1use crate::dates::{day_from_number, day_number, days_in_month, format_day, iso_week, parse_day, weekday_of};
12use crate::json;
13use serde_json::{Map, Value};
14
15pub const MAX_LENGTH: usize = 1000;
17pub const MAX_DEPTH: usize = 32;
18pub const MAX_STEPS: usize = 10_000;
19
20pub const DATE_UNITS: &[&str] = &["second", "minute", "hour", "day", "week", "month", "quarter", "year"];
22pub const CALENDAR_UNITS: &[&str] = &["day", "week", "month", "quarter", "year"];
23pub const DURATION_FORMATS: &[&str] = &["h:mm", "hh:mm"];
25
26#[derive(Debug, Clone, Copy, PartialEq, Eq)]
32pub enum FunctionEffect {
33 Pure,
34 Write,
35 Host,
36}
37
38impl FunctionEffect {
39 pub fn as_str(self) -> &'static str {
41 match self {
42 FunctionEffect::Pure => "pure",
43 FunctionEffect::Write => "write",
44 FunctionEffect::Host => "host",
45 }
46 }
47}
48
49pub const WRITE_FUNCTIONS: &[&str] = &["set", "records.upsert", "records.delete", "records.batch"];
51pub const HOST_FUNCTIONS: &[&str] = &["openUrl", "bind"];
53pub const HOST_FUNCTION_PREFIXES: &[&str] = &["actions.", "harness.", "exponential.", "tools.", "upstream."];
54
55pub fn function_effect(name: &str) -> FunctionEffect {
58 if WRITE_FUNCTIONS.contains(&name) {
59 return FunctionEffect::Write;
60 }
61 if HOST_FUNCTIONS.contains(&name) || HOST_FUNCTION_PREFIXES.iter().any(|p| name.starts_with(p)) {
62 return FunctionEffect::Host;
63 }
64 FunctionEffect::Pure
65}
66
67pub fn effect_words(name: &str, effect: FunctionEffect) -> String {
69 format!("{name} {}", if effect == FunctionEffect::Write { "writes" } else { "is a host function" })
70}
71
72#[derive(Clone, Copy, Default)]
74pub struct ExpressionEnv<'a> {
75 pub now: Option<f64>,
77 pub offset: Option<&'a (dyn Fn(f64) -> i32 + Send + Sync)>,
79 pub max_steps: Option<usize>,
81}
82
83impl std::fmt::Debug for ExpressionEnv<'_> {
84 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
85 f.debug_struct("ExpressionEnv").field("now", &self.now).field("max_steps", &self.max_steps).finish_non_exhaustive()
86 }
87}
88
89struct Env<'a> {
90 now: f64,
91 offset: Option<&'a (dyn Fn(f64) -> i32 + Send + Sync)>,
92}
93
94impl<'a> Env<'a> {
95 fn of(env: &ExpressionEnv<'a>) -> Env<'a> {
96 Env { now: env.now.unwrap_or_else(crate::format::now_ms).floor(), offset: env.offset }
97 }
98 fn off(&self, ms: f64) -> f64 {
99 self.offset.map_or(0.0, |o| f64::from(o(ms)))
100 }
101}
102
103type R<T> = Result<T, String>;
104
105pub fn type_name(v: &Value) -> &'static str {
111 match v {
112 Value::Null => "null",
113 Value::Number(_) => "number",
114 Value::String(_) => "string",
115 Value::Bool(_) => "bool",
116 Value::Array(_) => "list",
117 Value::Object(_) => "map",
118 }
119}
120
121fn num(v: f64) -> R<f64> {
123 if !v.is_finite() {
124 return Err("number out of range".into());
125 }
126 Ok(if v == 0.0 { 0.0 } else { v })
127}
128
129fn num_text(n: f64) -> String {
130 json::number_to_string(n)
131}
132
133fn deep_equal(a: &Value, b: &Value) -> bool {
134 match (a, b) {
135 (Value::Number(x), Value::Number(y)) => x.as_f64() == y.as_f64(),
136 (Value::Array(x), Value::Array(y)) => x.len() == y.len() && x.iter().zip(y).all(|(a, b)| deep_equal(a, b)),
137 (Value::Object(x), Value::Object(y)) => x.len() == y.len() && x.iter().all(|(k, v)| y.get(k).is_some_and(|w| deep_equal(v, w))),
138 _ => a == b,
139 }
140}
141
142fn round_half_away(x: f64) -> f64 {
144 let a = x.abs();
145 let f = a.floor();
146 let r = if a - f >= 0.5 { f + 1.0 } else { f };
147 if x < 0.0 {
148 -r
149 } else {
150 r
151 }
152}
153
154fn utf16_cmp(a: &str, b: &str) -> std::cmp::Ordering {
155 a.encode_utf16().cmp(b.encode_utf16())
156}
157
158const DAY_MS: f64 = 86_400_000.0;
163
164fn unit_ms(unit: &str) -> Option<f64> {
165 match unit {
166 "second" => Some(1000.0),
167 "minute" => Some(60_000.0),
168 "hour" => Some(3_600_000.0),
169 _ => None,
170 }
171}
172
173#[derive(Clone, Copy)]
174enum Temporal {
175 Date(i64),
176 Ts(f64),
177}
178
179pub fn parse_timestamp(text: &str) -> Option<f64> {
182 let b = text.as_bytes();
183 let digit = |i: usize| b.get(i).is_some_and(u8::is_ascii_digit);
184 let int = |from: usize, n: usize| b[from..from + n].iter().fold(0i64, |acc, c| acc * 10 + i64::from(c - b'0'));
185 if b.len() < 16 || b[10] != b'T' || !(digit(11) && digit(12) && b[13] == b':' && digit(14) && digit(15)) {
186 return None;
187 }
188 let day = parse_day(&text[..10]).filter(|_| text.is_char_boundary(10))?;
189 if !(b[..10].iter().enumerate().all(|(i, c)| if i == 4 || i == 7 { *c == b'-' } else { c.is_ascii_digit() })) {
190 return None;
191 }
192 let (h, mi) = (int(11, 2), int(14, 2));
193 let mut i = 16;
194 let mut s = 0;
195 let mut ms = 0;
196 if b.get(i) == Some(&b':') {
197 if !(digit(i + 1) && digit(i + 2)) {
198 return None;
199 }
200 s = int(i + 1, 2);
201 i += 3;
202 if b.get(i) == Some(&b'.') {
203 let start = i + 1;
204 let mut end = start;
205 while digit(end) {
206 end += 1;
207 }
208 if end == start {
209 return None;
210 }
211 let frac: String = text[start..end.min(start + 3)].chars().collect();
212 ms = format!("{frac:0<3}").parse::<i64>().ok()?;
213 i = end;
214 }
215 }
216 let mut offset = 0;
217 match b.get(i) {
218 None => {}
219 Some(b'Z') if i + 1 == b.len() => {}
220 Some(c @ (b'+' | b'-')) if b.len() == i + 6 && digit(i + 1) && digit(i + 2) && b[i + 3] == b':' && digit(i + 4) && digit(i + 5) => {
221 let (oh, om) = (int(i + 1, 2), int(i + 4, 2));
222 if oh > 23 || om > 59 {
223 return None;
224 }
225 offset = (if *c == b'-' { -1 } else { 1 }) * (oh * 60 + om);
226 }
227 _ => return None,
228 }
229 if h > 23 || mi > 59 || s > 59 {
230 return None;
231 }
232 Some((day_number(day.y, day.m, day.d) as f64) * DAY_MS + (h * 3_600_000 + mi * 60_000 + s * 1000 + ms - offset * 60_000) as f64)
233}
234
235pub fn format_timestamp(ms: f64) -> R<String> {
237 let day = (ms / DAY_MS).floor();
238 let tod = (ms - day * DAY_MS) as i64;
239 let d = day_from_number(day as i64);
240 if d.y < 0 || d.y > 9999 {
241 return Err("date out of range".into());
242 }
243 Ok(format!("{}T{:02}:{:02}:{:02}.{:03}Z", format_day(d), tod / 3_600_000, (tod / 60_000) % 60, (tod / 1000) % 60, tod % 1000))
244}
245
246fn day_text(day: i64) -> R<String> {
247 let d = day_from_number(day);
248 if d.y < 0 || d.y > 9999 {
249 return Err("date out of range".into());
250 }
251 Ok(format_day(d))
252}
253
254fn to_local(env: &Env, ms: f64) -> f64 {
256 ms + env.off(ms) * 60_000.0
257}
258
259fn from_local(env: &Env, wall: f64) -> f64 {
262 let first = env.off(wall);
263 let utc = wall - first * 60_000.0;
264 let second = env.off(utc);
265 if second == first {
266 utc
267 } else {
268 wall - second * 60_000.0
269 }
270}
271
272fn temporal(f: &str, v: &Value) -> R<Temporal> {
273 let Value::String(s) = v else { return Err(format!("{f} needs a date, got {}", type_name(v))) };
274 if s.chars().count() == 10 {
275 if let Some(d) = parse_day(s) {
276 return Ok(Temporal::Date(day_number(d.y, d.m, d.d)));
277 }
278 }
279 parse_timestamp(s).map(Temporal::Ts).ok_or_else(|| format!("not a date: {s}"))
280}
281
282fn floor_day(wall: f64) -> i64 {
283 (wall / DAY_MS).floor() as i64
284}
285
286fn wall_of(env: &Env, t: Temporal) -> f64 {
287 match t {
288 Temporal::Date(day) => day as f64 * DAY_MS,
289 Temporal::Ts(ms) => to_local(env, ms),
290 }
291}
292
293fn unit_of<'a>(f: &str, v: &'a Value, units: &[&str]) -> R<&'a str> {
294 let Value::String(s) = v else { return Err(format!("{f} needs a string, got {}", type_name(v))) };
295 if !units.contains(&s.as_str()) {
296 return Err(format!("unknown unit {s}"));
297 }
298 Ok(s)
299}
300
301fn add_months(day: i64, months: i64) -> i64 {
302 let c = day_from_number(day);
303 let total = c.y * 12 + (c.m - 1) + months;
304 let y = total.div_euclid(12);
305 let m = total - y * 12 + 1;
306 day_number(y, m, c.d.min(days_in_month(y, m)))
307}
308
309fn add_calendar(day: i64, n: i64, unit: &str) -> i64 {
310 match unit {
311 "day" => day + n,
312 "week" => day + 7 * n,
313 "month" => add_months(day, n),
314 "quarter" => add_months(day, n * 3),
315 _ => add_months(day, n * 12),
316 }
317}
318
319fn start_day(day: i64, unit: &str) -> i64 {
320 let c = day_from_number(day);
321 match unit {
322 "day" => day,
323 "week" => {
324 let wd = match weekday_of(c.y, c.m, c.d) {
325 0 => 7,
326 n => n as i64,
327 };
328 day - (wd - 1)
329 }
330 "month" => day_number(c.y, c.m, 1),
331 "quarter" => day_number(c.y, (c.m - 1) / 3 * 3 + 1, 1),
332 _ => day_number(c.y, 1, 1),
333 }
334}
335
336fn number(f: &str, v: &Value) -> R<f64> {
337 match v {
338 Value::Number(n) => Ok(n.as_f64().unwrap_or(0.0)),
339 _ => Err(format!("{f} needs a number, got {}", type_name(v))),
340 }
341}
342
343fn whole_number(f: &str, v: &Value) -> R<f64> {
344 let n = number(f, v)?;
345 if n.fract() != 0.0 {
346 return Err(format!("{f} needs a whole number, got {}", num_text(n)));
347 }
348 Ok(n)
349}
350
351fn text<'a>(f: &str, v: &'a Value) -> R<&'a str> {
352 match v {
353 Value::String(s) => Ok(s),
354 _ => Err(format!("{f} needs a string, got {}", type_name(v))),
355 }
356}
357
358fn date_add(args: &[Value], env: &Env) -> R<Value> {
359 let t = temporal("dateAdd", &args[0])?;
360 let n = whole_number("dateAdd", &args[1])?;
361 let unit = unit_of("dateAdd", &args[2], DATE_UNITS)?;
362 let out = match t {
363 Temporal::Date(day) => {
364 if unit_ms(unit).is_some() {
365 return Err(format!("unit {unit} needs a timestamp"));
366 }
367 day_text(add_calendar(day, n as i64, unit))?
368 }
369 Temporal::Ts(ms) => match unit_ms(unit) {
370 Some(u) => format_timestamp(num(ms + n * u)?)?,
371 None => {
372 let wall = to_local(env, ms);
373 let day = floor_day(wall);
374 let tod = wall - day as f64 * DAY_MS;
375 format_timestamp(from_local(env, add_calendar(day, n as i64, unit) as f64 * DAY_MS + tod))?
376 }
377 },
378 };
379 Ok(Value::String(out))
380}
381
382fn date_diff(args: &[Value], env: &Env) -> R<Value> {
383 let a = temporal("dateDiff", &args[0])?;
384 let b = temporal("dateDiff", &args[1])?;
385 let unit = unit_of("dateDiff", &args[2], DATE_UNITS)?;
386 if let Some(u) = unit_ms(unit) {
387 let instant = |t: Temporal| match t {
388 Temporal::Date(day) => from_local(env, day as f64 * DAY_MS),
389 Temporal::Ts(ms) => ms,
390 };
391 return Ok(json::number(num(((instant(b) - instant(a)) / u).trunc())?));
392 }
393 let (wa, wb) = (wall_of(env, a), wall_of(env, b));
394 if unit == "day" || unit == "week" {
395 let span = if unit == "day" { DAY_MS } else { 7.0 * DAY_MS };
396 return Ok(json::number(num(((wb - wa) / span).trunc())?));
397 }
398 let (da, db) = (floor_day(wa), floor_day(wb));
399 let (ca, cb) = (day_from_number(da), day_from_number(db));
400 let mut months = (cb.y - ca.y) * 12 + (cb.m - ca.m);
401 let rest_a = ca.d as f64 * DAY_MS + (wa - da as f64 * DAY_MS);
402 let rest_b = cb.d as f64 * DAY_MS + (wb - db as f64 * DAY_MS);
403 if months > 0 && rest_b < rest_a {
404 months -= 1;
405 } else if months < 0 && rest_b > rest_a {
406 months += 1;
407 }
408 let per = match unit {
409 "month" => 1.0,
410 "quarter" => 3.0,
411 _ => 12.0,
412 };
413 Ok(json::number(num((months as f64 / per).trunc())?))
414}
415
416fn start_end(f: &str, args: &[Value], env: &Env) -> R<Value> {
417 let t = temporal(f, &args[0])?;
418 let unit = unit_of(f, &args[1], CALENDAR_UNITS)?;
419 let day = match t {
420 Temporal::Date(day) => day,
421 Temporal::Ts(ms) => floor_day(to_local(env, ms)),
422 };
423 let is_date = matches!(t, Temporal::Date(_));
424 let out = if f == "startOf" {
425 let start = start_day(day, unit);
426 if is_date {
427 day_text(start)?
428 } else {
429 format_timestamp(from_local(env, start as f64 * DAY_MS))?
430 }
431 } else {
432 let next = add_calendar(start_day(day, unit), 1, unit);
433 if is_date {
434 day_text(next - 1)?
435 } else {
436 format_timestamp(from_local(env, next as f64 * DAY_MS) - 1.0)?
437 }
438 };
439 Ok(Value::String(out))
440}
441
442fn local_day(f: &str, v: &Value, env: &Env) -> R<crate::dates::CalendarDay> {
443 Ok(day_from_number(match temporal(f, v)? {
444 Temporal::Date(day) => day,
445 Temporal::Ts(ms) => floor_day(to_local(env, ms)),
446 }))
447}
448
449fn format_duration(args: &[Value]) -> R<Value> {
450 let h = number("formatDuration", &args[0])?;
451 let format = text("formatDuration", &args[1])?;
452 if !DURATION_FORMATS.contains(&format) {
453 return Err(format!("unknown duration format {format}"));
454 }
455 let total = round_half_away(h.abs() * 60.0);
456 if total > 1e15 {
457 return Err("number out of range".into());
458 }
459 let hours = (total / 60.0).floor() as i64;
460 let minutes = total as i64 - hours * 60;
461 let sign = if h < 0.0 && total > 0.0 { "-" } else { "" };
462 let hours = if format == "hh:mm" { format!("{hours:02}") } else { hours.to_string() };
463 Ok(Value::String(format!("{sign}{hours}:{minutes:02}")))
464}
465
466fn round(args: &[Value]) -> R<Value> {
467 let x = number("round", &args[0])?;
468 let digits = match args.get(1) {
469 None => 0.0,
470 Some(Value::Number(n)) => {
471 let d = n.as_f64().unwrap_or(f64::NAN);
472 if d.fract() != 0.0 || !(0.0..=10.0).contains(&d) {
473 return Err("round needs 0 to 10 digits".into());
474 }
475 d
476 }
477 Some(_) => return Err("round needs 0 to 10 digits".into()),
478 };
479 let mut p = 1.0;
480 for _ in 0..digits as usize {
481 p *= 10.0;
482 }
483 Ok(json::number(num(round_half_away(x * p) / p)?))
484}
485
486fn size(v: &Value) -> R<Value> {
487 Ok(json::number(match v {
488 Value::String(s) => s.chars().count() as f64,
489 Value::Array(a) => a.len() as f64,
490 Value::Object(m) => m.len() as f64,
491 _ => return Err(format!("size needs a list, string or map, got {}", type_name(v))),
492 }))
493}
494
495#[derive(Clone, Copy, PartialEq, Eq)]
496enum Param {
497 Number,
498 Text,
499 Temporal,
500 Sized,
501 Any,
502}
503
504struct Def {
505 params: &'static [Param],
506 min: usize,
507 variadic: bool,
508 args: &'static [&'static str],
509}
510
511use Param::{Any as A, Number as N, Sized as Z, Temporal as T, Text as S};
512
513pub const FUNCTION_NAMES: &[&str] = &["mul", "div", "mod", "round", "floor", "ceil", "abs", "min", "max", "today", "now", "dateAdd", "dateDiff", "startOf", "endOf", "weekday", "isoWeek", "hours", "formatDuration", "size", "orElse"];
516
517pub const CALL_NAMES: &[&str] = &["mul", "div", "mod", "round", "floor", "ceil", "abs", "min", "max", "today", "now", "dateAdd", "dateDiff", "startOf", "endOf", "weekday", "isoWeek", "hours", "formatDuration", "orElse", "expr"];
520
521fn def(name: &str) -> Option<Def> {
522 let d = |params: &'static [Param], args: &'static [&'static str]| Def { params, min: params.len(), variadic: false, args };
523 Some(match name {
524 "mul" | "div" | "mod" => d(&[N, N], &["a", "b"]),
525 "round" => Def { params: &[N, N], min: 1, variadic: false, args: &["value", "digits"] },
526 "floor" | "ceil" | "abs" => d(&[N], &["value"]),
527 "min" | "max" => Def { params: &[N], min: 1, variadic: true, args: &["values"] },
528 "today" | "now" => d(&[], &[]),
529 "dateAdd" => d(&[T, N, S], &["date", "n", "unit"]),
530 "dateDiff" => d(&[T, T, S], &["a", "b", "unit"]),
531 "startOf" | "endOf" => d(&[T, S], &["date", "unit"]),
532 "weekday" | "isoWeek" => d(&[T], &["date"]),
533 "hours" => d(&[N], &["minutes"]),
534 "formatDuration" => d(&[N, S], &["hours", "format"]),
535 "size" => d(&[Z], &["value"]),
536 "orElse" => d(&[A, A], &["value", "default"]),
538 _ => return None,
539 })
540}
541
542fn arity_error(name: &str, def: &Def, n: usize) -> Option<String> {
543 if def.variadic {
544 return (n < 1).then(|| format!("{name} takes at least 1 argument"));
545 }
546 let (min, max) = (def.min, def.params.len());
547 if n >= min && n <= max {
548 return None;
549 }
550 Some(if min == max { format!("{name} takes {min} argument{}", if min == 1 { "" } else { "s" }) } else { format!("{name} takes {min} or {max} arguments") })
551}
552
553fn run(name: &str, args: &[Value], env: &Env) -> R<Value> {
554 let n = |f: fn(f64) -> f64| -> R<Value> { Ok(json::number(num(f(number(name, &args[0])?))?)) };
555 match name {
556 "mul" => Ok(json::number(num(number("mul", &args[0])? * number("mul", &args[1])?)?)),
557 "div" => {
558 let (x, y) = (number("div", &args[0])?, number("div", &args[1])?);
559 if y == 0.0 {
560 return Err("division by zero".into());
561 }
562 Ok(json::number(num(x / y)?))
563 }
564 "mod" => {
565 let (x, y) = (number("mod", &args[0])?, number("mod", &args[1])?);
566 if y == 0.0 {
567 return Err("modulo by zero".into());
568 }
569 Ok(json::number(num(x % y)?))
570 }
571 "round" => round(args),
572 "floor" => n(f64::floor),
573 "ceil" => n(f64::ceil),
574 "abs" => n(f64::abs),
575 "min" | "max" => {
576 let mut out = if name == "min" { f64::INFINITY } else { f64::NEG_INFINITY };
577 for v in args {
578 let x = number(name, v)?;
579 out = if name == "min" { out.min(x) } else { out.max(x) };
580 }
581 Ok(json::number(num(out)?))
582 }
583 "today" => Ok(Value::String(day_text(floor_day(to_local(env, env.now)))?)),
584 "now" => Ok(Value::String(format_timestamp(env.now)?)),
585 "dateAdd" => date_add(args, env),
586 "dateDiff" => date_diff(args, env),
587 "startOf" | "endOf" => start_end(name, args, env),
588 "weekday" => {
589 let d = local_day("weekday", &args[0], env)?;
590 Ok(json::number(match weekday_of(d.y, d.m, d.d) {
591 0 => 7.0,
592 w => w as f64,
593 }))
594 }
595 "isoWeek" => {
596 let d = local_day("isoWeek", &args[0], env)?;
597 Ok(json::number(iso_week(d.y, d.m, d.d).1 as f64))
598 }
599 "hours" => Ok(json::number(num(number("hours", &args[0])? / 60.0)?)),
600 "formatDuration" => format_duration(args),
601 "size" => size(&args[0]),
602 "orElse" => Ok(if args[0].is_null() { args.get(1).cloned().unwrap_or(Value::Null) } else { args[0].clone() }),
603 _ => Err(format!("unknown function {name}")),
604 }
605}
606
607pub fn call_expression_function(name: &str, args: &Map<String, Value>, env: &ExpressionEnv) -> Option<Option<Value>> {
612 let fallback = || Some(args.get("default").cloned());
613 if name == "expr" {
614 let Some(Value::String(expr)) = args.get("expr") else { return fallback() };
615 let empty = Map::new();
616 let vars = args.get("vars").and_then(Value::as_object).unwrap_or(&empty);
617 return evaluate_expression(expr, vars, env).map_or_else(|_| fallback(), |v| Some(Some(v)));
618 }
619 if name == "size" {
620 return None;
621 }
622 let def = def(name)?;
623 let positional: Vec<Value> = if def.variadic {
624 args.get("values").and_then(Value::as_array).cloned().unwrap_or_default()
625 } else {
626 let mut list: Vec<Option<&Value>> = def.args.iter().map(|k| args.get(*k)).collect();
627 while list.len() > def.min && list.last().is_some_and(Option::is_none) {
628 list.pop();
629 }
630 list.into_iter().map(|v| v.cloned().unwrap_or(Value::Null)).collect()
631 };
632 if arity_error(name, &def, positional.len()).is_some() {
633 return fallback();
634 }
635 run(name, &positional, &Env::of(env)).map_or_else(|_| fallback(), |v| Some(Some(v)))
636}
637
638#[derive(Debug, Clone, PartialEq)]
644pub enum Node {
645 Lit(Value, usize),
646 Id(String, usize),
647 List(Vec<Node>, usize),
648 Field(Box<Node>, String, usize),
649 Index(Box<Node>, Box<Node>, usize),
650 Unary(char, Box<Node>, usize),
651 Bin(&'static str, Box<Node>, Box<Node>, usize),
652 Cond(Box<Node>, Box<Node>, Box<Node>, usize),
653 Call(String, Vec<Node>, usize),
654 Macro(String, Box<Node>, String, Box<Node>, usize),
655 Has(Box<Node>, usize),
657}
658
659#[derive(Debug, Clone, PartialEq)]
660enum Tok {
661 Num(f64),
662 Str(String),
663 Id(String),
664 Op(&'static str),
665 End,
666}
667
668#[derive(Debug, Clone)]
669struct Token {
670 t: Tok,
671 at: usize,
672 text: String,
673}
674
675const OPS2: &[&str] = &["==", "!=", "<=", ">=", "&&", "||"];
676const OPS1: &[&str] = &["+", "-", "*", "/", "%", "<", ">", "!", "(", ")", "[", "]", ",", ".", "?", ":"];
677
678fn is_digit(c: Option<&char>) -> bool {
679 c.is_some_and(char::is_ascii_digit)
680}
681fn is_id_start(c: Option<&char>) -> bool {
682 c.is_some_and(|c| c.is_ascii_alphabetic() || *c == '_')
683}
684fn is_id_part(c: Option<&char>) -> bool {
685 is_id_start(c) || is_digit(c)
686}
687
688fn tokenize(cs: &[char]) -> R<Vec<Token>> {
689 let mut out = Vec::new();
690 let mut i = 0;
691 while i < cs.len() {
692 let c = cs[i];
693 if matches!(c, ' ' | '\t' | '\n' | '\r') {
694 i += 1;
695 continue;
696 }
697 let at = i;
698 if c.is_ascii_digit() {
699 while is_digit(cs.get(i)) {
700 i += 1;
701 }
702 if cs.get(i) == Some(&'.') && is_digit(cs.get(i + 1)) {
703 i += 1;
704 while is_digit(cs.get(i)) {
705 i += 1;
706 }
707 }
708 if matches!(cs.get(i), Some('e' | 'E')) && (is_digit(cs.get(i + 1)) || (matches!(cs.get(i + 1), Some('+' | '-')) && is_digit(cs.get(i + 2)))) {
709 i += 2;
710 while is_digit(cs.get(i)) {
711 i += 1;
712 }
713 }
714 let text: String = cs[at..i].iter().collect();
715 let v = text.parse::<f64>().unwrap_or(f64::INFINITY);
716 out.push(Token { t: Tok::Num(v), at, text });
717 continue;
718 }
719 if c == '\'' || c == '"' {
720 i += 1;
721 let mut s = String::new();
722 loop {
723 let d = cs.get(i).copied();
724 match d {
725 None | Some('\n') => return Err(format!("at {at}: unterminated string")),
726 Some(d) if d == c => break,
727 Some('\\') => {
728 let e = cs.get(i + 1).copied();
729 let mapped = match e {
730 Some('\\') => '\\',
731 Some('\'') => '\'',
732 Some('"') => '"',
733 Some('n') => '\n',
734 Some('t') => '\t',
735 Some('r') => '\r',
736 _ => return Err(format!("at {i}: unknown escape \\{}", e.map(String::from).unwrap_or_default())),
737 };
738 s.push(mapped);
739 i += 2;
740 }
741 Some(d) => {
742 s.push(d);
743 i += 1;
744 }
745 }
746 }
747 i += 1;
748 out.push(Token { t: Tok::Str(s), at, text: cs[at..i].iter().collect() });
749 continue;
750 }
751 if is_id_start(Some(&c)) {
752 while is_id_part(cs.get(i)) {
753 i += 1;
754 }
755 let text: String = cs[at..i].iter().collect();
756 out.push(Token { t: Tok::Id(text.clone()), at, text });
757 continue;
758 }
759 let two: String = cs[i..(i + 2).min(cs.len())].iter().collect();
760 if let Some(op) = OPS2.iter().find(|o| **o == two) {
761 out.push(Token { t: Tok::Op(op), at, text: two });
762 i += 2;
763 continue;
764 }
765 if let Some(op) = OPS1.iter().find(|o| o.starts_with(c)) {
766 out.push(Token { t: Tok::Op(op), at, text: c.to_string() });
767 i += 1;
768 continue;
769 }
770 return Err(format!("at {at}: unexpected {c}"));
771 }
772 out.push(Token { t: Tok::End, at: cs.len(), text: "end".into() });
773 Ok(out)
774}
775
776const MACROS: &[&str] = &["map", "filter", "all", "exists"];
777
778struct Parser {
779 tokens: Vec<Token>,
780 i: usize,
781 depth: usize,
782 in_body: usize,
783}
784
785impl Parser {
786 fn peek(&self) -> &Token {
787 &self.tokens[self.i]
788 }
789 fn is_op(&self, v: &str) -> bool {
790 matches!(&self.peek().t, Tok::Op(o) if *o == v)
791 }
792 fn next(&mut self) -> Token {
793 let t = self.tokens[self.i].clone();
794 self.i += 1;
795 t
796 }
797 fn unexpected<T>(&self) -> R<T> {
798 let t = self.peek();
799 Err(format!("at {}: unexpected {}", t.at, t.text))
800 }
801 fn expect(&mut self, v: &str) -> R<Token> {
802 if !self.is_op(v) {
803 return self.unexpected();
804 }
805 Ok(self.next())
806 }
807 fn enter(&mut self, at: usize) -> R<()> {
808 self.depth += 1;
809 if self.depth > MAX_DEPTH {
810 return Err(format!("at {at}: expression is nested too deeply"));
811 }
812 Ok(())
813 }
814
815 fn parse_all(&mut self) -> R<Node> {
816 let node = self.expr()?;
817 if self.peek().t != Tok::End {
818 return self.unexpected();
819 }
820 Ok(node)
821 }
822
823 fn expr(&mut self) -> R<Node> {
824 crate::deep(|| self.expr_inner())
825 }
826
827 fn expr_inner(&mut self) -> R<Node> {
828 let at = self.peek().at;
829 self.enter(at)?;
830 let test = self.or()?;
831 let mut out = test;
832 if self.is_op("?") {
833 self.next();
834 let then = self.expr()?;
835 self.expect(":")?;
836 let otherwise = self.expr()?;
837 out = Node::Cond(Box::new(out), Box::new(then), Box::new(otherwise), at);
838 }
839 self.depth -= 1;
840 Ok(out)
841 }
842
843 fn binary(&mut self, ops: &[&'static str], operand: fn(&mut Parser) -> R<Node>) -> R<Node> {
844 let mut a = operand(self)?;
845 loop {
846 let t = self.peek().clone();
847 let Tok::Op(op) = t.t else { return Ok(a) };
848 let Some(op) = ops.iter().find(|o| **o == op) else { return Ok(a) };
849 self.next();
850 let b = operand(self)?;
851 a = Node::Bin(op, Box::new(a), Box::new(b), t.at);
852 }
853 }
854 fn or(&mut self) -> R<Node> {
855 self.binary(&["||"], Parser::and)
856 }
857 fn and(&mut self) -> R<Node> {
858 self.binary(&["&&"], Parser::rel)
859 }
860 fn rel(&mut self) -> R<Node> {
861 self.binary(&["==", "!=", "<", "<=", ">", ">="], Parser::add)
862 }
863 fn add(&mut self) -> R<Node> {
864 self.binary(&["+", "-"], Parser::mul)
865 }
866 fn mul(&mut self) -> R<Node> {
867 self.binary(&["*", "/", "%"], Parser::unary)
868 }
869
870 fn unary(&mut self) -> R<Node> {
871 let t = self.peek().clone();
872 if let Tok::Op(op @ ("!" | "-")) = t.t {
873 self.next();
874 self.enter(t.at)?;
875 let arg = crate::deep(|| self.unary())?;
876 self.depth -= 1;
877 return Ok(Node::Unary(op.chars().next().unwrap_or('!'), Box::new(arg), t.at));
878 }
879 self.postfix()
880 }
881
882 fn postfix(&mut self) -> R<Node> {
883 let mut node = self.primary()?;
884 loop {
885 if self.is_op(".") {
886 self.next();
887 let name = self.peek().clone();
888 let Tok::Id(field) = name.t.clone() else { return self.unexpected() };
889 self.next();
890 if self.is_op("(") {
891 if !MACROS.contains(&field.as_str()) {
892 if let Some(dotted) = dotted_name(&node) {
893 let full = format!("{dotted}.{field}");
894 let effect = function_effect(&full);
895 if effect != FunctionEffect::Pure {
896 return Err(format!("at {}: {}: an expression is pure, use an action step", root_at(&node), effect_words(&full, effect)));
897 }
898 }
899 return Err(format!("at {}: unknown method {field}", name.at));
900 }
901 if self.in_body > 0 {
902 return Err(format!("at {}: nested comprehensions are not allowed", name.at));
903 }
904 self.next();
905 let v = self.peek().clone();
906 let comma = matches!(self.tokens.get(self.i + 1).map(|t| &t.t), Some(Tok::Op(",")));
907 let Tok::Id(var) = v.t.clone() else { return Err(format!("at {}: {field} needs a name first", v.at)) };
908 if !comma {
909 return Err(format!("at {}: {field} needs a name first", v.at));
910 }
911 self.next();
912 self.next();
913 self.in_body += 1;
914 let body = self.expr()?;
915 self.in_body -= 1;
916 self.expect(")")?;
917 node = Node::Macro(field, Box::new(node), var, Box::new(body), name.at);
918 } else {
919 node = Node::Field(Box::new(node), field, name.at);
920 }
921 continue;
922 }
923 if self.is_op("[") {
924 let at = self.next().at;
925 let index = self.expr()?;
926 self.expect("]")?;
927 node = Node::Index(Box::new(node), Box::new(index), at);
928 continue;
929 }
930 return Ok(node);
931 }
932 }
933
934 fn list(&mut self, close: &str) -> R<Vec<Node>> {
935 let mut items = Vec::new();
936 if self.is_op(close) {
937 self.next();
938 return Ok(items);
939 }
940 loop {
941 items.push(self.expr()?);
942 if self.is_op(",") {
943 self.next();
944 if self.is_op(close) && close == "]" {
945 self.next();
946 return Ok(items);
947 }
948 continue;
949 }
950 self.expect(close)?;
951 return Ok(items);
952 }
953 }
954
955 fn primary(&mut self) -> R<Node> {
956 let t = self.peek().clone();
957 match t.t {
958 Tok::Num(v) => {
959 self.next();
960 if !v.is_finite() {
961 return Err(format!("at {}: number out of range", t.at));
962 }
963 Ok(Node::Lit(json::number(v), t.at))
964 }
965 Tok::Str(s) => {
966 self.next();
967 Ok(Node::Lit(Value::String(s), t.at))
968 }
969 Tok::Id(name) => {
970 self.next();
971 match name.as_str() {
972 "true" | "false" => return Ok(Node::Lit(Value::Bool(name == "true"), t.at)),
973 "null" => return Ok(Node::Lit(Value::Null, t.at)),
974 _ => {}
975 }
976 if name == "has" && self.is_op("(") {
978 self.next();
979 let arg = self.expr()?;
980 self.expect(")")?;
981 if !matches!(arg, Node::Field(..)) {
982 return Err(format!("at {}: has needs a field: has(a.b)", t.at));
983 }
984 return Ok(Node::Has(Box::new(arg), t.at));
985 }
986 if self.is_op("(") {
987 if def(&name).is_none() {
988 let effect = function_effect(&name);
989 if effect != FunctionEffect::Pure {
990 return Err(format!("at {}: {}: an expression is pure, use an action step", t.at, effect_words(&name, effect)));
991 }
992 return Err(format!("at {}: unknown function {name}", t.at));
993 }
994 self.next();
995 let args = self.list(")")?;
996 return Ok(Node::Call(name, args, t.at));
997 }
998 Ok(Node::Id(name, t.at))
999 }
1000 Tok::Op("(") => {
1001 self.next();
1002 let inner = self.expr()?;
1003 self.expect(")")?;
1004 Ok(inner)
1005 }
1006 Tok::Op("[") => {
1007 self.next();
1008 Ok(Node::List(self.list("]")?, t.at))
1009 }
1010 _ => self.unexpected(),
1011 }
1012 }
1013}
1014
1015fn dotted_name(node: &Node) -> Option<String> {
1017 match node {
1018 Node::Id(name, _) => Some(name.clone()),
1019 Node::Field(of, name, _) => dotted_name(of).map(|of| format!("{of}.{name}")),
1020 _ => None,
1021 }
1022}
1023
1024pub fn path_text(node: &Node) -> Option<String> {
1027 match node {
1028 Node::Id(name, _) => Some(name.clone()),
1029 Node::Field(of, name, _) => path_text(of).map(|of| format!("{of}.{name}")),
1030 Node::Index(of, index, _) => {
1031 let of = path_text(of)?;
1032 match literal(index) {
1033 Some(Value::Number(n)) => Some(format!("{of}[{}]", num_text(n.as_f64().unwrap_or(0.0)))),
1034 Some(Value::String(k)) => Some(format!("{of}['{k}']")),
1035 _ => None,
1036 }
1037 }
1038 _ => None,
1039 }
1040}
1041
1042fn node_at(node: &Node) -> usize {
1044 match node {
1045 Node::Lit(_, at) | Node::Id(_, at) | Node::List(_, at) | Node::Field(_, _, at) | Node::Index(_, _, at) | Node::Unary(_, _, at) | Node::Bin(_, _, _, at) | Node::Cond(_, _, _, at) | Node::Call(_, _, at) | Node::Macro(_, _, _, _, at) | Node::Has(_, at) => *at,
1046 }
1047}
1048
1049fn root_at(node: &Node) -> usize {
1050 match node {
1051 Node::Field(of, _, _) => root_at(of),
1052 _ => node_at(node),
1053 }
1054}
1055
1056pub fn parse_expression(source: &str) -> Result<Node, String> {
1059 let cs: Vec<char> = source.chars().collect();
1060 if cs.len() > MAX_LENGTH {
1061 return Err(format!("expression is longer than {MAX_LENGTH} characters"));
1062 }
1063 Parser { tokens: tokenize(&cs)?, i: 0, depth: 0, in_body: 0 }.parse_all()
1064}
1065
1066struct Scope<'a> {
1071 name: &'a str,
1072 value: &'a Value,
1073 parent: Option<&'a Scope<'a>>,
1074}
1075
1076struct Evaluator<'v> {
1077 vars: &'v Map<String, Value>,
1078 env: Env<'v>,
1079 steps: usize,
1080 max_steps: usize,
1081}
1082
1083impl Evaluator<'_> {
1084 fn step(&mut self) -> R<()> {
1085 self.steps += 1;
1086 if self.steps > self.max_steps {
1087 return Err("step limit exceeded".into());
1088 }
1089 Ok(())
1090 }
1091
1092 fn ev(&mut self, node: &Node, scope: Option<&Scope>) -> R<Value> {
1093 crate::deep(|| self.ev_inner(node, scope))
1094 }
1095
1096 fn ev_inner(&mut self, node: &Node, scope: Option<&Scope>) -> R<Value> {
1097 self.step()?;
1098 match node {
1099 Node::Lit(v, _) => Ok(v.clone()),
1100 Node::List(items, _) => items.iter().map(|i| self.ev(i, scope)).collect::<R<Vec<_>>>().map(Value::Array),
1101 Node::Id(name, _) => {
1102 let mut s = scope;
1103 while let Some(sc) = s {
1104 if sc.name == name {
1105 return Ok(sc.value.clone());
1106 }
1107 s = sc.parent;
1108 }
1109 self.vars.get(name).cloned().ok_or_else(|| format!("unknown name {name}"))
1110 }
1111 Node::Field(of, name, _) => match self.ev(of, scope)? {
1112 Value::Null => Ok(Value::Null),
1113 Value::Object(m) => Ok(m.get(name).cloned().unwrap_or(Value::Null)),
1114 other => Err(format!("no field {name} on {}", type_name(&other))),
1115 },
1116 Node::Has(field, _) => {
1117 let Node::Field(receiver, name, _) = field.as_ref() else { return Err("has needs a field: has(a.b)".into()) };
1118 match self.ev(receiver, scope)? {
1119 Value::Null => Ok(Value::Bool(false)),
1120 Value::Object(m) => Ok(Value::Bool(m.get(name).is_some_and(|v| !v.is_null()))),
1121 other => Err(format!("no field {name} on {}", type_name(&other))),
1122 }
1123 }
1124 Node::Index(of, index, _) => {
1125 let of = self.ev(of, scope)?;
1126 let index = self.ev(index, scope)?;
1127 match of {
1128 Value::Array(list) => {
1129 let Value::Number(n) = &index else { return Err(format!("a list index is a number, got {}", type_name(&index))) };
1130 let n = n.as_f64().unwrap_or(0.0);
1131 if n.fract() != 0.0 {
1132 return Err(format!("a list index is a whole number, got {}", num_text(n)));
1133 }
1134 if n < 0.0 || n >= list.len() as f64 {
1135 return Err(format!("index {} out of range", num_text(n)));
1136 }
1137 Ok(list[n as usize].clone())
1138 }
1139 Value::Object(m) => {
1140 let Value::String(k) = &index else { return Err(format!("a map key is a string, got {}", type_name(&index))) };
1141 Ok(m.get(k).cloned().unwrap_or(Value::Null))
1142 }
1143 Value::Null => Ok(Value::Null),
1144 other => Err(format!("cannot index {}", type_name(&other))),
1145 }
1146 }
1147 Node::Unary(op, arg, _) => {
1148 let v = self.ev(arg, scope)?;
1149 if *op == '-' {
1150 let Value::Number(n) = &v else { return Err(format!("cannot apply - to {}", type_name(&v))) };
1151 return Ok(json::number(num(-n.as_f64().unwrap_or(0.0))?));
1152 }
1153 let Value::Bool(b) = v else { return Err(format!("cannot apply ! to {}", type_name(&v))) };
1154 Ok(Value::Bool(!b))
1155 }
1156 Node::Cond(test, then, otherwise, _) => {
1157 let t = self.ev(test, scope)?;
1158 let Value::Bool(t) = t else { return Err(format!("cannot use {} as a condition", type_name(&t))) };
1159 self.ev(if t { then } else { otherwise }, scope)
1160 }
1161 Node::Bin(op, a, b, _) => self.binary(op, a, b, scope),
1162 Node::Call(name, args, _) => {
1163 let args = args.iter().map(|a| self.ev(a, scope)).collect::<R<Vec<_>>>()?;
1164 let d = def(name).ok_or_else(|| format!("unknown function {name}"))?;
1165 if let Some(e) = arity_error(name, &d, args.len()) {
1166 return Err(e);
1167 }
1168 run(name, &args, &self.env)
1169 }
1170 Node::Macro(f, of, var, body, _) => {
1171 let of = self.ev(of, scope)?;
1172 let Value::Array(list) = of else { return Err(format!("{f} needs a list, got {}", type_name(&of))) };
1173 let mut out = Vec::new();
1174 for item in &list {
1175 self.step()?;
1176 let inner = Scope { name: var, value: item, parent: scope };
1177 let v = self.ev(body, Some(&inner))?;
1178 if f == "map" {
1179 out.push(v);
1180 continue;
1181 }
1182 let Value::Bool(b) = v else { return Err(format!("{f} needs a bool, got {}", type_name(&v))) };
1183 match f.as_str() {
1184 "filter" if b => out.push(item.clone()),
1185 "all" if !b => return Ok(Value::Bool(false)),
1186 "exists" if b => return Ok(Value::Bool(true)),
1187 _ => {}
1188 }
1189 }
1190 Ok(match f.as_str() {
1191 "all" => Value::Bool(true),
1192 "exists" => Value::Bool(false),
1193 _ => Value::Array(out),
1194 })
1195 }
1196 }
1197 }
1198
1199 fn binary(&mut self, op: &str, a: &Node, b: &Node, scope: Option<&Scope>) -> R<Value> {
1200 if op == "&&" || op == "||" {
1201 let x = self.ev(a, scope)?;
1202 let Value::Bool(x) = x else { return Err(format!("cannot apply {op} to {}", type_name(&x))) };
1203 if (op == "&&" && !x) || (op == "||" && x) {
1204 return Ok(Value::Bool(x));
1205 }
1206 let y = self.ev(b, scope)?;
1207 let Value::Bool(_) = y else { return Err(format!("cannot apply {op} to {}", type_name(&y))) };
1208 return Ok(y);
1209 }
1210 let x = self.ev(a, scope)?;
1211 let y = self.ev(b, scope)?;
1212 match op {
1213 "==" => Ok(Value::Bool(deep_equal(&x, &y))),
1214 "!=" => Ok(Value::Bool(!deep_equal(&x, &y))),
1215 "<" | "<=" | ">" | ">=" => {
1216 let c = match (&x, &y) {
1217 (Value::Number(p), Value::Number(q)) => p.as_f64().partial_cmp(&q.as_f64()).unwrap_or(std::cmp::Ordering::Equal),
1218 (Value::String(p), Value::String(q)) => utf16_cmp(p, q),
1219 _ => return Err(format!("cannot compare {} and {}", type_name(&x), type_name(&y))),
1220 };
1221 use std::cmp::Ordering::*;
1222 Ok(Value::Bool(match op {
1223 "<" => c == Less,
1224 "<=" => c != Greater,
1225 ">" => c == Greater,
1226 _ => c != Less,
1227 }))
1228 }
1229 "+" => match (x, y) {
1230 (Value::Number(p), Value::Number(q)) => Ok(json::number(num(p.as_f64().unwrap_or(0.0) + q.as_f64().unwrap_or(0.0))?)),
1231 (Value::String(p), Value::String(q)) => Ok(Value::String(p + &q)),
1232 (Value::Array(mut p), Value::Array(q)) => {
1233 p.extend(q);
1234 Ok(Value::Array(p))
1235 }
1236 (x, y) => Err(format!("cannot apply + to {} and {}", type_name(&x), type_name(&y))),
1237 },
1238 _ => {
1239 let (Value::Number(p), Value::Number(q)) = (&x, &y) else { return Err(format!("cannot apply {op} to {} and {}", type_name(&x), type_name(&y))) };
1240 let (p, q) = (p.as_f64().unwrap_or(0.0), q.as_f64().unwrap_or(0.0));
1241 match op {
1242 "-" => Ok(json::number(num(p - q)?)),
1243 "*" => Ok(json::number(num(p * q)?)),
1244 _ => {
1245 if q == 0.0 {
1246 return Err(if op == "/" { "division by zero" } else { "modulo by zero" }.into());
1247 }
1248 Ok(json::number(num(if op == "/" { p / q } else { p % q })?))
1249 }
1250 }
1251 }
1252 }
1253 }
1254}
1255
1256pub fn evaluate_expression(source: &str, vars: &Map<String, Value>, env: &ExpressionEnv) -> Result<Value, String> {
1259 let ast = parse_expression(source)?;
1260 evaluate_node(&ast, vars, env)
1261}
1262
1263pub fn evaluate_node<'a>(ast: &Node, vars: &'a Map<String, Value>, env: &ExpressionEnv<'a>) -> Result<Value, String> {
1265 let mut e = Evaluator { vars, env: Env::of(env), steps: 0, max_steps: env.max_steps.unwrap_or(MAX_STEPS) };
1266 e.ev(ast, None)
1267}
1268
1269pub fn evaluate_counted<'a>(ast: &Node, vars: &'a Map<String, Value>, env: &ExpressionEnv<'a>) -> (Result<Value, String>, usize) {
1272 let mut e = Evaluator { vars, env: Env::of(env), steps: 0, max_steps: env.max_steps.unwrap_or(MAX_STEPS) };
1273 let result = e.ev(ast, None);
1274 (result, e.steps)
1275}
1276
1277pub fn is_optional(t: &Value) -> bool {
1284 matches!(t, Value::Object(m) if m.len() == 1 && m.contains_key("?"))
1285}
1286
1287pub fn optional_type(t: Value) -> Value {
1289 if is_optional(&t) || is(&t, "null") {
1290 return t;
1291 }
1292 let mut m = Map::new();
1293 m.insert("?".into(), t);
1294 Value::Object(m)
1295}
1296
1297pub fn required_type(t: &Value) -> Value {
1299 match t {
1300 Value::Object(m) if is_optional(t) => m["?"].clone(),
1301 _ => t.clone(),
1302 }
1303}
1304
1305pub fn type_display(t: &Value) -> String {
1308 match t {
1309 Value::Object(m) if is_optional(t) => format!("{}?", type_display(&m["?"])),
1310 Value::String(s) => s.clone(),
1311 Value::Array(_) => "list".into(),
1312 _ => "map".into(),
1313 }
1314}
1315
1316fn is_row(t: &Value) -> bool {
1317 t.is_object() && !is_optional(t)
1318}
1319fn is_textual(t: &Value) -> bool {
1320 matches!(t.as_str(), Some("string" | "date" | "timestamp"))
1321}
1322fn loose(t: &Value) -> bool {
1323 matches!(t.as_str(), Some("dyn" | "null"))
1324}
1325fn is(t: &Value, name: &str) -> bool {
1326 t.as_str() == Some(name)
1327}
1328fn ty(name: &str) -> Value {
1329 Value::String(name.into())
1330}
1331
1332fn accepts(param: Param, t: &Value) -> bool {
1333 if param == Param::Any || loose(t) {
1334 return true;
1335 }
1336 match param {
1337 Param::Number => is(t, "number"),
1338 Param::Text | Param::Temporal => is_textual(t),
1339 Param::Sized => is(t, "string") || is(t, "map") || t.is_array() || is_row(t),
1340 Param::Any => true,
1341 }
1342}
1343
1344fn param_name(p: Param) -> &'static str {
1345 match p {
1346 Param::Number => "a number",
1347 Param::Text => "a string",
1348 Param::Temporal => "a date",
1349 Param::Sized => "a list, string or map",
1350 Param::Any => "a value",
1351 }
1352}
1353
1354struct TypeScope<'a> {
1355 name: &'a str,
1356 ty: Value,
1357 parent: Option<&'a TypeScope<'a>>,
1358}
1359
1360fn literal(node: &Node) -> Option<&Value> {
1361 match node {
1362 Node::Lit(v, _) => Some(v),
1363 _ => None,
1364 }
1365}
1366
1367fn guards(node: &Node) -> (Vec<String>, Vec<String>) {
1371 match node {
1372 Node::Has(field, _) => (path_text(field).into_iter().collect(), Vec::new()),
1373 Node::Bin(op @ ("==" | "!="), a, b, _) => {
1374 let side = if matches!(a.as_ref(), Node::Lit(Value::Null, _)) {
1375 Some(b)
1376 } else if matches!(b.as_ref(), Node::Lit(Value::Null, _)) {
1377 Some(a)
1378 } else {
1379 None
1380 };
1381 match side.and_then(|s| path_text(s)) {
1382 None => (Vec::new(), Vec::new()),
1383 Some(p) if *op == "!=" => (vec![p], Vec::new()),
1384 Some(p) => (Vec::new(), vec![p]),
1385 }
1386 }
1387 Node::Bin("&&", a, b, _) => {
1388 let (mut yes, _) = guards(a);
1389 yes.extend(guards(b).0);
1390 (yes, Vec::new())
1391 }
1392 Node::Bin("||", a, b, _) => {
1393 let (_, mut no) = guards(a);
1394 no.extend(guards(b).1);
1395 (Vec::new(), no)
1396 }
1397 Node::Unary('!', arg, _) => {
1398 let (yes, no) = guards(arg);
1399 (no, yes)
1400 }
1401 _ => (Vec::new(), Vec::new()),
1402 }
1403}
1404
1405fn missing<T>(node: &Node) -> R<T> {
1407 let p = path_text(node).unwrap_or_else(|| format!("the value at {}", node_at(node)));
1408 Err(format!("{p} may be missing: guard it with has(), a null check or orElse()"))
1409}
1410
1411struct Checker<'t> {
1412 types: &'t Map<String, Value>,
1413 present: std::cell::RefCell<Vec<String>>,
1415}
1416
1417impl Checker<'_> {
1418 fn within(&self, paths: Vec<String>, node: &Node, scope: Option<&TypeScope>) -> R<Value> {
1420 let before = self.present.borrow().clone();
1421 self.present.borrow_mut().extend(paths);
1422 let out = self.tc(node, scope);
1423 *self.present.borrow_mut() = before;
1424 out
1425 }
1426
1427 fn need(&self, node: &Node, scope: Option<&TypeScope>) -> R<Value> {
1429 let t = self.tc(node, scope)?;
1430 if is_optional(&t) {
1431 return missing(node);
1432 }
1433 Ok(t)
1434 }
1435
1436 fn tc(&self, node: &Node, scope: Option<&TypeScope>) -> R<Value> {
1437 let t = crate::deep(|| self.tc_inner(node, scope))?;
1438 let p = if matches!(node, Node::Id(..) | Node::Field(..) | Node::Index(..)) { path_text(node) } else { None };
1439 Ok(match p {
1440 Some(p) if self.present.borrow().contains(&p) => required_type(&t),
1441 _ => t,
1442 })
1443 }
1444
1445 fn tc_inner(&self, node: &Node, scope: Option<&TypeScope>) -> R<Value> {
1446 match node {
1447 Node::Lit(v, _) => Ok(ty(type_name(v))),
1448 Node::List(items, _) => {
1449 let items = items.iter().map(|i| self.tc(i, scope)).collect::<R<Vec<_>>>()?;
1450 Ok(Value::Array(vec![match items.first() {
1451 None => ty("dyn"),
1452 Some(first) if items.iter().all(|t| t == first) => first.clone(),
1453 Some(_) => ty("dyn"),
1454 }]))
1455 }
1456 Node::Id(name, _) => {
1457 let mut s = scope;
1458 while let Some(sc) = s {
1459 if sc.name == name {
1460 return Ok(sc.ty.clone());
1461 }
1462 s = sc.parent;
1463 }
1464 self.types.get(name).cloned().ok_or_else(|| format!("unknown name {name}"))
1465 }
1466 Node::Field(..) | Node::Has(..) => {
1467 let has = matches!(node, Node::Has(..));
1468 let field = match node {
1469 Node::Has(field, _) => field.as_ref(),
1470 _ => node,
1471 };
1472 let Node::Field(of, name, _) = field else { return Err("has needs a field: has(a.b)".into()) };
1473 let raw = self.tc(of, scope)?;
1474 let of = required_type(&raw);
1475 let t = if matches!(of.as_str(), Some("dyn" | "map" | "null")) {
1476 ty("dyn")
1477 } else if let Value::Object(fields) = &of {
1478 fields.get(name).cloned().ok_or_else(|| format!("unknown field {name}"))?
1479 } else {
1480 return Err(format!("no field {name} on {}", type_display(&of)));
1481 };
1482 if has {
1483 return Ok(ty("bool"));
1484 }
1485 Ok(if is_optional(&raw) { optional_type(t) } else { t })
1486 }
1487 Node::Index(of, index, _) => {
1488 let raw = self.tc(of, scope)?;
1489 let of_t = required_type(&raw);
1490 let index_t = self.need(index, scope)?;
1491 let wrap = |t: Value| if is_optional(&raw) { optional_type(t) } else { t };
1492 if let Value::Array(item) = &of_t {
1493 if !loose(&index_t) && !is(&index_t, "number") {
1494 return Err(format!("a list index is a number, got {}", type_display(&index_t)));
1495 }
1496 return Ok(wrap(item.first().cloned().unwrap_or_else(|| ty("dyn"))));
1497 }
1498 if is_row(&of_t) || is(&of_t, "map") {
1499 if !loose(&index_t) && !is_textual(&index_t) {
1500 return Err(format!("a map key is a string, got {}", type_display(&index_t)));
1501 }
1502 if let (Value::Object(fields), Some(Value::String(key))) = (&of_t, literal(index)) {
1503 return fields.get(key).cloned().map(wrap).ok_or_else(|| format!("unknown field {key}"));
1504 }
1505 return Ok(wrap(ty("dyn")));
1506 }
1507 if loose(&of_t) {
1508 return Ok(wrap(ty("dyn")));
1509 }
1510 Err(format!("cannot index {}", type_display(&of_t)))
1511 }
1512 Node::Unary(op, arg, _) => {
1513 let t = self.need(arg, scope)?;
1514 let want = if *op == '-' { "number" } else { "bool" };
1515 if !loose(&t) && !is(&t, want) {
1516 return Err(format!("cannot apply {op} to {}", type_display(&t)));
1517 }
1518 Ok(ty(want))
1519 }
1520 Node::Cond(test, then, otherwise, _) => {
1521 let t = self.need(test, scope)?;
1522 if !loose(&t) && !is(&t, "bool") {
1523 return Err(format!("cannot use {} as a condition", type_display(&t)));
1524 }
1525 let (yes, no) = guards(test);
1526 let a = self.within(yes, then, scope)?;
1527 let b = self.within(no, otherwise, scope)?;
1528 Ok(if a == b {
1529 a
1530 } else if is(&a, "null") {
1531 optional_type(b)
1532 } else if is(&b, "null") {
1533 optional_type(a)
1534 } else if required_type(&a) == required_type(&b) {
1535 optional_type(required_type(&a))
1536 } else if is_optional(&a) || is_optional(&b) {
1537 optional_type(ty("dyn"))
1538 } else {
1539 ty("dyn")
1540 })
1541 }
1542 Node::Bin(op, a, b, _) => self.binary(op, a, b, scope),
1543 Node::Call(name, args, _) => self.call(name, args, scope),
1544 Node::Macro(f, of, var, body, _) => {
1545 let of = self.need(of, scope)?;
1546 if !of.is_array() && !loose(&of) {
1547 return Err(format!("{f} needs a list, got {}", type_display(&of)));
1548 }
1549 let item = of.as_array().and_then(|a| a.first().cloned()).unwrap_or_else(|| ty("dyn"));
1550 let inner = TypeScope { name: var, ty: item, parent: scope };
1551 let body_t = self.tc(body, Some(&inner))?;
1552 if f == "map" {
1553 return Ok(Value::Array(vec![body_t]));
1554 }
1555 if is_optional(&body_t) {
1556 return missing(body);
1557 }
1558 if !loose(&body_t) && !is(&body_t, "bool") {
1559 return Err(format!("{f} needs a bool, got {}", type_display(&body_t)));
1560 }
1561 Ok(if f == "filter" {
1562 if of.is_array() {
1563 of
1564 } else {
1565 Value::Array(vec![ty("dyn")])
1566 }
1567 } else {
1568 ty("bool")
1569 })
1570 }
1571 }
1572 }
1573
1574 fn binary(&self, op: &str, a: &Node, b: &Node, scope: Option<&TypeScope>) -> R<Value> {
1575 if op == "&&" || op == "||" {
1576 let x = self.need(a, scope)?;
1577 let (yes, no) = guards(a);
1578 let y = self.within(if op == "&&" { yes } else { no }, b, scope)?;
1579 if is_optional(&y) {
1580 return missing(b);
1581 }
1582 for t in [&x, &y] {
1583 if !loose(t) && !is(t, "bool") {
1584 return Err(format!("cannot apply {op} to {}", type_display(t)));
1585 }
1586 }
1587 return Ok(ty("bool"));
1588 }
1589 if op == "==" || op == "!=" {
1590 let x = required_type(&self.tc(a, scope)?);
1592 let y = required_type(&self.tc(b, scope)?);
1593 let (dx, dy) = (type_display(&x), type_display(&y));
1594 let comparable = loose(&x) || loose(&y) || (is_textual(&x) && is_textual(&y)) || dx == dy;
1595 if !comparable {
1596 return Err(format!("cannot compare {dx} and {dy}"));
1597 }
1598 return Ok(ty("bool"));
1599 }
1600 let x = self.need(a, scope)?;
1601 let y = self.need(b, scope)?;
1602 let (dx, dy) = (type_display(&x), type_display(&y));
1603 let numeric = |t: &Value| is(t, "number") || loose(t);
1604 match op {
1605 "<" | "<=" | ">" | ">=" => {
1606 let ok = |t: &Value| loose(t) || is(t, "number") || is_textual(t);
1607 if !ok(&x) || !ok(&y) || (is(&x, "number") && is_textual(&y)) || (is(&y, "number") && is_textual(&x)) {
1608 return Err(format!("cannot compare {dx} and {dy}"));
1609 }
1610 Ok(ty("bool"))
1611 }
1612 "+" => {
1613 if loose(&x) && loose(&y) {
1614 return Ok(ty("dyn"));
1615 }
1616 if numeric(&x) && numeric(&y) {
1617 return Ok(ty("number"));
1618 }
1619 if (is_textual(&x) || loose(&x)) && (is_textual(&y) || loose(&y)) {
1620 return Ok(ty("string"));
1621 }
1622 if (x.is_array() || loose(&x)) && (y.is_array() || loose(&y)) {
1623 return Ok(if x.is_array() && y.is_array() && x == y { x } else { Value::Array(vec![ty("dyn")]) });
1624 }
1625 Err(format!("cannot apply + to {dx} and {dy}"))
1626 }
1627 _ => {
1628 if !(numeric(&x) && numeric(&y)) {
1629 return Err(format!("cannot apply {op} to {dx} and {dy}"));
1630 }
1631 if (op == "/" || op == "%") && literal(b).is_some_and(|v| v.as_f64() == Some(0.0)) {
1632 return Err(if op == "/" { "division by zero" } else { "modulo by zero" }.into());
1633 }
1634 Ok(ty("number"))
1635 }
1636 }
1637 }
1638
1639 fn call(&self, name: &str, args: &[Node], scope: Option<&TypeScope>) -> R<Value> {
1640 let d = def(name).ok_or_else(|| format!("unknown function {name}"))?;
1641 if let Some(e) = arity_error(name, &d, args.len()) {
1642 return Err(e);
1643 }
1644 let types = args.iter().enumerate().map(|(i, a)| if name == "orElse" && i == 0 { self.tc(a, scope) } else { self.need(a, scope) }).collect::<R<Vec<_>>>()?;
1646 for (i, t) in types.iter().enumerate() {
1647 let param = d.params[i.min(d.params.len() - 1)];
1648 if !accepts(param, &required_type(t)) {
1649 return Err(format!("{name} needs {}, got {}", param_name(param), type_display(t)));
1650 }
1651 }
1652 if name == "orElse" {
1653 let (x, y) = (required_type(&types[0]), &types[1]);
1654 if !loose(&x) && !loose(y) && !(is_textual(&x) && is_textual(y)) && type_display(&x) != type_display(y) {
1655 return Err(format!("orElse needs a {} default, got {}", type_display(&x), type_display(y)));
1656 }
1657 }
1658 let lit = |i: usize| args.get(i).and_then(literal);
1659 let unit_at = match name {
1660 "dateAdd" | "dateDiff" => Some(2),
1661 "startOf" | "endOf" => Some(1),
1662 _ => None,
1663 };
1664 if let Some(Value::String(unit)) = unit_at.and_then(lit) {
1665 let units = if unit_at == Some(2) { DATE_UNITS } else { CALENDAR_UNITS };
1666 if !units.contains(&unit.as_str()) {
1667 return Err(format!("unknown unit {unit}"));
1668 }
1669 if name == "dateAdd" && is(&types[0], "date") && unit_ms(unit).is_some() {
1670 return Err(format!("unit {unit} needs a timestamp"));
1671 }
1672 }
1673 if name == "dateAdd" {
1674 if let Some(n) = lit(1).and_then(Value::as_f64) {
1675 if n.fract() != 0.0 {
1676 return Err(format!("dateAdd needs a whole number, got {}", num_text(n)));
1677 }
1678 }
1679 }
1680 if name == "round" && args.len() > 1 {
1681 if let Some(d) = lit(1).and_then(Value::as_f64) {
1682 if d.fract() != 0.0 || !(0.0..=10.0).contains(&d) {
1683 return Err("round needs 0 to 10 digits".into());
1684 }
1685 }
1686 }
1687 if name == "div" && lit(1).and_then(Value::as_f64) == Some(0.0) {
1688 return Err("division by zero".into());
1689 }
1690 if name == "mod" && lit(1).and_then(Value::as_f64) == Some(0.0) {
1691 return Err("modulo by zero".into());
1692 }
1693 if name == "formatDuration" {
1694 if let Some(Value::String(f)) = lit(1) {
1695 if !DURATION_FORMATS.contains(&f.as_str()) {
1696 return Err(format!("unknown duration format {f}"));
1697 }
1698 }
1699 }
1700 Ok(match name {
1701 "dateAdd" | "startOf" | "endOf" => {
1702 if is(&types[0], "date") || is(&types[0], "timestamp") {
1703 types[0].clone()
1704 } else {
1705 ty("dyn")
1706 }
1707 }
1708 "today" => ty("date"),
1709 "now" => ty("timestamp"),
1710 "formatDuration" => ty("string"),
1711 "orElse" => {
1712 let x = required_type(&types[0]);
1713 if is(&x, "null") {
1714 required_type(&types[1])
1715 } else {
1716 x
1717 }
1718 }
1719 _ => ty("number"),
1720 })
1721 }
1722}
1723
1724pub fn check_expression(source: &str, types: &Map<String, Value>) -> Result<Value, String> {
1729 let ast = parse_expression(source)?;
1730 Checker { types, present: std::cell::RefCell::new(Vec::new()) }.tc(&ast, None)
1731}