Improve and fix filesize formatting/display (#14397)

# Description

This PR cleans up the code surrounding formatting and displaying file
sizes.
- The `byte_unit` crate we use for file size units displays kilobytes as
`KB`, which is not the SI or ISO/IEC standard. Rather it should be `kB`,
so this fixes #8872. On some systems, `KB` actually means `KiB`, so this
avoids any potential confusion.
- The `byte_unit` crate, when displaying file sizes, casts integers to
floats which will lose precision for large file sizes. This PR adds a
custom `Display` implementation for `Filesize` that can give an exact
string representation of a `Filesize` for metric/SI units.
- This PR also removes the dependency on the `byte_unit` crate which
brought in several other dependencies.

Additionally, this PR makes some changes to the config for filesize
formatting (`$env.config.filesize`).
- The previous filesize config had the `metric` and `format` options. If
a metric (SI) unit was set in `format`, but `metric` was set to false,
then the `metric` option would take precedence and convert `format` to
the corresponding binary unit (or vice versa). E.g., `{ format: kB,
metric: false }` => `KiB`. Instead, this PR adds the `unit` option to
replace the `format` and `metric` options. `unit` can be set to a fixed
file size unit like `kB` or `KiB`, or it can be set to one of the
special options: `binary` or `metric`. These options tells nushell to
format file sizes using an appropriately scaled metric or binary unit
(examples below).
  ```nushell
  # precision = null

  # unit = kB
  1kB  # 1 kB
  1KiB # 1.024 kB
  
  # unit = KiB
  1kB  # 0.9765625 KiB
  1KiB # 1 KiB
  
  # unit = metric
  1000B     # 1 kB
  1024B     # 1.024 kB
  10_000MB  # 10 GB
  10_240MiB # 10.73741824 GB

  # unit = binary
  1000B     # 1000 B
  1024B     # 1 KiB
  10_000MB  # 9.313225746154785 GiB
  10_240MiB # 10 GiB
  ```
- In addition, this PR also adds the `precision` option to the filesize
config. It determines how many digits to show after the decimal point.
If set to null, then everything after the decimal point is shown.
- The default filesize config is `{ unit: metric, precision: 1 }`.

# User-Facing Changes

- Commands that use the config to format file sizes will follow the
changes described above (e.g., `table`, `into string`, `to text`, etc.).
- The file size unit/format passed to `format filesize` is now case
sensitive. An error with the valid units is shown if the case does not
match.
- `$env.config.filesize.format` and `$env.config.filesize.metric` are
deprecated and replaced by `$env.config.filesize.unit`.
- A new `$env.config.filesize.precision` option was added.

# Tests + Formatting

Mostly updated test expected outputs.

# After Submitting

This PR does not change the way NUON serializes file sizes, because that
would require changing the nu parser to be able to losslessly decode the
new, exact string representation introduced in this PR.

Similarly, this PR also does not change the file size parsing in any
way. Although the file size units provided to `format filesize` or the
filesize config are now case-sensitive, the same is not yet true for
file size literals in nushell code.
This commit is contained in:
Ian Manske
2025-01-23 06:24:51 +00:00
committed by GitHub
parent befeddad59
commit 93e121782c
23 changed files with 431 additions and 544 deletions

View File

@ -251,9 +251,9 @@ $env.config.bracketed_paste = true
# - If `FORCE_COLOR` is set, coloring is always enabled.
# - If `NO_COLOR` is set, coloring is disabled.
# - If `CLICOLOR` is set, its value (0 or 1) decides whether coloring is used.
# If none of these are set, it checks whether the standard output is a terminal
# If none of these are set, it checks whether the standard output is a terminal
# and enables coloring if it is.
# A value of `true` or `false` overrides this behavior, explicitly enabling or
# A value of `true` or `false` overrides this behavior, explicitly enabling or
# disabling ANSI coloring in Nushell's internal commands.
# When disabled, built-in commands will only use the default foreground color.
# Note: This setting does not affect the `ansi` command.
@ -371,17 +371,18 @@ $env.config.datetime_format.normal = "%m/%d/%y %I:%M:%S%p"
# ----------------
# Filesize Display
# ----------------
# filesize.metric (bool): When displaying filesize values ...
# true: Use the ISO-standard KB, MB, GB
# false: Use the Windows-standard KiB, MiB, GiB
$env.config.filesize.metric = false
# filesize.unit (string): One of either:
# - A filesize unit: "B", "kB", "KiB", "MB", "MiB", "GB", "GiB", "TB", "TiB", "PB", "PiB", "EB", or "EiB".
# - An automatically scaled unit: "metric" or "binary".
# "metric" will use units with metric (SI) prefixes like kB, MB, or GB.
# "binary" will use units with binary prefixes like KiB, MiB, or GiB.
# Otherwise, setting this to one of the filesize units will use that particular unit when displaying all file sizes.
$env.config.filesize.unit = 'metric'
# filesize.format (string): One of either:
# - The filesize units such as "KB", "KiB", etc. In this case, filesize values always display using
# this unit.
# - Or "auto": Filesizes are displayed using the closest unit. For example, 1_000_000_000b will display
# as 953.7 MiB (when `metric = false`) or 1.0GB (when `metric = true`)
$env.config.filesize.format = "auto"
# filesize.precision (int or nothing):
# The number of digits to display after the decimal point for file sizes.
# When set to `null`, all digits after the decimal point will be displayed.
$env.config.filesize.precision = 1
# ---------------------
# Miscellaneous Display
@ -942,4 +943,4 @@ path add "~/.local/bin"
path add ($env.CARGO_HOME | path join "bin")
# You can remove duplicate directories from the path using:
$env.PATH = ($env.PATH | uniq)
$env.PATH = ($env.PATH | uniq)