
2405fn cast_to_bool(v: Value) -> Result<Value, EvalError> {
2406 match v {
2407 Value::Bool(b) => Ok(Value::Bool(b)),
2408 Value::Int(n) => Ok(Value::Bool(n != 0)),
2409 Value::BigInt(n) => Ok(Value::Bool(n != 0)),
2410 Value::Text(s) => {
2411 // PG boolin accepts any unambiguous prefix of true/false/yes/no
2412 // plus on/off/1/0 (case-insensitive, trimmed); `o` alone is
2413 // ambiguous (on vs off) and errors.
2414 let lo = s.trim().to_ascii_lowercase();
2415 match lo.as_str() {
2416 "1" | "t" | "tr" | "tru" | "true" | "y" | "ye" | "yes" | "on" => {
2417 Ok(Value::Bool(true))
2418 }
2419 "0" | "f" | "fa" | "fal" | "fals" | "false" | "n" | "no" | "of" | "off" => {
2420 Ok(Value::Bool(false))
2421 }
2422 _ => Err(EvalError::TypeMismatch {
2423 detail: format!("invalid input syntax for type boolean: {:?}", s.trim()),
2424 }),
2425 }
2426 }
2427 // v7.39 (read01 round 113) — jsonb → boolean accepts only JSON
2428 // true/false; a JSON number/string/array/object errors.
2429 Value::Json(s) => match crate::conversions::jsonb_scalar_for_cast(&s, "boolean")? {
2430 crate::conversions::JsonbScalar::Bool(b) => Ok(Value::Bool(b)),
2431 crate::conversions::JsonbScalar::Numeric(_) => Err(
2432 crate::conversions::jsonb_cast_type_error("numeric", "boolean"),
2433 ),
2434 crate::conversions::JsonbScalar::Null => Ok(Value::Null),
2435 },
2436 other => Err(EvalError::TypeMismatch {
2437 detail: format!(
2438 "cannot cast {} to bool",
2439 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2440 ),
2441 }),
2442 }
2443}
2444
2445/// Parse a `Value::text("[1.0, 2.0, 3.0]")` into a `Value::vector(..)`. Mirrors
2446/// pgvector's `'[..]'::vector` cast. NULL casts as NULL.
2447pub fn cast_to_vector(v: Value) -> Result<Value<'static>, EvalError> {
2448 match v {
2449 Value::Null => Ok(Value::Null),
2450 Value::Vector(v) => Ok(Value::vector(v.into_owned())),
2451 Value::Text(s) => {
2452 parse_vector_text(&s)
2453 .map(Value::vector)
2454 .ok_or_else(|| EvalError::TypeMismatch {
2455 detail: format!("cannot parse {s:?} as a vector literal"),
2456 })
2457 }
2458 other => Err(EvalError::TypeMismatch {
2459 detail: format!(
2460 "::vector requires text input, got {}",
2461 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2462 ),
2463 }),
2464 }
2465}
2466
2467/// Parse `"[1.0, 2.0, -3]"` into `Vec<f32>`. Returns `None` on malformed input.
2468pub fn parse_vector_text(s: &str) -> Option<Vec<f32>> {
2469 let trimmed = s.trim();
2470 let inner = trimmed.strip_prefix('[')?.strip_suffix(']')?;
2471 let trimmed_inner = inner.trim();
2472 if trimmed_inner.is_empty() {
2473 return Some(Vec::new());
2474 }
2475 let mut out = Vec::new();
2476 for part in trimmed_inner.split(',') {
2477 let f: f32 = part.trim().parse().ok()?;
2478 out.push(f);
2479 }
2480 Some(out)
2481}
2482
2483#[cfg(test)]
2484mod round613_plain_named_targets {
2485 use super::*;
2486
2487 /// v7.39 (round 613) — the shortcut is only equivalent to walking the
2488 /// arm while these hold. Checked here rather than by eye, so a name that
2489 /// grows a special case above the resolve fails the gate instead of
2490 /// silently taking the wrong path.
2491 fn assert_no_arm_above_the_resolve_claims(name: &str) {
2492 assert!(
2493 !REG_MISC_TYPES.iter().any(|k| name.eq_ignore_ascii_case(k)),
2494 "{name} is a reg-misc type"
2495 );
2496 assert!(
2497 !CATALOG_SCALAR_TYPES
2498 .iter()
2499 .any(|k| name.eq_ignore_ascii_case(k)),
2500 "{name} is a catalog scalar"
2501 );
2502 assert!(
2503 !OPAQUE_TYPES.iter().any(|k| name.eq_ignore_ascii_case(k)),
2504 "{name} is a pseudotype"
2505 );
2506 for special in [
2507 "__bit_literal",
2508 "tid",
2509 "xid",
2510 "xid8",
2511 "jsonpath",
2512 "binary",
2513 "signed",
2514 "unsigned",
2515 ] {
2516 assert!(
2517 !name.eq_ignore_ascii_case(special),
2518 "{name} has its own arm ({special})"
2519 );
2520 }
2521 assert!(bit_cast_width(name).is_none(), "{name} is a bit spelling");
2522 assert!(
2523 temporal_typmod(name).is_none(),
2524 "{name} carries a temporal precision"
2525 );
2526 assert!(
2527 !is_bare_temporal_type(name),
2528 "{name} is a bare temporal type"
2529 );
2530 assert!(
2531 matches!(cast_catalog_scalar(name, &Value::text("x")), Ok(None)),
2532 "{name} is claimed by the catalog-scalar arm"
2533 );
2534 }
2535
2536 #[test]
2537 fn every_plain_target_resolves_to_the_type_the_table_claims() {
2538 for (name, dt) in PLAIN_NAMED_TARGETS {
2539 assert_eq!(
2540 crate::conversions::type_name_to_data_type(name),
2541 Some(*dt),
2542 "{name} does not resolve to the type the table gives it"
2543 );
2544 assert_eq!(plain_named_target(name), Some(*dt));
2545 // The spelling is matched without regard to case.
2546 assert_eq!(plain_named_target(&name.to_uppercase()), Some(*dt));
2547 assert_no_arm_above_the_resolve_claims(name);
2548 }
2549 }
2550
2551 #[test]
2552 fn every_typmod_head_is_plain_and_resolves_through_the_type_table() {
2553 for head in PLAIN_NAMED_HEADS {
2554 assert_no_arm_above_the_resolve_claims(head);
2555 for spelled in [alloc::format!("{head}(4)"), alloc::format!("{head}(10,2)")] {
2556 assert_no_arm_above_the_resolve_claims(&spelled);
2557 assert_eq!(
2558 plain_named_target(&spelled),
2559 crate::conversions::type_name_to_data_type(&spelled),
2560 "{spelled} takes a different type through the shortcut"
2561 );
2562 }
2563 // A bare head with no typmod only shortcuts when it is in the
2564 // exact table; the head list alone must not claim it.
2565 let bare = plain_named_target(head);
2566 let exact = PLAIN_NAMED_TARGETS
2567 .iter()
2568 .find(|(k, _)| head.eq_ignore_ascii_case(k))
2569 .map(|(_, dt)| *dt);
2570 assert_eq!(bare, exact, "{head} bare");
2571 }
2572 }
2573
2574 #[test]
2575 fn a_name_with_its_own_arm_is_not_shortcut() {
2576 for name in [
2577 "regproc",
2578 "aclitem",
2579 "anyarray",
2580 "tid",
2581 "xid",
2582 "jsonpath",
2583 "bit",
2584 "bit(4)",
2585 "timestamp",
2586 "timestamp(2)",
2587 "time(3)",
2588 "nosuchtype",
2589 "int4range",
2590 ] {
2591 assert_eq!(plain_named_target(name), None, "{name} was shortcut");
2592 }
2593 }
2594}