First take on cursor position #32

thomasballinger · 2014-12-03T02:27:08Z

This is an in-progress PR so there's something to talk about while I make progress. Things I can think of to worry about:

How should this be tested?
Anything different to do for signals received during this blocking call?
Reporting cursors current location erikrose/blessings#65 says Sun needs to be treated differently, and that this won't work in Windows. What's the right way to detect these situations?

I'll keep working on these but would appreciate thoughts.

jquast · 2014-12-03T04:53:24Z

blessed/terminal.py

@@ -280,6 +281,27 @@ def width(self):
        """
        return self._height_and_width().ws_col

+    @property
+    def cursor_position(self):


As much of a fan of @Property as I am, I want to use a function, get_cursor_position(), because we should indicate this is an expensive/blocking call, where-as a @property-backed attribute somewhat implies otherwise.

Please provide argument timeout=5 and honor it.

Please return the (invalid) value of (0,0) if:

self.keyboard_fd is None

HAS_TTY is False

or a response is not received within timeout

and document as such.

This is so that programs that use it (like the proposed bin/resize.py) may work even if stdin is not a tty.

jquast · 2014-12-03T05:47:35Z

how should this be tested?

For testing, lets port resize.c from x11.org's xterm in python in as bin/resize.py and bin/sunsize.py!

http://cvsweb.openbsd.org/cgi-bin/cvsweb/xenocara/app/xterm/resize.c?rev=1.10&content-type=text/x-cvsweb-markup

I know it looks scary but ignore all the cruft, its simple:

echo(term.move(998,998))
width, height = term.get_cursor_position()
echo('COLUMNS={0};\nLINES={1};\nexport COLUMNS LINES;\n'.format(width, height))

This can then be manually tested, that it reports the same COLUMNS and LINES that your terminal emulator already has.

Secondly, this will have to be "mocked", but not in the mock module sense, just in that the build agents are not terminal emulators and we will have to mock one. You see many tests around keyboard input that spawns a pty and reads its output and sends input, you'll do the same, look for the "report cursor position" sequence, send one manually into the child process, which then asserts the child's (height, width) returned is what was sent.

Anything different to do for signals received during this blocking call?

I don't think so, hardcoding _intr_continue=True is fine until somebody can report a use-case where they need otherwise. Can you think of an edge case?

erikrose#65 says Sun needs to be treated differently

Don't worry about it, trust the values of term.u7 (request) and term.u6 (response)

and that this won't work in Windows.

Nope, we'll cross that moat when we get to it -- don't worry about it.

jquast · 2014-12-03T06:04:44Z

blessed/terminal.py

+            first = 0
+            keystrokes = []
+            while True:
+                keystrokes.append(self.inkey(_intr_continue=True))


please conditionally check return value of keystrokes it not None, as you will be adding a timeout parameter. Make sure to mark the start-time and compute the remaining time for each call to inkey(), You will have to use some kind of '_timeleft' function to compute remaining time like inkey() does.

jquast · 2014-12-04T05:07:27Z

Doing some research

forget kind argument and sun support.
use term.u6 and term.u7:

#   u7  cursor position request (equiv. to VT100/ANSI/ECMA-48 DSR 6)
#   u6  cursor position report (equiv. to ANSI/ECMA-48 CPR)
#
# The cursor position request (<u7>) string should elicit a cursor position
# report.  A typical value (for VT100 terminals) is \E[6n.
#
# The cursor position report (<u6>) string must contain two scanf(3)-style 
# %d format elements.  The first of these must correspond to the Y coordinate
# and the second to the %d.  If the string contains the sequence %i, it is
# taken as an instruction to decrement each value after reading it (this is
# the inverse sense from the cup string).  The typical CPR value is
# \E[%i%d;%dR (on VT100/ANSI/ECMA-48-compatible terminals).

thomasballinger · 2014-12-07T16:01:47Z

This is terrific, thanks! Unfortunately I probably won't be able to work on this for about a week, but will get back to it then.

jquast · 2015-02-28T21:49:11Z

Well there is plenty of time, now. I'd really like to dive into and help with your curtsies project (i've been watching and seeing your issues/pr's... i see what you're doing, but no comment) I'm currently focused on getting blessed merged back into blessings.

The blessed project will become very little code: only "imports blessings, monkey patch the previous behavior/methods for compatibility, recommend blessings to be used instead".

Erik transferred blessed code into his project as a branch we're working through some weeks ago, so in effect, "blessed" is in feature freeze. I proposed we get this specific feature in before the transfer, but erik declined it. This feature will have to be a part of some later release of blessings.

I will transfer this and other tickets after the merge.

As you're looking at using blessed for keyboard, you should know that I fought a tough battle over some of the api's and won some and lost some. Erik declined the inkey, cbreak and raw methods, preferring instead to rename them to keystroke, and keystroke_input(raw=False), there may be even more in the keyboard.py module, for a time he renamed the Keystroke class to Key but I think I got that reversed. There is some danger that it won't be directly compatible with the underlying code.

(It's too bad, really I asked for his feedback almost 2 years ago on these methods and their names before blessed was born, and he didn't have any opinion to share then! I have over 267 uses of inkey currently across projects, and that's only me!)

jquast · 2015-02-28T21:52:18Z

Also, Good luck with your pycon talk!

thomasballinger · 2015-02-28T22:42:40Z

Thanks for the updates, and great work on doing this merge with Eric. Thanks for the great feedback on the PR a while ago, sorry I didn't make any progress on it after you gave great specific feedback.

jquast · 2015-10-10T05:47:54Z

blessed/terminal.py

+                    self._keyboard_buf.extendleft(
+                            u''.join(keystrokes)[:m.start()])
+                    return tuple(int(x) for x in m.groups())
+


man... lot of mojo here without any test cases !

jquast · 2015-10-10T05:54:26Z

blessed/terminal.py

+                keystrokes.append(self.inkey(_intr_continue=True))
+                m = re.search('\x1b[[]([0-9]+);([0-9]+)R', u''.join(keystrokes))
+                if m:
+                    self._keyboard_buf.extendleft(


this may warrant the need for a _ungetch(ucs) method, where ucs is characters to be pushed back into the keyboard buffer.

jquast · 2015-10-11T00:26:55Z

resuming

jquast · 2015-10-11T13:01:12Z

This PR superseded by #72

@thomasballinger

This PR succeeds #32 by @thomasballinger. major -------- - **new method**, ``.get_location(timeout=None)``. - **new method**, ``.ungetch()`` which pushes data back for next ``inkey()``. - **new example** program, ``bin/resize.py``, which required such condition. - **new example** program, ``bin/detect-multibyte.py`` which uses this method. - enhancement: now possible to use stdin when stream=stderr minor -------- - method-local function, ``time_left()``, moved to ``keyboard.py`` as ``_time_left()``. - new function in keyboard.py, ``_read_until``, something like a mini-pexpect. - new private state value ``_line_buffered``, used to determine whether the terminal has already entered cbreak or raw mode. This may become public later if we also use termios routines, instead as a read-only \@Property.

First take on cursor position

48ae54f

jquast reviewed Dec 3, 2014
View reviewed changes

jquast added M enhancement needs-changelog needs-code needs-tests labels Oct 10, 2015

jquast reviewed Oct 10, 2015
View reviewed changes

jquast removed the M label Oct 10, 2015

jquast mentioned this pull request Oct 10, 2015

add method get_cursor_position() #67

Closed

jquast reviewed Oct 10, 2015
View reviewed changes

jquast self-assigned this Oct 11, 2015

jquast mentioned this pull request Oct 11, 2015

add Terminal.get_location() method #72

Closed

jquast closed this Oct 11, 2015

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

First take on cursor position #32

First take on cursor position #32

thomasballinger commented Dec 3, 2014

jquast Dec 3, 2014

jquast Dec 3, 2014

jquast commented Dec 3, 2014

jquast Dec 3, 2014

jquast commented Dec 4, 2014

thomasballinger commented Dec 7, 2014

jquast commented Feb 28, 2015

jquast commented Feb 28, 2015

thomasballinger commented Feb 28, 2015

jquast Oct 10, 2015

jquast Oct 10, 2015

jquast commented Oct 11, 2015

jquast commented Oct 11, 2015

First take on cursor position #32

First take on cursor position #32

Conversation

thomasballinger commented Dec 3, 2014

jquast Dec 3, 2014

Choose a reason for hiding this comment

jquast Dec 3, 2014

Choose a reason for hiding this comment

jquast commented Dec 3, 2014

jquast Dec 3, 2014

Choose a reason for hiding this comment

jquast commented Dec 4, 2014

thomasballinger commented Dec 7, 2014

jquast commented Feb 28, 2015

jquast commented Feb 28, 2015

thomasballinger commented Feb 28, 2015

jquast Oct 10, 2015

Choose a reason for hiding this comment

jquast Oct 10, 2015

Choose a reason for hiding this comment

jquast commented Oct 11, 2015

jquast commented Oct 11, 2015