nushell（nushell 搜索）

Nushell - Modern Structured Shell

This skill activates when working with Nushell (Nu), writing Nu scripts, working with structured data pipelines, or configuring the Nu environment.

When to Use This Skill

Activate when:

• Writing Nushell scripts or commands
• Working with structured data in pipelines
• Converting from bash/zsh to Nushell
• Configuring Nushell environment
• Processing JSON, CSV, YAML, or other structured data
• Creating custom commands or modules

What is Nushell?

Current stable: 0.111.0 (pre-1.0, breaking changes possible between minor versions)

Nushell is a modern shell that:

• Treats data as structured (not just text streams)
• Works cross-platform (Windows, macOS, Linux)
• Provides clear error messages and IDE support
• Combines shell and programming language features
• Has built-in data format support (JSON, CSV, YAML, TOML, XML, etc.)

Installation

# macOS
brew install nushell

# Linux (cargo)
cargo install nu

# Windows
winget install nushell

# Via mise (recommended for project-level management)
# Add to mise.toml:
# [tools]
# "github:nushell/nushell" = "latest"
mise install github:nushell/nushell

# Or download from https://www.nushell.sh/

Basic Concepts

Everything is Data

Unlike traditional shells where everything is text, Nu works with structured data:

# Traditional shell (text output)
ls | grep ".txt"

# Nushell (structured data)
ls | where name =~ ".txt"

Pipeline Philosophy

Data flows through pipelines as structured tables/records:

# Each command outputs structured data
ls | where size > 1kb | sort-by modified | reverse

Data Types

Basic Types

# Integers
42
-10

# Floats
3.14
-2.5

# Strings
"hello"
'world'

# Booleans
true
false

# Null
null

Collections

# Lists
[1 2 3 4 5]
["apple" "banana" "cherry"]

# Records (like objects/dicts)
{name: "Alice", age: 30, city: "NYC"}

# Tables (list of records)
[
  {name: "Alice", age: 30}
  {name: "Bob", age: 25}
]

Ranges

# Number ranges
1..10
1..2..10  # Step by 2

# Use in commands
1..5 | each { |i| $i * 2 }

Working with Files and Directories

Navigation

# Change directory
cd /path/to/dir

# List files (returns structured table)
ls

# List with details
ls | select name size modified

# Filter files
ls | where type == file
ls | where size > 1mb
ls | where name =~ "\.txt$"

File Operations

# Create file
"hello" | save hello.txt

# Read file
open hello.txt

# Append to file
"world" | save -a hello.txt

# Copy
cp source.txt dest.txt

# Move/rename
mv old.txt new.txt

# Remove
rm file.txt
rm -r directory/

# Create directory
mkdir new-dir

File Content

# Read as string
open file.txt

# Read structured data
open data.json
open config.toml
open data.csv

# Write structured data
{name: "Alice", age: 30} | to json | save user.json
[{a: 1} {a: 2}] | to csv | save data.csv

Pipeline Operations

Filtering

# Filter with where
ls | where size > 1mb
ls | where type == dir
ls | where name =~ "test"

# Multiple conditions
ls | where size > 1kb and type == file

Selecting Columns

# Select specific columns
ls | select name size

# Rename columns
ls | select name size | rename file bytes

Sorting

# Sort by column
ls | sort-by size
ls | sort-by modified

# Reverse sort
ls | sort-by size | reverse

# Multiple columns
ls | sort-by type size

Transforming Data

# Map over items with each
1..5 | each { |i| $i * 2 }

# Update column
ls | update name { |row| $row.name | str upcase }

# Insert column
ls | insert size_kb { |row| $row.size / 1000 }

# Upsert (update or insert)
ls | upsert type_upper { |row| $row.type | str upcase }

Aggregation

# Count items
ls | length

# Sum
[1 2 3 4 5] | math sum

# Average
[1 2 3 4 5] | math avg

# Min/Max
ls | get size | math max
ls | get size | math min

# Group by
ls | group-by type

Variables

Variable Assignment

# Let (immutable by default)
let name = "Alice"
let age = 30
let colors = ["red" "green" "blue"]

# Mut (mutable)
mut counter = 0
$counter = $counter + 1

Using Variables

# Reference with $
let name = "Alice"
print $"Hello, ($name)!"

# In pipelines
let threshold = 1mb
ls | where size > $threshold

Environment Variables

# Get environment variable
$env.PATH
$env.HOME

# Set environment variable
$env.MY_VAR = "value"

# Load from file
load-env { API_KEY: "secret" }

String Operations

String Interpolation

# String interpolation with ()
let name = "Alice"
print $"Hello, ($name)!"

# With expressions
let x = 5
print $"Result: (5 * $x)"

String Methods

# Case conversion
"hello" | str upcase  # HELLO
"WORLD" | str downcase  # world

# Trimming
"  spaces  " | str trim

# Replace
"hello world" | str replace "world" "nu"

# Contains
"hello world" | str contains "world"  # true

# Split
"a,b,c" | split row ","

Conditionals

If Expressions

# If-else
if $age >= 18 {
  print "Adult"
} else {
  print "Minor"
}

# If-else if-else
if $score >= 90 {
  "A"
} else if $score >= 80 {
  "B"
} else {
  "C"
}

# Ternary-style with match
let status = if $is_active { "active" } else { "inactive" }

Match (Pattern Matching)

# Match expression
match $value {
  1 => "one"
  2 => "two"
  _ => "other"
}

# With conditions
match $age {
  0..17 => "minor"
  18..64 => "adult"
  _ => "senior"
}

Loops

For Loop

# Loop over range
for i in 1..5 {
  print $i
}

# Loop over list
for name in ["Alice" "Bob" "Charlie"] {
  print $"Hello, ($name)"
}

# Loop over files
for file in (ls | where type == file) {
  print $file.name
}

While Loop

# While loop
mut i = 0
while $i < 5 {
  print $i
  $i = $i + 1
}

Each (Functional)

# Transform each item
1..5 | each { |i| $i * 2 }

# With index
["a" "b" "c"] | enumerate | each { |item|
  print $"($item.index): ($item.item)"
}

Custom Commands

Defining Commands

# Simple command
def greet [name: string] {
  print $"Hello, ($name)!"
}

greet "Alice"

# With return value
def add [a: int, b: int] {
  $a + $b
}

let result = add 5 3

# With default values
def greet [name: string = "World"] {
  print $"Hello, ($name)!"
}

Command Parameters

# Required parameters
def copy [source: path, dest: path] {
  cp $source $dest
}

# Optional parameters
def greet [
  name: string
  --loud (-l)  # Flag
  --repeat (-r): int = 1  # Named parameter with default
] {
  let message = if $loud {
    $name | str upcase
  } else {
    $name
  }

  1..$repeat | each { print $"Hello, ($message)!" }
}

# Usage
greet "Alice"
greet "Bob" --loud
greet "Charlie" --repeat 3

Pipeline Commands

# Accept pipeline input
def filter-large [] {
  where size > 1mb
}

# Usage
ls | filter-large

# Accept and transform pipeline
def double [] {
  each { |value| $value * 2 }
}

[1 2 3] | double

Working with Structured Data

JSON

# Read JSON
let data = open data.json

# Parse JSON string
let obj = '{"name": "Alice", "age": 30}' | from json

# Write JSON
{name: "Alice", age: 30} | to json | save user.json

# Pretty print JSON
{name: "Alice", age: 30} | to json -i 2

CSV

# Read CSV
let data = open data.csv

# Convert to CSV
[{a: 1, b: 2} {a: 3, b: 4}] | to csv

# Save CSV
ls | select name size | to csv | save files.csv

YAML/TOML

# Read YAML
let config = open config.yaml

# Read TOML
let config = open config.toml

# Write YAML
{key: "value"} | to yaml | save config.yaml

# Write TOML
{key: "value"} | to toml | save config.toml

Working with Tables

# Create table
let users = [
  {name: "Alice", age: 30, city: "NYC"}
  {name: "Bob", age: 25, city: "LA"}
  {name: "Charlie", age: 35, city: "NYC"}
]

# Query table
$users | where age > 25
$users | where city == "NYC"
$users | select name age

# Add column
$users | insert country { "USA" }

# Group and count
$users | group-by city | transpose city users

Modules

Creating Modules

# utils.nu
export def greet [name: string] {
  print $"Hello, ($name)!"
}

export def add [a: int, b: int] {
  $a + $b
}

Using Modules

# Import module
use utils.nu

# Use exported commands
utils greet "Alice"
utils add 5 3

# Import specific commands
use utils.nu [greet add]

greet "Alice"
add 5 3

# Import with alias
use utils.nu *

Configuration

Config File Location

# View config
config nu

# Edit config
config nu | open

# Config location
$nu.config-path

Common Configurations

# config.nu
$env.config = {
  show_banner: false

  ls: {
    use_ls_colors: true
    clickable_links: true
  }

  table: {
    mode: rounded
    index_mode: auto
  }

  completions: {
    quick: true
    partial: true
  }

  history: {
    max_size: 10000
    sync_on_enter: true
    file_format: "sqlite"
  }
}

Environment Setup

# env.nu
$env.PATH = ($env.PATH | split row (char esep) | append '/custom/bin')
$env.EDITOR = "nvim"

# Load completions
use completions/git.nu *

Common Patterns

File Processing

# Process all JSON files
ls *.json | each { |file|
  let data = open $file.name
  print $"Processing ($file.name): ($data | length) items"
}

# Batch rename files
ls *.txt | each { |file|
  let new_name = ($file.name | str replace ".txt" ".md")
  mv $file.name $new_name
}

Data Transformation

# CSV to JSON
open data.csv | to json | save data.json

# Filter and transform
open users.json
| where active == true
| select name email
| to csv
| save active_users.csv

# Merge data
let users = open users.json
let orders = open orders.json
$users | merge $orders

HTTP Requests

# GET request
http get https://api.example.com/users

# POST request
http post https://api.example.com/users {
  name: "Alice"
  email: "alice@example.com"
}

# With headers
http get -H [Authorization "Bearer token"] https://api.example.com/data

System Commands

# Run external command
^ls -la

# Capture output
let output = (^git status)

# Check if command exists
which git

# Get command path
which git | get path

Error Handling

Try-Catch

# Try expression
try {
  open missing.txt
} catch {
  print "File not found"
}

# With error value
try {
  open missing.txt
} catch { |err|
  print $"Error: ($err)"
}

Null Handling

# Default value
let value = ($env.MY_VAR? | default "default_value")

# Null propagation
let length = ($value | get name? | str length)

Scripting

Script Files

#!/usr/bin/env nu

# Script: process_logs.nu
# Description: Process log files and generate report

def main [log_dir: path] {
  let errors = (
    ls $"($log_dir)/*.log"
    | each { |file| open $file.name | lines }
    | flatten
    | where $it =~ "ERROR"
  )

  print $"Found ($errors | length) errors"
  $errors | save error_report.txt
}

Make executable:

chmod +x process_logs.nu
./process_logs.nu /var/log

Script Parameters

# With parameters
def main [
  input: path
  --output (-o): path = "output.txt"
  --verbose (-v)
] {
  if $verbose {
    print $"Processing ($input)..."
  }

  let data = open $input
  $data | save $output

  if $verbose {
    print "Done!"
  }
}

Comparison with Bash

Common Operations

# Bash
find . -name "*.txt" | wc -l

# Nushell
ls **/*.txt | length

# Bash
cat file.json | jq '.users[] | select(.age > 25) | .name'

# Nushell
open file.json | get users | where age > 25 | get name

# Bash
for file in *.txt; do
  mv "$file" "${file%.txt}.md"
done

# Nushell
ls *.txt | each { |f| mv $f.name ($f.name | str replace ".txt" ".md") }

Best Practices

• Use structured data: Leverage Nu's strength in handling structured data
• Pipeline composition: Build complex operations from simple pipeline stages
• Type annotations: Add types to custom command parameters for clarity
• Error handling: Use try-catch for operations that might fail
• Modules for reuse: Organize reusable commands in modules
• Configuration: Customize Nu to fit your workflow
• External commands: Use ^ prefix when calling external commands explicitly

Common Pitfalls

String vs Bare Words

# Bare word (interpreted as string in some contexts)
echo hello

# Explicit string (clearer)
echo "hello"

External Commands

# Wrong - Nu tries to parse as Nu command
ls -la

# Right - Explicitly call external command
^ls -la

Variable Scope

# Variables are scoped to blocks
if true {
  let x = 5
}
# $x not available here

# Use mut outside for wider scope
mut x = 0
if true {
  $x = 5
}
print $x  # Works

Key Principles

• Structured data first: Think in terms of tables and records, not text
• Pipeline composition: Chain simple operations to build complex workflows
• Type safety: Leverage Nu's type system for reliable scripts
• Cross-platform: Write scripts that work on all platforms
• Interactive and scriptable: Same syntax works in REPL and scripts
• Clear errors: Nu provides helpful error messages for debugging