[slfan1989]

Which issue does this PR close?

• Closes Decimal parsing loses positive exponents and accepts oversized values #3654.

What changes are included in this PR?

Apply positive exponents before extracting the unscaled decimal value, so 1e3 produces 1000. Validate precision before converting the coefficient to i128, so oversized inputs return DataInvalid instead of silently producing an incorrect value. JSON decimal parsing also validates the declared precision after rounding to the target scale.

Are these changes tested?

Added three regression tests covering positive exponents, the 38-digit precision boundary, and precision validation at the target JSON scale. All three failed before the fix and passed after it.

AI Disclosure

[Copilot]

🟢 Approval recommended

No active review issues are reported for the decimal parsing and precision-validation changes.

No open issues 🎉

What changed in this PR

Fixes decimal parsing in Iceberg’s value layer to address #3654: positive exponents are preserved, and oversized coefficients are rejected before conversion to i128.

Changes:

• Centralizes checked decimal coefficient extraction with scale and precision validation.
• Validates JSON decimal precision after rounding to the target scale.
• Adds regression coverage for exponent handling, oversized values, and JSON scale conversion.

File	Description
crates/​iceberg/​src/​writer/​file_writer/​parquet_writer.rs	Updates test calls for fallible coefficient extraction.
crates/​iceberg/​src/​spec/​values/​tests.rs	Adds decimal parsing and precision regression tests.
crates/​iceberg/​src/​spec/​values/​literal.rs	Uses checked extraction for string and JSON decimals.
crates/​iceberg/​src/​spec/​values/​decimal_utils.rs	Applies exponents and validates precision before conversion.
crates/​iceberg/​src/​spec/​values/​datum.rs	Routes decimal constructors through shared validation.

_{Review effort: Balanced}