1use std::any::Any;
21use std::str::FromStr;
22
23use chrono::{DateTime, NaiveDate, NaiveDateTime, NaiveTime, TimeZone, Utc};
24use ordered_float::OrderedFloat;
25use serde_json::{Map as JsonMap, Number, Value as JsonValue};
26use uuid::Uuid;
27
28use super::Map;
29use super::decimal_utils::{
30 decimal_from_str_exact, decimal_mantissa, decimal_rescale, try_decimal_from_i128_with_scale,
31};
32use super::primitive::PrimitiveLiteral;
33use super::struct_value::Struct;
34use super::temporal::{date, time, timestamp, timestamptz};
35use crate::error::Result;
36use crate::spec::datatypes::{PrimitiveType, Type};
37use crate::{Error, ErrorKind};
38
39#[derive(Clone, Debug, PartialEq, Eq, Hash)]
41pub enum Literal {
42 Primitive(PrimitiveLiteral),
44 Struct(Struct),
48 List(Vec<Option<Literal>>),
52 Map(Map),
56}
57
58impl Literal {
59 pub fn bool<T: Into<bool>>(t: T) -> Self {
69 Self::Primitive(PrimitiveLiteral::Boolean(t.into()))
70 }
71
72 pub fn bool_from_str<S: AsRef<str>>(s: S) -> Result<Self> {
83 let v = s.as_ref().parse::<bool>().map_err(|e| {
84 Error::new(ErrorKind::DataInvalid, "Can't parse string to bool.").with_source(e)
85 })?;
86 Ok(Self::Primitive(PrimitiveLiteral::Boolean(v)))
87 }
88
89 pub fn int<T: Into<i32>>(t: T) -> Self {
99 Self::Primitive(PrimitiveLiteral::Int(t.into()))
100 }
101
102 pub fn long<T: Into<i64>>(t: T) -> Self {
112 Self::Primitive(PrimitiveLiteral::Long(t.into()))
113 }
114
115 pub fn float<T: Into<f32>>(t: T) -> Self {
129 Self::Primitive(PrimitiveLiteral::Float(OrderedFloat(t.into())))
130 }
131
132 pub fn double<T: Into<f64>>(t: T) -> Self {
146 Self::Primitive(PrimitiveLiteral::Double(OrderedFloat(t.into())))
147 }
148
149 pub fn date(days: i32) -> Self {
151 Self::Primitive(PrimitiveLiteral::Int(days))
152 }
153
154 pub fn date_from_str<S: AsRef<str>>(s: S) -> Result<Self> {
166 let t = s.as_ref().parse::<NaiveDate>().map_err(|e| {
167 Error::new(
168 ErrorKind::DataInvalid,
169 format!("Can't parse date from string: {}", s.as_ref()),
170 )
171 .with_source(e)
172 })?;
173
174 Ok(Self::date(date::date_from_naive_date(t)))
175 }
176
177 pub fn date_from_ymd(year: i32, month: u32, day: u32) -> Result<Self> {
190 let t = NaiveDate::from_ymd_opt(year, month, day).ok_or_else(|| {
191 Error::new(
192 ErrorKind::DataInvalid,
193 format!("Can't create date from year: {year}, month: {month}, day: {day}"),
194 )
195 })?;
196
197 Ok(Self::date(date::date_from_naive_date(t)))
198 }
199
200 pub fn time(value: i64) -> Self {
202 Self::Primitive(PrimitiveLiteral::Long(value))
203 }
204
205 fn time_from_naive_time(t: NaiveTime) -> Self {
207 let duration = t - date::unix_epoch().time();
208 let micro_secs = duration.num_microseconds().unwrap();
210
211 Literal::time(micro_secs)
212 }
213
214 pub fn time_from_str<S: AsRef<str>>(s: S) -> Result<Self> {
232 let t = s.as_ref().parse::<NaiveTime>().map_err(|e| {
233 Error::new(
234 ErrorKind::DataInvalid,
235 format!("Can't parse time from string: {}", s.as_ref()),
236 )
237 .with_source(e)
238 })?;
239
240 Ok(Self::time_from_naive_time(t))
241 }
242
243 pub fn time_from_hms_micro(hour: u32, min: u32, sec: u32, micro: u32) -> Result<Self> {
255 let t = NaiveTime::from_hms_micro_opt(hour, min, sec, micro)
256 .ok_or_else(|| Error::new(
257 ErrorKind::DataInvalid,
258 format!("Can't create time from hour: {hour}, min: {min}, second: {sec}, microsecond: {micro}"),
259 ))?;
260 Ok(Self::time_from_naive_time(t))
261 }
262
263 pub fn timestamp(value: i64) -> Self {
265 Self::Primitive(PrimitiveLiteral::Long(value))
266 }
267
268 pub fn timestamptz(value: i64) -> Self {
270 Self::Primitive(PrimitiveLiteral::Long(value))
271 }
272
273 pub(crate) fn timestamp_nano(value: i64) -> Self {
275 Self::Primitive(PrimitiveLiteral::Long(value))
276 }
277
278 pub(crate) fn timestamptz_nano(value: i64) -> Self {
280 Self::Primitive(PrimitiveLiteral::Long(value))
281 }
282
283 pub fn timestamp_from_datetime<T: TimeZone>(dt: DateTime<T>) -> Self {
285 Self::timestamp(dt.with_timezone(&Utc).timestamp_micros())
286 }
287
288 pub fn timestamptz_from_datetime<T: TimeZone>(dt: DateTime<T>) -> Self {
290 Self::timestamptz(dt.with_timezone(&Utc).timestamp_micros())
291 }
292
293 pub fn timestamp_from_str<S: AsRef<str>>(s: S) -> Result<Self> {
317 let dt = DateTime::<Utc>::from_str(s.as_ref()).map_err(|e| {
318 Error::new(ErrorKind::DataInvalid, "Can't parse datetime.").with_source(e)
319 })?;
320
321 Ok(Self::timestamp_from_datetime(dt))
322 }
323
324 pub fn timestamptz_from_str<S: AsRef<str>>(s: S) -> Result<Self> {
326 let dt = DateTime::<Utc>::from_str(s.as_ref()).map_err(|e| {
327 Error::new(ErrorKind::DataInvalid, "Can't parse datetime.").with_source(e)
328 })?;
329
330 Ok(Self::timestamptz_from_datetime(dt))
331 }
332
333 pub fn string<S: ToString>(s: S) -> Self {
335 Self::Primitive(PrimitiveLiteral::String(s.to_string()))
336 }
337
338 pub fn uuid(uuid: Uuid) -> Self {
340 Self::Primitive(PrimitiveLiteral::UInt128(uuid.as_u128()))
341 }
342
343 pub fn uuid_from_str<S: AsRef<str>>(s: S) -> Result<Self> {
356 let uuid = Uuid::parse_str(s.as_ref()).map_err(|e| {
357 Error::new(
358 ErrorKind::DataInvalid,
359 format!("Can't parse uuid from string: {}", s.as_ref()),
360 )
361 .with_source(e)
362 })?;
363 Ok(Self::uuid(uuid))
364 }
365
366 pub fn fixed<I: IntoIterator<Item = u8>>(input: I) -> Self {
378 Literal::Primitive(PrimitiveLiteral::Binary(input.into_iter().collect()))
379 }
380
381 pub fn binary<I: IntoIterator<Item = u8>>(input: I) -> Self {
393 Literal::Primitive(PrimitiveLiteral::Binary(input.into_iter().collect()))
394 }
395
396 pub fn decimal(decimal: i128) -> Self {
398 Self::Primitive(PrimitiveLiteral::Int128(decimal))
399 }
400
401 pub fn decimal_from_str<S: AsRef<str>>(s: S) -> Result<Self> {
413 let decimal = decimal_from_str_exact(s.as_ref())?;
414 Ok(Self::decimal(decimal_mantissa(&decimal)))
415 }
416
417 pub fn as_primitive_literal(&self) -> Option<PrimitiveLiteral> {
419 match self {
420 Literal::Primitive(primitive) => Some(primitive.clone()),
421 _ => None,
422 }
423 }
424
425 pub fn try_from_json(value: JsonValue, data_type: &Type) -> Result<Option<Self>> {
429 match data_type {
430 Type::Primitive(primitive) => match (primitive, value) {
431 (PrimitiveType::Boolean, JsonValue::Bool(bool)) => {
432 Ok(Some(Literal::Primitive(PrimitiveLiteral::Boolean(bool))))
433 }
434 (PrimitiveType::Int, JsonValue::Number(number)) => {
435 Ok(Some(Literal::Primitive(PrimitiveLiteral::Int(
436 number
437 .as_i64()
438 .ok_or(Error::new(
439 crate::ErrorKind::DataInvalid,
440 "Failed to convert json number to int",
441 ))?
442 .try_into()?,
443 ))))
444 }
445 (PrimitiveType::Long, JsonValue::Number(number)) => Ok(Some(Literal::Primitive(
446 PrimitiveLiteral::Long(number.as_i64().ok_or(Error::new(
447 crate::ErrorKind::DataInvalid,
448 "Failed to convert json number to long",
449 ))?),
450 ))),
451 (PrimitiveType::Float, JsonValue::Number(number)) => Ok(Some(Literal::Primitive(
452 PrimitiveLiteral::Float(OrderedFloat(number.as_f64().ok_or(Error::new(
453 crate::ErrorKind::DataInvalid,
454 "Failed to convert json number to float",
455 ))? as f32)),
456 ))),
457 (PrimitiveType::Double, JsonValue::Number(number)) => Ok(Some(Literal::Primitive(
458 PrimitiveLiteral::Double(OrderedFloat(number.as_f64().ok_or(Error::new(
459 crate::ErrorKind::DataInvalid,
460 "Failed to convert json number to double",
461 ))?)),
462 ))),
463 (PrimitiveType::Date, JsonValue::String(s)) => {
464 Ok(Some(Literal::Primitive(PrimitiveLiteral::Int(
465 date::date_to_days(&NaiveDate::parse_from_str(&s, "%Y-%m-%d")?),
466 ))))
467 }
468 (PrimitiveType::Date, JsonValue::Number(number)) => {
469 Ok(Some(Literal::Primitive(PrimitiveLiteral::Int(
470 number
471 .as_i64()
472 .ok_or(Error::new(
473 crate::ErrorKind::DataInvalid,
474 "Failed to convert json number to date (days since epoch)",
475 ))?
476 .try_into()?,
477 ))))
478 }
479 (PrimitiveType::Time, JsonValue::String(s)) => {
480 Ok(Some(Literal::Primitive(PrimitiveLiteral::Long(
481 time::time_to_microseconds(&NaiveTime::parse_from_str(&s, "%H:%M:%S%.f")?),
482 ))))
483 }
484 (PrimitiveType::Timestamp, JsonValue::String(s)) => Ok(Some(Literal::Primitive(
485 PrimitiveLiteral::Long(timestamp::datetime_to_microseconds(
486 &NaiveDateTime::parse_from_str(&s, "%Y-%m-%dT%H:%M:%S%.f")?,
487 )),
488 ))),
489 (PrimitiveType::Timestamptz, JsonValue::String(s)) => {
490 Ok(Some(Literal::Primitive(PrimitiveLiteral::Long(
491 timestamptz::datetimetz_to_microseconds(&Utc.from_utc_datetime(
492 &NaiveDateTime::parse_from_str(&s, "%Y-%m-%dT%H:%M:%S%.f+00:00")?,
493 )),
494 ))))
495 }
496 (PrimitiveType::String, JsonValue::String(s)) => {
497 Ok(Some(Literal::Primitive(PrimitiveLiteral::String(s))))
498 }
499 (PrimitiveType::Uuid, JsonValue::String(s)) => Ok(Some(Literal::Primitive(
500 PrimitiveLiteral::UInt128(Uuid::parse_str(&s)?.as_u128()),
501 ))),
502 (PrimitiveType::Fixed(_), JsonValue::String(_)) => todo!(),
503 (PrimitiveType::Binary, JsonValue::String(_)) => todo!(),
504 (
505 PrimitiveType::Decimal {
506 precision: _,
507 scale,
508 },
509 JsonValue::String(s),
510 ) => {
511 let decimal = decimal_from_str_exact(&s)?;
512 let rescaled = decimal_rescale(decimal, *scale);
513 Ok(Some(Literal::Primitive(PrimitiveLiteral::Int128(
514 decimal_mantissa(&rescaled),
515 ))))
516 }
517 (_, JsonValue::Null) => Ok(None),
518 (i, j) => Err(Error::new(
519 crate::ErrorKind::DataInvalid,
520 format!("The json value {j} doesn't fit to the iceberg type {i}."),
521 )),
522 },
523 Type::Struct(schema) => {
524 if let JsonValue::Object(mut object) = value {
525 Ok(Some(Literal::Struct(Struct::from_iter(
526 schema.fields().iter().map(|field| {
527 object.remove(&field.id.to_string()).and_then(|value| {
528 Literal::try_from_json(value, &field.field_type)
529 .and_then(|value| {
530 value.ok_or(Error::new(
531 ErrorKind::DataInvalid,
532 "Key of map cannot be null",
533 ))
534 })
535 .ok()
536 })
537 }),
538 ))))
539 } else {
540 Err(Error::new(
541 crate::ErrorKind::DataInvalid,
542 "The json value for a struct type must be an object.",
543 ))
544 }
545 }
546 Type::List(list) => {
547 if let JsonValue::Array(array) = value {
548 Ok(Some(Literal::List(
549 array
550 .into_iter()
551 .map(|value| {
552 Literal::try_from_json(value, &list.element_field.field_type)
553 })
554 .collect::<Result<Vec<_>>>()?,
555 )))
556 } else {
557 Err(Error::new(
558 crate::ErrorKind::DataInvalid,
559 "The json value for a list type must be an array.",
560 ))
561 }
562 }
563 Type::Map(map) => {
564 if let JsonValue::Object(mut object) = value {
565 if let (Some(JsonValue::Array(keys)), Some(JsonValue::Array(values))) =
566 (object.remove("keys"), object.remove("values"))
567 {
568 Ok(Some(Literal::Map(Map::from_iter(
569 keys.into_iter()
570 .zip(values.into_iter())
571 .map(|(key, value)| {
572 Ok((
573 Literal::try_from_json(key, &map.key_field.field_type)
574 .and_then(|value| {
575 value.ok_or(Error::new(
576 ErrorKind::DataInvalid,
577 "Key of map cannot be null",
578 ))
579 })?,
580 Literal::try_from_json(value, &map.value_field.field_type)?,
581 ))
582 })
583 .collect::<Result<Vec<_>>>()?,
584 ))))
585 } else {
586 Err(Error::new(
587 crate::ErrorKind::DataInvalid,
588 "The json value for a list type must be an array.",
589 ))
590 }
591 } else {
592 Err(Error::new(
593 crate::ErrorKind::DataInvalid,
594 "The json value for a list type must be an array.",
595 ))
596 }
597 }
598 }
599 }
600
601 pub fn try_into_json(self, r#type: &Type) -> Result<JsonValue> {
605 match (self, r#type) {
606 (Literal::Primitive(prim), Type::Primitive(prim_type)) => match (prim_type, prim) {
607 (PrimitiveType::Boolean, PrimitiveLiteral::Boolean(val)) => {
608 Ok(JsonValue::Bool(val))
609 }
610 (PrimitiveType::Int, PrimitiveLiteral::Int(val)) => {
611 Ok(JsonValue::Number((val).into()))
612 }
613 (PrimitiveType::Long, PrimitiveLiteral::Long(val)) => {
614 Ok(JsonValue::Number((val).into()))
615 }
616 (PrimitiveType::Float, PrimitiveLiteral::Float(val)) => {
617 match Number::from_f64(val.0 as f64) {
618 Some(number) => Ok(JsonValue::Number(number)),
619 None => Ok(JsonValue::Null),
620 }
621 }
622 (PrimitiveType::Double, PrimitiveLiteral::Double(val)) => {
623 match Number::from_f64(val.0) {
624 Some(number) => Ok(JsonValue::Number(number)),
625 None => Ok(JsonValue::Null),
626 }
627 }
628 (PrimitiveType::Date, PrimitiveLiteral::Int(val)) => {
629 Ok(JsonValue::String(date::days_to_date(val).to_string()))
630 }
631 (PrimitiveType::Time, PrimitiveLiteral::Long(val)) => Ok(JsonValue::String(
632 time::microseconds_to_time(val).to_string(),
633 )),
634 (PrimitiveType::Timestamp, PrimitiveLiteral::Long(val)) => Ok(JsonValue::String(
635 timestamp::microseconds_to_datetime(val)
636 .format("%Y-%m-%dT%H:%M:%S%.f")
637 .to_string(),
638 )),
639 (PrimitiveType::Timestamptz, PrimitiveLiteral::Long(val)) => Ok(JsonValue::String(
640 timestamptz::microseconds_to_datetimetz(val)
641 .format("%Y-%m-%dT%H:%M:%S%.f+00:00")
642 .to_string(),
643 )),
644 (PrimitiveType::TimestampNs, PrimitiveLiteral::Long(val)) => Ok(JsonValue::String(
645 timestamp::nanoseconds_to_datetime(val)
646 .format("%Y-%m-%dT%H:%M:%S%.f")
647 .to_string(),
648 )),
649 (PrimitiveType::TimestamptzNs, PrimitiveLiteral::Long(val)) => {
650 Ok(JsonValue::String(
651 timestamptz::nanoseconds_to_datetimetz(val)
652 .format("%Y-%m-%dT%H:%M:%S%.f+00:00")
653 .to_string(),
654 ))
655 }
656 (PrimitiveType::String, PrimitiveLiteral::String(val)) => {
657 Ok(JsonValue::String(val.clone()))
658 }
659 (_, PrimitiveLiteral::UInt128(val)) => {
660 Ok(JsonValue::String(Uuid::from_u128(val).to_string()))
661 }
662 (_, PrimitiveLiteral::Binary(val)) => Ok(JsonValue::String(val.iter().fold(
663 String::new(),
664 |mut acc, x| {
665 acc.push_str(&format!("{x:x}"));
666 acc
667 },
668 ))),
669 (_, PrimitiveLiteral::Int128(val)) => match r#type {
670 Type::Primitive(PrimitiveType::Decimal {
671 precision: _precision,
672 scale,
673 }) => {
674 let decimal = try_decimal_from_i128_with_scale(val, *scale)?;
675 Ok(JsonValue::String(decimal.to_string()))
676 }
677 _ => Err(Error::new(
678 ErrorKind::DataInvalid,
679 "The iceberg type for decimal literal must be decimal.",
680 ))?,
681 },
682 _ => Err(Error::new(
683 ErrorKind::DataInvalid,
684 "The iceberg value doesn't fit to the iceberg type.",
685 )),
686 },
687 (Literal::Struct(s), Type::Struct(struct_type)) => {
688 let mut id_and_value = Vec::with_capacity(struct_type.fields().len());
689 for (value, field) in s.into_iter().zip(struct_type.fields()) {
690 let json = match value {
691 Some(val) => val.try_into_json(&field.field_type)?,
692 None => JsonValue::Null,
693 };
694 id_and_value.push((field.id.to_string(), json));
695 }
696 Ok(JsonValue::Object(JsonMap::from_iter(id_and_value)))
697 }
698 (Literal::List(list), Type::List(list_type)) => Ok(JsonValue::Array(
699 list.into_iter()
700 .map(|opt| match opt {
701 Some(literal) => literal.try_into_json(&list_type.element_field.field_type),
702 None => Ok(JsonValue::Null),
703 })
704 .collect::<Result<Vec<JsonValue>>>()?,
705 )),
706 (Literal::Map(map), Type::Map(map_type)) => {
707 let mut object = JsonMap::with_capacity(2);
708 let mut json_keys = Vec::with_capacity(map.len());
709 let mut json_values = Vec::with_capacity(map.len());
710 for (key, value) in map.into_iter() {
711 json_keys.push(key.try_into_json(&map_type.key_field.field_type)?);
712 json_values.push(match value {
713 Some(literal) => literal.try_into_json(&map_type.value_field.field_type)?,
714 None => JsonValue::Null,
715 });
716 }
717 object.insert("keys".to_string(), JsonValue::Array(json_keys));
718 object.insert("values".to_string(), JsonValue::Array(json_values));
719 Ok(JsonValue::Object(object))
720 }
721 (value, r#type) => Err(Error::new(
722 ErrorKind::DataInvalid,
723 format!("The iceberg value {value:?} doesn't fit to the iceberg type {type}."),
724 )),
725 }
726 }
727
728 pub fn into_any(self) -> Box<dyn Any> {
730 match self {
731 Literal::Primitive(prim) => match prim {
732 PrimitiveLiteral::Boolean(any) => Box::new(any),
733 PrimitiveLiteral::Int(any) => Box::new(any),
734 PrimitiveLiteral::Long(any) => Box::new(any),
735 PrimitiveLiteral::Float(any) => Box::new(any),
736 PrimitiveLiteral::Double(any) => Box::new(any),
737 PrimitiveLiteral::Binary(any) => Box::new(any),
738 PrimitiveLiteral::String(any) => Box::new(any),
739 PrimitiveLiteral::UInt128(any) => Box::new(any),
740 PrimitiveLiteral::Int128(any) => Box::new(any),
741 PrimitiveLiteral::AboveMax | PrimitiveLiteral::BelowMin => unimplemented!(),
742 },
743 _ => unimplemented!(),
744 }
745 }
746}