This is an automated email from the ASF dual-hosted git repository. github-merge-queue[bot] pushed a commit to branch gh-readonly-queue/main/pr-3358-b154603a710d730d2f43f8af4db257cc137237ef in repository https://gitbox.apache.org/repos/asf/iceberg-rust.git
commit b2e1d13767643e1ca9f2ace6fdf31ffbc2154f9a Author: JeonDaehong <[email protected]> AuthorDate: Tue Oct 13 08:52:20 2026 +0000 fix(spec): format float and double partition values like iceberg-java (#3358) * fix(spec): format float and double partition values like iceberg-java * test(spec): move float to_human_string test to datum tests --- crates/iceberg/src/spec/partition.rs | 26 +++++++++++++++ crates/iceberg/src/spec/values/datum.rs | 57 ++++++++++++++++++++++++++++++++- crates/iceberg/src/spec/values/tests.rs | 35 ++++++++++++++++++++ 3 files changed, 117 insertions(+), 1 deletion(-) diff --git a/crates/iceberg/src/spec/partition.rs b/crates/iceberg/src/spec/partition.rs index 17aa3de96..f3c9f9e54 100644 --- a/crates/iceberg/src/spec/partition.rs +++ b/crates/iceberg/src/spec/partition.rs @@ -1951,6 +1951,32 @@ mod tests { ); } + #[test] + fn test_partition_to_path_float_matches_java() { + let schema = Schema::builder() + .with_fields(vec![ + NestedField::required(1, "d", Type::Primitive(PrimitiveType::Double)).into(), + NestedField::required(2, "f", Type::Primitive(PrimitiveType::Float)).into(), + ]) + .build() + .unwrap(); + + let spec = PartitionSpec::builder(schema.clone()) + .add_partition_field("d", "d", Transform::Identity) + .unwrap() + .add_partition_field("f", "f", Transform::Identity) + .unwrap() + .build() + .unwrap(); + + let data = Struct::from_iter([Some(Literal::double(1.0)), Some(Literal::float(1.0e7_f32))]); + + assert_eq!( + spec.partition_to_path(&data, schema.into()), + "d=1.0/f=1.0E7" + ); + } + #[test] fn test_partition_to_path_escaped_strings() { let schema = Schema::builder() diff --git a/crates/iceberg/src/spec/values/datum.rs b/crates/iceberg/src/spec/values/datum.rs index c4c835f61..a501466b3 100644 --- a/crates/iceberg/src/spec/values/datum.rs +++ b/crates/iceberg/src/spec/values/datum.rs @@ -18,7 +18,7 @@ //! Typed literals with validation use std::cmp::Ordering; -use std::fmt::{Display, Formatter}; +use std::fmt::{Display, Formatter, LowerExp}; use std::str::FromStr; use chrono::{DateTime, NaiveDate, NaiveDateTime, NaiveTime, TimeZone, Utc}; @@ -1226,11 +1226,66 @@ impl Datum { /// Returns a human-readable string representation of this literal. /// /// For string literals, this returns the raw string value without quotes. + /// Float and double literals are formatted like Java's `Float.toString` / + /// `Double.toString` (JDK 19+). /// For all other literals, it falls back to [`to_string()`](ToString::to_string). pub fn to_human_string(&self) -> String { match self.literal() { PrimitiveLiteral::String(s) => s.to_string(), + PrimitiveLiteral::Float(val) => human_float(val.0), + PrimitiveLiteral::Double(val) => human_float(val.0), _ => self.to_string(), } } } + +/// Formats a float like Java's `Float.toString` / `Double.toString` (JDK 19+): +/// plain decimal when `1e-3 <= |value| < 1e7`, otherwise `<d>.<ddd>E<exp>`, +/// always with at least one digit after the point. +fn human_float<T: Float + LowerExp + FromStr>(value: T) -> String { + if value.is_nan() { + return "NaN".to_string(); + } + let sign = if value.is_sign_negative() { "-" } else { "" }; + if value.is_infinite() { + return format!("{sign}Infinity"); + } + if value.is_zero() { + return format!("{sign}0.0"); + } + + let abs = value.abs(); + // `{:e}` gives the fewest digits that round-trip. Java uses at least two + // digits and, among decimals of that length that round-trip, the closest + // one, preferring an even last digit on ties (2097152.2 for 2097152.25f32, + // 4.9E-324 for f64::from_bits(1)). Exact formatting picks exactly that. + let shortest = format!("{abs:e}"); + let precision = shortest.find('e').unwrap().saturating_sub(2).max(1); + let closest = format!("{abs:.precision$e}"); + let scientific = if closest.parse::<T>().is_ok_and(|v| v == abs) { + closest + } else { + shortest + }; + let (mantissa, exponent) = scientific.split_once('e').unwrap(); + let exponent: i32 = exponent.parse().unwrap(); + let digits = mantissa.replace('.', ""); + let digits = digits.trim_end_matches('0'); + + if (-3..7).contains(&exponent) { + if exponent < 0 { + let zeros = "0".repeat((-exponent - 1) as usize); + format!("{sign}0.{zeros}{digits}") + } else { + let int_len = exponent as usize + 1; + if digits.len() > int_len { + format!("{sign}{}.{}", &digits[..int_len], &digits[int_len..]) + } else { + format!("{sign}{digits:0<int_len$}.0") + } + } + } else { + let fraction = if digits.len() > 1 { &digits[1..] } else { "0" }; + format!("{sign}{}.{fraction}E{exponent}", &digits[..1]) + } +} diff --git a/crates/iceberg/src/spec/values/tests.rs b/crates/iceberg/src/spec/values/tests.rs index 63ff9f636..f21e78ef3 100644 --- a/crates/iceberg/src/spec/values/tests.rs +++ b/crates/iceberg/src/spec/values/tests.rs @@ -1986,3 +1986,38 @@ fn test_raw_literal_project_by_name_rejects_missing_required_field() { let err = raw.project_by_name(&expected).unwrap_err(); assert_eq!(err.kind(), ErrorKind::DataInvalid); } + +/// Expected values are the output of Java's `Double.toString` / `Float.toString` (JDK 21). +#[test] +fn test_datum_to_human_string_float_matches_java() { + let double = |value: f64| Datum::double(value).to_human_string(); + assert_eq!(double(1.0), "1.0"); + assert_eq!(double(100.0), "100.0"); + assert_eq!(double(0.0), "0.0"); + assert_eq!(double(-0.0), "-0.0"); + assert_eq!(double(1.5), "1.5"); + assert_eq!(double(0.001), "0.001"); + assert_eq!(double(9999999.0), "9999999.0"); + assert_eq!(double(1.0e7), "1.0E7"); + assert_eq!(double(1.0e-5), "1.0E-5"); + assert_eq!(double(1.23456789e20), "1.23456789E20"); + assert_eq!(double(1.0e23), "1.0E23"); + assert_eq!(double(f64::INFINITY), "Infinity"); + assert_eq!(double(f64::NEG_INFINITY), "-Infinity"); + assert_eq!(double(f64::NAN), "NaN"); + assert_eq!(double(f64::MAX), "1.7976931348623157E308"); + // Java uses at least two significant digits. + assert_eq!(double(f64::from_bits(1)), "4.9E-324"); + // Halfway between two 17-digit decimals: Java picks the even one. + assert_eq!(double(2f64.powi(-25)), "2.9802322387695312E-8"); + + let float = |value: f32| Datum::float(value).to_human_string(); + assert_eq!(float(1.0), "1.0"); + assert_eq!(float(-0.0), "-0.0"); + assert_eq!(float(0.1), "0.1"); + assert_eq!(float(1.0e7), "1.0E7"); + assert_eq!(float(f32::NEG_INFINITY), "-Infinity"); + assert_eq!(float(f32::from_bits(1)), "1.4E-45"); + // 2097152.25, halfway between 2097152.2 and 2097152.3. + assert_eq!(float(f32::from_bits(0x4a00_0001)), "2097152.2"); +}
