[ara-stock]

Motivation

cast to-base / to-hex / to-dec return the input string unchanged whenever the input base equals the output base, because to_base short-circuits with return Ok(value.to_string()). The input is parsed (and validated) but the parsed value is thrown away, so the output is not normalized and is inconsistent with every other path:

$ cast to-hex --base-in 16 ff
ff                      # no 0x prefix (cast to-hex 255 prints 0xff)
$ cast to-hex 0XFF
0XFF
$ cast to-hex 0x00ff
0x00ff
$ cast to-hex +0x10
+0x10
$ cast to-dec 0010
0010
$ cast to-base 0b0011 bin
0b0011
$ cast to-hex -- -0x1
-0x1                    # but `cast to-hex -- -1` prints the two's complement 0xfff...f

(reproduced with cast 1.8.3 and on master a57bf18)

The negative case is the most surprising: the same number gives a two's-complement hex word when written in decimal, but a signed hex string when written in hex.

Solution

Drop the base_in == base_out early return in to_base, so the parsed NumberWithBase is always formatted with {:#?}. The input is already parsed before the early return, so this only changes the output, which is now the same as converting from any other base:

$ cast to-hex --base-in 16 ff
0xff
$ cast to-hex 0XFF
0xff
$ cast to-hex 0x00ff
0xff
$ cast to-hex +0x10
0x10
$ cast to-dec 0010
10
$ cast to-base 0b0011 bin
0b11
$ cast to-hex -- -0x1
0xffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff

Added to_base_same_base_normalizes in crates/cast/tests/cli/conversions.rs. It fails with the args.rs change reverted and passes with it.

Note: leading zeros in a same-base input are no longer kept (0x00ff -> 0xff), matching what happens for a different-base input (cast to-hex 255 -> 0xff). cast to-uint256 / cast pad remain the way to get fixed-width output.

Tests run locally: cargo test -p cast --test cli -- conversions:: (46 passed), cargo clippy -p cast --all-targets (no warnings in cast), rustfmt check on the changed files.

PR Checklist

• Added Tests
• Added Documentation
• Breaking changes

I used an AI assistant while preparing this; I reproduced the bug and ran the tests myself.

🤖 Generated with Claude Code

[stevencartavia]

should we keep same-base output unchanged? @mattsse @grandizzy