How do I delete specific history entries in Nushell?

Question

Given a history like this in Nushell, how do I delete specific entries; e.g. entries 6, 8, and 10?

nu > history
╭────┬────────────────────╮
│  # │      command       │
├────┼────────────────────┤
│  0 │ history --clear    │
│  1 │ php --version      │
│  2 │ composer --version │
│  3 │ node --version     │
│  4 │ npm --version      │
│  5 │ composer --version │
│  6 │ history            │
│  7 │ php --version      │
│  8 │ history            │
│  9 │ php --version      │
│ 10 │ history            │
│ 11 │ composer --version │
╰────┴────────────────────╯

Answer 1

nu's history command does not (yet) provide a functionality to delete items similar to bash's history -d. However, you can query where the history file is located using $nu.history-path, then use drop nth to delete the lines in question.

open $nu.history-path | lines | drop nth 6 8 10 | save -f $nu.history-path

Answer 2

Based on the code that I can read here and on the documentation here, it appears that an option like this is not currently available.

However, the code indicates that the history.txt file is located at ~/.config/nushell. Utilizing this information, it is possible to accomplish what you asked by using my script below:

nushell_history_manager.py

import os
import sys

def delete_lines(file_path, line_numbers):
    # open the file in read mode
    with open(file_path, 'r') as f:
        # read all the lines and store them in a list
        lines = f.readlines()
    # open the file in write mode
    with open(file_path, 'w') as f:
        for i, line in enumerate(lines):
            # check if the current line number is not in the list of line numbers to delete
            if i+1 not in line_numbers:
                # if it's not, write the line to the file
                f.write(line)

def print_table(file_path):
    # open the file in read mode
    with open(file_path, 'r') as f:
        # read all the lines and store them in a list
        lines = f.readlines()
    # print the table header
    print("╭──────┬───────────────────────────────╮")
    print("│  ##  │            command            │")
    print("├──────┼───────────────────────────────┤")
    for i, line in enumerate(lines):
        # print each line number and the corresponding command
        print(f"│ {i+1:3}  │ {line.strip():26}    │")
    # print the table footer
    print("╰──────┴───────────────────────────────╯")

if __name__ == '__main__':
    # set the file path to the history.txt file in the nushell config directory
    file_path = os.path.expanduser('~/.config/nushell/history.txt')
    # print the initial contents of the file in a table format
    print_table(file_path)
    # ask the user to enter line numbers to delete
    line_numbers_str = input("Enter line numbers to delete (separated by commas): ")
    # convert the entered line numbers to a list of integers
    line_numbers = list(map(int, line_numbers_str.split(',')))
    # delete the specified lines from the file
    delete_lines(file_path, line_numbers)
    # print the updated contents of the file in a table format
    print_table(file_path)

Usage:

python nushell_history_manager.py

Runtime:

╭──────┬────────────────────╮
│  ##  │      command       │
├──────┼────────────────────┤
│   1  │ history --clear    │
│   2  │ php --version      │
│   3  │ composer --version │
│   4  │ node --version     │
│   5  │ npm --version      │
│   6  │ composer --version │
│   7  │ history            │
│   8  │ php --version      │
│   9  │ history            │
│  10  │ php --version      │
│  11  │ history            │
│  12  │ composer --version │
╰──────┴────────────────────╯

Enter line numbers to delete (separated by commas): 7,9,11

╭──────┬────────────────────╮
│  ##  │      command       │
├──────┼────────────────────┤
│   1  │ history --clear    │
│   2  │ php --version      │
│   3  │ composer --version │
│   4  │ node --version     │
│   5  │ npm --version      │
│   6  │ composer --version │
│   7  │ php --version      │
│   8  │ php --version      │
│   9  │ composer --version │
╰──────┴────────────────────╯

Answer 3

While @pmf response does EXACTLY what you ask, a more generic form is shown below. Here you don't even have to know the lines of the file that are duplicates:

export def dedupeLines [filepath: string = $"($nu.history-path)"] {
   open $filepath | lines | uniq | save --force $filepath  
}

In your specific case, filepath = $nu.history-path.

Executing the command below will accomplish your request

dedupeLines # since $nu.history-path is the default filepath for the command

Also for the future, Discords nushell has many questions and answers and the fellows there are extremely helpful and prompt in responding to queries. I had actually asked your question there before too.

I don't know if the python solution lines of code could be reduced, but it is interesting what nushell accomplishes in a single line when compared to the python solution.