Tasks Shell Input installs statistics

Tasks: Shell Input

This extension aims to extend the possibilities of input in task execution. Currently, VSCode supports 3 types of inputs for your tasks:

• promptString
• pickString
• command

None of them allows to get an input from a system command for example. This extension executes a shell command in your OS and each line of the output will be used as a possible input for your task.

Usage example:

{
  "version": "2.0.0",
  "tasks": [
    {
      "label": "Echo Project File",
      "type": "shell",
      "command": "echo ${input:inputTest}",
      "problemMatcher": []
    }
  ],
  "inputs": [
    {
      "id": "inputTest",
      "type": "command",
      "command": "shellCommand.execute",
      "args": {
          "command": "cat ${file}",
          "cwd": "${workspaceFolder}",
          "env": {
              "WORKSPACE": "${workspaceFolder[0]}",
              "FILE": "${file}",
              "PROJECT": "${workspaceFolderBasename}"
          }
      }
    }
  ]
}

Available Commands

The extension makes two commands available:

1. shellCommand.execute: Runs a system command and shows a dropdown with the returned options. See The command shellCommand.execute
2. shellCommand.promptString: This is a reimplementation of the built-in promptString, but this version can be used in the command of shellCommand.execute. See The command shellCommand.promptString

The command shellCommand.execute

Field Separator

By default the extension returns the exact string value that was produced by the shell command and then shown and selected in 'Quick Pick' dialog. However, sometimes it is useful to show more descriptive information than the internal string value that is returned. This can be done by specifying a fieldSeparator and making the shell command return lines containing multiple fields separated by that value. Supported fields are:

<value>|<label>|<description>|<detail>

Here, <value> is what is returned as input variable and is not shown in the UI. Instead, <label> and <description> are shown on a single line and <detail> is rendered on a separate line. These fields can also include icons such as $(git-merge). All fields except <value> are optional and can be omitted. fieldSeparator can be any string value.

Next example shows a process picker very similar to the built-in ${command:pickProcess}:

{
  "version": "2.0.0",
  "tasks": [
    {
      "label": "Echo Process ID",
      "type": "shell",
      "command": "echo ${input:processID}",
      "problemMatcher": []
    }
  ],
  "inputs": [
    {
      "id": "processID",
      "type": "command",
      "command": "shellCommand.execute",
      "args": {
          "command": "ps axww --no-headers k comm -o '%p|%c|%p|%a' | sed -e 's/^\\s*//' -e 's/\\s*|\\s*/|/g'",
          "fieldSeparator": "|",
          "description": "Select the process to attach to"
      }
    }
  ]
}

VSCode renders it like this:

Arguments for shellCommand.execute

• command: the system command to be executed (must be in PATH). If given as an array, the elements are joined by spaces.
• commandArgs: if provided, command is interpreted as the binary to run and commandArgs are the arguments. This is useful if the binary you want to run has spaces (like C:\Program Files\*). This translates to child_process.execFileSync(command, commandArgs).
• stdin: if provided, this string is sent to the standard input of the command
• stdinResolveVars: whether to perform variable substitution (see below) for stdin (default true).
• cwd: the directory from within it will be executed
• env: key-value pairs to use as environment variables (it won't append the variables to the current existing ones. It will replace instead)
• useFirstResult: skip 'Quick Pick' dialog and use first result returned from the command
• useSingleResult: skip 'Quick Pick' dialog and use the single result if only one returned from the command
• rememberPrevious: remember the value you previously selected and default to it the next time (default false) (:warning: need taskId or rememberAs to be set)
• allowCustomValues: If true, it's possible to enter a new value that is not part of the command output. Has no effect with useFirstResult.
• multiselect: If true, it's possible to select multiple values. They are joined by multiselectSeparator. Has no effect with useFirstResult.
• multiselectSeparator: The string with which to join multiple options when multiselect is true (default " "). Has no effect without multiselect.
• warnOnStderr: If true, a warning message is shown if the command outputs anything on stderr (default: true). Has no effect if stdio is not stdout.
• filterEmptyResults: If true, empty lines output on stdio are filtered out (default: true).
• taskId: Unique id to use for storing the last-used value.
• rememberAs: Specify the taskId of another input to share history with that input.
• fieldSeparator: the string that separates value, label, description and detail fields
• description: shown as a placeholder in 'Quick Pick', provides context for the input
• maxBuffer: largest amount of data in bytes allowed on stdout. Default is 1024 * 1024. If exceeded ENOBUFS error will be displayed
• defaultOptions: if the command doesn't return anything, the list provided will be set for the user to choose from
• stdio: specifies whether to get the results from stdout, stderr, or both. Default is stdout.

As of today, the extension supports variable substitution for:…