Reference¶

Application¶

class prompt_toolkit.application.AppSession(input: Input | None = None, output: Output | None = None)¶

An AppSession is an interactive session, usually connected to one terminal. Within one such session, interaction with many applications can happen, one after the other.

The input/output device is not supposed to change during one session.

Warning: Always use the create_app_session function to create an instance, so that it gets activated correctly.

Parameters:

input – Use this as a default input for all applications running in this session, unless an input is passed to the Application explicitly.
output – Use this as a default output.

class prompt_toolkit.application.Application(layout: Layout | None = None, style: BaseStyle | None = None, include_default_pygments_style: FilterOrBool = True, style_transformation: StyleTransformation | None = None, key_bindings: KeyBindingsBase | None = None, clipboard: Clipboard | None = None, full_screen: bool = False, color_depth: ColorDepth | Callable[[], ColorDepth | None] | None = None, mouse_support: FilterOrBool = False, enable_page_navigation_bindings: None | FilterOrBool = None, paste_mode: FilterOrBool = False, editing_mode: EditingMode = EditingMode.EMACS, erase_when_done: bool = False, reverse_vi_search_direction: FilterOrBool = False, min_redraw_interval: float | int | None = None, max_render_postpone_time: float | int | None = 0.01, refresh_interval: float | None = None, terminal_size_polling_interval: float | None = 0.5, cursor: AnyCursorShapeConfig = None, on_reset: ApplicationEventHandler[_AppResult] | None = None, on_invalidate: ApplicationEventHandler[_AppResult] | None = None, before_render: ApplicationEventHandler[_AppResult] | None = None, after_render: ApplicationEventHandler[_AppResult] | None = None, input: Input | None = None, output: Output | None = None)¶

The main Application class! This glues everything together.

Parameters:

layout – A Layout instance.
key_bindings – KeyBindingsBase instance for the key bindings.
clipboard – Clipboard to use.
full_screen – When True, run the application on the alternate screen buffer.
color_depth – Any ColorDepth value, a callable that returns a ColorDepth or None for default.
erase_when_done – (bool) Clear the application output when it finishes.
reverse_vi_search_direction – Normally, in Vi mode, a ‘/’ searches forward and a ‘?’ searches backward. In Readline mode, this is usually reversed.
min_redraw_interval –
Number of seconds to wait between redraws. Use this for applications where invalidate is called a lot. This could cause a lot of terminal output, which some terminals are not able to process.

None means that every invalidate will be scheduled right away (which is usually fine).

When one invalidate is called, but a scheduled redraw of a previous invalidate call has not been executed yet, nothing will happen in any case.
max_render_postpone_time – When there is high CPU (a lot of other scheduled calls), postpone the rendering max x seconds. ‘0’ means: don’t postpone. ‘.5’ means: try to draw at least twice a second.
refresh_interval – Automatically invalidate the UI every so many seconds. When None (the default), only invalidate when invalidate has been called.
terminal_size_polling_interval – Poll the terminal size every so many seconds. Useful if the applications runs in a thread other then then main thread where SIGWINCH can’t be handled, or on Windows.

Filters:

Parameters:

mouse_support – (Filter or boolean). When True, enable mouse support.
paste_mode – Filter or boolean.
editing_mode – EditingMode.
enable_page_navigation_bindings – When True, enable the page navigation key bindings. These include both Emacs and Vi bindings like page-up, page-down and so on to scroll through pages. Mostly useful for creating an editor or other full screen applications. Probably, you don’t want this for the implementation of a REPL. By default, this is enabled if full_screen is set.

Callbacks (all of these should accept an Application object as input.)

Parameters:

on_reset – Called during reset.
on_invalidate – Called when the UI has been invalidated.
before_render – Called right before rendering.
after_render – Called right after rendering.

I/O: (Note that the preferred way to change the input/output is by creating an AppSession with the required input/output objects. If you need multiple applications running at the same time, you have to create a separate AppSession using a with create_app_session(): block.

Parameters:

input – Input instance.
output – Output instance. (Probably Vt100_Output or Win32Output.)

Usage:

app = Application(…) app.run()

# Or await app.run_async()

async cancel_and_wait_for_background_tasks() → None¶

Cancel all background tasks, and wait for the cancellation to complete. If any of the background tasks raised an exception, this will also propagate the exception.

(If we had nurseries like Trio, this would be the __aexit__ of a nursery.)

property color_depth: ColorDepth¶

The active ColorDepth.

The current value is determined as follows:

If a color depth was given explicitly to this application, use that value.
Otherwise, fall back to the color depth that is reported by the Output implementation. If the Output class was created using output.defaults.create_output, then this value is coming from the $PROMPT_TOOLKIT_COLOR_DEPTH environment variable.

cpr_not_supported_callback() → None¶: Called when we don’t receive the cursor position response in time.

create_background_task(coroutine: Coroutine[Any, Any, None]) → Task[None]¶

Start a background task (coroutine) for the running application. When the Application terminates, unfinished background tasks will be cancelled.

Given that we still support Python versions before 3.11, we can’t use task groups (and exception groups), because of that, these background tasks are not allowed to raise exceptions. If they do, we’ll call the default exception handler from the event loop.

If at some point, we have Python 3.11 as the minimum supported Python version, then we can use a TaskGroup (with the lifetime of Application.run_async(), and run run the background tasks in there.

This is not threadsafe.

property current_buffer: Buffer¶

The currently focused Buffer.

(This returns a dummy Buffer when none of the actual buffers has the focus. In this case, it’s really not practical to check for None values or catch exceptions every time.)

property current_search_state: SearchState¶: Return the current SearchState. (The one for the focused BufferControl.)

exit(result: _AppResult | None = None, exception: BaseException | type[BaseException] | None = None, style: str = '') → None¶

Exit application.

Note

If Application.exit is called before Application.run() is called, then the Application won’t exit (because the Application.future doesn’t correspond to the current run). Use a pre_run hook and an event to synchronize the closing if there’s a chance this can happen.

Parameters:

result – Set this result for the application.
exception – Set this exception as the result for an application. For a prompt, this is often EOFError or KeyboardInterrupt.
style – Apply this style on the whole content when quitting, often this is ‘class:exiting’ for a prompt. (Used when erase_when_done is not set.)

get_used_style_strings() → list[str]¶: Return a list of used style strings. This is helpful for debugging, and for writing a new Style.

invalidate() → None¶: Thread safe way of sending a repaint trigger to the input event loop.

property invalidated: bool¶: True when a redraw operation has been scheduled.

property is_running: bool¶: True when the application is currently active/running.

key_processor¶: The InputProcessor instance.

print_text(text: AnyFormattedText, style: BaseStyle | None = None) → None¶

Print a list of (style_str, text) tuples to the output. (When the UI is running, this method has to be called through run_in_terminal, otherwise it will destroy the UI.)

Parameters:

text – List of (style_str, text) tuples.
style – Style class to use. Defaults to the active style in the CLI.

quoted_insert¶: Quoted insert. This flag is set if we go into quoted insert mode.

render_counter¶: Render counter. This one is increased every time the UI is rendered. It can be used as a key for caching certain information during one rendering.

reset() → None¶: Reset everything, for reading the next input.

run(pre_run: Callable[[], None] | None = None, set_exception_handler: bool = True, handle_sigint: bool = True, in_thread: bool = False, inputhook: Callable[[InputHookContext], None] | None = None) → _AppResult¶

A blocking ‘run’ call that waits until the UI is finished.

This will run the application in a fresh asyncio event loop.

Parameters:

pre_run – Optional callable, which is called right after the “reset” of the application.
set_exception_handler – When set, in case of an exception, go out of the alternate screen and hide the application, display the exception, and wait for the user to press ENTER.
in_thread – When true, run the application in a background thread, and block the current thread until the application terminates. This is useful if we need to be sure the application won’t use the current event loop (asyncio does not support nested event loops). A new event loop will be created in this background thread, and that loop will also be closed when the background thread terminates. When this is used, it’s especially important to make sure that all asyncio background tasks are managed through get_appp().create_background_task(), so that unfinished tasks are properly cancelled before the event loop is closed. This is used for instance in ptpython.
handle_sigint – Handle SIGINT signal. Call the key binding for Keys.SIGINT. (This only works in the main thread.)

async run_async(pre_run: Callable[[], None] | None = None, set_exception_handler: bool = True, handle_sigint: bool = True, slow_callback_duration: float = 0.5) → _AppResult¶

Run the prompt_toolkit Application until exit() has been called. Return the value that was passed to exit().

This is the main entry point for a prompt_toolkit Application and usually the only place where the event loop is actually running.

Parameters:

pre_run – Optional callable, which is called right after the “reset” of the application.
set_exception_handler – When set, in case of an exception, go out of the alternate screen and hide the application, display the exception, and wait for the user to press ENTER.
handle_sigint – Handle SIGINT signal if possible. This will call the <sigint> key binding when a SIGINT is received. (This only works in the main thread.)
slow_callback_duration – Display warnings if code scheduled in the asyncio event loop takes more time than this. The asyncio default of 0.1 is sometimes not sufficient on a slow system, because exceptionally, the drawing of the app, which happens in the event loop, can take a bit longer from time to time.

async run_system_command(command: str, wait_for_enter: bool = True, display_before_text: AnyFormattedText = '', wait_text: str = 'Press ENTER to continue...') → None¶

Run system command (While hiding the prompt. When finished, all the output will scroll above the prompt.)

Parameters:

command – Shell command to be executed.
wait_for_enter – FWait for the user to press enter, when the command is finished.
display_before_text – If given, text to be displayed before the command executes.

Returns:

A Future object.

suspend_to_background(suspend_group: bool = True) → None¶

(Not thread safe – to be called from inside the key bindings.) Suspend process.

Parameters:: suspend_group – When true, suspend the whole process group. (This is the default, and probably what you want.)

timeoutlen¶: Like Vim’s timeoutlen option. This can be None or a float. For instance, suppose that we have a key binding AB and a second key binding A. If the uses presses A and then waits, we don’t handle this binding yet (unless it was marked ‘eager’), because we don’t know what will follow. This timeout is the maximum amount of time that we wait until we call the handlers anyway. Pass None to disable this timeout.

ttimeoutlen¶: When to flush the input (For flushing escape keys.) This is important on terminals that use vt100 input. We can’t distinguish the escape key from for instance the left-arrow key, if we don’t know what follows after “x1b”. This little timer will consider “x1b” to be escape if nothing did follow in this time span. This seems to work like the ttimeoutlen option in Vim.

vi_state¶: Vi state. (For Vi key bindings.)

class prompt_toolkit.application.DummyApplication¶: When no Application is running, get_app() will run an instance of this DummyApplication instead.

prompt_toolkit.application.create_app_session(input: Input | None = None, output: Output | None = None) → Generator[AppSession, None, None]¶

Create a separate AppSession.

This is useful if there can be multiple individual AppSession’s going on. Like in the case of a Telnet/SSH server.

prompt_toolkit.application.get_app() → Application[Any]¶

Get the current active (running) Application. An Application is active during the Application.run_async() call.

We assume that there can only be one Application active at the same time. There is only one terminal window, with only one stdin and stdout. This makes the code significantly easier than passing around the Application everywhere.

If no Application is running, then return by default a DummyApplication. For practical reasons, we prefer to not raise an exception. This way, we don’t have to check all over the place whether an actual Application was returned.

(For applications like pymux where we can have more than one Application, we’ll use a work-around to handle that.)

prompt_toolkit.application.get_app_or_none() → Application[Any] | None¶: Get the current active (running) Application, or return None if no application is running.

prompt_toolkit.application.in_terminal(render_cli_done: bool = False) → AsyncGenerator[None, None]¶

Asynchronous context manager that suspends the current application and runs the body in the terminal.

async def f():
    async with in_terminal():
        call_some_function()
        await call_some_async_function()

prompt_toolkit.application.run_in_terminal(func: Callable[[], _T], render_cli_done: bool = False, in_executor: bool = False) → Awaitable[_T]¶

Run function on the terminal above the current application or prompt.

What this does is first hiding the prompt, then running this callable (which can safely output to the terminal), and then again rendering the prompt which causes the output of this function to scroll above the prompt.

func is supposed to be a synchronous function. If you need an asynchronous version of this function, use the in_terminal context manager directly.

Parameters:

func – The callable to execute.
render_cli_done – When True, render the interface in the ‘Done’ state first, then execute the function. If False, erase the interface first.
in_executor – When True, run in executor. (Use this for long blocking functions, when you don’t want to block the event loop.)

Returns:

A Future.

prompt_toolkit.application.set_app(app: Application[Any]) → Generator[None, None, None]¶

Context manager that sets the given Application active in an AppSession.

This should only be called by the Application itself. The application will automatically be active while its running. If you want the application to be active in other threads/coroutines, where that’s not the case, use contextvars.copy_context(), or use Application.context to run it in the appropriate context.

Formatted text¶

Many places in prompt_toolkit can take either plain text, or formatted text. For instance the prompt() function takes either plain text or formatted text for the prompt. The FormattedTextControl can also take either plain text or formatted text.

In any case, there is an input that can either be just plain text (a string), an HTML object, an ANSI object or a sequence of (style_string, text) tuples. The to_formatted_text() conversion function takes any of these and turns all of them into such a tuple sequence.

class prompt_toolkit.formatted_text.ANSI(value: str)¶

ANSI formatted text. Take something ANSI escaped text, for use as a formatted string. E.g.

ANSI('\x1b[31mhello \x1b[32mworld')

Characters between \001 and \002 are supposed to have a zero width when printed, but these are literally sent to the terminal output. This can be used for instance, for inserting Final Term prompt commands. They will be translated into a prompt_toolkit ‘[ZeroWidthEscape]’ fragment.

format(*args: str, **kwargs: str) → ANSI¶: Like str.format, but make sure that the arguments are properly escaped. (No ANSI escapes can be injected.)

class prompt_toolkit.formatted_text.FormattedText(iterable=(), /)¶

A list of (style, text) tuples.

(In some situations, this can also be (style, text, mouse_handler) tuples.)

class prompt_toolkit.formatted_text.HTML(value: str)¶

HTML formatted text. Take something HTML-like, for use as a formatted string.

# Turn something into red.
HTML('<style fg="ansired" bg="#00ff44">...</style>')

# Italic, bold, underline and strike.
HTML('<i>...</i>')
HTML('<b>...</b>')
HTML('<u>...</u>')
HTML('<s>...</s>')

All HTML elements become available as a “class” in the style sheet. E.g. <username>...</username> can be styled, by setting a style for username.

format(*args: object, **kwargs: object) → HTML¶: Like str.format, but make sure that the arguments are properly escaped.

class prompt_toolkit.formatted_text.PygmentsTokens(token_list: list[tuple[Token, str]])¶: Turn a pygments token list into a list of prompt_toolkit text fragments ((style_str, text) tuples).

class prompt_toolkit.formatted_text.Template(text: str)¶

Template for string interpolation with formatted text.

Example:

Template(' ... {} ... ').format(HTML(...))

Parameters:: text – Plain text.

prompt_toolkit.formatted_text.fragment_list_len(fragments: StyleAndTextTuples) → int¶

Return the amount of characters in this text fragment list.

Parameters:: fragments – List of (style_str, text) or (style_str, text, mouse_handler) tuples.

prompt_toolkit.formatted_text.fragment_list_to_text(fragments: StyleAndTextTuples) → str¶

Concatenate all the text parts again.

Parameters:: fragments – List of (style_str, text) or (style_str, text, mouse_handler) tuples.

prompt_toolkit.formatted_text.fragment_list_width(fragments: StyleAndTextTuples) → int¶

Return the character width of this text fragment list. (Take double width characters into account.)

Parameters:: fragments – List of (style_str, text) or (style_str, text, mouse_handler) tuples.

prompt_toolkit.formatted_text.is_formatted_text(value: object) → TypeGuard[AnyFormattedText]¶: Check whether the input is valid formatted text (for use in assert statements). In case of a callable, it doesn’t check the return type.

prompt_toolkit.formatted_text.merge_formatted_text(items: Iterable[AnyFormattedText]) → AnyFormattedText¶: Merge (Concatenate) several pieces of formatted text together.

prompt_toolkit.formatted_text.split_lines(fragments: Iterable[OneStyleAndTextTuple]) → Iterable[StyleAndTextTuples]¶

Take a single list of (style_str, text) tuples and yield one such list for each line. Just like str.split, this will yield at least one item.

Parameters:: fragments – Iterable of (style_str, text) or (style_str, text, mouse_handler) tuples.

prompt_toolkit.formatted_text.to_formatted_text(value: AnyFormattedText, style: str = '', auto_convert: bool = False) → FormattedText¶

Convert the given value (which can be formatted text) into a list of text fragments. (Which is the canonical form of formatted text.) The outcome is always a FormattedText instance, which is a list of (style, text) tuples.

It can take a plain text string, an HTML or ANSI object, anything that implements __pt_formatted_text__ or a callable that takes no arguments and returns one of those.

Parameters:

style – An additional style string which is applied to all text fragments.
auto_convert – If True, also accept other types, and convert them to a string first.

prompt_toolkit.formatted_text.to_plain_text(value: AnyFormattedText) → str¶: Turn any kind of formatted text back into plain text.

Buffer¶

Data structures for the Buffer. It holds the text, cursor position, history, etc…

class prompt_toolkit.buffer.Buffer(completer: Completer | None = None, auto_suggest: AutoSuggest | None = None, history: History | None = None, validator: Validator | None = None, tempfile_suffix: str | Callable[[], str] = '', tempfile: str | Callable[[], str] = '', name: str = '', complete_while_typing: Filter | bool = False, validate_while_typing: Filter | bool = False, enable_history_search: Filter | bool = False, document: Document | None = None, accept_handler: Callable[[Buffer], bool] | None = None, read_only: Filter | bool = False, multiline: Filter | bool = True, max_number_of_completions: int = 10000, on_text_changed: Callable[[Buffer], None] | None = None, on_text_insert: Callable[[Buffer], None] | None = None, on_cursor_position_changed: Callable[[Buffer], None] | None = None, on_completions_changed: Callable[[Buffer], None] | None = None, on_suggestion_set: Callable[[Buffer], None] | None = None)¶

The core data structure that holds the text and cursor position of the current input line and implements all text manipulations on top of it. It also implements the history, undo stack and the completion state.

Parameters:

completer – Completer instance.
history – History instance.
tempfile_suffix – The tempfile suffix (extension) to be used for the “open in editor” function. For a Python REPL, this would be “.py”, so that the editor knows the syntax highlighting to use. This can also be a callable that returns a string.
tempfile – For more advanced tempfile situations where you need control over the subdirectories and filename. For a Git Commit Message, this would be “.git/COMMIT_EDITMSG”, so that the editor knows the syntax highlighting to use. This can also be a callable that returns a string.
name – Name for this buffer. E.g. DEFAULT_BUFFER. This is mostly useful for key bindings where we sometimes prefer to refer to a buffer by their name instead of by reference.
accept_handler –
Called when the buffer input is accepted. (Usually when the user presses enter.) The accept handler receives this Buffer as input and should return True when the buffer text should be kept instead of calling reset.

In case of a PromptSession for instance, we want to keep the text, because we will exit the application, and only reset it during the next run.
max_number_of_completions – Never display more than this number of completions, even when the completer can produce more (limited by default to 10k for performance).

Events:

Parameters:

on_text_changed – When the buffer text changes. (Callable or None.)
on_text_insert – When new text is inserted. (Callable or None.)
on_cursor_position_changed – When the cursor moves. (Callable or None.)
on_completions_changed – When the completions were changed. (Callable or None.)
on_suggestion_set – When an auto-suggestion text has been set. (Callable or None.)

Filters:

Parameters:

complete_while_typing – Filter or bool. Decide whether or not to do asynchronous autocompleting while typing.
validate_while_typing – Filter or bool. Decide whether or not to do asynchronous validation while typing.
enable_history_search – Filter or bool to indicate when up-arrow partial string matching is enabled. It is advised to not enable this at the same time as complete_while_typing, because when there is an autocompletion found, the up arrows usually browse through the completions, rather than through the history.
read_only – Filter. When True, changes will not be allowed.
multiline – Filter or bool. When not set, pressing Enter will call the accept_handler. Otherwise, pressing Esc-Enter is required.

append_to_history() → None¶: Append the current input to the history.

apply_completion(completion: Completion) → None¶: Insert a given completion.

apply_search(search_state: SearchState, include_current_position: bool = True, count: int = 1) → None¶: Apply search. If something is found, set working_index and cursor_position.

auto_down(count: int = 1, go_to_start_of_line_if_history_changes: bool = False) → None¶: If we’re not on the last line (of a multiline input) go a line down, otherwise go forward in history. (If nothing is selected.)

auto_up(count: int = 1, go_to_start_of_line_if_history_changes: bool = False) → None¶: If we’re not on the first line (of a multiline input) go a line up, otherwise go back in history. (If nothing is selected.)

cancel_completion() → None¶: Cancel completion, go back to the original text.

complete_next(count: int = 1, disable_wrap_around: bool = False) → None¶: Browse to the next completions. (Does nothing if there are no completion.)

complete_previous(count: int = 1, disable_wrap_around: bool = False) → None¶: Browse to the previous completions. (Does nothing if there are no completion.)

copy_selection(_cut: bool = False) → ClipboardData¶

Copy selected text and return ClipboardData instance.

Notice that this doesn’t store the copied data on the clipboard yet. You can store it like this:

data = buffer.copy_selection()
get_app().clipboard.set_data(data)

cursor_down(count: int = 1) → None¶: (for multiline edit). Move cursor to the next line.

cursor_up(count: int = 1) → None¶: (for multiline edit). Move cursor to the previous line.

cut_selection() → ClipboardData¶: Delete selected text and return ClipboardData instance.

delete(count: int = 1) → str¶: Delete specified number of characters and Return the deleted text.

delete_before_cursor(count: int = 1) → str¶: Delete specified number of characters before cursor and return the deleted text.

property document: Document¶: Return Document instance from the current text, cursor position and selection state.

document_for_search(search_state: SearchState) → Document¶: Return a Document instance that has the text/cursor position for this search, if we would apply it. This will be used in the BufferControl to display feedback while searching.

get_search_position(search_state: SearchState, include_current_position: bool = True, count: int = 1) → int¶: Get the cursor position for this search. (This operation won’t change the working_index. It’s won’t go through the history. Vi text objects can’t span multiple items.)

go_to_completion(index: int | None) → None¶: Select a completion from the list of current completions.

go_to_history(index: int) → None¶: Go to this item in the history.

history_backward(count: int = 1) → None¶: Move backwards through history.

history_forward(count: int = 1) → None¶

Move forwards through the history.

Parameters:: count – Amount of items to move forward.

insert_line_above(copy_margin: bool = True) → None¶: Insert a new line above the current one.

insert_line_below(copy_margin: bool = True) → None¶: Insert a new line below the current one.

insert_text(data: str, overwrite: bool = False, move_cursor: bool = True, fire_event: bool = True) → None¶

Insert characters at cursor position.

Parameters:: fire_event – Fire on_text_insert event. This is mainly used to trigger autocompletion while typing.

property is_returnable: bool¶: True when there is something handling accept.

join_next_line(separator: str = ' ') → None¶: Join the next line to the current one by deleting the line ending after the current line.

join_selected_lines(separator: str = ' ') → None¶: Join the selected lines.

load_history_if_not_yet_loaded() → None¶

Create task for populating the buffer history (if not yet done).

Note:

This needs to be called from within the event loop of the
application, because history loading is async, and we need to be
sure the right event loop is active. Therefor, we call this method
in the `BufferControl.create_content`.

There are situations where prompt_toolkit applications are created
in one thread, but will later run in a different thread (Ptpython
is one example. The REPL runs in a separate thread, in order to
prevent interfering with a potential different event loop in the
main thread. The REPL UI however is still created in the main
thread.) We could decide to not support creating prompt_toolkit
objects in one thread and running the application in a different
thread, but history loading is the only place where it matters, and
this solves it.

newline(copy_margin: bool = True) → None¶: Insert a line ending at the current position.

open_in_editor(validate_and_handle: bool = False) → Task[None]¶

Open code in editor.

This returns a future, and runs in a thread executor.

paste_clipboard_data(data: ClipboardData, paste_mode: PasteMode = PasteMode.EMACS, count: int = 1) → None¶: Insert the data from the clipboard.

reset(document: Document | None = None, append_to_history: bool = False) → None¶

Parameters:: append_to_history – Append current input to history first.

save_to_undo_stack(clear_redo_stack: bool = True) → None¶: Safe current state (input text and cursor position), so that we can restore it by calling undo.

set_document(value: Document, bypass_readonly: bool = False) → None¶

Set Document instance. Like the document property, but accept an bypass_readonly argument.

Parameters:: bypass_readonly – When True, don’t raise an EditReadOnlyBuffer exception, even when the buffer is read-only.

Warning

When this buffer is read-only and bypass_readonly was not passed, the EditReadOnlyBuffer exception will be caught by the KeyProcessor and is silently suppressed. This is important to keep in mind when writing key bindings, because it won’t do what you expect, and there won’t be a stack trace. Use try/finally around this function if you need some cleanup code.

start_completion(select_first: bool = False, select_last: bool = False, insert_common_part: bool = False, complete_event: CompleteEvent | None = None) → None¶: Start asynchronous autocompletion of this buffer. (This will do nothing if a previous completion was still in progress.)

start_history_lines_completion() → None¶: Start a completion based on all the other lines in the document and the history.

start_selection(selection_type: SelectionType = SelectionType.CHARACTERS) → None¶: Take the current cursor position as the start of this selection.

swap_characters_before_cursor() → None¶: Swap the last two characters before the cursor.

transform_current_line(transform_callback: Callable[[str], str]) → None¶

Apply the given transformation function to the current line.

Parameters:: transform_callback – callable that takes a string and return a new string.

transform_lines(line_index_iterator: Iterable[int], transform_callback: Callable[[str], str]) → str¶

Transforms the text on a range of lines. When the iterator yield an index not in the range of lines that the document contains, it skips them silently.

To uppercase some lines:

new_text = transform_lines(range(5,10), lambda text: text.upper())

Parameters:

line_index_iterator – Iterator of line numbers (int)
transform_callback – callable that takes the original text of a line, and return the new text for this line.

Returns:

The new text.

transform_region(from_: int, to: int, transform_callback: Callable[[str], str]) → None¶

Transform a part of the input string.

Parameters:

from – (int) start position.
to – (int) end position.
transform_callback – Callable which accepts a string and returns the transformed string.

validate(set_cursor: bool = False) → bool¶

Returns True if valid.

Parameters:: set_cursor – Set the cursor position, if an error was found.

validate_and_handle() → None¶: Validate buffer and handle the accept action.

yank_last_arg(n: int | None = None) → None¶: Like yank_nth_arg, but if no argument has been given, yank the last word by default.

yank_nth_arg(n: int | None = None, _yank_last_arg: bool = False) → None¶

Pick nth word from previous history entry (depending on current yank_nth_arg_state) and insert it at current position. Rotate through history if called repeatedly. If no n has been given, take the first argument. (The second word.)

Parameters:: n – (None or int), The index of the word from the previous line to take.

class prompt_toolkit.buffer.CompletionState(original_document: Document, completions: list[Completion] | None = None, complete_index: int | None = None)¶

Immutable class that contains a completion state.

complete_index¶: Position in the completions array. This can be None to indicate “no completion”, the original text.

completions¶: List of all the current Completion instances which are possible at this point.

property current_completion: Completion | None¶: Return the current completion, or return None when no completion is selected.

go_to_index(index: int | None) → None¶

Create a new CompletionState object with the new index.

When index is None deselect the completion.

new_text_and_position() → tuple[str, int]¶: Return (new_text, new_cursor_position) for this completion.

original_document¶: Document as it was when the completion started.

exception prompt_toolkit.buffer.EditReadOnlyBuffer¶: Attempt editing of read-only Buffer.

prompt_toolkit.buffer.indent(buffer: Buffer, from_row: int, to_row: int, count: int = 1) → None¶: Indent text of a Buffer object.

prompt_toolkit.buffer.reshape_text(buffer: Buffer, from_row: int, to_row: int) → None¶: Reformat text, taking the width into account. to_row is included. (Vi ‘gq’ operator.)

prompt_toolkit.buffer.unindent(buffer: Buffer, from_row: int, to_row: int, count: int = 1) → None¶: Unindent text of a Buffer object.

Selection¶

Data structures for the selection.

class prompt_toolkit.selection.SelectionState(original_cursor_position: int = 0, type: SelectionType = SelectionType.CHARACTERS)¶

State of the current selection.

Parameters:

original_cursor_position – int
type – SelectionType

class prompt_toolkit.selection.SelectionType(value)¶

Type of selection.

BLOCK = 'BLOCK'¶: A block selection. (Visual-Block in Vi.)

CHARACTERS = 'CHARACTERS'¶: Characters. (Visual in Vi.)

LINES = 'LINES'¶: Whole lines. (Visual-Line in Vi.)

Clipboard¶

class prompt_toolkit.clipboard.Clipboard¶

Abstract baseclass for clipboards. (An implementation can be in memory, it can share the X11 or Windows keyboard, or can be persistent.)

abstractmethod get_data() → ClipboardData¶: Return clipboard data.

rotate() → None¶: For Emacs mode, rotate the kill ring.

abstractmethod set_data(data: ClipboardData) → None¶

Set data to the clipboard.

Parameters:: data – ClipboardData instance.

set_text(text: str) → None¶: Shortcut for setting plain text on clipboard.

class prompt_toolkit.clipboard.ClipboardData(text: str = '', type: SelectionType = SelectionType.CHARACTERS)¶

Text on the clipboard.

Parameters:

text – string
type – SelectionType

class prompt_toolkit.clipboard.DummyClipboard¶: Clipboard implementation that doesn’t remember anything.

class prompt_toolkit.clipboard.DynamicClipboard(get_clipboard: Callable[[], Clipboard | None])¶

Clipboard class that can dynamically returns any Clipboard.

Parameters:: get_clipboard – Callable that returns a Clipboard instance.

class prompt_toolkit.clipboard.InMemoryClipboard(data: ClipboardData | None = None, max_size: int = 60)¶

Default clipboard implementation. Just keep the data in memory.

This implements a kill-ring, for Emacs mode.

class prompt_toolkit.clipboard.pyperclip.PyperclipClipboard¶: Clipboard that synchronizes with the Windows/Mac/Linux system clipboard, using the pyperclip module.

Auto completion¶

class prompt_toolkit.completion.CompleteEvent(text_inserted: bool = False, completion_requested: bool = False)¶

Event that called the completer.

Parameters:

text_inserted – When True, it means that completions are requested because of a text insert. (Buffer.complete_while_typing.)
completion_requested – When True, it means that the user explicitly pressed the Tab key in order to view the completions.

These two flags can be used for instance to implement a completer that shows some completions when Tab has been pressed, but not automatically when the user presses a space. (Because of complete_while_typing.)

completion_requested¶: Used explicitly requested completion by pressing ‘tab’.

text_inserted¶: Automatic completion while typing.

class prompt_toolkit.completion.Completer¶

Base class for completer implementations.

abstractmethod get_completions(document: Document, complete_event: CompleteEvent) → Iterable[Completion]¶

This should be a generator that yields Completion instances.

If the generation of completions is something expensive (that takes a lot of time), consider wrapping this Completer class in a ThreadedCompleter. In that case, the completer algorithm runs in a background thread and completions will be displayed as soon as they arrive.

Parameters:

document – Document instance.
complete_event – CompleteEvent instance.

async get_completions_async(document: Document, complete_event: CompleteEvent) → AsyncGenerator[Completion, None]¶

Asynchronous generator for completions. (Probably, you won’t have to override this.)

Asynchronous generator of Completion objects.

class prompt_toolkit.completion.Completion(text: str, start_position: int = 0, display: AnyFormattedText | None = None, display_meta: AnyFormattedText | None = None, style: str = '', selected_style: str = '')¶

Parameters:

text – The new string that will be inserted into the document.
start_position – Position relative to the cursor_position where the new text will start. The text will be inserted between the start_position and the original cursor position.
display – (optional string or formatted text) If the completion has to be displayed differently in the completion menu.
display_meta – (Optional string or formatted text) Meta information about the completion, e.g. the path or source where it’s coming from. This can also be a callable that returns a string.
style – Style string.
selected_style – Style string, used for a selected completion. This can override the style parameter.

property display_meta: StyleAndTextTuples¶: Return meta-text. (This is lazy when using a callable).

property display_meta_text: str¶: The ‘meta’ field as plain text.

property display_text: str¶: The ‘display’ field as plain text.

new_completion_from_position(position: int) → Completion¶: (Only for internal use!) Get a new completion by splitting this one. Used by Application when it needs to have a list of new completions after inserting the common prefix.

class prompt_toolkit.completion.ConditionalCompleter(completer: Completer, filter: Filter | bool)¶

Wrapper around any other completer that will enable/disable the completions depending on whether the received condition is satisfied.

Parameters:

completer – Completer instance.
filter – Filter instance.

class prompt_toolkit.completion.DeduplicateCompleter(completer: Completer)¶

Wrapper around a completer that removes duplicates. Only the first unique completions are kept.

Completions are considered to be a duplicate if they result in the same document text when they would be applied.

class prompt_toolkit.completion.DummyCompleter¶: A completer that doesn’t return any completion.

class prompt_toolkit.completion.DynamicCompleter(get_completer: Callable[[], Completer | None])¶

Completer class that can dynamically returns any Completer.

Parameters:: get_completer – Callable that returns a Completer instance.

class prompt_toolkit.completion.ExecutableCompleter¶: Complete only executable files in the current path.

class prompt_toolkit.completion.FuzzyCompleter(completer: Completer, WORD: bool = False, pattern: str | None = None, enable_fuzzy: Filter | bool = True)¶

Fuzzy completion. This wraps any other completer and turns it into a fuzzy completer.

If the list of words is: [“leopard” , “gorilla”, “dinosaur”, “cat”, “bee”] Then trying to complete “oar” would yield “leopard” and “dinosaur”, but not the others, because they match the regular expression ‘o.*a.*r’. Similar, in another application “djm” could expand to “django_migrations”.

The results are sorted by relevance, which is defined as the start position and the length of the match.

Notice that this is not really a tool to work around spelling mistakes, like what would be possible with difflib. The purpose is rather to have a quicker or more intuitive way to filter the given completions, especially when many completions have a common prefix.

Fuzzy algorithm is based on this post: https://blog.amjith.com/fuzzyfinder-in-10-lines-of-python

Parameters:

completer – A Completer instance.
WORD – When True, use WORD characters.
pattern – Regex pattern which selects the characters before the cursor that are considered for the fuzzy matching.
enable_fuzzy – (bool or Filter) Enabled the fuzzy behavior. For easily turning fuzzyness on or off according to a certain condition.

class prompt_toolkit.completion.FuzzyWordCompleter(words: Sequence[str] | Callable[[], Sequence[str]], meta_dict: dict[str, str] | None = None, WORD: bool = False)¶

Fuzzy completion on a list of words.

(This is basically a WordCompleter wrapped in a FuzzyCompleter.)

Parameters:

words – List of words or callable that returns a list of words.
meta_dict – Optional dict mapping words to their meta-information.
WORD – When True, use WORD characters.

class prompt_toolkit.completion.NestedCompleter(options: dict[str, Completer | None], ignore_case: bool = True)¶

Completer which wraps around several other completers, and calls any the one that corresponds with the first word of the input.

By combining multiple NestedCompleter instances, we can achieve multiple hierarchical levels of autocompletion. This is useful when WordCompleter is not sufficient.

If you need multiple levels, check out the from_nested_dict classmethod.

classmethod from_nested_dict(data: Mapping[str, Any | Set[str] | None | Completer]) → NestedCompleter¶

Create a NestedCompleter, starting from a nested dictionary data structure, like this:

data = {
    'show': {
        'version': None,
        'interfaces': None,
        'clock': None,
        'ip': {'interface': {'brief'}}
    },
    'exit': None
    'enable': None
}

The value should be None if there is no further completion at some point. If all values in the dictionary are None, it is also possible to use a set instead.

Values in this data structure can be a completers as well.

class prompt_toolkit.completion.PathCompleter(only_directories: bool = False, get_paths: Callable[[], list[str]] | None = None, file_filter: Callable[[str], bool] | None = None, min_input_len: int = 0, expanduser: bool = False)¶

Complete for Path variables.

Parameters:

get_paths – Callable which returns a list of directories to look into when the user enters a relative path.
file_filter – Callable which takes a filename and returns whether this file should show up in the completion. None when no filtering has to be done.
min_input_len – Don’t do autocompletion when the input string is shorter.

class prompt_toolkit.completion.ThreadedCompleter(completer: Completer)¶

Wrapper that runs the get_completions generator in a thread.

(Use this to prevent the user interface from becoming unresponsive if the generation of completions takes too much time.)

The completions will be displayed as soon as they are produced. The user can already select a completion, even if not all completions are displayed.

async get_completions_async(document: Document, complete_event: CompleteEvent) → AsyncGenerator[Completion, None]¶: Asynchronous generator of completions.

class prompt_toolkit.completion.WordCompleter(words: Sequence[str] | Callable[[], Sequence[str]], ignore_case: bool = False, display_dict: Mapping[str, AnyFormattedText] | None = None, meta_dict: Mapping[str, AnyFormattedText] | None = None, WORD: bool = False, sentence: bool = False, match_middle: bool = False, pattern: Pattern[str] | None = None)¶

Simple autocompletion on a list of words.

Parameters:

words – List of words or callable that returns a list of words.
ignore_case – If True, case-insensitive completion.
meta_dict – Optional dict mapping words to their meta-text. (This should map strings to strings or formatted text.)
WORD – When True, use WORD characters.
sentence – When True, don’t complete by comparing the word before the cursor, but by comparing all the text before the cursor. In this case, the list of words is just a list of strings, where each string can contain spaces. (Can not be used together with the WORD option.)
match_middle – When True, match not only the start, but also in the middle of the word.
pattern – Optional compiled regex for finding the word before the cursor to complete. When given, use this regex pattern instead of default one (see document._FIND_WORD_RE)

prompt_toolkit.completion.get_common_complete_suffix(document: Document, completions: Sequence[Completion]) → str¶: Return the common prefix for all completions.

prompt_toolkit.completion.merge_completers(completers: Sequence[Completer], deduplicate: bool = False) → Completer¶

Combine several completers into one.

Parameters:: deduplicate – If True, wrap the result in a DeduplicateCompleter so that completions that would result in the same text will be deduplicated.

Document¶

The Document that implements all the text operations/querying.

class prompt_toolkit.document.Document(text: str = '', cursor_position: int | None = None, selection: SelectionState | None = None)¶

This is a immutable class around the text and cursor position, and contains methods for querying this data, e.g. to give the text before the cursor.

This class is usually instantiated by a Buffer object, and accessed as the document property of that class.

Parameters:

text – string
cursor_position – int
selection – SelectionState

property char_before_cursor: str¶: Return character before the cursor or an empty string.

property current_char: str¶: Return character under cursor or an empty string.

property current_line: str¶: Return the text on the line where the cursor is. (when the input consists of just one line, it equals text.

property current_line_after_cursor: str¶: Text from the cursor until the end of the line.

property current_line_before_cursor: str¶: Text from the start of the line until the cursor.

property cursor_position: int¶: The document cursor position.

property cursor_position_col: int¶: Current column. (0-based.)

property cursor_position_row: int¶: Current row. (0-based.)

cut_selection() → tuple[Document, ClipboardData]¶: Return a (Document, ClipboardData) tuple, where the document represents the new document when the selection is cut, and the clipboard data, represents whatever has to be put on the clipboard.

empty_line_count_at_the_end() → int¶: Return number of empty lines at the end of the document.

end_of_paragraph(count: int = 1, after: bool = False) → int¶: Return the end of the current paragraph. (Relative cursor position.)

find(sub: str, in_current_line: bool = False, include_current_position: bool = False, ignore_case: bool = False, count: int = 1) → int | None¶

Find text after the cursor, return position relative to the cursor position. Return None if nothing was found.

Parameters:: count – Find the n-th occurrence.

find_all(sub: str, ignore_case: bool = False) → list[int]¶: Find all occurrences of the substring. Return a list of absolute positions in the document.

find_backwards(sub: str, in_current_line: bool = False, ignore_case: bool = False, count: int = 1) → int | None¶

Find text before the cursor, return position relative to the cursor position. Return None if nothing was found.

Parameters:: count – Find the n-th occurrence.

find_boundaries_of_current_word(WORD: bool = False, include_leading_whitespace: bool = False, include_trailing_whitespace: bool = False) → tuple[int, int]¶: Return the relative boundaries (startpos, endpos) of the current word under the cursor. (This is at the current line, because line boundaries obviously don’t belong to any word.) If not on a word, this returns (0,0)

find_enclosing_bracket_left(left_ch: str, right_ch: str, start_pos: int | None = None) → int | None¶

Find the left bracket enclosing current position. Return the relative position to the cursor position.

When start_pos is given, don’t look past the position.

find_enclosing_bracket_right(left_ch: str, right_ch: str, end_pos: int | None = None) → int | None¶

Find the right bracket enclosing current position. Return the relative position to the cursor position.

When end_pos is given, don’t look past the position.

find_matching_bracket_position(start_pos: int | None = None, end_pos: int | None = None) → int¶

Return relative cursor position of matching [, (, { or < bracket.

When start_pos or end_pos are given. Don’t look past the positions.

find_next_matching_line(match_func: Callable[[str], bool], count: int = 1) → int | None¶: Look downwards for empty lines. Return the line index, relative to the current line.

find_next_word_beginning(count: int = 1, WORD: bool = False) → int | None¶: Return an index relative to the cursor position pointing to the start of the next word. Return None if nothing was found.

find_next_word_ending(include_current_position: bool = False, count: int = 1, WORD: bool = False) → int | None¶: Return an index relative to the cursor position pointing to the end of the next word. Return None if nothing was found.

find_previous_matching_line(match_func: Callable[[str], bool], count: int = 1) → int | None¶: Look upwards for empty lines. Return the line index, relative to the current line.

find_previous_word_beginning(count: int = 1, WORD: bool = False) → int | None¶: Return an index relative to the cursor position pointing to the start of the previous word. Return None if nothing was found.

find_previous_word_ending(count: int = 1, WORD: bool = False) → int | None¶: Return an index relative to the cursor position pointing to the end of the previous word. Return None if nothing was found.

find_start_of_previous_word(count: int = 1, WORD: bool = False, pattern: Pattern[str] | None = None) → int | None¶

Return an index relative to the cursor position pointing to the start of the previous word. Return None if nothing was found.

Parameters:: pattern – (None or compiled regex). When given, use this regex pattern.

get_column_cursor_position(column: int) → int¶: Return the relative cursor position for this column at the current line. (It will stay between the boundaries of the line in case of a larger number.)

get_cursor_down_position(count: int = 1, preferred_column: int | None = None) → int¶

Return the relative cursor position (character index) where we would be if the user pressed the arrow-down button.

Parameters:: preferred_column – When given, go to this column instead of staying at the current column.

get_cursor_left_position(count: int = 1) → int¶: Relative position for cursor left.

get_cursor_right_position(count: int = 1) → int¶: Relative position for cursor_right.

get_cursor_up_position(count: int = 1, preferred_column: int | None = None) → int¶

Return the relative cursor position (character index) where we would be if the user pressed the arrow-up button.

Parameters:: preferred_column – When given, go to this column instead of staying at the current column.

get_end_of_document_position() → int¶: Relative position for the end of the document.

get_end_of_line_position() → int¶: Relative position for the end of this line.

get_start_of_document_position() → int¶: Relative position for the start of the document.

get_start_of_line_position(after_whitespace: bool = False) → int¶: Relative position for the start of this line.

get_word_before_cursor(WORD: bool = False, pattern: Pattern[str] | None = None) → str¶

Give the word before the cursor. If we have whitespace before the cursor this returns an empty string.

Parameters:: pattern – (None or compiled regex). When given, use this regex pattern.

get_word_under_cursor(WORD: bool = False) → str¶: Return the word, currently below the cursor. This returns an empty string when the cursor is on a whitespace region.

has_match_at_current_position(sub: str) → bool¶: True when this substring is found at the cursor position.

insert_after(text: str) → Document¶: Create a new document, with this text inserted after the buffer. It keeps selection ranges and cursor position in sync.

insert_before(text: str) → Document¶: Create a new document, with this text inserted before the buffer. It keeps selection ranges and cursor position in sync.

property is_cursor_at_the_end: bool¶: True when the cursor is at the end of the text.

property is_cursor_at_the_end_of_line: bool¶: True when the cursor is at the end of this line.

last_non_blank_of_current_line_position() → int¶: Relative position for the last non blank character of this line.

property leading_whitespace_in_current_line: str¶: The leading whitespace in the left margin of the current line.

property line_count: int¶: Return the number of lines in this document. If the document ends with a trailing n, that counts as the beginning of a new line.

property lines: list[str]¶: Array of all the lines.

property lines_from_current: list[str]¶: Array of the lines starting from the current line, until the last line.

property on_first_line: bool¶: True when we are at the first line.

property on_last_line: bool¶: True when we are at the last line.

paste_clipboard_data(data: ClipboardData, paste_mode: PasteMode = PasteMode.EMACS, count: int = 1) → Document¶

Return a new Document instance which contains the result if we would paste this data at the current cursor position.

Parameters:

paste_mode – Where to paste. (Before/after/emacs.)
count – When >1, Paste multiple times.

property selection: SelectionState | None¶: SelectionState object.

selection_range() → tuple[int, int]¶

Return (from, to) tuple of the selection. start and end position are included.

This doesn’t take the selection type into account. Use selection_ranges instead.

selection_range_at_line(row: int) → tuple[int, int] | None¶

If the selection spans a portion of the given line, return a (from, to) tuple.

The returned upper boundary is not included in the selection, so (0, 0) is an empty selection. (0, 1), is a one character selection.

Returns None if the selection doesn’t cover this line at all.

selection_ranges() → Iterable[tuple[int, int]]¶

Return a list of (from, to) tuples for the selection or none if nothing was selected. The upper boundary is not included.

This will yield several (from, to) tuples in case of a BLOCK selection. This will return zero ranges, like (8,8) for empty lines in a block selection.

start_of_paragraph(count: int = 1, before: bool = False) → int¶: Return the start of the current paragraph. (Relative cursor position.)

property text: str¶: The document text.

translate_index_to_position(index: int) → tuple[int, int]¶: Given an index for the text, return the corresponding (row, col) tuple. (0-based. Returns (0, 0) for index=0.)

translate_row_col_to_index(row: int, col: int) → int¶

Given a (row, col) tuple, return the corresponding index. (Row and col params are 0-based.)

Negative row/col values are turned into zero.

Enums¶

prompt_toolkit.enums.DEFAULT_BUFFER = 'DEFAULT_BUFFER'¶: Name of the default buffer.

prompt_toolkit.enums.SEARCH_BUFFER = 'SEARCH_BUFFER'¶: Name of the search buffer.

prompt_toolkit.enums.SYSTEM_BUFFER = 'SYSTEM_BUFFER'¶: Name of the system buffer.

History¶

Implementations for the history of a Buffer.

NOTE: There is no DynamicHistory:: This doesn’t work well, because the Buffer needs to be able to attach an event handler to the event when a history entry is loaded. This loading can be done asynchronously and making the history swappable would probably break this.

class prompt_toolkit.history.DummyHistory¶: History object that doesn’t remember anything.

class prompt_toolkit.history.FileHistory(filename: str | bytes | PathLike[str] | PathLike[bytes])¶: History class that stores all strings in a file.

class prompt_toolkit.history.History¶

Base History class.

This also includes abstract methods for loading/storing history.

append_string(string: str) → None¶: Add string to the history.

get_strings() → list[str]¶: Get the strings from the history that are loaded so far. (In order. Oldest item first.)

async load() → AsyncGenerator[str, None]¶

Load the history and yield all the entries in reverse order (latest, most recent history entry first).

This method can be called multiple times from the Buffer to repopulate the history when prompting for a new input. So we are responsible here for both caching, and making sure that strings that were were appended to the history will be incorporated next time this method is called.

abstractmethod load_history_strings() → Iterable[str]¶

This should be a generator that yields str instances.

It should yield the most recent items first, because they are the most important. (The history can already be used, even when it’s only partially loaded.)

abstractmethod store_string(string: str) → None¶: Store the string in persistent storage.

class prompt_toolkit.history.InMemoryHistory(history_strings: Sequence[str] | None = None)¶

History class that keeps a list of all strings in memory.

In order to prepopulate the history, it’s possible to call either append_string for all items or pass a list of strings to __init__ here.

class prompt_toolkit.history.ThreadedHistory(history: History)¶

Wrapper around History implementations that run the load() generator in a thread.

Use this to increase the start-up time of prompt_toolkit applications. History entries are available as soon as they are loaded. We don’t have to wait for everything to be loaded.

async load() → AsyncGenerator[str, None]¶: Like History.load(), but call `self.load_history_strings() in a background thread.

Keys¶

class prompt_toolkit.keys.Keys(value)¶

List of keys for use in key bindings.

Note that this is an “StrEnum”, all values can be compared against strings.

Style¶

Styling for prompt_toolkit applications.

class prompt_toolkit.styles.AdjustBrightnessStyleTransformation(min_brightness: Callable[[], float] | float = 0.0, max_brightness: Callable[[], float] | float = 1.0)¶

Adjust the brightness to improve the rendering on either dark or light backgrounds.

For dark backgrounds, it’s best to increase min_brightness. For light backgrounds it’s best to decrease max_brightness. Usually, only one setting is adjusted.

This will only change the brightness for text that has a foreground color defined, but no background color. It works best for 256 or true color output.

Note

Notice that there is no universal way to detect whether the application is running in a light or dark terminal. As a developer of an command line application, you’ll have to make this configurable for the user.

Parameters:

min_brightness – Float between 0.0 and 1.0 or a callable that returns a float.
max_brightness – Float between 0.0 and 1.0 or a callable that returns a float.

class prompt_toolkit.styles.Attrs(color, bgcolor, bold, underline, strike, italic, blink, reverse, hidden)¶

bgcolor: str | None¶: Alias for field number 1

blink: bool | None¶: Alias for field number 6

bold: bool | None¶: Alias for field number 2

color: str | None¶: Alias for field number 0

hidden: bool | None¶: Alias for field number 8

italic: bool | None¶: Alias for field number 5

reverse: bool | None¶: Alias for field number 7

strike: bool | None¶: Alias for field number 4

underline: bool | None¶: Alias for field number 3

class prompt_toolkit.styles.BaseStyle¶

Abstract base class for prompt_toolkit styles.

abstractmethod get_attrs_for_style_str(style_str: str, default: Attrs = ('', '', False, False, False, False, False, False, False)) → Attrs¶

Return Attrs for the given style string.

Parameters:

style_str – The style string. This can contain inline styling as well as classnames (e.g. “class:title”).
default – Attrs to be used if no styling was defined.

abstractmethod invalidation_hash() → Hashable¶: Invalidation hash for the style. When this changes over time, the renderer knows that something in the style changed, and that everything has to be redrawn.

abstract property style_rules: list[tuple[str, str]]¶: The list of style rules, used to create this style. (Required for DynamicStyle and _MergedStyle to work.)

class prompt_toolkit.styles.ConditionalStyleTransformation(style_transformation: StyleTransformation, filter: Filter | bool)¶: Apply the style transformation depending on a condition.

class prompt_toolkit.styles.DummyStyle¶: A style that doesn’t style anything.

class prompt_toolkit.styles.DummyStyleTransformation¶: Don’t transform anything at all.

class prompt_toolkit.styles.DynamicStyle(get_style: Callable[[], BaseStyle | None])¶

Style class that can dynamically returns an other Style.

Parameters:: get_style – Callable that returns a Style instance.

class prompt_toolkit.styles.DynamicStyleTransformation(get_style_transformation: Callable[[], StyleTransformation | None])¶

StyleTransformation class that can dynamically returns any StyleTransformation.

Parameters:: get_style_transformation – Callable that returns a StyleTransformation instance.

class prompt_toolkit.styles.Priority(value)¶

The priority of the rules, when a style is created from a dictionary.

In a Style, rules that are defined later will always override previous defined rules, however in a dictionary, the key order was arbitrary before Python 3.6. This means that the style could change at random between rules.

We have two options:

DICT_KEY_ORDER: This means, iterate through the dictionary, and take
the key/value pairs in order as they come. This is a good option if you have Python >3.6. Rules at the end will override rules at the beginning.
MOST_PRECISE: keys that are defined with most precision will get higher priority. (More precise means: more elements.)

class prompt_toolkit.styles.Style(style_rules: list[tuple[str, str]])¶

Create a Style instance from a list of style rules.

The style_rules is supposed to be a list of (‘classnames’, ‘style’) tuples. The classnames are a whitespace separated string of class names and the style string is just like a Pygments style definition, but with a few additions: it supports ‘reverse’ and ‘blink’.

Later rules always override previous rules.

Usage:

Style([
    ('title', '#ff0000 bold underline'),
    ('something-else', 'reverse'),
    ('class1 class2', 'reverse'),
])

The from_dict classmethod is similar, but takes a dictionary as input.

classmethod from_dict(style_dict: dict[str, str], priority: Priority = Priority.DICT_KEY_ORDER) → Style¶

Parameters:

style_dict – Style dictionary.
priority – Priority value.

get_attrs_for_style_str(style_str: str, default: Attrs = ('', '', False, False, False, False, False, False, False)) → Attrs¶: Get Attrs for the given style string.

class prompt_toolkit.styles.StyleTransformation¶

Base class for any style transformation.

invalidation_hash() → Hashable¶: When this changes, the cache should be invalidated.

abstractmethod transform_attrs(attrs: Attrs) → Attrs¶

Take an Attrs object and return a new Attrs object.

Remember that the color formats can be either “ansi…” or a 6 digit lowercase hexadecimal color (without ‘#’ prefix).

class prompt_toolkit.styles.SwapLightAndDarkStyleTransformation¶

Turn dark colors into light colors and the other way around.

This is meant to make color schemes that work on a dark background usable on a light background (and the other way around).

Notice that this doesn’t swap foreground and background like “reverse” does. It turns light green into dark green and the other way around. Foreground and background colors are considered individually.

Also notice that when <reverse> is used somewhere and no colors are given in particular (like what is the default for the bottom toolbar), then this doesn’t change anything. This is what makes sense, because when the ‘default’ color is chosen, it’s what works best for the terminal, and reverse works good with that.

transform_attrs(attrs: Attrs) → Attrs¶: Return the Attrs used when opposite luminosity should be used.

prompt_toolkit.styles.merge_style_transformations(style_transformations: Sequence[StyleTransformation]) → StyleTransformation¶: Merge multiple transformations together.

prompt_toolkit.styles.merge_styles(styles: list[BaseStyle]) → _MergedStyle¶: Merge multiple Style objects.

prompt_toolkit.styles.pygments_token_to_classname(token: Token) → str¶

Turn e.g. Token.Name.Exception into ‘pygments.name.exception’.

(Our Pygments lexer will also turn the tokens that pygments produces in a prompt_toolkit list of fragments that match these styling rules.)

prompt_toolkit.styles.style_from_pygments_cls(pygments_style_cls: type[PygmentsStyle]) → Style¶

Shortcut to create a Style instance from a Pygments style class and a style dictionary.

Example:

from prompt_toolkit.styles.from_pygments import style_from_pygments_cls
from pygments.styles import get_style_by_name
style = style_from_pygments_cls(get_style_by_name('monokai'))

Parameters:: pygments_style_cls – Pygments style class to start from.

prompt_toolkit.styles.style_from_pygments_dict(pygments_dict: dict[Token, str]) → Style¶: Create a Style instance from a Pygments style dictionary. (One that maps Token objects to style strings.)

Shortcuts¶

class prompt_toolkit.shortcuts.CompleteStyle(value)¶: How to display autocompletions for the prompt.

class prompt_toolkit.shortcuts.ProgressBar(title: AnyFormattedText = None, formatters: Sequence[Formatter] | None = None, bottom_toolbar: AnyFormattedText = None, style: BaseStyle | None = None, key_bindings: KeyBindings | None = None, cancel_callback: Callable[[], None] | None = None, file: TextIO | None = None, color_depth: ColorDepth | None = None, output: Output | None = None, input: Input | None = None)¶

Progress bar context manager.

Usage

with ProgressBar(...) as pb:
    for item in pb(data):
        ...

Parameters:

title – Text to be displayed above the progress bars. This can be a callable or formatted text as well.
formatters – List of Formatter instances.
bottom_toolbar – Text to be displayed in the bottom toolbar. This can be a callable or formatted text.
style – prompt_toolkit.styles.BaseStyle instance.
key_bindings – KeyBindings instance.
cancel_callback – Callback function that’s called when control-c is pressed by the user. This can be used for instance to start “proper” cancellation if the wrapped code supports it.
file – The file object used for rendering, by default sys.stderr is used.
color_depth – prompt_toolkit ColorDepth instance.
output – Output instance.
input – Input instance.

class prompt_toolkit.shortcuts.PromptSession(message: AnyFormattedText = '', *, multiline: FilterOrBool = False, wrap_lines: FilterOrBool = True, is_password: FilterOrBool = False, vi_mode: bool = False, editing_mode: EditingMode = EditingMode.EMACS, complete_while_typing: FilterOrBool = True, validate_while_typing: FilterOrBool = True, enable_history_search: FilterOrBool = False, search_ignore_case: FilterOrBool = False, lexer: Lexer | None = None, enable_system_prompt: FilterOrBool = False, enable_suspend: FilterOrBool = False, enable_open_in_editor: FilterOrBool = False, validator: Validator | None = None, completer: Completer | None = None, complete_in_thread: bool = False, reserve_space_for_menu: int = 8, complete_style: CompleteStyle = CompleteStyle.COLUMN, auto_suggest: AutoSuggest | None = None, style: BaseStyle | None = None, style_transformation: StyleTransformation | None = None, swap_light_and_dark_colors: FilterOrBool = False, color_depth: ColorDepth | None = None, cursor: AnyCursorShapeConfig = None, include_default_pygments_style: FilterOrBool = True, history: History | None = None, clipboard: Clipboard | None = None, prompt_continuation: PromptContinuationText | None = None, rprompt: AnyFormattedText = None, bottom_toolbar: AnyFormattedText = None, mouse_support: FilterOrBool = False, input_processors: list[Processor] | None = None, placeholder: AnyFormattedText | None = None, key_bindings: KeyBindingsBase | None = None, erase_when_done: bool = False, tempfile_suffix: str | Callable[[], str] | None = '.txt', tempfile: str | Callable[[], str] | None = None, refresh_interval: float = 0, input: Input | None = None, output: Output | None = None, interrupt_exception: type[BaseException] = <class 'KeyboardInterrupt'>, eof_exception: type[BaseException] = <class 'EOFError'>)¶

PromptSession for a prompt application, which can be used as a GNU Readline replacement.

This is a wrapper around a lot of prompt_toolkit functionality and can be a replacement for raw_input.

All parameters that expect “formatted text” can take either just plain text (a unicode object), a list of (style_str, text) tuples or an HTML object.

Example usage:

s = PromptSession(message='>')
text = s.prompt()

Parameters:

message – Plain text or formatted text to be shown before the prompt. This can also be a callable that returns formatted text.
multiline – bool or Filter. When True, prefer a layout that is more adapted for multiline input. Text after newlines is automatically indented, and search/arg input is shown below the input, instead of replacing the prompt.
wrap_lines – bool or Filter. When True (the default), automatically wrap long lines instead of scrolling horizontally.
is_password – Show asterisks instead of the actual typed characters.
editing_mode – EditingMode.VI or EditingMode.EMACS.
vi_mode – bool, if True, Identical to editing_mode=EditingMode.VI.
complete_while_typing – bool or Filter. Enable autocompletion while typing.
validate_while_typing – bool or Filter. Enable input validation while typing.
enable_history_search – bool or Filter. Enable up-arrow parting string matching.
search_ignore_case – Filter. Search case insensitive.
lexer – Lexer to be used for the syntax highlighting.
validator – Validator instance for input validation.
completer – Completer instance for input completion.
complete_in_thread – bool or Filter. Run the completer code in a background thread in order to avoid blocking the user interface. For CompleteStyle.READLINE_LIKE, this setting has no effect. There we always run the completions in the main thread.
reserve_space_for_menu – Space to be reserved for displaying the menu. (0 means that no space needs to be reserved.)
auto_suggest – AutoSuggest instance for input suggestions.
style – Style instance for the color scheme.
include_default_pygments_style – bool or Filter. Tell whether the default styling for Pygments lexers has to be included. By default, this is true, but it is recommended to be disabled if another Pygments style is passed as the style argument, otherwise, two Pygments styles will be merged.
style_transformation – StyleTransformation instance.
swap_light_and_dark_colors – bool or Filter. When enabled, apply SwapLightAndDarkStyleTransformation. This is useful for switching between dark and light terminal backgrounds.
enable_system_prompt – bool or Filter. Pressing Meta+’!’ will show a system prompt.
enable_suspend – bool or Filter. Enable Control-Z style suspension.
enable_open_in_editor – bool or Filter. Pressing ‘v’ in Vi mode or C-X C-E in emacs mode will open an external editor.
history – History instance.
clipboard – Clipboard instance. (e.g. InMemoryClipboard)
rprompt – Text or formatted text to be displayed on the right side. This can also be a callable that returns (formatted) text.
bottom_toolbar – Formatted text or callable which is supposed to return formatted text.
prompt_continuation – Text that needs to be displayed for a multiline prompt continuation. This can either be formatted text or a callable that takes a prompt_width, line_number and wrap_count as input and returns formatted text. When this is None (the default), then prompt_width spaces will be used.
complete_style – CompleteStyle.COLUMN, CompleteStyle.MULTI_COLUMN or CompleteStyle.READLINE_LIKE.
mouse_support – bool or Filter to enable mouse support.
placeholder – Text to be displayed when no input has been given yet. Unlike the default parameter, this won’t be returned as part of the output ever. This can be formatted text or a callable that returns formatted text.
refresh_interval – (number; in seconds) When given, refresh the UI every so many seconds.
input – Input object. (Note that the preferred way to change the input/output is by creating an AppSession.)
output – Output object.
interrupt_exception – The exception type that will be raised when there is a keyboard interrupt (control-c keypress).
eof_exception – The exception type that will be raised when there is an end-of-file/exit event (control-d keypress).

prompt(message: AnyFormattedText | None = None, *, editing_mode: EditingMode | None = None, refresh_interval: float | None = None, vi_mode: bool | None = None, lexer: Lexer | None = None, completer: Completer | None = None, complete_in_thread: bool | None = None, is_password: bool | None = None, key_bindings: KeyBindingsBase | None = None, bottom_toolbar: AnyFormattedText | None = None, style: BaseStyle | None = None, color_depth: ColorDepth | None = None, cursor: AnyCursorShapeConfig | None = None, include_default_pygments_style: FilterOrBool | None = None, style_transformation: StyleTransformation | None = None, swap_light_and_dark_colors: FilterOrBool | None = None, rprompt: AnyFormattedText | None = None, multiline: FilterOrBool | None = None, prompt_continuation: PromptContinuationText | None = None, wrap_lines: FilterOrBool | None = None, enable_history_search: FilterOrBool | None = None, search_ignore_case: FilterOrBool | None = None, complete_while_typing: FilterOrBool | None = None, validate_while_typing: FilterOrBool | None = None, complete_style: CompleteStyle | None = None, auto_suggest: AutoSuggest | None = None, validator: Validator | None = None, clipboard: Clipboard | None = None, mouse_support: FilterOrBool | None = None, input_processors: list[Processor] | None = None, placeholder: AnyFormattedText | None = None, reserve_space_for_menu: int | None = None, enable_system_prompt: FilterOrBool | None = None, enable_suspend: FilterOrBool | None = None, enable_open_in_editor: FilterOrBool | None = None, tempfile_suffix: str | Callable[[], str] | None = None, tempfile: str | Callable[[], str] | None = None, default: str | Document = '', accept_default: bool = False, pre_run: Callable[[], None] | None = None, set_exception_handler: bool = True, handle_sigint: bool = True, in_thread: bool = False, inputhook: InputHook | None = None) → _T¶

Display the prompt.

The first set of arguments is a subset of the PromptSession class itself. For these, passing in None will keep the current values that are active in the session. Passing in a value will set the attribute for the session, which means that it applies to the current, but also to the next prompts.

Note that in order to erase a Completer, Validator or AutoSuggest, you can’t use None. Instead pass in a DummyCompleter, DummyValidator or DummyAutoSuggest instance respectively. For a Lexer you can pass in an empty SimpleLexer.

Additional arguments, specific for this prompt:

Parameters:

default – The default input text to be shown. (This can be edited by the user).
accept_default – When True, automatically accept the default value without allowing the user to edit the input.
pre_run – Callable, called at the start of Application.run.
in_thread – Run the prompt in a background thread; block the current thread. This avoids interference with an event loop in the current thread. Like Application.run(in_thread=True).

This method will raise KeyboardInterrupt when control-c has been pressed (for abort) and EOFError when control-d has been pressed (for exit).

prompt_toolkit.shortcuts.button_dialog(title: AnyFormattedText = '', text: AnyFormattedText = '', buttons: list[tuple[str, _T]] = [], style: BaseStyle | None = None) → Application[_T]¶: Display a dialog with button choices (given as a list of tuples). Return the value associated with button.

prompt_toolkit.shortcuts.clear() → None¶: Clear the screen.

prompt_toolkit.shortcuts.clear_title() → None¶: Erase the current title.

prompt_toolkit.shortcuts.confirm(message: str = 'Confirm?', suffix: str = ' (y/n) ') → bool¶: Display a confirmation prompt that returns True/False.

prompt_toolkit.shortcuts.create_confirm_session(message: str, suffix: str = ' (y/n) ') → PromptSession[bool]¶: Create a PromptSession object for the ‘confirm’ function.

prompt_toolkit.shortcuts.input_dialog(title: AnyFormattedText = '', text: AnyFormattedText = '', ok_text: str = 'OK', cancel_text: str = 'Cancel', completer: Completer | None = None, validator: Validator | None = None, password: FilterOrBool = False, style: BaseStyle | None = None, default: str = '') → Application[str]¶: Display a text input box. Return the given text, or None when cancelled.

prompt_toolkit.shortcuts.message_dialog(title: AnyFormattedText = '', text: AnyFormattedText = '', ok_text: str = 'Ok', style: BaseStyle | None = None) → Application[None]¶: Display a simple message box and wait until the user presses enter.

prompt_toolkit.shortcuts.print_formatted_text(*values: Any, sep: str = ' ', end: str = '\n', file: TextIO | None = None, flush: bool = False, style: BaseStyle | None = None, output: Output | None = None, color_depth: ColorDepth | None = None, style_transformation: StyleTransformation | None = None, include_default_pygments_style: bool = True) → None¶

print_formatted_text(*values, sep=' ', end='\n', file=None, flush=False, style=None, output=None)

Print text to stdout. This is supposed to be compatible with Python’s print function, but supports printing of formatted text. You can pass a FormattedText, HTML or ANSI object to print formatted text.

Print HTML as follows:

print_formatted_text(HTML('<i>Some italic text</i> <ansired>This is red!</ansired>'))

style = Style.from_dict({
    'hello': '#ff0066',
    'world': '#884444 italic',
})
print_formatted_text(HTML('<hello>Hello</hello> <world>world</world>!'), style=style)

Print a list of (style_str, text) tuples in the given style to the output. E.g.:

style = Style.from_dict({
    'hello': '#ff0066',
    'world': '#884444 italic',
})
fragments = FormattedText([
    ('class:hello', 'Hello'),
    ('class:world', 'World'),
])
print_formatted_text(fragments, style=style)

If you want to print a list of Pygments tokens, wrap it in PygmentsTokens to do the conversion.

If a prompt_toolkit Application is currently running, this will always print above the application or prompt (similar to patch_stdout). So, print_formatted_text will erase the current application, print the text, and render the application again.

Parameters:

values – Any kind of printable object, or formatted string.
sep – String inserted between values, default a space.
end – String appended after the last value, default a newline.
style – Style instance for the color scheme.
include_default_pygments_style – bool. Include the default Pygments style when set to True (the default).

prompt_toolkit.shortcuts.progress_dialog(title: AnyFormattedText = '', text: AnyFormattedText = '', run_callback: Callable[[Callable[[int], None], Callable[[str], None]], None] = <function <lambda>>, style: BaseStyle | None = None) → Application[None]¶

Parameters:: run_callback – A function that receives as input a set_percentage function and it does the work.

prompt_toolkit.shortcuts.prompt(message: AnyFormattedText | None = None, *, history: History | None = None, editing_mode: EditingMode | None = None, refresh_interval: float | None = None, vi_mode: bool | None = None, lexer: Lexer | None = None, completer: Completer | None = None, complete_in_thread: bool | None = None, is_password: bool | None = None, key_bindings: KeyBindingsBase | None = None, bottom_toolbar: AnyFormattedText | None = None, style: BaseStyle | None = None, color_depth: ColorDepth | None = None, cursor: AnyCursorShapeConfig = None, include_default_pygments_style: FilterOrBool | None = None, style_transformation: StyleTransformation | None = None, swap_light_and_dark_colors: FilterOrBool | None = None, rprompt: AnyFormattedText | None = None, multiline: FilterOrBool | None = None, prompt_continuation: PromptContinuationText | None = None, wrap_lines: FilterOrBool | None = None, enable_history_search: FilterOrBool | None = None, search_ignore_case: FilterOrBool | None = None, complete_while_typing: FilterOrBool | None = None, validate_while_typing: FilterOrBool | None = None, complete_style: CompleteStyle | None = None, auto_suggest: AutoSuggest | None = None, validator: Validator | None = None, clipboard: Clipboard | None = None, mouse_support: FilterOrBool | None = None, input_processors: list[Processor] | None = None, placeholder: AnyFormattedText | None = None, reserve_space_for_menu: int | None = None, enable_system_prompt: FilterOrBool | None = None, enable_suspend: FilterOrBool | None = None, enable_open_in_editor: FilterOrBool | None = None, tempfile_suffix: str | Callable[[], str] | None = None, tempfile: str | Callable[[], str] | None = None, default: str = '', accept_default: bool = False, pre_run: Callable[[], None] | None = None, set_exception_handler: bool = True, handle_sigint: bool = True, in_thread: bool = False, inputhook: InputHook | None = None) → str¶

Display the prompt.

Additional arguments, specific for this prompt:

Parameters:

default – The default input text to be shown. (This can be edited by the user).
accept_default – When True, automatically accept the default value without allowing the user to edit the input.
pre_run – Callable, called at the start of Application.run.
in_thread – Run the prompt in a background thread; block the current thread. This avoids interference with an event loop in the current thread. Like Application.run(in_thread=True).

This method will raise KeyboardInterrupt when control-c has been pressed (for abort) and EOFError when control-d has been pressed (for exit).

prompt_toolkit.shortcuts.radiolist_dialog(title: AnyFormattedText = '', text: AnyFormattedText = '', ok_text: str = 'Ok', cancel_text: str = 'Cancel', values: Sequence[tuple[_T, AnyFormattedText]] | None = None, default: _T | None = None, style: BaseStyle | None = None) → Application[_T]¶

Display a simple list of element the user can choose amongst.

Only one element can be selected at a time using Arrow keys and Enter. The focus can be moved between the list and the Ok/Cancel button with tab.

prompt_toolkit.shortcuts.set_title(text: str) → None¶: Set the terminal title.

prompt_toolkit.shortcuts.yes_no_dialog(title: AnyFormattedText = '', text: AnyFormattedText = '', yes_text: str = 'Yes', no_text: str = 'No', style: BaseStyle | None = None) → Application[bool]¶: Display a Yes/No dialog. Return a boolean.

Formatter classes for the progress bar. Each progress bar consists of a list of these formatters.

class prompt_toolkit.shortcuts.progress_bar.formatters.Bar(start: str = '[', end: str = ']', sym_a: str = '=', sym_b: str = '>', sym_c: str = ' ', unknown: str = '#')¶: Display the progress bar itself.

class prompt_toolkit.shortcuts.progress_bar.formatters.Formatter¶: Base class for any formatter.

class prompt_toolkit.shortcuts.progress_bar.formatters.IterationsPerSecond¶: Display the iterations per second.

class prompt_toolkit.shortcuts.progress_bar.formatters.Label(width: None | int | Dimension | Callable[[], Any] = None, suffix: str = '')¶

Display the name of the current task.

Parameters:

width – If a width is given, use this width. Scroll the text if it doesn’t fit in this width.
suffix – String suffix to be added after the task name, e.g. ‘: ‘. If no task name was given, no suffix will be added.

class prompt_toolkit.shortcuts.progress_bar.formatters.Percentage¶: Display the progress as a percentage.

class prompt_toolkit.shortcuts.progress_bar.formatters.Progress¶: Display the progress as text. E.g. “8/20”

class prompt_toolkit.shortcuts.progress_bar.formatters.Rainbow(formatter: Formatter)¶: For the fun. Add rainbow colors to any of the other formatters.

class prompt_toolkit.shortcuts.progress_bar.formatters.SpinningWheel¶: Display a spinning wheel.

class prompt_toolkit.shortcuts.progress_bar.formatters.Text(text: AnyFormattedText, style: str = '')¶: Display plain text.

class prompt_toolkit.shortcuts.progress_bar.formatters.TimeElapsed¶: Display the elapsed time.

class prompt_toolkit.shortcuts.progress_bar.formatters.TimeLeft¶: Display the time left.

prompt_toolkit.shortcuts.progress_bar.formatters.create_default_formatters() → list[Formatter]¶: Return the list of default formatters.

Validation¶

Input validation for a Buffer. (Validators will be called before accepting input.)

class prompt_toolkit.validation.ConditionalValidator(validator: Validator, filter: Filter | bool)¶: Validator that can be switched on/off according to a filter. (This wraps around another validator.)

class prompt_toolkit.validation.DummyValidator¶: Validator class that accepts any input.

class prompt_toolkit.validation.DynamicValidator(get_validator: Callable[[], Validator | None])¶

Validator class that can dynamically returns any Validator.

Parameters:: get_validator – Callable that returns a Validator instance.

class prompt_toolkit.validation.ThreadedValidator(validator: Validator)¶

Wrapper that runs input validation in a thread. (Use this to prevent the user interface from becoming unresponsive if the input validation takes too much time.)

async validate_async(document: Document) → None¶: Run the validate function in a thread.

exception prompt_toolkit.validation.ValidationError(cursor_position: int = 0, message: str = '')¶

Error raised by Validator.validate().

Parameters:

cursor_position – The cursor position where the error occurred.
message – Text.

class prompt_toolkit.validation.Validator¶

Abstract base class for an input validator.

A validator is typically created in one of the following two ways:

Either by overriding this class and implementing the validate method.
Or by passing a callable to Validator.from_callable.

If the validation takes some time and needs to happen in a background thread, this can be wrapped in a ThreadedValidator.

classmethod from_callable(validate_func: Callable[[str], bool], error_message: str = 'Invalid input', move_cursor_to_end: bool = False) → Validator¶

Create a validator from a simple validate callable. E.g.:

def is_valid(text):
    return text in ['hello', 'world']
Validator.from_callable(is_valid, error_message='Invalid input')

Parameters:

validate_func – Callable that takes the input string, and returns True if the input is valid input.
error_message – Message to be displayed if the input is invalid.
move_cursor_to_end – Move the cursor to the end of the input, if the input is invalid.

abstractmethod validate(document: Document) → None¶

Validate the input. If invalid, this should raise a ValidationError.

Parameters:: document – Document instance.

async validate_async(document: Document) → None¶: Return a Future which is set when the validation is ready. This function can be overloaded in order to provide an asynchronous implementation.

Auto suggestion¶

Fish-style like auto-suggestion.

While a user types input in a certain buffer, suggestions are generated (asynchronously.) Usually, they are displayed after the input. When the cursor presses the right arrow and the cursor is at the end of the input, the suggestion will be inserted.

If you want the auto suggestions to be asynchronous (in a background thread), because they take too much time, and could potentially block the event loop, then wrap the AutoSuggest instance into a ThreadedAutoSuggest.

class prompt_toolkit.auto_suggest.AutoSuggest¶

Base class for auto suggestion implementations.

abstractmethod get_suggestion(buffer: Buffer, document: Document) → Suggestion | None¶

Return None or a Suggestion instance.

We receive both Buffer and Document. The reason is that auto suggestions are retrieved asynchronously. (Like completions.) The buffer text could be changed in the meantime, but document contains the buffer document like it was at the start of the auto suggestion call. So, from here, don’t access buffer.text, but use document.text instead.

Parameters:

buffer – The Buffer instance.
document – The Document instance.

async get_suggestion_async(buff: Buffer, document: Document) → Suggestion | None¶: Return a Future which is set when the suggestions are ready. This function can be overloaded in order to provide an asynchronous implementation.

class prompt_toolkit.auto_suggest.AutoSuggestFromHistory¶: Give suggestions based on the lines in the history.

class prompt_toolkit.auto_suggest.ConditionalAutoSuggest(auto_suggest: AutoSuggest, filter: bool | Filter)¶: Auto suggest that can be turned on and of according to a certain condition.

class prompt_toolkit.auto_suggest.DummyAutoSuggest¶: AutoSuggest class that doesn’t return any suggestion.

class prompt_toolkit.auto_suggest.DynamicAutoSuggest(get_auto_suggest: Callable[[], AutoSuggest | None])¶

Validator class that can dynamically returns any Validator.

Parameters:: get_validator – Callable that returns a Validator instance.

class prompt_toolkit.auto_suggest.Suggestion(text: str)¶

Suggestion returned by an auto-suggest algorithm.

Parameters:: text – The suggestion text.

class prompt_toolkit.auto_suggest.ThreadedAutoSuggest(auto_suggest: AutoSuggest)¶

Wrapper that runs auto suggestions in a thread. (Use this to prevent the user interface from becoming unresponsive if the generation of suggestions takes too much time.)

async get_suggestion_async(buff: Buffer, document: Document) → Suggestion | None¶: Run the get_suggestion function in a thread.

Renderer¶

Renders the command line on the console. (Redraws parts of the input line that were changed.)

class prompt_toolkit.renderer.Renderer(style: BaseStyle, output: Output, full_screen: bool = False, mouse_support: Filter | bool = False, cpr_not_supported_callback: Callable[[], None] | None = None)¶

Typical usage:

output = Vt100_Output.from_pty(sys.stdout)
r = Renderer(style, output)
r.render(app, layout=...)

clear() → None¶: Clear screen and go to 0,0

erase(leave_alternate_screen: bool = True) → None¶

Hide all output and put the cursor back at the first line. This is for instance used for running a system command (while hiding the CLI) and later resuming the same CLI.)

Parameters:: leave_alternate_screen – When True, and when inside an alternate screen buffer, quit the alternate screen.

property height_is_known: bool¶: True when the height from the cursor until the bottom of the terminal is known. (It’s often nicer to draw bottom toolbars only if the height is known, in order to avoid flickering when the CPR response arrives.)

property last_rendered_screen: Screen | None¶: The Screen class that was generated during the last rendering. This can be None.

render(app: Application[Any], layout: Layout, is_done: bool = False) → None¶

Render the current interface to the output.

Parameters:: is_done – When True, put the cursor at the end of the interface. We won’t print any changes to this part.

report_absolute_cursor_row(row: int) → None¶: To be called when we know the absolute cursor position. (As an answer of a “Cursor Position Request” response.)

request_absolute_cursor_position() → None¶

Get current cursor position.

We do this to calculate the minimum available height that we can consume for rendering the prompt. This is the available space below te cursor.

For vt100: Do CPR request. (answer will arrive later.) For win32: Do API call. (Answer comes immediately.)

property rows_above_layout: int¶: Return the number of rows visible in the terminal above the layout.

async wait_for_cpr_responses(timeout: int = 1) → None¶: Wait for a CPR response.

property waiting_for_cpr: bool¶: Waiting for CPR flag. True when we send the request, but didn’t got a response.

prompt_toolkit.renderer.print_formatted_text(output: Output, formatted_text: AnyFormattedText, style: BaseStyle, style_transformation: StyleTransformation | None = None, color_depth: ColorDepth | None = None) → None¶: Print a list of (style_str, text) tuples in the given style to the output.

Lexers¶

Lexer interface and implementations. Used for syntax highlighting.

class prompt_toolkit.lexers.DynamicLexer(get_lexer: Callable[[], Lexer | None])¶

Lexer class that can dynamically returns any Lexer.

Parameters:: get_lexer – Callable that returns a Lexer instance.

class prompt_toolkit.lexers.Lexer¶

Base class for all lexers.

invalidation_hash() → Hashable¶: When this changes, lex_document could give a different output. (Only used for DynamicLexer.)

abstractmethod lex_document(document: Document) → Callable[[int], StyleAndTextTuples]¶

Takes a Document and returns a callable that takes a line number and returns a list of (style_str, text) tuples for that line.

XXX: Note that in the past, this was supposed to return a list: of (Token, text) tuples, just like a Pygments lexer.

class prompt_toolkit.lexers.PygmentsLexer(pygments_lexer_cls: type[PygmentsLexerCls], sync_from_start: FilterOrBool = True, syntax_sync: SyntaxSync | None = None)¶

Lexer that calls a pygments lexer.

Example:

from pygments.lexers.html import HtmlLexer
lexer = PygmentsLexer(HtmlLexer)

Note: Don’t forget to also load a Pygments compatible style. E.g.:

from prompt_toolkit.styles.from_pygments import style_from_pygments_cls
from pygments.styles import get_style_by_name
style = style_from_pygments_cls(get_style_by_name('monokai'))

Parameters:

pygments_lexer_cls – A Lexer from Pygments.
sync_from_start – Start lexing at the start of the document. This will always give the best results, but it will be slow for bigger documents. (When the last part of the document is display, then the whole document will be lexed by Pygments on every key stroke.) It is recommended to disable this for inputs that are expected to be more than 1,000 lines.
syntax_sync – SyntaxSync object.

classmethod from_filename(filename: str, sync_from_start: Filter | bool = True) → Lexer¶: Create a Lexer from a filename.

lex_document(document: Document) → Callable[[int], StyleAndTextTuples]¶: Create a lexer function that takes a line number and returns the list of (style_str, text) tuples as the Pygments lexer returns for that line.

class prompt_toolkit.lexers.RegexSync(pattern: str)¶

Synchronize by starting at a line that matches the given regex pattern.

classmethod from_pygments_lexer_cls(lexer_cls: PygmentsLexerCls) → RegexSync¶: Create a RegexSync instance for this Pygments lexer class.

get_sync_start_position(document: Document, lineno: int) → tuple[int, int]¶: Scan backwards, and find a possible position to start.

class prompt_toolkit.lexers.SimpleLexer(style: str = '')¶

Lexer that doesn’t do any tokenizing and returns the whole input as one token.

Parameters:: style – The style string for this lexer.

class prompt_toolkit.lexers.SyncFromStart¶: Always start the syntax highlighting from the beginning.

class prompt_toolkit.lexers.SyntaxSync¶

Syntax synchronizer. This is a tool that finds a start position for the lexer. This is especially important when editing big documents; we don’t want to start the highlighting by running the lexer from the beginning of the file. That is very slow when editing.

abstractmethod get_sync_start_position(document: Document, lineno: int) → tuple[int, int]¶

Return the position from where we can start lexing as a (row, column) tuple.

Parameters:

document – Document instance that contains all the lines.
lineno – The line that we want to highlight. (We need to return this line, or an earlier position.)

Layout¶

Command line layout definitions¶

The layout of a command line interface is defined by a Container instance. There are two main groups of classes here. Containers and controls:

A container can contain other containers or controls, it can have multiple children and it decides about the dimensions.
A control is responsible for rendering the actual content to a screen. A control can propose some dimensions, but it’s the container who decides about the dimensions – or when the control consumes more space – which part of the control will be visible.

Container classes:

- Container (Abstract base class)
   |- HSplit (Horizontal split)
   |- VSplit (Vertical split)
   |- FloatContainer (Container which can also contain menus and other floats)
   `- Window (Container which contains one actual control

Control classes:

- UIControl (Abstract base class)
   |- FormattedTextControl (Renders formatted text, or a simple list of text fragments)
   `- BufferControl (Renders an input buffer.)

Usually, you end up wrapping every control inside a Window object, because that’s the only way to render it in a layout.

There are some prepared toolbars which are ready to use:

- SystemToolbar (Shows the 'system' input buffer, for entering system commands.)
- ArgToolbar (Shows the input 'arg', for repetition of input commands.)
- SearchToolbar (Shows the 'search' input buffer, for incremental search.)
- CompletionsToolbar (Shows the completions of the current buffer.)
- ValidationToolbar (Shows validation errors of the current buffer.)

And one prepared menu:

CompletionsMenu

The layout class itself¶

class prompt_toolkit.layout.Layout(container: AnyContainer, focused_element: FocusableElement | None = None)¶

The layout for a prompt_toolkit Application. This also keeps track of which user control is focused.

Parameters:

container – The “root” container for the layout.
focused_element – element to be focused initially. (Can be anything the focus function accepts.)

property buffer_has_focus: bool¶: Return True if the currently focused control is a BufferControl. (For instance, used to determine whether the default key bindings should be active or not.)

property current_buffer: Buffer | None¶: The currently focused Buffer or None.

property current_control: UIControl¶: Get the UIControl to currently has the focus.

property current_window: Window¶: Return the Window object that is currently focused.

find_all_windows() → Generator[Window, None, None]¶: Find all the UIControl objects in this layout.

focus(value: FocusableElement) → None¶

Focus the given UI element.

value can be either:

a UIControl
a Buffer instance or the name of a Buffer
a Window
Any container object. In this case we will focus the Window from this container that was focused most recent, or the very first focusable Window of the container.

focus_last() → None¶: Give the focus to the last focused control.

focus_next() → None¶: Focus the next visible/focusable Window.

focus_previous() → None¶: Focus the previous visible/focusable Window.

get_buffer_by_name(buffer_name: str) → Buffer | None¶: Look in the layout for a buffer with the given name. Return None when nothing was found.

get_focusable_windows() → Iterable[Window]¶: Return all the Window objects which are focusable (in the ‘modal’ area).

get_parent(container: Container) → Container | None¶: Return the parent container for the given container, or None, if it wasn’t found.

get_visible_focusable_windows() → list[Window]¶: Return a list of Window objects that are focusable.

has_focus(value: FocusableElement) → bool¶: Check whether the given control has the focus. :param value: UIControl or Window instance.

property is_searching: bool¶: True if we are searching right now.

property previous_control: UIControl¶: Get the UIControl to previously had the focus.

property search_target_buffer_control: BufferControl | None¶: Return the BufferControl in which we are searching or None.

update_parents_relations() → None¶: Update child->parent relationships mapping.

walk() → Iterable[Container]¶: Walk through all the layout nodes (and their children) and yield them.

walk_through_modal_area() → Iterable[Container]¶: Walk through all the containers which are in the current ‘modal’ part of the layout.

class prompt_toolkit.layout.InvalidLayoutError¶

class prompt_toolkit.layout.walk(container: Container, skip_hidden: bool = False)¶: Walk through layout, starting at this container.

Containers¶

class prompt_toolkit.layout.Container¶

Base class for user interface layout.

abstractmethod get_children() → list[Container]¶: Return the list of child Container objects.

get_key_bindings() → KeyBindingsBase | None¶: Returns a KeyBindings object. These bindings become active when any user control in this container has the focus, except if any containers between this container and the focused user control is modal.

is_modal() → bool¶: When this container is modal, key bindings from parent containers are not taken into account if a user control in this container is focused.

abstractmethod preferred_height(width: int, max_available_height: int) → Dimension¶: Return a Dimension that represents the desired height for this container.

abstractmethod preferred_width(max_available_width: int) → Dimension¶: Return a Dimension that represents the desired width for this container.

abstractmethod reset() → None¶: Reset the state of this container and all the children. (E.g. reset scroll offsets, etc…)

abstractmethod write_to_screen(screen: Screen, mouse_handlers: MouseHandlers, write_position: WritePosition, parent_style: str, erase_bg: bool, z_index: int | None) → None¶

Write the actual content to the screen.

Parameters:

screen – Screen
mouse_handlers – MouseHandlers.
parent_style – Style string to pass to the Window object. This will be applied to all content of the windows. VSplit and HSplit can use it to pass their style down to the windows that they contain.
z_index – Used for propagating z_index from parent to child.

class prompt_toolkit.layout.HSplit(children: Sequence[AnyContainer], window_too_small: Container | None = None, align: VerticalAlign = VerticalAlign.JUSTIFY, padding: AnyDimension = 0, padding_char: str | None = None, padding_style: str = '', width: AnyDimension = None, height: AnyDimension = None, z_index: int | None = None, modal: bool = False, key_bindings: KeyBindingsBase | None = None, style: str | Callable[[], str] = '')¶

Several layouts, one stacked above/under the other.

+--------------------+
|                    |
+--------------------+
|                    |
+--------------------+

By default, this doesn’t display a horizontal line between the children, but if this is something you need, then create a HSplit as follows:

HSplit(children=[ ... ], padding_char='-',
       padding=1, padding_style='#ffff00')

Parameters:

children – List of child Container objects.
window_too_small – A Container object that is displayed if there is not enough space for all the children. By default, this is a “Window too small” message.
align – VerticalAlign value.
width – When given, use this width instead of looking at the children.
height – When given, use this height instead of looking at the children.
z_index – (int or None) When specified, this can be used to bring element in front of floating elements. None means: inherit from parent.
style – A style string.
modal – True or False.
key_bindings – None or a KeyBindings object.
padding – (Dimension or int), size to be used for the padding.
padding_char – Character to be used for filling in the padding.
padding_style – Style to applied to the padding.

write_to_screen(screen: Screen, mouse_handlers: MouseHandlers, write_position: WritePosition, parent_style: str, erase_bg: bool, z_index: int | None) → None¶

Render the prompt to a Screen instance.

Parameters:: screen – The Screen class to which the output has to be written.

class prompt_toolkit.layout.VSplit(children: Sequence[AnyContainer], window_too_small: Container | None = None, align: HorizontalAlign = HorizontalAlign.JUSTIFY, padding: AnyDimension = 0, padding_char: str | None = None, padding_style: str = '', width: AnyDimension = None, height: AnyDimension = None, z_index: int | None = None, modal: bool = False, key_bindings: KeyBindingsBase | None = None, style: str | Callable[[], str] = '')¶

Several layouts, one stacked left/right of the other.

+---------+----------+
|         |          |
|         |          |
+---------+----------+

By default, this doesn’t display a vertical line between the children, but if this is something you need, then create a HSplit as follows:

VSplit(children=[ ... ], padding_char='|',
       padding=1, padding_style='#ffff00')

Parameters:

children – List of child Container objects.
window_too_small – A Container object that is displayed if there is not enough space for all the children. By default, this is a “Window too small” message.
align – HorizontalAlign value.
width – When given, use this width instead of looking at the children.
height – When given, use this height instead of looking at the children.
z_index – (int or None) When specified, this can be used to bring element in front of floating elements. None means: inherit from parent.
style – A style string.
modal – True or False.
key_bindings – None or a KeyBindings object.
padding – (Dimension or int), size to be used for the padding.
padding_char – Character to be used for filling in the padding.
padding_style – Style to applied to the padding.

write_to_screen(screen: Screen, mouse_handlers: MouseHandlers, write_position: WritePosition, parent_style: str, erase_bg: bool, z_index: int | None) → None¶

Render the prompt to a Screen instance.

Parameters:: screen – The Screen class to which the output has to be written.

class prompt_toolkit.layout.FloatContainer(content: AnyContainer, floats: list[Float], modal: bool = False, key_bindings: KeyBindingsBase | None = None, style: str | Callable[[], str] = '', z_index: int | None = None)¶

Container which can contain another container for the background, as well as a list of floating containers on top of it.

Example Usage:

FloatContainer(content=Window(...),
               floats=[
                   Float(xcursor=True,
                        ycursor=True,
                        content=CompletionsMenu(...))
               ])

Parameters:: z_index – (int or None) When specified, this can be used to bring element in front of floating elements. None means: inherit from parent. This is the z_index for the whole Float container as a whole.

preferred_height(width: int, max_available_height: int) → Dimension¶: Return the preferred height of the float container. (We don’t care about the height of the floats, they should always fit into the dimensions provided by the container.)

class prompt_toolkit.layout.Float(content: AnyContainer, top: int | None = None, right: int | None = None, bottom: int | None = None, left: int | None = None, width: int | Callable[[], int] | None = None, height: int | Callable[[], int] | None = None, xcursor: bool = False, ycursor: bool = False, attach_to_window: AnyContainer | None = None, hide_when_covering_content: bool = False, allow_cover_cursor: bool = False, z_index: int = 1, transparent: bool = False)¶

Float for use in a FloatContainer. Except for the content parameter, all other options are optional.

Parameters:

content – Container instance.
width – Dimension or callable which returns a Dimension.
height – Dimension or callable which returns a Dimension.
left – Distance to the left edge of the FloatContainer.
right – Distance to the right edge of the FloatContainer.
top – Distance to the top of the FloatContainer.
bottom – Distance to the bottom of the FloatContainer.
attach_to_window – Attach to the cursor from this window, instead of the current window.
hide_when_covering_content – Hide the float when it covers content underneath.
allow_cover_cursor – When False, make sure to display the float below the cursor. Not on top of the indicated position.
z_index – Z-index position. For a Float, this needs to be at least one. It is relative to the z_index of the parent container.
transparent – Filter indicating whether this float needs to be drawn transparently.

class prompt_toolkit.layout.Window(content: UIControl | None = None, width: AnyDimension = None, height: AnyDimension = None, z_index: int | None = None, dont_extend_width: FilterOrBool = False, dont_extend_height: FilterOrBool = False, ignore_content_width: FilterOrBool = False, ignore_content_height: FilterOrBool = False, left_margins: Sequence[Margin] | None = None, right_margins: Sequence[Margin] | None = None, scroll_offsets: ScrollOffsets | None = None, allow_scroll_beyond_bottom: FilterOrBool = False, wrap_lines: FilterOrBool = False, get_vertical_scroll: Callable[[Window], int] | None = None, get_horizontal_scroll: Callable[[Window], int] | None = None, always_hide_cursor: FilterOrBool = False, cursorline: FilterOrBool = False, cursorcolumn: FilterOrBool = False, colorcolumns: None | list[ColorColumn] | Callable[[], list[ColorColumn]] = None, align: WindowAlign | Callable[[], WindowAlign] = WindowAlign.LEFT, style: str | Callable[[], str] = '', char: None | str | Callable[[], str] = None, get_line_prefix: GetLinePrefixCallable | None = None)¶

Container that holds a control.

Parameters:

content – UIControl instance.
width – Dimension instance or callable.
height – Dimension instance or callable.
z_index – When specified, this can be used to bring element in front of floating elements.
dont_extend_width – When True, don’t take up more width then the preferred width reported by the control.
dont_extend_height – When True, don’t take up more width then the preferred height reported by the control.
ignore_content_width – A bool or Filter instance. Ignore the UIContent width when calculating the dimensions.
ignore_content_height – A bool or Filter instance. Ignore the UIContent height when calculating the dimensions.
left_margins – A list of Margin instance to be displayed on the left. For instance: NumberedMargin can be one of them in order to show line numbers.
right_margins – Like left_margins, but on the other side.
scroll_offsets – ScrollOffsets instance, representing the preferred amount of lines/columns to be always visible before/after the cursor. When both top and bottom are a very high number, the cursor will be centered vertically most of the time.
allow_scroll_beyond_bottom – A bool or Filter instance. When True, allow scrolling so far, that the top part of the content is not visible anymore, while there is still empty space available at the bottom of the window. In the Vi editor for instance, this is possible. You will see tildes while the top part of the body is hidden.
wrap_lines – A bool or Filter instance. When True, don’t scroll horizontally, but wrap lines instead.
get_vertical_scroll – Callable that takes this window instance as input and returns a preferred vertical scroll. (When this is None, the scroll is only determined by the last and current cursor position.)
get_horizontal_scroll – Callable that takes this window instance as input and returns a preferred vertical scroll.
always_hide_cursor – A bool or Filter instance. When True, never display the cursor, even when the user control specifies a cursor position.
cursorline – A bool or Filter instance. When True, display a cursorline.
cursorcolumn – A bool or Filter instance. When True, display a cursorcolumn.
colorcolumns – A list of ColorColumn instances that describe the columns to be highlighted, or a callable that returns such a list.
align – WindowAlign value or callable that returns an WindowAlign value. alignment of content.
style – A style string. Style to be applied to all the cells in this window. (This can be a callable that returns a string.)
char – (string) Character to be used for filling the background. This can also be a callable that returns a character.
get_line_prefix – None or a callable that returns formatted text to be inserted before a line. It takes a line number (int) and a wrap_count and returns formatted text. This can be used for implementation of line continuations, things like Vim “breakindent” and so on.

preferred_height(width: int, max_available_height: int) → Dimension¶: Calculate the preferred height for this window.

preferred_width(max_available_width: int) → Dimension¶: Calculate the preferred width for this window.

write_to_screen(screen: Screen, mouse_handlers: MouseHandlers, write_position: WritePosition, parent_style: str, erase_bg: bool, z_index: int | None) → None¶: Write window to screen. This renders the user control, the margins and copies everything over to the absolute position at the given screen.

class prompt_toolkit.layout.WindowAlign(value)¶

Alignment of the Window content.

Note that this is different from HorizontalAlign and VerticalAlign, which are used for the alignment of the child containers in respectively VSplit and HSplit.

class prompt_toolkit.layout.ConditionalContainer(content: AnyContainer, filter: FilterOrBool)¶

Wrapper around any other container that can change the visibility. The received filter determines whether the given container should be displayed or not.

Parameters:

content – Container instance.
filter – Filter instance.

class prompt_toolkit.layout.DynamicContainer(get_container: Callable[[], AnyContainer])¶

Container class that dynamically returns any Container.

Parameters:: get_container – Callable that returns a Container instance or any widget with a __pt_container__ method.

Container widget that exposes a larger virtual screen to its content and displays it in a vertical scrollbale region.

Typically this is wrapped in a large HSplit container. Make sure in that case to not specify a height dimension of the HSplit, so that it will scale according to the content.

Note

If you want to display a completion menu for widgets in this ScrollablePane, then it’s still a good practice to use a FloatContainer with a CompletionsMenu in a Float at the top-level of the layout hierarchy, rather then nesting a FloatContainer in this ScrollablePane. (Otherwise, it’s possible that the completion menu is clipped.)

Parameters:

content – The content container.
scrolloffset – Try to keep the cursor within this distance from the top/bottom (left/right offset is not used).
keep_cursor_visible – When True, automatically scroll the pane so that the cursor (of the focused window) is always visible.
keep_focused_window_visible – When True, automatically scroll the pane so that the focused window is visible, or as much visible as possible if it doesn’t completely fit the screen.
max_available_height – Always constraint the height to this amount for performance reasons.
width – When given, use this width instead of looking at the children.
height – When given, use this height instead of looking at the children.
show_scrollbar – When True display a scrollbar on the right.

write_to_screen(screen: Screen, mouse_handlers: MouseHandlers, write_position: WritePosition, parent_style: str, erase_bg: bool, z_index: int | None) → None¶

Render scrollable pane content.

This works by rendering on an off-screen canvas, and copying over the visible region.

class prompt_toolkit.layout.ScrollOffsets(top: int | Callable[[], int] = 0, bottom: int | Callable[[], int] = 0, left: int | Callable[[], int] = 0, right: int | Callable[[], int] = 0)¶

Scroll offsets for the Window class.

Note that left/right offsets only make sense if line wrapping is disabled.

class prompt_toolkit.layout.ColorColumn(position: int, style: str = 'class:color-column')¶: Column for a Window to be colored.

class prompt_toolkit.layout.to_container(container: AnyContainer)¶: Make sure that the given object is a Container.

class prompt_toolkit.layout.to_window(container: AnyContainer)¶: Make sure that the given argument is a Window.

class prompt_toolkit.layout.is_container(value: object)¶: Checks whether the given value is a container object (for use in assert statements).

class prompt_toolkit.layout.HorizontalAlign(value)¶: Alignment for VSplit.

class prompt_toolkit.layout.VerticalAlign(value)¶: Alignment for HSplit.

Controls¶

class prompt_toolkit.layout.BufferControl(buffer: Buffer | None = None, input_processors: list[Processor] | None = None, include_default_input_processors: bool = True, lexer: Lexer | None = None, preview_search: FilterOrBool = False, focusable: FilterOrBool = True, search_buffer_control: None | SearchBufferControl | Callable[[], SearchBufferControl] = None, menu_position: Callable[[], int | None] | None = None, focus_on_click: FilterOrBool = False, key_bindings: KeyBindingsBase | None = None)¶

Control for visualizing the content of a Buffer.

Parameters:

buffer – The Buffer object to be displayed.
input_processors – A list of Processor objects.
include_default_input_processors – When True, include the default processors for highlighting of selection, search and displaying of multiple cursors.
lexer – Lexer instance for syntax highlighting.
preview_search – bool or Filter: Show search while typing. When this is True, probably you want to add a HighlightIncrementalSearchProcessor as well. Otherwise only the cursor position will move, but the text won’t be highlighted.
focusable – bool or Filter: Tell whether this control is focusable.
focus_on_click – Focus this buffer when it’s click, but not yet focused.
key_bindings – a KeyBindings object.

create_content(width: int, height: int, preview_search: bool = False) → UIContent¶: Create a UIContent.

get_invalidate_events() → Iterable[Event[object]]¶: Return the Window invalidate events.

get_key_bindings() → KeyBindingsBase | None¶: When additional key bindings are given. Return these.

mouse_handler(mouse_event: MouseEvent) → NotImplementedOrNone¶: Mouse handler for this control.

preferred_width(max_available_width: int) → int | None¶

This should return the preferred width.

Note: We don’t specify a preferred width according to the content,: because it would be too expensive. Calculating the preferred width can be done by calculating the longest line, but this would require applying all the processors to each line. This is unfeasible for a larger document, and doing it for small documents only would result in inconsistent behavior.

property search_state: SearchState¶: Return the SearchState for searching this BufferControl. This is always associated with the search control. If one search bar is used for searching multiple BufferControls, then they share the same SearchState.

class prompt_toolkit.layout.SearchBufferControl(buffer: Buffer | None = None, input_processors: list[Processor] | None = None, lexer: Lexer | None = None, focus_on_click: FilterOrBool = False, key_bindings: KeyBindingsBase | None = None, ignore_case: FilterOrBool = False)¶

BufferControl which is used for searching another BufferControl.

Parameters:: ignore_case – Search case insensitive.

class prompt_toolkit.layout.DummyControl¶

A dummy control object that doesn’t paint any content.

Useful for filling a Window. (The fragment and char attributes of the Window class can be used to define the filling.)

class prompt_toolkit.layout.FormattedTextControl(text: AnyFormattedText = '', style: str = '', focusable: FilterOrBool = False, key_bindings: KeyBindingsBase | None = None, show_cursor: bool = True, modal: bool = False, get_cursor_position: Callable[[], Point | None] | None = None)¶

Control that displays formatted text. This can be either plain text, an HTML object an ANSI object, a list of (style_str, text) tuples or a callable that takes no argument and returns one of those, depending on how you prefer to do the formatting. See prompt_toolkit.layout.formatted_text for more information.

(It’s mostly optimized for rather small widgets, like toolbars, menus, etc…)

When this UI control has the focus, the cursor will be shown in the upper left corner of this control by default. There are two ways for specifying the cursor position:

Pass a get_cursor_position function which returns a Point instance with the current cursor position.
If the (formatted) text is passed as a list of (style, text) tuples and there is one that looks like ('[SetCursorPosition]', ''), then this will specify the cursor position.

Mouse support:

The list of fragments can also contain tuples of three items, looking like: (style_str, text, handler). When mouse support is enabled and the user clicks on this fragment, then the given handler is called. That handler should accept two inputs: (Application, MouseEvent) and it should either handle the event or return NotImplemented in case we want the containing Window to handle this event.

Parameters:

focusable – bool or Filter: Tell whether this control is focusable.
text – Text or formatted text to be displayed.
style – Style string applied to the content. (If you want to style the whole Window, pass the style to the Window instead.)
key_bindings – a KeyBindings object.
get_cursor_position – A callable that returns the cursor position as a Point instance.

mouse_handler(mouse_event: MouseEvent) → NotImplementedOrNone¶

Handle mouse events.

(When the fragment list contained mouse handlers and the user clicked on on any of these, the matching handler is called. This handler can still return NotImplemented in case we want the Window to handle this particular event.)

preferred_height(width: int, max_available_height: int, wrap_lines: bool, get_line_prefix: GetLinePrefixCallable | None) → int | None¶: Return the preferred height for this control.

preferred_width(max_available_width: int) → int¶: Return the preferred width for this control. That is the width of the longest line.

class prompt_toolkit.layout.UIControl¶

Base class for all user interface controls.

abstractmethod create_content(width: int, height: int) → UIContent¶

Generate the content for this user control.

Returns a UIContent instance.

get_invalidate_events() → Iterable[Event[object]]¶: Return a list of Event objects. This can be a generator. (The application collects all these events, in order to bind redraw handlers to these events.)

get_key_bindings() → KeyBindingsBase | None¶

The key bindings that are specific for this user control.

Return a KeyBindings object if some key bindings are specified, or None otherwise.

is_focusable() → bool¶: Tell whether this user control is focusable.

mouse_handler(mouse_event: MouseEvent) → NotImplementedOrNone¶

Handle mouse events.

When NotImplemented is returned, it means that the given event is not handled by the UIControl itself. The Window or key bindings can decide to handle this event as scrolling or changing focus.

Parameters:: mouse_event – MouseEvent instance.

move_cursor_down() → None¶: Request to move the cursor down. This happens when scrolling down and the cursor is completely at the top.

move_cursor_up() → None¶: Request to move the cursor up.

class prompt_toolkit.layout.UIContent(get_line: Callable[[int], StyleAndTextTuples] = <function UIContent.<lambda>>, line_count: int = 0, cursor_position: Point | None = None, menu_position: Point | None = None, show_cursor: bool = True)¶

Content generated by a user control. This content consists of a list of lines.

Parameters:

get_line – Callable that takes a line number and returns the current line. This is a list of (style_str, text) tuples.
line_count – The number of lines.
cursor_position – a Point for the cursor position.
menu_position – a Point for the menu position.
show_cursor – Make the cursor visible.

get_height_for_line(lineno: int, width: int, get_line_prefix: GetLinePrefixCallable | None, slice_stop: int | None = None) → int¶

Return the height that a given line would need if it is rendered in a space with the given width (using line wrapping).

Parameters:

get_line_prefix – None or a Window.get_line_prefix callable that returns the prefix to be inserted before this line.
slice_stop – Wrap only “line[:slice_stop]” and return that partial result. This is needed for scrolling the window correctly when line wrapping.

Returns:

The computed height.

Other¶

Sizing¶

class prompt_toolkit.layout.Dimension(min: int | None = None, max: int | None = None, weight: int | None = None, preferred: int | None = None)¶

Specified dimension (width/height) of a user control or window.

The layout engine tries to honor the preferred size. If that is not possible, because the terminal is larger or smaller, it tries to keep in between min and max.

Parameters:

min – Minimum size.
max – Maximum size.
weight – For a VSplit/HSplit, the actual size will be determined by taking the proportion of weights from all the children. E.g. When there are two children, one with a weight of 1, and the other with a weight of 2, the second will always be twice as big as the first, if the min/max values allow it.
preferred – Preferred size.

classmethod exact(amount: int) → Dimension¶: Return a Dimension with an exact size. (min, max and preferred set to amount).

is_zero() → bool¶: True if this Dimension represents a zero size.

classmethod zero() → Dimension¶: Create a dimension that represents a zero size. (Used for ‘invisible’ controls.)

Margins¶

class prompt_toolkit.layout.Margin¶

Base interface for a margin.

abstractmethod create_margin(window_render_info: WindowRenderInfo, width: int, height: int) → StyleAndTextTuples¶

Creates a margin. This should return a list of (style_str, text) tuples.

Parameters:

window_render_info – WindowRenderInfo instance, generated after rendering and copying the visible part of the UIControl into the Window.
width – The width that’s available for this margin. (As reported by get_width().)
height – The height that’s available for this margin. (The height of the Window.)

abstractmethod get_width(get_ui_content: Callable[[], UIContent]) → int¶

Return the width that this margin is going to consume.

Parameters:: get_ui_content – Callable that asks the user control to create a UIContent instance. This can be used for instance to obtain the number of lines.

class prompt_toolkit.layout.NumberedMargin(relative: Filter | bool = False, display_tildes: Filter | bool = False)¶

Margin that displays the line numbers.

Parameters:

relative – Number relative to the cursor position. Similar to the Vi ‘relativenumber’ option.
display_tildes – Display tildes after the end of the document, just like Vi does.

class prompt_toolkit.layout.ScrollbarMargin(display_arrows: Filter | bool = False, up_arrow_symbol: str = '^', down_arrow_symbol: str = 'v')¶

Margin displaying a scrollbar.

Parameters:: display_arrows – Display scroll up/down arrows.

class prompt_toolkit.layout.ConditionalMargin(margin: Margin, filter: Filter | bool)¶: Wrapper around other Margin classes to show/hide them.

class prompt_toolkit.layout.PromptMargin(get_prompt: Callable[[], StyleAndTextTuples], get_continuation: None | Callable[[int, int, bool], StyleAndTextTuples] = None)¶

[Deprecated]

Create margin that displays a prompt. This can display one prompt at the first line, and a continuation prompt (e.g, just dots) on all the following lines.

This PromptMargin implementation has been largely superseded in favor of the get_line_prefix attribute of Window. The reason is that a margin is always a fixed width, while get_line_prefix can return a variable width prefix in front of every line, making it more powerful, especially for line continuations.

Parameters:

get_prompt – Callable returns formatted text or a list of (style_str, type) tuples to be shown as the prompt at the first line.
get_continuation – Callable that takes three inputs. The width (int), line_number (int), and is_soft_wrap (bool). It should return formatted text or a list of (style_str, type) tuples for the next lines of the input.

get_width(get_ui_content: Callable[[], UIContent]) → int¶: Width to report to the Window.

Completion Menus¶

class prompt_toolkit.layout.CompletionsMenu(max_height: int | None = None, scroll_offset: int | Callable[[], int] = 0, extra_filter: Filter | bool = True, display_arrows: Filter | bool = False, z_index: int = 100000000)¶

class prompt_toolkit.layout.MultiColumnCompletionsMenu(min_rows: int = 3, suggested_max_column_width: int = 30, show_meta: Filter | bool = True, extra_filter: Filter | bool = True, z_index: int = 100000000)¶: Container that displays the completions in several columns. When show_meta (a Filter) evaluates to True, it shows the meta information at the bottom.

Processors¶

Processors are little transformation blocks that transform the fragments list from a buffer before the BufferControl will render it to the screen.

They can insert fragments before or after, or highlight fragments by replacing the fragment types.

class prompt_toolkit.layout.processors.AfterInput(text: AnyFormattedText, style: str = '')¶

Insert text after the input.

Parameters:

text – This can be either plain text or formatted text (or a callable that returns any of those).
style – style to be applied to this prompt/prefix.

class prompt_toolkit.layout.processors.AppendAutoSuggestion(style: str = 'class:auto-suggestion')¶: Append the auto suggestion to the input. (The user can then press the right arrow the insert the suggestion.)

class prompt_toolkit.layout.processors.BeforeInput(text: AnyFormattedText, style: str = '')¶

Insert text before the input.

Parameters:

text – This can be either plain text or formatted text (or a callable that returns any of those).
style – style to be applied to this prompt/prefix.

class prompt_toolkit.layout.processors.ConditionalProcessor(processor: Processor, filter: Filter | bool)¶

Processor that applies another processor, according to a certain condition. Example:

# Create a function that returns whether or not the processor should
# currently be applied.
def highlight_enabled():
    return true_or_false

# Wrapped it in a `ConditionalProcessor` for usage in a `BufferControl`.
BufferControl(input_processors=[
    ConditionalProcessor(HighlightSearchProcessor(),
                         Condition(highlight_enabled))])

Parameters:

processor – Processor instance.
filter – Filter instance.

class prompt_toolkit.layout.processors.DisplayMultipleCursors¶: When we’re in Vi block insert mode, display all the cursors.

class prompt_toolkit.layout.processors.DummyProcessor¶: A Processor that doesn’t do anything.

class prompt_toolkit.layout.processors.DynamicProcessor(get_processor: Callable[[], Processor | None])¶

Processor class that dynamically returns any Processor.

Parameters:: get_processor – Callable that returns a Processor instance.

class prompt_toolkit.layout.processors.HighlightIncrementalSearchProcessor¶

Highlight the search terms that are used for highlighting the incremental search. The style class ‘incsearch’ will be applied to the content.

Important: this requires the preview_search=True flag to be set for the BufferControl. Otherwise, the cursor position won’t be set to the search match while searching, and nothing happens.

class prompt_toolkit.layout.processors.HighlightMatchingBracketProcessor(chars: str = '[](){}<>', max_cursor_distance: int = 1000)¶

When the cursor is on or right after a bracket, it highlights the matching bracket.

Parameters:: max_cursor_distance – Only highlight matching brackets when the cursor is within this distance. (From inside a Processor, we can’t know which lines will be visible on the screen. But we also don’t want to scan the whole document for matching brackets on each key press, so we limit to this value.)

class prompt_toolkit.layout.processors.HighlightSearchProcessor¶

Processor that highlights search matches in the document. Note that this doesn’t support multiline search matches yet.

The style classes ‘search’ and ‘search.current’ will be applied to the content.

class prompt_toolkit.layout.processors.HighlightSelectionProcessor¶: Processor that highlights the selection in the document.

class prompt_toolkit.layout.processors.PasswordProcessor(char: str = '*')¶

Processor that masks the input. (For passwords.)

Parameters:: char – (string) Character to be used. “*” by default.

class prompt_toolkit.layout.processors.Processor¶

Manipulate the fragments for a given line in a BufferControl.

abstractmethod apply_transformation(transformation_input: TransformationInput) → Transformation¶

Apply transformation. Returns a Transformation instance.

Parameters:: transformation_input – TransformationInput object.

class prompt_toolkit.layout.processors.ReverseSearchProcessor¶

Process to display the “(reverse-i-search)`…`:…” stuff around the search buffer.

Note: This processor is meant to be applied to the BufferControl that contains the search buffer, it’s not meant for the original input.

class prompt_toolkit.layout.processors.ShowArg¶

Display the ‘arg’ in front of the input.

This was used by the PromptSession, but now it uses the Window.get_line_prefix function instead.

class prompt_toolkit.layout.processors.ShowLeadingWhiteSpaceProcessor(get_char: Callable[[], str] | None = None, style: str = 'class:leading-whitespace')¶

Make leading whitespace visible.

Parameters:: get_char – Callable that returns one character.

class prompt_toolkit.layout.processors.ShowTrailingWhiteSpaceProcessor(get_char: Callable[[], str] | None = None, style: str = 'class:training-whitespace')¶

Make trailing whitespace visible.

Parameters:: get_char – Callable that returns one character.

class prompt_toolkit.layout.processors.TabsProcessor(tabstop: int | Callable[[], int] = 4, char1: str | Callable[[], str] = '|', char2: str | Callable[[], str] = '┈', style: str = 'class:tab')¶

Render tabs as spaces (instead of ^I) or make them visible (for instance, by replacing them with dots.)

Parameters:

tabstop – Horizontal space taken by a tab. (int or callable that returns an int).
char1 – Character or callable that returns a character (text of length one). This one is used for the first space taken by the tab.
char2 – Like char1, but for the rest of the space.

class prompt_toolkit.layout.processors.Transformation(fragments: StyleAndTextTuples, source_to_display: SourceToDisplay | None = None, display_to_source: DisplayToSource | None = None)¶

Transformation result, as returned by Processor.apply_transformation().

Important: Always make sure that the length of document.text is equal to: the length of all the text in fragments!

Parameters:

fragments – The transformed fragments. To be displayed, or to pass to the next processor.
source_to_display – Cursor position transformation from original string to transformed string.
display_to_source – Cursor position transformed from source string to original string.

class prompt_toolkit.layout.processors.TransformationInput(buffer_control: BufferControl, document: Document, lineno: int, source_to_display: SourceToDisplay, fragments: StyleAndTextTuples, width: int, height: int, get_line: Callable[[int], StyleAndTextTuples] | None = None)¶

Parameters:

buffer_control – BufferControl instance.
lineno – The number of the line to which we apply the processor.
source_to_display – A function that returns the position in the fragments for any position in the source string. (This takes previous processors into account.)
fragments – List of fragments that we can transform. (Received from the previous processor.)
get_line – Optional ; a callable that returns the fragments of another line in the current buffer; This can be used to create processors capable of affecting transforms across multiple lines.

prompt_toolkit.layout.processors.merge_processors(processors: list[Processor]) → Processor¶: Merge multiple Processor objects into one.

Utils¶

prompt_toolkit.layout.utils.explode_text_fragments(fragments: Iterable[_T]) → _ExplodedList[_T]¶

Turn a list of (style_str, text) tuples into another list where each string is exactly one character.

It should be fine to call this function several times. Calling this on a list that is already exploded, is a null operation.

Parameters:: fragments – List of (style, text) tuples.

Screen¶

class prompt_toolkit.layout.screen.Char(char: str = ' ', style: str = '')¶

Represent a single character in a Screen.

This should be considered immutable.

Parameters:

char – A single character (can be a double-width character).
style – A style string. (Can contain classnames.)

class prompt_toolkit.layout.screen.Screen(default_char: Char | None = None, initial_width: int = 0, initial_height: int = 0)¶

Two dimensional buffer of Char instances.

append_style_to_content(style_str: str) → None¶: For all the characters in the screen. Set the style string to the given style_str.

cursor_positions: dict[Window, Point]¶: Position of the cursor.

draw_all_floats() → None¶: Draw all float functions in order of z-index.

draw_with_z_index(z_index: int, draw_func: Callable[[], None]) → None¶: Add a draw-function for a Window which has a >= 0 z_index. This will be postponed until draw_all_floats is called.

fill_area(write_position: WritePosition, style: str = '', after: bool = False) → None¶: Fill the content of this area, using the given style. The style is prepended before whatever was here before.

get_cursor_position(window: Window) → Point¶: Get the cursor position for a given window. Returns a Point.

get_menu_position(window: Window) → Point¶: Get the menu position for a given window. (This falls back to the cursor position if no menu position was set.)

menu_positions: dict[Window, Point]¶: (Optional) Where to position the menu. E.g. at the start of a completion. (We can’t use the cursor position, because we don’t want the completion menu to change its position when we browse through all the completions.)

set_cursor_position(window: Window, position: Point) → None¶: Set the cursor position for a given window.

set_menu_position(window: Window, position: Point) → None¶: Set the cursor position for a given window.

show_cursor¶: Visibility of the cursor.

width¶: Currently used width/height of the screen. This will increase when data is written to the screen.

zero_width_escapes: defaultdict[int, defaultdict[int, str]]¶: Escape sequences to be injected.

Widgets¶

Collection of reusable components for building full screen applications. These are higher level abstractions on top of the prompt_toolkit.layout module.

Most of these widgets implement the __pt_container__ method, which makes it possible to embed these in the layout like any other container.

class prompt_toolkit.widgets.Box(body: AnyContainer, padding: AnyDimension = None, padding_left: AnyDimension = None, padding_right: AnyDimension = None, padding_top: AnyDimension = None, padding_bottom: AnyDimension = None, width: AnyDimension = None, height: AnyDimension = None, style: str = '', char: None | str | Callable[[], str] = None, modal: bool = False, key_bindings: KeyBindings | None = None)¶

Add padding around a container.

This also makes sure that the parent can provide more space than required by the child. This is very useful when wrapping a small element with a fixed size into a VSplit or HSplit object. The HSplit and VSplit try to make sure to adapt respectively the width and height, possibly shrinking other elements. Wrapping something in a Box makes it flexible.

Parameters:

body – Another container object.
padding – The margin to be used around the body. This can be overridden by padding_left, padding_right`, padding_top and padding_bottom.
style – A style string.
char – Character to be used for filling the space around the body. (This is supposed to be a character with a terminal width of 1.)

class prompt_toolkit.widgets.Button(text: str, handler: Callable[[], None] | None = None, width: int = 12, left_symbol: str = '<', right_symbol: str = '>')¶

Clickable button.

Parameters:

text – The caption for the button.
handler – None or callable. Called when the button is clicked. No parameters are passed to this callable. Use for instance Python’s functools.partial to pass parameters to this callable if needed.
width – Width of the button.

class prompt_toolkit.widgets.Checkbox(text: AnyFormattedText = '', checked: bool = False)¶

Backward compatibility util: creates a 1-sized CheckboxList

Parameters:: text – the text

class prompt_toolkit.widgets.Frame(body: AnyContainer, title: AnyFormattedText = '', style: str = '', width: AnyDimension = None, height: AnyDimension = None, key_bindings: KeyBindings | None = None, modal: bool = False)¶

Draw a border around any container, optionally with a title text.

Changing the title and body of the frame is possible at runtime by assigning to the body and title attributes of this class.

Parameters:

body – Another container object.
title – Text to be displayed in the top of the frame (can be formatted text).
style – Style string to be applied to this widget.

class prompt_toolkit.widgets.HorizontalLine¶: A simple horizontal line with a height of 1.

class prompt_toolkit.widgets.Label(text: AnyFormattedText, style: str = '', width: AnyDimension = None, dont_extend_height: bool = True, dont_extend_width: bool = False, align: WindowAlign | Callable[[], WindowAlign] = WindowAlign.LEFT, wrap_lines: FilterOrBool = True)¶

Widget that displays the given text. It is not editable or focusable.

Parameters:

text – Text to display. Can be multiline. All value types accepted by prompt_toolkit.layout.FormattedTextControl are allowed, including a callable.
style – A style string.
width – When given, use this width, rather than calculating it from the text size.
dont_extend_width – When True, don’t take up more width than preferred, i.e. the length of the longest line of the text, or value of width parameter, if given. True by default
dont_extend_height – When True, don’t take up more width than the preferred height, i.e. the number of lines of the text. False by default.

class prompt_toolkit.widgets.MenuContainer(body: AnyContainer, menu_items: list[MenuItem], floats: list[Float] | None = None, key_bindings: KeyBindingsBase | None = None)¶

Parameters:

floats – List of extra Float objects to display.
menu_items – List of MenuItem objects.

class prompt_toolkit.widgets.RadioList(values: Sequence[tuple[_T, AnyFormattedText]], default: _T | None = None)¶

List of radio buttons. Only one can be checked at the same time.

Parameters:: values – List of (value, label) tuples.

class prompt_toolkit.widgets.SearchToolbar(search_buffer: Buffer | None = None, vi_mode: bool = False, text_if_not_searching: AnyFormattedText = '', forward_search_prompt: AnyFormattedText = 'I-search: ', backward_search_prompt: AnyFormattedText = 'I-search backward: ', ignore_case: FilterOrBool = False)¶

Parameters:

vi_mode – Display ‘/’ and ‘?’ instead of I-search.
ignore_case – Search case insensitive.

class prompt_toolkit.widgets.Shadow(body: AnyContainer)¶

Draw a shadow underneath/behind this container. (This applies class:shadow the the cells under the shadow. The Style should define the colors for the shadow.)

Parameters:: body – Another container object.

class prompt_toolkit.widgets.SystemToolbar(prompt: AnyFormattedText = 'Shell command: ', enable_global_bindings: FilterOrBool = True)¶

Toolbar for a system prompt.

Parameters:: prompt – Prompt to be displayed to the user.

class prompt_toolkit.widgets.TextArea(text: str = '', multiline: FilterOrBool = True, password: FilterOrBool = False, lexer: Lexer | None = None, auto_suggest: AutoSuggest | None = None, completer: Completer | None = None, complete_while_typing: FilterOrBool = True, validator: Validator | None = None, accept_handler: BufferAcceptHandler | None = None, history: History | None = None, focusable: FilterOrBool = True, focus_on_click: FilterOrBool = False, wrap_lines: FilterOrBool = True, read_only: FilterOrBool = False, width: AnyDimension = None, height: AnyDimension = None, dont_extend_height: FilterOrBool = False, dont_extend_width: FilterOrBool = False, line_numbers: bool = False, get_line_prefix: GetLinePrefixCallable | None = None, scrollbar: bool = False, style: str = '', search_field: SearchToolbar | None = None, preview_search: FilterOrBool = True, prompt: AnyFormattedText = '', input_processors: list[Processor] | None = None, name: str = '')¶

A simple input field.

This is a higher level abstraction on top of several other classes with sane defaults.

This widget does have the most common options, but it does not intend to cover every single use case. For more configurations options, you can always build a text area manually, using a Buffer, BufferControl and Window.

Buffer attributes:

Parameters:

text – The initial text.
multiline – If True, allow multiline input.
completer – Completer instance for auto completion.
complete_while_typing – Boolean.
accept_handler – Called when Enter is pressed (This should be a callable that takes a buffer as input).
history – History instance.
auto_suggest – AutoSuggest instance for input suggestions.

BufferControl attributes:

Parameters:

password – When True, display using asterisks.
focusable – When True, allow this widget to receive the focus.
focus_on_click – When True, focus after mouse click.
input_processors – None or a list of Processor objects.
validator – None or a Validator object.

Window attributes:

Parameters:

lexer – Lexer instance for syntax highlighting.
wrap_lines – When True, don’t scroll horizontally, but wrap lines.
width – Window width. (Dimension object.)
height – Window height. (Dimension object.)
scrollbar – When True, display a scroll bar.
style – A style string.
dont_extend_width – When True, don’t take up more width then the preferred width reported by the control.
dont_extend_height – When True, don’t take up more width then the preferred height reported by the control.
get_line_prefix – None or a callable that returns formatted text to be inserted before a line. It takes a line number (int) and a wrap_count and returns formatted text. This can be used for implementation of line continuations, things like Vim “breakindent” and so on.

Other attributes:

Parameters:: search_field – An optional SearchToolbar object.

property accept_handler: Callable[[Buffer], bool] | None¶: The accept handler. Called when the user accepts the input.

property document: Document¶: The Buffer document (text + cursor position).

property text: str¶: The Buffer text.

class prompt_toolkit.widgets.VerticalLine¶: A simple vertical line with a width of 1.

Filters¶

Filters decide whether something is active or not (they decide about a boolean state). This is used to enable/disable features, like key bindings, parts of the layout and other stuff. For instance, we could have a HasSearch filter attached to some part of the layout, in order to show that part of the user interface only while the user is searching.

Filters are made to avoid having to attach callbacks to all event in order to propagate state. However, they are lazy, they don’t automatically propagate the state of what they are observing. Only when a filter is called (it’s actually a callable), it will calculate its value. So, its not really reactive programming, but it’s made to fit for this framework.

Filters can be chained using & and | operations, and inverted using the ~ operator, for instance:

filter = has_focus('default') & ~ has_selection

class prompt_toolkit.filters.Always¶: Always enable feature.

class prompt_toolkit.filters.Condition(func: Callable[[], bool])¶

Turn any callable into a Filter. The callable is supposed to not take any arguments.

This can be used as a decorator:

@Condition
def feature_is_active():  # `feature_is_active` becomes a Filter.
    return True

Parameters:: func – Callable which takes no inputs and returns a boolean.

class prompt_toolkit.filters.Filter¶

Base class for any filter to activate/deactivate a feature, depending on a condition.

The return value of __call__ will tell if the feature should be active.

prompt_toolkit.filters.HasFocus(value: FocusableElement) → Condition¶: Enable when this buffer has the focus.

prompt_toolkit.filters.InEditingMode(editing_mode: EditingMode) → Condition¶: Check whether a given editing mode is active. (Vi or Emacs.)

class prompt_toolkit.filters.Never¶: Never enable feature.

prompt_toolkit.filters.has_focus(value: FocusableElement) → Condition¶: Enable when this buffer has the focus.

prompt_toolkit.filters.in_editing_mode(editing_mode: EditingMode) → Condition¶: Check whether a given editing mode is active. (Vi or Emacs.)

prompt_toolkit.filters.is_true(value: Filter | bool) → bool¶

Test whether value is True. In case of a Filter, call it.

Parameters:: value – Boolean or Filter instance.

prompt_toolkit.filters.to_filter(bool_or_filter: Filter | bool) → Filter¶: Accept both booleans and Filters as input and turn it into a Filter.

class prompt_toolkit.filters.Filter

Base class for any filter to activate/deactivate a feature, depending on a condition.

The return value of __call__ will tell if the feature should be active.

class prompt_toolkit.filters.Condition(func: Callable[[], bool])

Turn any callable into a Filter. The callable is supposed to not take any arguments.

This can be used as a decorator:

@Condition
def feature_is_active():  # `feature_is_active` becomes a Filter.
    return True

Parameters:: func – Callable which takes no inputs and returns a boolean.

prompt_toolkit.filters.utils.is_true(value: Filter | bool) → bool¶

Test whether value is True. In case of a Filter, call it.

Parameters:: value – Boolean or Filter instance.

prompt_toolkit.filters.utils.to_filter(bool_or_filter: Filter | bool) → Filter¶: Accept both booleans and Filters as input and turn it into a Filter.

Filters that accept a Application as argument.

prompt_toolkit.filters.app.has_focus(value: FocusableElement) → Condition¶: Enable when this buffer has the focus.

prompt_toolkit.filters.app.in_editing_mode(editing_mode: EditingMode) → Condition¶: Check whether a given editing mode is active. (Vi or Emacs.)

Key binding¶

class prompt_toolkit.key_binding.ConditionalKeyBindings(key_bindings: KeyBindingsBase, filter: Filter | bool = True)¶

Wraps around a KeyBindings. Disable/enable all the key bindings according to the given (additional) filter.:

@Condition
def setting_is_true():
    return True  # or False

registry = ConditionalKeyBindings(key_bindings, setting_is_true)

When new key bindings are added to this object. They are also enable/disabled according to the given filter.

Parameters:

registries – List of KeyBindings objects.
filter – Filter object.

class prompt_toolkit.key_binding.DynamicKeyBindings(get_key_bindings: Callable[[], KeyBindingsBase | None])¶

KeyBindings class that can dynamically returns any KeyBindings.

Parameters:: get_key_bindings – Callable that returns a KeyBindings instance.

class prompt_toolkit.key_binding.KeyBindings¶

A container for a set of key bindings.

Example usage:

kb = KeyBindings()

@kb.add('c-t')
def _(event):
    print('Control-T pressed')

@kb.add('c-a', 'c-b')
def _(event):
    print('Control-A pressed, followed by Control-B')

@kb.add('c-x', filter=is_searching)
def _(event):
    print('Control-X pressed')  # Works only if we are searching.

add(*keys: Keys | str, filter: FilterOrBool = True, eager: FilterOrBool = False, is_global: FilterOrBool = False, save_before: Callable[[KeyPressEvent], bool] = <function KeyBindings.<lambda>>, record_in_macro: FilterOrBool = True) → Callable[[T], T]¶

Decorator for adding a key bindings.

Parameters:

filter – Filter to determine when this key binding is active.
eager – Filter or bool. When True, ignore potential longer matches when this key binding is hit. E.g. when there is an active eager key binding for Ctrl-X, execute the handler immediately and ignore the key binding for Ctrl-X Ctrl-E of which it is a prefix.
is_global – When this key bindings is added to a Container or Control, make it a global (always active) binding.
save_before – Callable that takes an Event and returns True if we should save the current buffer, before handling the event. (That’s the default.)
record_in_macro – Record these key bindings when a macro is being recorded. (True by default.)

add_binding(*keys: Keys | str, filter: FilterOrBool = True, eager: FilterOrBool = False, is_global: FilterOrBool = False, save_before: Callable[[KeyPressEvent], bool] = <function KeyBindings.<lambda>>, record_in_macro: FilterOrBool = True) → Callable[[T], T]¶

Decorator for adding a key bindings.

Parameters:

filter – Filter to determine when this key binding is active.
eager – Filter or bool. When True, ignore potential longer matches when this key binding is hit. E.g. when there is an active eager key binding for Ctrl-X, execute the handler immediately and ignore the key binding for Ctrl-X Ctrl-E of which it is a prefix.
is_global – When this key bindings is added to a Container or Control, make it a global (always active) binding.
save_before – Callable that takes an Event and returns True if we should save the current buffer, before handling the event. (That’s the default.)
record_in_macro – Record these key bindings when a macro is being recorded. (True by default.)

get_bindings_for_keys(keys: Tuple[Keys | str, ...]) → list[Binding]¶

Return a list of key bindings that can handle this key. (This return also inactive bindings, so the filter still has to be called, for checking it.)

Parameters:: keys – tuple of keys.

get_bindings_starting_with_keys(keys: Tuple[Keys | str, ...]) → list[Binding]¶

Return a list of key bindings that handle a key sequence starting with keys. (It does only return bindings for which the sequences are longer than keys. And like get_bindings_for_keys, it also includes inactive bindings.)

Parameters:: keys – tuple of keys.

remove(*args: Keys | str | KeyHandlerCallable) → None¶

Remove a key binding.

This expects either a function that was given to add method as parameter or a sequence of key bindings.

Raises ValueError when no bindings was found.

Usage:

remove(handler)  # Pass handler.
remove('c-x', 'c-a')  # Or pass the key bindings.

remove_binding(*args: Keys | str | KeyHandlerCallable) → None¶

Remove a key binding.

This expects either a function that was given to add method as parameter or a sequence of key bindings.

Raises ValueError when no bindings was found.

Usage:

remove(handler)  # Pass handler.
remove('c-x', 'c-a')  # Or pass the key bindings.

class prompt_toolkit.key_binding.KeyBindingsBase¶

Interface for a KeyBindings.

abstract property bindings: list[Binding]¶: List of Binding objects. (These need to be exposed, so that KeyBindings objects can be merged together.)

abstractmethod get_bindings_for_keys(keys: Tuple[Keys | str, ...]) → list[Binding]¶

Return a list of key bindings that can handle these keys. (This return also inactive bindings, so the filter still has to be called, for checking it.)

Parameters:: keys – tuple of keys.

abstractmethod get_bindings_starting_with_keys(keys: Tuple[Keys | str, ...]) → list[Binding]¶

Parameters:: keys – tuple of keys.

prompt_toolkit.key_binding.merge_key_bindings(bindings: Sequence[KeyBindingsBase]) → _MergedKeyBindings¶

Merge multiple Keybinding objects together.

Usage:

bindings = merge_key_bindings([bindings1, bindings2, ...])

Default key bindings.:

key_bindings = load_key_bindings()
app = Application(key_bindings=key_bindings)

prompt_toolkit.key_binding.defaults.load_key_bindings() → KeyBindingsBase¶: Create a KeyBindings object that contains the default key bindings.

class prompt_toolkit.key_binding.vi_state.ViState¶

Mutable class to hold the state of the Vi navigation.

property input_mode: InputMode¶: Get InputMode.

last_character_find: CharacterFind | None¶: None or CharacterFind instance. (This is used to repeat the last search in Vi mode, by pressing the ‘n’ or ‘N’ in navigation mode.)

named_registers: dict[str, ClipboardData]¶: Named registers. Maps register name (e.g. ‘a’) to ClipboardData instances.

reset() → None¶: Reset state, go back to the given mode. INSERT by default.

tilde_operator¶: When true, make ~ act as an operator.

waiting_for_digraph¶: Waiting for digraph.

An KeyProcessor receives callbacks for the keystrokes parsed from the input in the InputStream instance.

The KeyProcessor will according to the implemented keybindings call the correct callbacks when new key presses are feed through feed.

class prompt_toolkit.key_binding.key_processor.KeyPress(key: Keys | str, data: str | None = None)¶

Parameters:

key – A Keys instance or text (one character).
data – The received string on stdin. (Often vt100 escape codes.)

class prompt_toolkit.key_binding.key_processor.KeyPressEvent(key_processor_ref: ReferenceType[KeyProcessor], arg: str | None, key_sequence: list[KeyPress], previous_key_sequence: list[KeyPress], is_repeat: bool)¶

Key press event, delivered to key bindings.

Parameters:

key_processor_ref – Weak reference to the KeyProcessor.
arg – Repetition argument.
key_sequence – List of KeyPress instances.
previouskey_sequence – Previous list of KeyPress instances.
is_repeat – True when the previous event was delivered to the same handler.

property app: Application[Any]¶: The current Application object.

append_to_arg_count(data: str) → None¶

Add digit to the input argument.

Parameters:: data – the typed digit as string

property arg: int¶: Repetition argument.

property arg_present: bool¶: True if repetition argument was explicitly provided.

property cli: Application[Any]¶: For backward-compatibility.

property current_buffer: Buffer¶: The current buffer.

is_repeat¶: True when the previous key sequence was handled by the same handler.

class prompt_toolkit.key_binding.key_processor.KeyProcessor(key_bindings: KeyBindingsBase)¶

Statemachine that receives KeyPress instances and according to the key bindings in the given KeyBindings, calls the matching handlers.

p = KeyProcessor(key_bindings)

# Send keys into the processor.
p.feed(KeyPress(Keys.ControlX, ''))
p.feed(KeyPress(Keys.ControlC, '')

# Process all the keys in the queue.
p.process_keys()

# Now the ControlX-ControlC callback will be called if this sequence is
# registered in the key bindings.

Parameters:: key_bindings – KeyBindingsBase instance.

empty_queue() → list[KeyPress]¶: Empty the input queue. Return the unprocessed input.

feed(key_press: KeyPress, first: bool = False) → None¶

Add a new KeyPress to the input queue. (Don’t forget to call process_keys in order to process the queue.)

Parameters:: first – If true, insert before everything else.

feed_multiple(key_presses: list[KeyPress], first: bool = False) → None¶

Parameters:: first – If true, insert before everything else.

process_keys() → None¶

Process all the keys in the input_queue. (To be called after feed.)

Note: because of the feed/process_keys separation, it is: possible to call feed from inside a key binding. This function keeps looping until the queue is empty.

send_sigint() → None¶: Send SIGINT. Immediately call the SIGINT key handler.

Eventloop¶

prompt_toolkit.eventloop.call_soon_threadsafe(func: Callable[[], None], max_postpone_time: float | None = None, loop: AbstractEventLoop | None = None) → None¶

Wrapper around asyncio’s call_soon_threadsafe.

This takes a max_postpone_time which can be used to tune the urgency of the method.

Asyncio runs tasks in first-in-first-out. However, this is not what we want for the render function of the prompt_toolkit UI. Rendering is expensive, but since the UI is invalidated very often, in some situations we render the UI too often, so much that the rendering CPU usage slows down the rest of the processing of the application. (Pymux is an example where we have to balance the CPU time spend on rendering the UI, and parsing process output.) However, we want to set a deadline value, for when the rendering should happen. (The UI should stay responsive).

prompt_toolkit.eventloop.get_traceback_from_context(context: dict[str, Any]) → TracebackType | None¶: Get the traceback object from the context.

prompt_toolkit.eventloop.run_in_executor_with_context(func: Callable[[...], _T], *args: Any, loop: AbstractEventLoop | None = None) → Awaitable[_T]¶

Run a function in an executor, but make sure it uses the same contextvars. This is required so that the function will see the right application.

Similar to PyOS_InputHook of the Python API, we can plug in an input hook in the asyncio event loop.

The way this works is by using a custom ‘selector’ that runs the other event loop until the real selector is ready.

It’s the responsibility of this event hook to return when there is input ready. There are two ways to detect when input is ready:

The inputhook itself is a callable that receives an InputHookContext. This callable should run the other event loop, and return when the main loop has stuff to do. There are two ways to detect when to return:

Call the input_is_ready method periodically. Quit when this returns True.
Add the fileno as a watch to the external eventloop. Quit when file descriptor becomes readable. (But don’t read from it.)

Note that this is not the same as checking for sys.stdin.fileno(). The eventloop of prompt-toolkit allows thread-based executors, for example for asynchronous autocompletion. When the completion for instance is ready, we also want prompt-toolkit to gain control again in order to display that.

class prompt_toolkit.eventloop.inputhook.InputHookContext(fileno: int, input_is_ready: Callable[[], bool])¶: Given as a parameter to the inputhook.

class prompt_toolkit.eventloop.inputhook.InputHookSelector(selector: BaseSelector, inputhook: Callable[[InputHookContext], None])¶

Usage:

selector = selectors.SelectSelector() loop = asyncio.SelectorEventLoop(InputHookSelector(selector, inputhook)) asyncio.set_event_loop(loop)

close() → None¶: Clean up resources.

prompt_toolkit.eventloop.inputhook.new_eventloop_with_inputhook(inputhook: Callable[[InputHookContext], None]) → AbstractEventLoop¶: Create a new event loop with the given inputhook.

prompt_toolkit.eventloop.inputhook.set_eventloop_with_inputhook(inputhook: Callable[[InputHookContext], None]) → AbstractEventLoop¶: Create a new event loop with the given inputhook, and activate it.

prompt_toolkit.eventloop.utils.call_soon_threadsafe(func: Callable[[], None], max_postpone_time: float | None = None, loop: AbstractEventLoop | None = None) → None¶

Wrapper around asyncio’s call_soon_threadsafe.

This takes a max_postpone_time which can be used to tune the urgency of the method.

prompt_toolkit.eventloop.utils.get_traceback_from_context(context: dict[str, Any]) → TracebackType | None¶: Get the traceback object from the context.

prompt_toolkit.eventloop.utils.run_in_executor_with_context(func: Callable[[...], _T], *args: Any, loop: AbstractEventLoop | None = None) → Awaitable[_T]¶

Run a function in an executor, but make sure it uses the same contextvars. This is required so that the function will see the right application.

Input¶

class prompt_toolkit.input.DummyInput¶

Input for use in a DummyApplication

If used in an actual application, it will make the application render itself once and exit immediately, due to an EOFError.

class prompt_toolkit.input.Input¶

Abstraction for any input.

An instance of this class can be given to the constructor of a Application and will also be passed to the EventLoop.

abstractmethod attach(input_ready_callback: Callable[[], None]) → ContextManager[None]¶: Return a context manager that makes this input active in the current event loop.

close() → None¶: Close input.

abstract property closed: bool¶: Should be true when the input stream is closed.

abstractmethod cooked_mode() → ContextManager[None]¶: Context manager that turns the input into cooked mode.

abstractmethod detach() → ContextManager[None]¶: Return a context manager that makes sure that this input is not active in the current event loop.

abstractmethod fileno() → int¶: Fileno for putting this in an event loop.

flush() → None¶: The event loop can call this when the input has to be flushed.

flush_keys() → list[KeyPress]¶: Flush the underlying parser. and return the pending keys. (Used for vt100 input.)

abstractmethod raw_mode() → ContextManager[None]¶: Context manager that turns the input into raw mode.

abstractmethod read_keys() → list[KeyPress]¶: Return a list of Key objects which are read/parsed from the input.

abstractmethod typeahead_hash() → str¶: Identifier for storing type ahead key presses.

prompt_toolkit.input.create_input(stdin: TextIO | None = None, always_prefer_tty: bool = False) → Input¶

Create the appropriate Input object for the current os/environment.

Parameters:: always_prefer_tty – When set, if sys.stdin is connected to a Unix pipe, check whether sys.stdout or sys.stderr are connected to a pseudo terminal. If so, open the tty for reading instead of reading for sys.stdin. (We can open stdout or stderr for reading, this is how a $PAGER works.)

prompt_toolkit.input.create_pipe_input() → ContextManager[PipeInput]¶

Create an input pipe. This is mostly useful for unit testing.

Usage:

with create_pipe_input() as input:
    input.send_text('inputdata')

Breaking change: In prompt_toolkit 3.0.28 and earlier, this was returning the PipeInput directly, rather than through a context manager.

class prompt_toolkit.input.vt100.Vt100Input(stdin: TextIO)¶

Vt100 input for Posix systems. (This uses a posix file descriptor that can be registered in the event loop.)

attach(input_ready_callback: Callable[[], None]) → ContextManager[None]¶: Return a context manager that makes this input active in the current event loop.

detach() → ContextManager[None]¶: Return a context manager that makes sure that this input is not active in the current event loop.

flush_keys() → list[KeyPress]¶: Flush pending keys and return them. (Used for flushing the ‘escape’ key.)

read_keys() → list[KeyPress]¶: Read list of KeyPress.

class prompt_toolkit.input.vt100.cooked_mode(fileno: int)¶

The opposite of raw_mode, used when we need cooked mode inside a raw_mode block. Used in Application.run_in_terminal.:

with cooked_mode(stdin):
    ''' the pseudo-terminal stdin is now used in cooked mode. '''

class prompt_toolkit.input.vt100.raw_mode(fileno: int)¶

with raw_mode(stdin):
    ''' the pseudo-terminal stdin is now used in raw mode '''

We ignore errors when executing tcgetattr fails.

Parser for VT100 input stream.

class prompt_toolkit.input.vt100_parser.Vt100Parser(feed_key_callback: Callable[[KeyPress], None])¶

Parser for VT100 input stream. Data can be fed through the feed method and the given callback will be called with KeyPress objects.

def callback(key):
    pass
i = Vt100Parser(callback)
i.feed('data...')

Attr feed_key_callback:: Function that will be called when a key is parsed.

feed(data: str) → None¶

Feed the input stream.

Parameters:: data – Input string (unicode).

feed_and_flush(data: str) → None¶: Wrapper around feed and flush.

flush() → None¶

Flush the buffer of the input stream.

This will allow us to handle the escape key (or maybe meta) sooner. The input received by the escape key is actually the same as the first characters of e.g. Arrow-Up, so without knowing what follows the escape sequence, we don’t know whether escape has been pressed, or whether it’s something else. This flush function should be called after a timeout, and processes everything that’s still in the buffer as-is, so without assuming any characters will follow.

Mappings from VT100 (ANSI) escape sequences to the corresponding prompt_toolkit keys.

We are not using the terminfo/termcap databases to detect the ANSI escape sequences for the input. Instead, we recognize 99% of the most common sequences. This works well, because in practice, every modern terminal is mostly Xterm compatible.

Some useful docs: - Mintty: https://github.com/mintty/mintty/blob/master/wiki/Keycodes.md

Output¶

class prompt_toolkit.output.ColorDepth(value)¶

Possible color depth values for the output.

DEPTH_1_BIT = 'DEPTH_1_BIT'¶: One color only.

DEPTH_24_BIT = 'DEPTH_24_BIT'¶: 24 bit True color.

DEPTH_4_BIT = 'DEPTH_4_BIT'¶: ANSI Colors.

DEPTH_8_BIT = 'DEPTH_8_BIT'¶: The default.

classmethod default() → ColorDepth¶: Return the default color depth for the default output.

classmethod from_env() → ColorDepth | None¶

Return the color depth if the $PROMPT_TOOLKIT_COLOR_DEPTH environment variable has been set.

This is a way to enforce a certain color depth in all prompt_toolkit applications.

class prompt_toolkit.output.DummyOutput¶

For testing. An output class that doesn’t render anything.

fileno() → int¶: There is no sensible default for fileno().

class prompt_toolkit.output.Output¶

Base class defining the output interface for a Renderer.

Actual implementations are Vt100_Output and Win32Output.

ask_for_cpr() → None¶: Asks for a cursor position report (CPR). (VT100 only.)

bell() → None¶: Sound bell.

abstractmethod clear_title() → None¶: Clear title again. (or restore previous title.)

abstractmethod cursor_backward(amount: int) → None¶: Move cursor amount place backward.

abstractmethod cursor_down(amount: int) → None¶: Move cursor amount place down.

abstractmethod cursor_forward(amount: int) → None¶: Move cursor amount place forward.

abstractmethod cursor_goto(row: int = 0, column: int = 0) → None¶: Move cursor position.

abstractmethod cursor_up(amount: int) → None¶: Move cursor amount place up.

abstractmethod disable_autowrap() → None¶: Disable auto line wrapping.

disable_bracketed_paste() → None¶: For vt100 only.

abstractmethod disable_mouse_support() → None¶: Disable mouse.

abstractmethod enable_autowrap() → None¶: Enable auto line wrapping.

enable_bracketed_paste() → None¶: For vt100 only.

abstractmethod enable_mouse_support() → None¶: Enable mouse.

abstractmethod encoding() → str¶: Return the encoding for this output, e.g. ‘utf-8’. (This is used mainly to know which characters are supported by the output the data, so that the UI can provide alternatives, when required.)

abstractmethod enter_alternate_screen() → None¶: Go to the alternate screen buffer. (For full screen applications).

abstractmethod erase_down() → None¶: Erases the screen from the current line down to the bottom of the screen.

abstractmethod erase_end_of_line() → None¶: Erases from the current cursor position to the end of the current line.

abstractmethod erase_screen() → None¶: Erases the screen with the background color and moves the cursor to home.

abstractmethod fileno() → int¶: Return the file descriptor to which we can write for the output.

abstractmethod flush() → None¶: Write to output stream and flush.

abstractmethod get_default_color_depth() → ColorDepth¶

Get default color depth for this output.

This value will be used if no color depth was explicitly passed to the Application.

Note

If the $PROMPT_TOOLKIT_COLOR_DEPTH environment variable has been set, then outputs.defaults.create_output will pass this value to the implementation as the default_color_depth, which is returned here. (This is not used when the output corresponds to a prompt_toolkit SSH/Telnet session.)

get_rows_below_cursor_position() → int¶: For Windows only.

abstractmethod get_size() → Size¶: Return the size of the output window.

abstractmethod hide_cursor() → None¶: Hide cursor.

abstractmethod quit_alternate_screen() → None¶: Leave the alternate screen buffer.

abstractmethod reset_attributes() → None¶: Reset color and styling attributes.

reset_cursor_key_mode() → None¶

For vt100 only. Put the terminal in normal cursor mode (instead of application mode).

See: https://vt100.net/docs/vt100-ug/chapter3.html

abstractmethod reset_cursor_shape() → None¶: Reset cursor shape.

property responds_to_cpr: bool¶

True if the Application can expect to receive a CPR response after calling ask_for_cpr (this will come back through the corresponding Input).

This is used to determine the amount of available rows we have below the cursor position. In the first place, we have this so that the drop down autocompletion menus are sized according to the available space.

On Windows, we don’t need this, there we have get_rows_below_cursor_position.

scroll_buffer_to_prompt() → None¶: For Win32 only.

abstractmethod set_attributes(attrs: Attrs, color_depth: ColorDepth) → None¶: Set new color and styling attributes.

abstractmethod set_cursor_shape(cursor_shape: CursorShape) → None¶: Set cursor shape to block, beam or underline.

abstractmethod set_title(title: str) → None¶: Set terminal title.

abstractmethod show_cursor() → None¶: Show cursor.

abstractmethod write(data: str) → None¶: Write text (Terminal escape sequences will be removed/escaped.)

abstractmethod write_raw(data: str) → None¶: Write text.

prompt_toolkit.output.create_output(stdout: TextIO | StdoutProxy | None = None, always_prefer_tty: bool = False) → Output¶

Return an Output instance for the command line.

Parameters:

stdout – The stdout object
always_prefer_tty –
When set, look for sys.stderr if sys.stdout is not a TTY. Useful if sys.stdout is redirected to a file, but we still want user input and output on the terminal.

By default, this is False. If sys.stdout is not a terminal (maybe it’s redirected to a file), then a PlainTextOutput will be returned. That way, tools like print_formatted_text will write plain text into that file.

Output for vt100 terminals.

A lot of thanks, regarding outputting of colors, goes to the Pygments project: (We don’t rely on Pygments anymore, because many things are very custom, and everything has been highly optimized.) http://pygments.org/

class prompt_toolkit.output.vt100.Vt100_Output(stdout: TextIO, get_size: Callable[[], Size], term: str | None = None, default_color_depth: ColorDepth | None = None, enable_bell: bool = True, enable_cpr: bool = True)¶

Parameters:

get_size – A callable which returns the Size of the output terminal.
stdout – Any object with has a write and flush method + an ‘encoding’ property.
term – The terminal environment variable. (xterm, xterm-256color, linux, …)
enable_cpr – When True (the default), send “cursor position request” escape sequences to the output in order to detect the cursor position. That way, we can properly determine how much space there is available for the UI (especially for drop down menus) to render. The Renderer will still try to figure out whether the current terminal does respond to CPR escapes. When False, never attempt to send CPR requests.

ask_for_cpr() → None¶: Asks for a cursor position report (CPR).

bell() → None¶: Sound bell.

cursor_goto(row: int = 0, column: int = 0) → None¶: Move cursor position.

encoding() → str¶: Return encoding used for stdout.

erase_down() → None¶: Erases the screen from the current line down to the bottom of the screen.

erase_end_of_line() → None¶: Erases from the current cursor position to the end of the current line.

erase_screen() → None¶: Erases the screen with the background color and moves the cursor to home.

fileno() → int¶: Return file descriptor.

flush() → None¶: Write to output stream and flush.

classmethod from_pty(stdout: TextIO, term: str | None = None, default_color_depth: ColorDepth | None = None, enable_bell: bool = True) → Vt100_Output¶: Create an Output class from a pseudo terminal. (This will take the dimensions by reading the pseudo terminal attributes.)

get_default_color_depth() → ColorDepth¶

Return the default color depth for a vt100 terminal, according to the our term value.

We prefer 256 colors almost always, because this is what most terminals support these days, and is a good default.

reset_cursor_key_mode() → None¶: For vt100 only. Put the terminal in cursor mode (instead of application mode).

reset_cursor_shape() → None¶: Reset cursor shape.

set_attributes(attrs: Attrs, color_depth: ColorDepth) → None¶

Create new style and output.

Parameters:: attrs – Attrs instance.

set_title(title: str) → None¶: Set terminal title.

write(data: str) → None¶: Write text to output. (Removes vt100 escape codes. – used for safely writing text.)

write_raw(data: str) → None¶: Write raw data to output.

Data structures¶

class prompt_toolkit.layout.WindowRenderInfo(window: Window, ui_content: UIContent, horizontal_scroll: int, vertical_scroll: int, window_width: int, window_height: int, configured_scroll_offsets: ScrollOffsets, visible_line_to_row_col: dict[int, tuple[int, int]], rowcol_to_yx: dict[tuple[int, int], tuple[int, int]], x_offset: int, y_offset: int, wrap_lines: bool)¶

Render information for the last render time of this control. It stores mapping information between the input buffers (in case of a BufferControl) and the actual render position on the output screen.

(Could be used for implementation of the Vi ‘H’ and ‘L’ key bindings as well as implementing mouse support.)

Parameters:

ui_content – The original UIContent instance that contains the whole input, without clipping. (ui_content)
horizontal_scroll – The horizontal scroll of the Window instance.
vertical_scroll – The vertical scroll of the Window instance.
window_width – The width of the window that displays the content, without the margins.
window_height – The height of the window that displays the content.
configured_scroll_offsets – The scroll offsets as configured for the Window instance.
visible_line_to_row_col – Mapping that maps the row numbers on the displayed screen (starting from zero for the first visible line) to (row, col) tuples pointing to the row and column of the UIContent.
rowcol_to_yx – Mapping that maps (row, column) tuples representing coordinates of the UIContent to (y, x) absolute coordinates at the rendered screen.

property applied_scroll_offsets: ScrollOffsets¶: Return a ScrollOffsets instance that indicates the actual offset. This can be less than or equal to what’s configured. E.g, when the cursor is completely at the top, the top offset will be zero rather than what’s configured.

property bottom_visible: bool¶: True when the bottom of the buffer is visible.

center_visible_line(before_scroll_offset: bool = False, after_scroll_offset: bool = False) → int¶: Like first_visible_line, but for the center visible line.

property content_height: int¶: The full height of the user control.

property cursor_position: Point¶: Return the cursor position coordinates, relative to the left/top corner of the rendered screen.

property displayed_lines: list[int]¶: List of all the visible rows. (Line numbers of the input buffer.) The last line may not be entirely visible.

first_visible_line(after_scroll_offset: bool = False) → int¶: Return the line number (0 based) of the input document that corresponds with the first visible line.

property full_height_visible: bool¶: True when the full height is visible (There is no vertical scroll.)

get_height_for_line(lineno: int) → int¶: Return the height of the given line. (The height that it would take, if this line became visible.)

property input_line_to_visible_line: dict[int, int]¶: Return the dictionary mapping the line numbers of the input buffer to the lines of the screen. When a line spans several rows at the screen, the first row appears in the dictionary.

last_visible_line(before_scroll_offset: bool = False) → int¶: Like first_visible_line, but for the last visible line.

property top_visible: bool¶: True when the top of the buffer is visible.

property vertical_scroll_percentage: int¶: Vertical scroll as a percentage. (0 means: the top is visible, 100 means: the bottom is visible.)

class prompt_toolkit.data_structures.Point(x, y)¶

x: int¶: Alias for field number 0

y: int¶: Alias for field number 1

class prompt_toolkit.data_structures.Size(rows, columns)¶

columns: int¶: Alias for field number 1

rows: int¶: Alias for field number 0

Patch stdout¶

patch_stdout¶

This implements a context manager that ensures that print statements within it won’t destroy the user interface. The context manager will replace sys.stdout by something that draws the output above the current prompt, rather than overwriting the UI.

Usage:

with patch_stdout(application):
    ...
    application.run()
    ...

Multiple applications can run in the body of the context manager, one after the other.

class prompt_toolkit.patch_stdout.StdoutProxy(sleep_between_writes: float = 0.2, raw: bool = False)¶

File-like object, which prints everything written to it, output above the current application/prompt. This class is compatible with other file objects and can be used as a drop-in replacement for sys.stdout or can for instance be passed to logging.StreamHandler.

The current application, above which we print, is determined by looking what application currently runs in the AppSession that is active during the creation of this instance.

This class can be used as a context manager.

In order to avoid having to repaint the prompt continuously for every little write, a short delay of sleep_between_writes seconds will be added between writes in order to bundle many smaller writes in a short timespan.

close() → None¶

Stop StdoutProxy proxy.

This will terminate the write thread, make sure everything is flushed and wait for the write thread to finish.

flush() → None¶: Flush buffered output.

prompt_toolkit.patch_stdout.patch_stdout(raw: bool = False) → Generator[None, None, None]¶

Replace sys.stdout by an _StdoutProxy instance.

Writing to this proxy will make sure that the text appears above the prompt, and that it doesn’t destroy the output from the renderer. If no application is curring, the behavior should be identical to writing to sys.stdout directly.

Warning: If a new event loop is installed using asyncio.set_event_loop(),: then make sure that the context manager is applied after the event loop is changed. Printing to stdout will be scheduled in the event loop that’s active when the context manager is created.

Parameters:: raw – (bool) When True, vt100 terminal escape sequences are not removed/escaped.