like piping to column -t. I generally pipe aligned input to choose, then pipe the output to column. This is common enough that I'd like a short option for it, and I imagine others also work this way. Maybe when passed -t?

I'm happy to open a PR if you like this idea.

It would nice to to able to install it with cargo install choose

Debian package would be nice. I use Fedora most of the time but sometimes need to connect to a Ubuntu server (ew)

I would like to integrate similar functionality to the cut Unix utility and, of course, this tool to a Rust cli app. However, this crate is just a binary, but I need a library. Do you know of any rust libraries with similar functionality to this? Would you ever consider splitting this project into a library and then a front-end cli tool?

In grep I can print what it finds between values using easy to remember syntax for perl-regexp. Could choose be extended to also find text between matches that is easier to use than perl-regexp?

choose -m string1 string2 to print the value found after matching string1 and stopping at string2.

choose -M string1 string2 to print the value including the matched string1 and string2. adding 0 1 2 ... could then print specific characters inside the matched result.

choose -a string1 will look for string1 and print all the text on that line after matching string1.

choose -A string1 will look for string1 and print all the text on that line including string1 that matched. adding 0 1 2 ... could then print specific characters inside the matched result.

Here are some grep examples:

Grep from match to end

Example 1

Text to match

$ gdu --version
Version:         v5.20.0
Built time:      Sat Oct 22 10:48:31 PM CEST 2022
Built user:      dundee

gdu --version | grep -oP '(?<=Version:\t\s).*'
Output:
v5.20.0

Example 2

Text to match

$ openssl x509 -noout -enddate -in /etc/ssl/certs/COMODO_Certification_Authority.pem
notAfter=Dec 31 23:59:59 2029 GMT

Keep only value

openssl x509 -noout -enddate -in /etc/ssl/certs/COMODO_Certification_Authority.pem | grep -oP '(?<=notAfter=).*'
Dec 31 23:59:59 2029 GMT

Grep between matches

Text to match

docker inspect 9512b532dcaf1 | grep tls
                "/etc/dockers/conf/web/tls:/etc/ssl/nginx:ro",
                "Source": "/etc/dockers/conf/web/tls",

docker inspect 9512b532dcaf1 | grep -oP '(?<="Source": ").*(tls)'
/etc/dockers/conf/web/tls

vs

docker inspect 9512b532dcaf1 | grep -oP '(?<="Source": ").*(?=tls)'
/etc/dockers/conf/web

This project is quite similar to one I came across a couple months ago: fex (field extraction). It has a different syntax for selecting fields, and only supports single character delimiters, but it also has some nice features:

Non-greedy String Splitting

Given the input a:b::c:::d, a standard selection of fields 0:2 would print a:b:c, whilst a non-greedy selection would print a:b:. This would probably be hard to implement due to regex delimiters.

Multi-Pass selections

Field selections could have multiple stages, so that each split string could be split with another delimiter and another field selected within. There's a great example on the project for splitting web-server access logs like so.

194.25.90.180 - - [19/Jan/2015:12:58:19 +0000] "GET /run?command=ls HTTP/1.1" 200 115 "-" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/39.0.2171.95 Safari/537.36"

Piping this to fex 1 '"2 2', split the line by spaces and select field 1, then split the line by double-quotes, select field 2, split that by spaces, and select field 2 from the result. This gives:

194.25.90.180 /run?command=ls

Select fields that match regex

Any field that matches the regular expression /regexp/ would be selected.

Nice program!
Can have Window Version?

I'd like choose to take an option under which it treats the first row as headers, and allows selecting columns by name. This saves counting, and is especially nice in scripts where it's more clear what is going on and somewhat more resilient to changes in the input format. Not sure if it should be exact-match, prefix, case-insensitive....

I'm happy to open a PR if you like this idea.

Trying to pipe into head results in repeated Failed to write to output: Broken pipe (os error 32) messages.

Hi,

I work on quite old linux machines and it is not possible to install anything.
Could you provide musl binaries for linux (in the same way ripgrep and fd do)

Many thanks

Hi, Im appreciate your nice works.
I just want to request new feature.
Could you support escape character?
For example:

• I want to set --field-separator as , for .csv file.
• But there is escaped data, like "hello, this is me","123",smaple.

I want to choose first field: hello, this is me but i got a just hello.

Thank you.

choose helpfully offers compatibility with Python's style of 0-indexing, inclusive start, and exclusive end, with -x (which the docs make sound like it uses an exclusive range, while it uses an inclusive start):

$ s=(1 2 3 4 5 6)
$ print $s | choose 2:4
3 4 5
$ print $s | choose -x 2:4
3 4

But when I'm in a shell, using ZSH tools, cut, etc., I expect 1-indexing, inclusive start, inclusive end:

$ s=(1 2 3 4 5 6)
$ print $s[2,4]
2 3 4
$ print $s | cut -d' ' -f2-4
2 3 4

Since the context I'd use choose in is the same I'd use ZSH and cut in, it would lessen the mental load and chance of user error if I could get it to match the indexing and slicing of ZSH (default config) and cut.

Here's the same data as a table, where s is an array of the numbers 1 through 6:

It would be nice to have an inline syntax for explicitly inclusive and explicitly exclusive ranges, not just a separate -x option that changes the behavior of :. That way, it'd be possible to specify both inclusive and exclusive ranges on the same command line, and to not have to go back and change options when typing ranges.

For example, using Rust's syntax:

(A different syntax would be fine, as long as it's self-explanatory.)

Hello! I'm trying package to debian, but have some errors. Do you know what it could be?

test tests::test_process_escapes::test_unicode_crab_emoji ... FAILED test tests::test_process_escapes::test_unicode_u7fff ... FAILED failures: ---- tests::test_process_escapes::test_unicode_crab_emoji stdout ---- thread 'tests::test_process_escapes::test_unicode_crab_emoji' panicked at 'not implemented: escape_ unicode is not yet implemented', src/lib.rs:50:5 stack backtrace: 0: rust_begin_unwind at /usr/src/rustc-1.59.0/library/std/src/panicking.rs:498:5 1: core::panicking::panic_fmt at /usr/src/rustc-1.59.0/library/core/src/panicking.rs:116:14 2: backslash::escape_unicode at /usr/share/cargo/registry/backslash-0.2.0/src/lib.rs:50:5 3: backslash::tests::test_process_escapes::test_unicode_crab_emoji at /usr/share/cargo/registry/backslash-0.2.0/src/lib.rs:219:17 4: backslash::tests::test_process_escapes::test_unicode_crab_emoji::{{closure}} at /usr/share/cargo/registry/backslash-0.2.0/src/lib.rs:216:9 5: core::ops::function::FnOnce::call_once at /usr/src/rustc-1.59.0/library/core/src/ops/function.rs:227:5 6: core::ops::function::FnOnce::call_once at /usr/src/rustc-1.59.0/library/core/src/ops/function.rs:227:5 note: Some details are omitted, run with RUST_BACKTRACE=full for a verbose backtrace. ---- tests::test_process_escapes::test_unicode_u7fff stdout ---- thread 'tests::test_process_escapes::test_unicode_u7fff' panicked at 'not implemented: escape_unico de is not yet implemented', src/lib.rs:50:5 stack backtrace:

I've come across several issues with inclusivity / exclusivity:

readme mistake

The readme.md says:

choose :3              # print the beginning of the line to the 3rd item,
                       # exclusive

Which I take to mean that the item with index 3 won't be included when :3 is used, i.e:
echo 0 1 2 3 | choose :3 should output 0 1 2. However, it outputs 0 1 2 3. Is the readme.md wrong? Did you intend to add a -x there and forgot? Or is that a bug?

python-like

Claiming choose is Python like and then requiring -x for exclusivity is wrong in my opinion, to aim to be more Python-like, choose should be exclusive by default.

Python:

In [1]: [0,1,2,3][:3]
Out[1]: [0, 1, 2]

Choose:

$ echo 0 1 2 3 | choose :3
0 1 2 3

Empty results when using -x

Since I want choose to be more Python-like, I created an alias alias c='choose -x' so it's always exclusive by default. However, when I try to use it like this:

$ echo 0 1 2 3 | c 2

Nothing is output, as shown.

What do you think?

version: commit 97b3abf
environment: linux x86_64

When an input has no newline at EOF, --field-separator option does not work correctly.

$ /bin/echo 'a:b' | target/debug/choose -f ':' 1
b

$ /bin/echo -n 'a:b' | target/debug/choose -f ':' 1
(expected "b", but no output)

Moreover, if an input contains non-ASCII characters, choose can panic:

$ /bin/echo -n 'あ:い' | target/debug/choose -f ':' 1
thread 'main' panicked at 'byte index 6 is not a char boundary; it is inside 'い' (bytes 4..7) of `あ:い`', src/main.rs:77:22
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

I guess this code causes the issue:

This code seems to trim the last character unconditionally, but it will have to check if a line actually has a newline.

$ identify 11095193_p0.jpg 
11095193_p0.jpg JPEG 1200x1200 1200x1200+0+0 8-bit sRGB 695510B 0.000u 0:00.002
$ identify 11095193_p0.jpg | gawk '{split($3,sizes,"x"); print $1,sizes[1]/sizes[2]}'
11095193_p0.jpg 1

I'd like to use choose to replace awk because the filename sometime include whitespace that lead to error, how to rewrite this by choose

Hi,

although the compile process went smooth on my macOS, is there any chance to push this as the brew package? The installation would be much easier...

Thanks,
Lubos

$ cargo build --release
error: failed to parse lock file at: /home/pete/stuff/choose/Cargo.lock

Caused by:
invalid serialized PackageId for key package.dependencies

Would be nice to have a argument to filter by the nth line, similar as the nth element of each line.

$echo "a b c d
e f g
h i j" | choose 1 -l 1
f

what is rust

'\n' as output separator doesn't work. It works literally. Other system utilities, like tr, understand that this is newline.

❯ echo "Hello world" | choose : -o '\n'
Hello\nworld

❯ echo "Hello world" | tr ' ' '\n'
Hello
world

Empty results when using -x

Since I want choose to be more Python-like, I created an alias alias c='choose -x' so it's always exclusive by default. However, when I try to use it like this:

$ echo 0 1 2 3 | c 2

Nothing is output, as shown.

What do you think?

Should be overridden by an inclusive flag on the cli.

This ruins pre-formatted tables and requires extra chaining to return the output to its original state, meaning choose is not a transparent translation utility as much as a complete refactoring.

Examples:

➜ sdiff <(choose 0:4 -i mnemonics.tmp | head ) <(head mnemonics.tmp )
0x2813: 0x0a, 0, 2, 2                                         | 0x2813: 0x0a,  0,   2,            2
0xb537: 0x2d, 6, 17, 34                                       | 0xb537: 0x2d,  6,  17,           34
0x460c: 0x11, 2, 5, 170                                       | 0x460c: 0x11,  2,   5,          170
0x4615: 0x11, 2, 5, 850                                       | 0x4615: 0x11,  2,   5,          850
0xd814: 0x36, 9, 29, 24650                                    | 0xd814: 0x36,  9,  29,        24650
0xa901: 0x2a, 5, 13, 320450                                   | 0xa901: 0x2a,  5,  13,       320450
0xf7f4: 0x3d, 11, 37, 11856650                                | 0xf7f4: 0x3d, 11,  37,     11856650
0x2800: 0x0a, 0, 2, 23713300                                  | 0x2800: 0x0a,  0,   2,     23713300
0xda06: 0x36, 9, 29, 687685700                                | 0xda06: 0x36,  9,  29,    687685700
0x4244: 0x10, 1, 3, 2063057100                                | 0x4244: 0x10,  1,   3,   2063057100

even with columns -t it is slightly off in terms of justification:

➜ sdiff <(choose 0:4 -i mnemonics.tmp | head | column -t) <(head mnemonics.tmp )
0x2813:  0x0a,  0,   2,   2                                   | 0x2813: 0x0a,  0,   2,            2
0xb537:  0x2d,  6,   17,  34                                  | 0xb537: 0x2d,  6,  17,           34
0x460c:  0x11,  2,   5,   170                                 | 0x460c: 0x11,  2,   5,          170
0x4615:  0x11,  2,   5,   850                                 | 0x4615: 0x11,  2,   5,          850
0xd814:  0x36,  9,   29,  24650                               | 0xd814: 0x36,  9,  29,        24650
0xa901:  0x2a,  5,   13,  320450                              | 0xa901: 0x2a,  5,  13,       320450
0xf7f4:  0x3d,  11,  37,  11856650                            | 0xf7f4: 0x3d, 11,  37,     11856650
0x2800:  0x0a,  0,   2,   23713300                            | 0x2800: 0x0a,  0,   2,     23713300
0xda06:  0x36,  9,   29,  687685700                           | 0xda06: 0x36,  9,  29,    687685700
0x4244:  0x10,  1,   3,   2063057100                          | 0x4244: 0x10,  1,   3,   2063057100

reformatting is technically possible with column -t -R 3,4,5, but -R is unsupported on default builds of column. Additionally, this requires specifying all column numbers to be right-justified, which is infeasible for files with large numbers of right-justified columns

➜ choose --version
choose 1.3.0

Hi!
Cool project, I like it!
And easy to use, too, but with tab completions it is even easier. I added some for fish shell:

complete choose --no-files

#flags:
complete choose -s c -l character-wise -d "Choose fields by character number"
complete choose -s d -l debug -d "Activate debug mode"
complete choose -s x -l exclusive -d "Use exclusive ranges, similar to array indexing in many programming languages"
complete choose -x -s h -l help -d "Prints help information"
complete choose -s n -l non-greedy -d "Use non-greedy field separators"
complete choose -l one-indexed -d "Index from 1 instead of 0"
complete choose -s V -l version -d "Prints version information"

#options:
complete choose -x -s f -l field-separator -d "Specify field separator other than whitespace, using Rust `regex` syntax"
complete choose -r -s i -l input -d "Specify input file"
complete choose -x -s o -l output-field-separator -d "Specify output field separator"

Do you want to bundle them with the project, or would you like me to create a pull request in the fish shell repo instead?

The readme.md says:

choose :3              # print the beginning of the line to the 3rd item,
                       # exclusive

Without the -x flag, it shouldn't be exclusive.

Hi! Cool project, very nice alternative to awk/cut.

It would be nice to be able to use choose on an endless stream, passing on output as new lines arrives on stdin. Some examples:

ping 8.8.8.8 | choose -2 | choose -f "=" 1
tail -f log.log | choose 0

Of course using the approach of printing the match after each line is highly inefficient working on files. Maybe it could be a setting --release-output which by default is not used when reading files, but used when reading from stdin?

Issue environment:

If I open a rather long file with many lines,

cat file | choose 0 | less

after existing less, choose will print out the error message with many
Failed to write to output: Broken pipe (os error 32)

Please add an aarch64 linux version as well to your CI.

I'd like to package this for the AUR, but since the project has no license it makes it hard for me to do so.

Personally, I'd recommend something like the ISC or MIT license.

Can I select the 'n' field from 'x' line on multiline input ?

Desired example
$ cat demo.txt
This is a
multiline test

$ cat demo.txt | choose 1 {--line 1} # Desired line selection
test # Desired output

Please check the test result below.

~% echo "1/2/3/4/5" | choose -f '/' -o '/' 1:-1
2/3/4/5
~% echo "1/2/3/4/5" | choose -f '/' -o '/' 1:-2
2/3/4
~% echo "1/2/3/4/5" | choose -f '/' -o '/' 1:-3
2/3
~% echo "1/2/3/4/5" | choose -f '/' -o '/' 1:-4

~%

I expected "2" from the last run. But nothing returned.

Hey there

Would you mind adding a pre-build binary to your GH release ? I never tried building & publishing Rust based code before, but from a past experience using rust based tool, here's a starter.

name: Publish binary

on: 
  push:

jobs:
  ebook:
    name: Build and upload
    runs-on: ubuntu-latest
    steps:
      # Install Rust & its package manager Cargo
      - name: Install cargo
        uses: actions-rs/toolchain@v1
        with:
          toolchain: stable
          profile: minimal

      # Cache installation assets
      - name: Cache cargo registry
        uses: actions/cache@v1
        with:
          path: ~/.cargo/registry
          key: ${{ runner.os }}-cargo-registry-${{ hashFiles('**/Cargo.lock') }}
      - name: Cache cargo index
        uses: actions/cache@v1
        with:
          path: ~/.cargo/git
          key: ${{ runner.os }}-cargo-index-${{ hashFiles('**/Cargo.lock') }}
      - name: Cache cargo build
        uses: actions/cache@v1
        with:
          path: target
          key: ${{ runner.os }}-cargo-build-target-${{ hashFiles('**/Cargo.lock') }}

      # BUILD HERE
      - name: build
        run: cargo ??? 

      # Publish
      - name: Create Release
        id: create-release
        uses: actions/[email protected]
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        with:
          tag_name: ${{ github.ref }}
          release_name: Release ${{ github.ref }}
          draft: false
          prerelease: false

      - name: Upload to release
        id: upload-release-asset
        uses: actions/[email protected]
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        with:
          upload_url: ${{ steps.create-release.outputs.upload_url }}
          asset_path: ./PATH
          asset_name: NAME

Best regards,