Autopage is a Python library to
automatically display terminal output from a program in a pager (like less
)
whenever you need it, and never when you don't. And it only takes one line of
code.
You know how some CLI programs like git
(and a handful of others, including
man
and systemctl
) automatically pipe their output to less
? Except not if
there's less than one screen's worth of data. And if you redirect the output to
a file or a pipe, it does the right thing instead. Colours are preserved. Don't
you wish all programs worked like that? Now at least all of your Python
programs can.
© 2020-2022 by Zane Bitter
Open Source licensed under the terms of the Apache Software License, version 2.0.
Autopage is available from PyPI. The
easiest way to install (preferably in a virtualenv
virtual environment) is
with pip
:
$ pip install autopage
Autopage is packaged for Fedora 35 and later. To install:
# dnf install python3-autopage
Copr repositories are available for older versions of Fedora and EPEL. Before attempting to install on those versions, first enable the copr repository:
# dnf copr enable zaneb/autopage
Autopage is packaged for Ubuntu jammy and later, and for Debian bookworm. To install:
# apt-get install python3-autopage
A PPA is available for older versions of Ubuntu. Before attempting to install on those versions, first enable the PPA:
# add-apt-repository ppa:zaneb/autopage
# apt-get update
Autopage is packaged for Gentoo. To install:
# emerge dev-python/autopage
The AutoPager
class provides a context manager that furnishes the output
stream to write to. Here is a basic example that reads from stdin and outputs
to a pager connected to stdout:
import sys
import autopage
with autopage.AutoPager() as out:
for l in sys.stdin:
out.write(l)
If you are explicitly passing a stream to write to (rather than directly
referencing a global variable such as sys.stdout
then you may be able to add
automatic paging support with only a single line of code.
If your program uses the argparse
module from the standard library, you can
ensure that the help output is automatically paged when possible by changing
the import statement to:
from autopage import argparse
If you don't control the module that imports argparse
, you can instead call
autopage.argparse.monkey_patch()
to patch the module directly. This function
can also be used as a context manager.
The default pager command (autopage.command.DefaultPager()
) allows the end
user to override the pager command by setting the PAGER
environment variable.
To disable this behaviour, pass
pager_command=autopage.command.PlatformPager()
to use the default pager for
the current platform, or pass a specific pager from autopage.command
. The
default pager command is less
on most platforms. On AIX the default pager
command is more
, and on Windows more.com
.
The end user can also override the settings for less
by setting the LESS
environment variable. If not specified, the settings are determined by the
allow_color
and line_buffering
options. By default ANSI control characters
for setting colours are respected and the pager will not run if there is less
than a full screen of text to display.
Normally output streams are buffered so that data is written to the output file
only when the buffer becomes full. This is efficient and generally works fine
as long as the data is being produced as fast as it can be consumed. However,
when the data is streaming at a slower rate than it could be displayed (e.g.
log output from something like tail -f
) this results in a large delay between
data being produced and consumed. If you have ever tried to grep a streaming
log and pipe the output to a pager then you are familiar with how
unsatisfactory this is.
The solution is to flush the output buffer after each line is written, which is
known as line
buffering. The
AutoPager
class supports a line_buffering
argument to enable or disable
line buffering. The default is to use the line buffering mode already
configured for the output stream (which is usually to disable line buffering).
When reading from an input stream (which may be a file, pipe, or the console)
and optionally processing the data before outputting it again, the convenience
function line_buffer_from_input()
returns the optimal line buffering setting
for a given input stream (sys.stdin
by default).
import sys
import autopage
with autopage.AutoPager(line_buffering=autopage.line_buffer_from_input()) as o:
for l in sys.stdin:
o.write(l)
By default, when the pager exits it will leave the latest displayed output on
screen in the terminal. This can be changed by passing True
for the
reset_on_exit
argument to the AutoPager
class. If this option is set, the
terminal will be cleared when the pager exits, returning to its position prior
to starting the pager (as is the case by default when running less
manually
from the command line).
Programs may wish to return a different exit code if they are interrupted by
the user (either with Ctrl-C or by closing the pager) than if they ran to
completion. The exceptions generated when the pager is closed prematurely are
suppressed, so the AutoPager
class offers the exit_code()
method to provide
a suitable exit code for the program. This also takes into account other
exceptions that bubble up through the context manager.
import sys
import autopage
def process(input_stream, output_stream):
pager = autopage.AutoPager(
output_stream,
line_buffering=autopage.line_buffer_from_input(input_stream),
allow_color=True,
reset_on_exit=True,
errors=autopage.ErrorStrategy.REPLACE,
)
try:
with pager as out:
for l in input_stream:
out.write(l)
except Exception as exc:
sys.stderr.write(f'{str(exc)}\n')
except KeyboardInterrupt:
pass
return pager.exit_code()
sys.exit(process(sys.stdin, sys.stdout))