A cross platform package to do curses-like operations, plus higher level APIs and widgets to create text UIs and ASCII art animations

Describe the solution you'd like I would really like to have some barcharts, but vertically.

Describe alternatives you've considered The only possibility to implement this with acsiimatics I can imagine is drawing it completely manually

Issues fixed by this PR

We had ~40% worst case CPU load without the fix in MPF. With this fix worst case CPU reduces to ~30%.

What does this implement/fix?

Cache the result of split_text for labels. This hardly changes and we invalidate the cache if it would.

Any other comments?

I guess it would make sense in some other widgets too. Text for example.

PR is on top of #213 to prevent merge conflicts.

I'm using Mac OS X and when I press the "delete" key (in Macs this is deleting to the left, so actually backspace) in a Text widget, its size changes and starts to behave strangely. The actual delete key (this is fn+delete in Macs) works fine. The key code of "delete" is 127 and I haven't seen this being handled in screen.py.

Describe the bug

I'm packaging this for NixOS: https://github.com/NixOS/nixpkgs/pull/54404

However there are test failures as can be seen here: https://github.com/NixOS/nixpkgs/pull/54404#issuecomment-456238198

To Reproduce

Using the PR, run nix-env -f ./default.nix -iA 'python36.pkgs.asciimatics'.

System details (please complete the following information):

• system: "x86_64-linux"
• host os: Linux 4.17.6, Form Bay ML, noversion
• multi-user?: yes
• sandbox: yes
• version: nix-env (Nix) 2.0.4
• nixpkgs: /nix/nixpkgs

Hello, great project I was thinking, how about something like a listview ...i know that the treeview is pretty close (and great) but it would be great if we could have something like dos commander style

Code works perfectly on Python 2.7, but in Python 3.6 crashes the first time Screen.refresh() is called.

Traceback (most recent call last):
  File "/Users/anthonyshaw/repo/retox-2/retox/__main__.py", line 66, in main
    screen.refresh()
  File "/Users/anthonyshaw/repo/retox-2/src/asciimatics/asciimatics/screen.py", line 2129, in refresh
    super(_CursesScreen, self).refresh()
  File "/Users/anthonyshaw/repo/retox-2/src/asciimatics/asciimatics/screen.py", line 1216, in refresh
    self._change_colours(new_cell[1], new_cell[2], new_cell[3])
  File "/Users/anthonyshaw/repo/retox-2/src/asciimatics/asciimatics/screen.py", line 2225, in _change_colours
    self._safe_write(self._a_normal)
  File "/Users/anthonyshaw/repo/retox-2/src/asciimatics/asciimatics/screen.py", line 2085, in _safe_write
    sys.stdout.write(msg)
  File "/Users/anthonyshaw/repo/retox-2/bin/../lib/python3.6/codecs.py", line 377, in write
    self.stream.write(data)
  File "/Users/anthonyshaw/repo/retox-2/bin/../lib/python3.6/codecs.py", line 376, in write
    data, consumed = self.encode(object, self.errors)
TypeError: utf_8_encode() argument 1 must be str, not bytes

Any help/tips on debugging would be appreciated.

It would be preferable for left and right to switch focus to the nearest widget rather than the bottom of the nearest widget's column. I have an app that is a grid of equal size buttons, which this behavior makes it unusable.

Even if this is something you are not interested in doing. Can you point me in the right direction on getting this done per app? I subclassed Layout->process_event, I see the code for KEY_LEFT and KEY_RIGHT, but I am not understanding how it works. find_next_widget looks like where the magic happens. First value is the column difference. I would think start_at would be how far down the column to check for the next widget. That doesn't seem to be it tho. Even if it is, I'd need to get the row? somewhere.

Chinese characters cannot display correctly, no exceptions or errors, just display nothing for those characters. code like this

    def _quit(self):
        self._scene.add_effect(
            PopUpDialog(self._screen,
                        u"你確定嗎？",
                        [u"是", u"否"],
                        on_close=self._quit_on_yes))

For my app, I really needed to be able to support getting verbatim output from other terminal apps and showing it to the user. These other commands might print color text, and I wanted the color to be preserved and shown to the user as-is. I've basically been approaching my asciimatics app as if it were a web app in the terminal; if I were doing a web app, I would be able to use any HTML along with the framework widgets, and I had not considered that a similar way of working with asciimatics and raw ansii codes would be difficult.

I dug around in the asciimatics code for a while and decided that implementing this as a widget would be... pretty difficult, at least for me. My workaround below might be useful for the library documentation.

First we create a new exception to hold the external command to be run and scene to resume once it's finished:

class ExternalCallException(Exception):
    def __init__(self, last_scene, command):
        self.last_scene = last_scene
        self.command = command

When the external command should be run, we raise the exception like so (self is the widget object):

raise ExternalCallException(self._scene, 'my command here')

Then we add an extra except clause under the one handling screen resizes:

    except ResizeScreenError as e:
        last_scene = e.scene
    # new code below
    except ExternalCallException as e:
        last_scene = e.last_scene
        try:
            # -K means quit on ^C; -R means print ansii codes as-is
            process = Popen(e.command + " | less -K -R", shell=True)
            process.wait()
        except KeyboardInterrupt:
            # let less handle this, -K will exit cleanly
            pass

With this, the external command is piped to less, which temporarily controls the whole screen and shows the command's output in color. When the user quits less, the app reappears using the same mechanism as when the screen is resized.

Describe the bug A large number of tests fail on Python 2.7 & 3.7 . I have confirmed that Python 3.7 works on Travis.

To Reproduce The OBS project at https://build.opensuse.org/package/show/home:jayvdb:py-wheels/python-asciimatics has a .spec which makes this very reproducible.

Build log for openSUSE tumbleweed amd64, which includes the Python 2.7 errors:

_log.txt

I've run the same locally on my machine with pytest and it also fails.

Expected behavior Tess pass.

System details (please complete the following information):

• openSUSE (all versions)
• Asciimatics 1.10.0

Issues fixed by this PR

#186

What does this implement/fix?

Added a new RegEx option to the FileBrowser to allow filtering of files

Any other comments?

If you need me to change anything, please let me know.

When I run this script in my Powershell: https://stackoverflow.com/questions/20541456/list-of-all-colors-available-for-powershell I can see there is a DarkGray color in my system.

However, asciimatics doesn't have that in the Screen class (or anywhere else according to my search). Is there any way to access DarkGray somehow?

(Incidentally, do you know a shell that supports more colors than Powershell on Windows? Or is the limitation underlying?)

Describe the bug Rain test fails after 2038-01-19

To Reproduce

on openSUSE, I build the package with

osc co openSUSE:Factory/python-asciimatics
osc build --vm-type=kvm --noservice --clean --build-opt=--vm-custom-opt="-rtc base=2038-01-20T00:00:00" --alternative-project=home:bmwiedemann:reproducible openSUSE_Tumbleweed

Expected behavior

tests should continue to pass in future

Screenshots

 =================================== FAILURES ===================================
 ___________________________ TestParticles.test_rain ____________________________
 
 self = <tests.test_particles.TestParticles testMethod=test_rain>
 
     def test_rain(self):
         """     
         Test that Rain works as expected.
         """     
         screen = MagicMock(spec=Screen, colours=8)
         canvas = Canvas(screen, 10, 40, 0, 0)
         effect = Rain(canvas, 200)
 >       self.check_effect(canvas,
                           effect,
                           lambda value: self.assertIn(chr(value[0]), ' `\\v'))

 tests/test_particles.py:98:
 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
 tests/test_particles.py:43: in check_effect
     self.assertTrue(changed, "failed at step %d %s" % (i, view))
 E   AssertionError: False is not true : failed at step 0

==== 1 failed, 149 passed, 41 skipped, 1 deselected, 966 warnings in 3.67s =====

System details (please complete the following information):

• OS and version: openSUSE-Tumbleweed-20220906
• Python version: 3.10
• Python distribution: cpython
• Asciimatics version 1.14.0

Additional context See also https://en.wikipedia.org/wiki/Year_2038_problem

As part of my work on reproducible builds for openSUSE, I check that software still gives identical build results in the future. The usual offset is +16 years, because that is how long I expect some software will be used in some places. This showed up failing tests in our package build. See https://reproducible-builds.org/ for why this matters.

As per our conversation 2022-02-06 at https://gitter.im/asciimatics/Lobby, I am requesting scrollwheel support in Asciimatics.

Main technical thrust: Modify Screen to recognize scrollwheel events and handle it in the Frame

I will begin work on this soon and submit a PR when I get it working.

=======================

Jeff Wright @jeffwright13 Feb 06 06:35 @peterbrittain - is there a way to modify the scrolling behavior? When I use the scroll wheel on my mouse, it seems to want to scroll an entire page at a time, instead of what I have come to expect in a normal GUI: N lines at a time. The result is a very jumpy experience. I didn't see anything in the docs about configuring scroll behavior. peterbrittain @peterbrittain Feb 06 06:46 That's because asciimatics doesn't support scroll wheels (due to patchy support in Linux). I suspect what you're seeing is the terminal intercepting the event and scrolling the whole buffer. You'll be able to tell if this is the case, because asciimatics will still only draw on the original screen that was visible when you started the app. Jeff Wright @jeffwright13 Feb 06 06:55 So nothing to be done then? peterbrittain @peterbrittain Feb 06 07:08 Mouse support has improved in ncurses 6. Do you know what version your system uses? We'd also need to see if there is a setting on your terminal to force it to pass the input to the running app (instead of consuming it itself). Jeff Wright @jeffwright13 Feb 06 07:14

I am running (currently) on a Macbook Pro (2019/Intel). Latest version of OSX. I usually run Pytest (and hence Asciimatics) out of iTerm2, although sometimes also from my IDE (VSCode, with Integrated Terminal). My shell is latest zsh.

$ tic -V ncurses 5.7.20081102

BUT:

$ brew upgrade ncurses Running brew update --preinstall... ==> Auto-updated Homebrew! Updated 2 taps (homebrew/core and homebrew/cask). ==> Updated Formulae Updated 84 formulae. ==> Updated Casks Updated 36 casks.

Warning: ncurses 6.3 already installed

peterbrittain @peterbrittain Feb 06 08:02 Ok... with a newer version of ncurses, you then need to enable scroll wheel reporting. Looks like there's a setting in iterm2: https://iterm2.com/documentation-preferences-profiles-terminal.html Finally, we'd need to tweak asciimatics Screen to recognize the events and handle it in the Frame. All pretty self-contained changes, so I can explain them to you if you're up for submitting a PR. Jeff Wright @jeffwright13 Feb 06 09:16 Screen Shot 2022-02-06 at 9.16.13 AM.png Screen Shot 2022-02-06 at 9.16.38 AM.png Looks like I already had that setting enabled ("Enable mouse reporting") Sure, I could submit a PR. Would need some guidance on where in the code to make changes. peterbrittain @peterbrittain Feb 06 14:57 Ok... There are 2 basic changes. First is to _CursesScreen.get_event(). There's already logic to handle other buttons in there. We just need to find what events fire when you scroll the mouse wheel. You could either add some logging to the function, or use something like this: https://stackoverflow.com/questions/11893034/mouse-wheel-in-python-curses. According to the curses maintainer, it should look like mouse buttons 4 and 5. See https://stackoverflow.com/questions/42356352/mouse-scroll-up-ncurses-c Once you have those events translated into equivalent asciimatics MouseEvents, you just need to act on them in the scrollbar. That's in scrollbar.py. simply get the current position and add/subtract a percentage to the position on the relevant button. Jeff Wright @jeffwright13 Feb 06 20:54

I made the following changes to screen.py.

                # Some Linux modes only report clicks, so check for any
                # button down or click events.
                if (bstate & curses.BUTTON1_PRESSED != 0 or
                        bstate & curses.BUTTON1_CLICKED != 0):
                    buttons |= MouseEvent.LEFT_CLICK
                if (bstate & curses.BUTTON3_PRESSED != 0 or
                        bstate & curses.BUTTON3_CLICKED != 0):
                    buttons |= MouseEvent.RIGHT_CLICK
                if bstate & curses.BUTTON1_DOUBLE_CLICKED != 0:
                    buttons |= MouseEvent.DOUBLE_CLICK
                if bstate & curses.BUTTON4_PRESSED != 0:
                    print("BUTTON4_PRESSED")
                if bstate & curses.BUTTON5_PRESSED != 0:
                    print("BUTTON5_PRESSED")
                if bstate & curses.BUTTON4_CLICKED != 0:
                    print("BUTTON4_CLICKED")
                if bstate & curses.BUTTON5_CLICKED != 0:
                    print("BUTTON5_CLICKED")
                return MouseEvent(x, y, buttons)

This resulted in an exception when the scroll wheel was rolled:

Traceback (most recent call last): File "/Users/jwr003/coding/pytest-fold/pytest_fold/tui.py", line 186, in main() File "/Users/jwr003/coding/pytest-fold/pytest_fold/tui.py", line 179, in main Screen.wrapper(demo, catch_interrupt=True, arguments=[last_scene]) File "/Users/jwr003/coding/pytest-fold/venv/lib/python3.9/site-packages/asciimatics/screen.py", line 1393, in wrapper return func(screen, *arguments) File "/Users/jwr003/coding/pytest-fold/pytest_fold/tui.py", line 166, in demo screen.play( File "/Users/jwr003/coding/pytest-fold/venv/lib/python3.9/site-packages/asciimatics/screen.py", line 1566, in play self.draw_next_frame(repeat=repeat) File "/Users/jwr003/coding/pytest-fold/venv/lib/python3.9/site-packages/asciimatics/screen.py", line 1649, in draw_next_frame event = self.get_event() File "/Users/jwr003/coding/pytest-fold/venv/lib/python3.9/site-packages/asciimatics/screen.py", line 2494, in get_event if bstate & curses.BUTTON5_PRESSED != 0: AttributeError: module 'curses' has no attribute 'BUTTON5_PRESSED'

I am on a Mac. I saw the following comment in _curses.py:

# Darwin ncurses doesn't provide BUTTON5_* constants

Maybe that's the issue with the crash above. peterbrittain @peterbrittain Feb 07 00:58 Yeah - I wondered if we'd start hitting issue like this. We can create the constants ourselves if needed. For now, I suggest you just log the bstate in hex and compare it to https://github.com/mirror/ncurses/blob/d30f99439fcc8d4bb4c38e5c4afb4f6555fc6ad4/include/curses.tail Note that print doesn't work in alternate screens (like asciimatics uses). Also, if you don't like doing the bit maths, you can probably use this definition instead: https://github.com/mattn/pdcurses/blob/master/curses.h Jeff Wright @jeffwright13 Feb 07 11:14 The print was meant for initial diagnosis to see what button the system registered when I scrolled. Definitely will remove before issuing a PR! peterbrittain @peterbrittain Feb 07 11:57 :-)

Is your feature request related to a problem? Please describe. Full disclosure, I'm not a python power user, so I may well be missing something obvious with python itself.

I'm building a video game, and have the engine mostly done in curses. I wanted to add fancy animations, so I started trying to port things over to asciimatics. However, I'm finding it hard to follow the docs in some places. For a specific example, I'm finding it hard to sort out how to use wait_for_input() in conjunction with get_event() as outlined in the docs, but there are no examples to pull from. It's not clear to me how I should even begin to get a user's input and then process it using these two methods.

I can see the timeout of wait_for_input() is working in my script, but when I press any character the app just exits immediately. From the explanation in the docs, it made it seem as though I could just keep typing and the app would block execution until something is entered (presumably just hitting enter, but the docs aren't clear here either). I don't think this is a bug, but rather my misunderstanding of how it is supposed to function.

The one example I've tried to follow that sort of has what I'm looking for is forms.py, but I don't see that there is any get_event() or wait_for_input() calls. None of the examples I've dug through seem to have these calls in them.

Describe the solution you'd like Some example of using wait_for_input() and get_event(), and expand on the docs here: https://asciimatics.readthedocs.io/en/stable/io.html#input

Would it be possible, at some point, to get an example that allows the user to input a word or short phrase in one frame, then display that entered text as FigletText in its own, separate frame?

Additional context In my particular application, I've got several frames on-screen where one of those frames is a single-line input. Like a text adventure game, I want to have the user input a simple command then process it to change the db and then update the display with new info.

I've been able to sort out a lot of other details from the docs, but the user input handling is one where I am a little lost. In curses, I was able to do what I needed to do with the textpad.Textbox object with relative ease. I thought that I might be able to use asciimatics's Text widget, but so far have been unsuccessful.

EDIT:

I'm currently running python 3.10 with asciimatics v1.13.0, and I am being careful to dig through examples in the 1.13.0 tag.

Is your feature request related to a problem? Please describe. I'm trying to display in a label on a bar chart for the current rendered value.

Describe the solution you'd like A method to display the currently displayed value

Describe alternatives you've considered I've tried to use the label but it is not updating on each re render.