Filter/search prompt Choices

[badranx]

How to achieve the same behavior of prompt_toolkit.shortcuts.choice but with an extra text input field that allows user to filter/search choices. Is there a library based on prompt-toolkit that support that already?

Thanks

[ZeLoExE]

You can do this by combining a Buffer for the search input with a dynamic FormattedTextControl to render only matching choices.

Example:

from prompt_toolkit import prompt
from prompt_toolkit.completion import WordCompleter

choices = [
    "Google",
    "Cloudflare",
    "Quad9",
    "AdGuard",
    "OpenDNS",
]

completer = WordCompleter(
    choices,
    ignore_case=True,
)

result = prompt(
    "Pick a provider: ",
    completer=completer,
)

print(result)

For a more custom UI with live filtering and keyboard navigation, the usual approach is:

• use a Buffer to capture the search text
• filter the choices when the buffer changes
• render the results with FormattedTextControl
• handle selection with key bindings

Also, prompt_toolkit.shortcuts.radiolist_dialog can help with selection UIs, but for a VS Code-style Ctrl+P search you still need to add the filtering layer yourself.

[ZeLoExE]

You can do this by combining a Buffer for the search input with a dynamic FormattedTextControl to render only matching choices.

Example:

from prompt_toolkit import prompt
from prompt_toolkit.completion import WordCompleter

choices = [
    "Google",
    "Cloudflare",
    "Quad9",
    "AdGuard",
    "OpenDNS",
]

completer = WordCompleter(
    choices,
    ignore_case=True,
)

result = prompt(
    "Pick a provider: ",
    completer=completer,
)

print(result)

For a more custom UI with live filtering and keyboard navigation, the usual approach is:

• use a Buffer to capture the search text
• filter the choices when the buffer changes
• render the results with FormattedTextControl
• handle selection with key bindings

Also, prompt_toolkit.shortcuts.radiolist_dialog can help with selection UIs, but for a VS Code-style Ctrl+P search you still need to add the filtering layer yourself.

I should also mention that if you are looking for a ready-made solution, you may want to check projects built on top of prompt-toolkit such as questionary or InquirerPy. They provide interactive selection prompts and can be extended with filtering/search behavior.

For a fully custom Ctrl+P-style picker, building it directly with prompt-toolkit primitives gives more control.

[tayfuryldz]

There are two different UI models here, and prompt_toolkit currently provides the building blocks for both but not a single choice(..., searchable=True) shortcut.

If you are happy with completion-menu behavior, the smallest solution is a normal prompt with FuzzyWordCompleter. prompt_toolkit already implements fuzzy scoring/highlighting there, so you don't need to maintain your own filtering algorithm.

If you specifically need the choice shortcut's persistent list/navigation UI plus a search field, I would compose it rather than force a Completer into ChoiceInput: use a Buffer/BufferControl for the query and a RadioList (or FormattedTextControl for more control) for the filtered choices. Recompute the visible values when the query buffer changes, preserve the selected value when it is still present, and use shared key bindings/focus management for Up/Down/Enter.

The existing answer's WordCompleter example is prefix completion, not fuzzy choice filtering. Current prompt_toolkit actually ships FuzzyCompleter and FuzzyWordCompleter, so for a command-palette-like input where the completion menu is acceptable, use those directly.

For a VS Code Ctrl+P-style full-screen selector, the custom two-control composition remains the appropriate route: one editable Buffer owns the search text; the list owns selection. Keeping those states separate also avoids making the selected choice overwrite the query text.