🙋 Asking

Asking is a Python package for asking questions and getting answers.

asciicast

Getting started

How does Asking work?

In a nutshell, your Python script invokes Asking with two inputs:

  1. A script to follow.
  2. A dictionary to read default values out of and populate with new values.

When the script ends, Asking returns some state to describe the success of the interaction, then the populated dictionary is yours to do with as you please.

Installation

Asking requires Python 3.8 or later.

Install Asking via pip:

pip install asking

Quick-start example

Download the sample script to your working directory then run asking to perform it:

asking sample.asking.yaml

Note that this will perform the script, but will not load or save any responses. Performing a script via the CLI application is intended only for testing.

Scripts

Script structure

A script is essentially a dictionary of stages, and a stage is a list of actions.

A script always starts with a stage named "start", and actions are invoked sequentially in order.

Asking for a value

start:
  - ask:
      question: Name?
      key: user.name
      branches:
        - response: "^.+$"
          then:
            - stop: true

This script asks the user for their name then exits.

The script contains just one stage, named "start". The stage contains just one action: an "ask" action.

The "ask" action contains a "question", which is printed to the screen. The "key" describes where in the dictionary the user's answer should be saved. "user.name" describes the "name" key of the "user" sub dictionary.

The "branches" describe how to react to the user's answer. "response" is a regular expression. Asking checks the user's answer against each expression in order, and uses the first one that matches. In this case, there's only one choice. "then" describes the actions to invoke on that branch and, in this case, we "stop" and send truthiness back to the host application.

For example:

from asking import demo

script = """
start:
  - ask:
      question: Name?
      key: user.name
      branches:
        - response: "^.+$"
          then:
            - stop: true
"""

# demo() runs non-interactively with preset directions and
# is intended for demonstrating only. See the usage
# documentation for implementation guidance.

demo(script, directions={"user": {"name": "Bobby Pringles"}})

Name?
: Bobby Pringles

Stop reason: True
Responses:
{
  "user": {
    "name": "Bobby Pringles"
  }
}

Offering the previous value as the default

start:
  - ask:
      question: Name?
      key: user.name
      recall: true
      branches:
        - response: "^.+$"
          then:
            - stop: true

This example is identical to the previous except "recall" is now truthy. Asking will read the dictionary's current "key" value and offer it as the default value.

For example:

from asking import demo

script = """
start:
  - ask:
      question: Name?
      key: user.name
      recall: true
      branches:
        - response: "^.+$"
          then:
            - stop: true
"""

# demo() runs non-interactively with preset directions and
# is intended for demonstrating only. See the usage
# documentation for implementation guidance.

demo(
    script,
    responses={
        "user": {
            "name": "Bobby Sprinkles",
        },
    },
    directions={
        "user": {
            "name": "Bobby Pringles",
        },
    },
)

Name? (default: Bobby Sprinkles)
: Bobby Pringles

Stop reason: True
Responses:
{
  "user": {
    "name": "Bobby Pringles"
  }
}

Offering multiple choice responses

start:
  - ask:
      question: Cake?
      branches:
        - response: y
          then:
            - stop: true
        - response: n
          then:
            - stop: false

In this example, the responses are plain strings rather than regular expressions. Asking will present all the available options.

If the user enters "y" then Asking will stop and return truthiness to the host application. If the user enters "n" then Asking will stop and return falsiness.

For example:

from asking import demo

script = """
start:
  - ask:
      question: Cake?
      key: cake
      branches:
        - response: y
          then:
            - stop: true
        - response: n
          then:
            - stop: false
"""

# demo() runs non-interactively with preset directions and
# is intended for demonstrating only. See the usage
# documentation for implementation guidance.

demo(script, directions={"cake": "n"})

Cake? (y/n)
: n

Stop reason: False
Responses:
{
  "cake": "n"
}

Offering multiple choice responses with a default

start:
  - ask:
      question: Cake?
      branches:
        - response: [y, ""]
          then:
            - stop: true
        - response: n
          then:
            - stop: false

This is identical to the previous example, except now the first checked response is a list of options rather than a single string. Asking will use the first branch if the user hits enter without entering a value.

For example:

from asking import demo

script = """
start:
  - ask:
      question: Cake?
      key: cake
      branches:
        - response: [y, ""]
          then:
            - stop: true
        - response: n
          then:
            - stop: false
"""

# demo() runs non-interactively with preset directions and
# is intended for demonstrating only. See the usage
# documentation for implementation guidance.

demo(script, directions={"cake": ""})

Cake? (Y/n)
: 

Stop reason: True
Responses:
{
  "cake": ""
}

Dynamic values

start:
  - text: Today's date is {today}.
  - ask:
      question: Is this correct?
      branches:
        - response: y
          then:
            - stop: true
        - response: n
          then:
            - stop: false

To set some text value at runtime, specify the key in braces and include the value in the state when performing the script:

from asking import ask, State, YamlStringLoader
from datetime import date


loader = YamlStringLoader("""
start:
  - text: Today's date is {today}.
  - ask:
      question: Is this correct?
      key: correct
      branches:
        - response: y
          then:
            - stop: true
        - response: n
          then:
            - stop: false
""")

state = State(
    {},
    color=False,
    directions={
      "correct": "y",
    },
    references={
        "today": date.today(),
    },
)

ask(loader, state)

Today's date is 2021-11-30.

Is this correct? (y/n)
: y

Multiple stages

start:
  - ask:
      question: Cake?
      key: cake
      branches:
        - response: [y, ""]
          then:
            - goto: cake
        - response: n
          then:
            - goto: tea

cake:
  - ask:
      question: Which cake?
      key: which_cake
      branches:
        - response: "^.+$"
          then:
            - goto: tea

tea:
  - ask:
      question: Tea?
      key: tea
      branches:
        - response: [y, ""]
          then:
            - goto: final
        - response: n
          then:
            - goto: final

final:
  - text: Thank you for your order!
  - stop: true

This example presents multiple questions. If the user wants cake then they're asked which one. If they don't want cake, or if they answered which cake they want, then they're asked if they want tea. Either way, the script ends with a thank-you message then stops successfully.

For example:

from asking import demo

script = """
start:
  - ask:
      question: Cake?
      key: cake
      branches:
        - response: [y, ""]
          then:
            - goto: cake
        - response: n
          then:
            - goto: tea

cake:
  - ask:
      question: Which cake?
      key: which_cake
      branches:
        - response: "^.+$"
          then:
            - goto: tea

tea:
  - ask:
      question: Tea?
      key: tea
      branches:
        - response: [y, ""]
          then:
            - goto: final
        - response: n
          then:
            - goto: final

final:
  - text: Thank you for your order!
  - stop: true
"""

# demo() runs non-interactively with preset directions and
# is intended for demonstrating only. See the usage
# documentation for implementation guidance.

demo(
    script,
    directions={
        "cake": "",
        "which_cake": "Battenberg",
        "tea": "",
    },
)

Cake? (Y/n)
: 

Which cake?
: Battenberg

Tea? (Y/n)
: 

Thank you for your order!

Stop reason: True
Responses:
{
  "cake": "",
  "tea": "",
  "which_cake": "Battenberg"
}

Usage

Loaders

Asking offers the following options for loading a YAML script:

States

A State instance must be made available to your script. This describes the dictionary to read/write and any dynamic values.

Performing a script

Essentially, a script is performed by:

  1. Constructing a loader.
  2. Constructing a state, including your responses dictionary.
  3. Passing the loader and state into ask().
from asking import ask, State, YamlResourceLoader

responses = {}

loader = YamlResourceLoader(__package__, "sample.asking.yml")
state = State(responses)

stop_reason = ask(loader, state)

When ask() is complete, it will return whatever value the performance's "stop" action returned, and the responses dictionary will be populated with the user's answers.

Running non-interactively and unit testing

By default, Asking will -- naturally -- ask users for input.

To test your script without human interaction, you can pass directions into the state. This is a dictionary containing the value to respond with for each question key.

For example:

from asking import ask, State, FileLoader
from pprint import pprint


responses = {}

loader = FileLoader("sample.asking.yml")
state = State(
    responses,
    color=False,
    directions={
        "ready": "y",
        "user": {
            "name": "Bobby Pringles",
            "smell": "Sulphur",
        },
        "save": "y",
    },
)

stop_reason = ask(loader, state)

print("Stop reason:", stop_reason)
pprint(responses)

Asking Demo

Welcome to the Asking demo!

Are you ready? (y/N)
: y

Great! Let's get started!

What's your name?
: Bobby Pringles

What's your favourite smell?
: Sulphur

Here's your configuration:

{
  "ready": "y",
  "user": {
    "name": "Bobby Pringles",
    "smell": "Sulphur"
  }
}

Okay to save? (y/n)
: y

Stop reason: True
{'ready': 'y',
 'save': 'y',
 'user': {'name': 'Bobby Pringles', 'smell': 'Sulphur'}}

These directions also allow you to invoke Asking in unit tests, and assert on stop reasons and dictionary values for given inputs.

Actions

ask

ask:
  branches:
    - response: string or string list, required
      then:
        - action
  key:      string
  question: string, required
  recall:   boolean, default=False

"ask" asks a question.

key describes the path in the responses dictionary to read/write the value. Use . as a path separator.

recall indicates whether to load the response dictionary's current value as the default answer.

branches describes the possible reactions to the user's answer.

response can be a string value, list of string values, or a regular expression. The first branch with a matching response will be followed.

then describes the list of actions to perform when the branch is followed.

goto

goto: stage

When a "goto" action is encountered then execution will jump immediately to the specified stage.

responses

responses: json

The "responses" action prints the current values of the response dictionary.

stop

stop: any

The "stop" action immediately stops the script and returns the reason to the host application.

The reason is required but can be any value; even a list or dictionary.

text

text: string

The "text" action simply prints a line of text.

title

title: string

The "title" action prints a line of text formatted as a title.

Project

Contributing

To contribute a bug report, enhancement or feature request, please raise an issue at github.com/cariad/asking/issues.

If you want to contribute a code change, please raise an issue first so we can chat about the direction you want to take.

Licence

Asking is released at github.com/cariad/asking under the MIT Licence.

See LICENSE for more information.

Author

Hello! 👋 I'm Cariad Eccleston and I'm a freelance DevOps and backend engineer. My contact details are available on my personal wiki at cariad.earth.

Please consider supporting my open source projects by sponsoring me on GitHub.

Acknowledgements