nushell — consolidated language guide — DedSec Project

Nushell

Category: Scripting. Extension: .nu. Run: nu lesson.nu with Nushell installed.

Nushell pipes structured values between commands. A list of integers remains a list rather than becoming whitespace-separated text. This matters when names, paths or records contain spaces.

Reproducible calculation

Save the following as lesson.nu:

let values = [2 4 6]
let doubled = ($values | each {|n| $n * 2 })
print ($doubled | math sum)

Run nu lesson.nu. Expected output: 24. each transforms each integer, while math sum aggregates the resulting list. The parentheses pass the complete pipeline result to print.

Debug the type, not the formatting

In the interactive shell, run [2 4 6] | describe and help each. Replace the list with ["2" "4"] and investigate the type mismatch before adding conversions. Do not parse a human-formatted table from an external command when a structured alternative exists.

Exercise and edge cases

Create records with fields name and points. Filter records with points >= 5 and count them. Repeat with an empty list, one record and a record with a missing field. Explain which input should be rejected rather than silently defaulted.

Platform troubleshooting

Verify nu --version before executing scripts. Windows and Linux builds exist; Termux support depends on available packages and CPU architecture. A missing executable usually indicates installation or PATH, while a syntax error can mean POSIX-shell syntax was copied into Nushell. Quote paths containing spaces and test any file-modifying pipeline in a temporary directory first.

NuShell: applied exercises

Reproduce the starter's computation

00_starter.nu takes a list of integers, doubles each using each, and passes the results to print. Start with the purely numeric case [1 2 3] | each {|n| $n * 2 }. Predict the three values 2, 4 and 6 before running. Inspect the pipeline value and avoid assuming it is one unstructured string; Nushell processes typed values. You can inspect a value's type using describe.

Work with a finite table

In a scratch Nushell session create this literal table and query it:

let jobs = [[name size]; [alpha 4] [beta 9] [gamma 12]]
$jobs | where size > 5 | get name

The expected name values are beta and gamma. The semicolon separates column headers from rows. The where stage retains records with size above five; get name extracts a list of cell values rather than preserving the whole table. Compare with select name, which keeps tabular records. Record the Nushell version, the describe result and the visible values. This recipe has not been run in the connected shell.

A safe failed lookup

Try $jobs | get nonexistent and capture the actual cell-path error. This is structurally different from a table that has a name column but no matching rows. Also create a row with a string in the size column and observe what the comparison operator does under the installed version; do not claim every Nushell release silently coerces strings to integers.

Extending the pipeline

Add a delta row with size 5 and verify that the strict > 5 excludes it while >= 5 includes it. Next sort by size and explain whether the projected names preserve the same ordering. If you process file metadata with ls, limit the exercise to a disposable directory: real path names or permission errors introduce unrelated variables.

Completion evidence

Submit the starter result, typed table result, the failing column lookup and a boundary test at size 5. Explain the distinctions among list, table and record, and why errors in a command pipeline deserve capture rather than being hidden by a generic fallback. Reference: https://www.nushell.sh/book/

NuShell — Focused mastery pathway

Foundation exercise

Create a Nushell table from a test CSV and inspect typed columns rather than converting every row to an unstructured string.

Problem-solving exercise

Use where, select and group-by to summarize rows; verify the output schema and observe how null fields differ from empty strings.

Independent project

Build a read-only inventory pipeline for one temporary directory and export results to JSON. Compare pipeline errors with a legitimate empty table; avoid assuming Bash syntax.