title |
---|
Interfacing with your OS |
The ANSI Common Lisp standard doesn't mention this topic. (Keep in mind that it was written at a time where Lisp Machines were at their peak. On these boxes Lisp was your operating system!) So almost everything that can be said here depends on your OS and your implementation. There are, however, some widely used libraries, which either come with your Common Lisp implementation, or are easily available through Quicklisp. These include:
- ASDF3, which is included with almost all Common Lisp implementations, includes Utilities for Implementation- and OS- Portability (UIOP).
- osicat
- unix-opts is a command-line argument parser, similar to Python's
argparse
.
UIOP comes with a function that'll allow you to look at Unix/Linux environment variables on a lot of different CL implementations:
* (uiop:getenv "HOME")
"/home/edi"
Below is an example implementation, where we can see /feature flags/ used to run code on specific implementations:
* (defun my-getenv (name &optional default)
"Obtains the current value of the POSIX environment variable NAME."
(declare (type (or string symbol) name))
(let ((name (string name)))
(or #+abcl (ext:getenv name)
#+ccl (ccl:getenv name)
#+clisp (ext:getenv name)
#+cmu (unix:unix-getenv name) ; since CMUCL 20b
#+ecl (si:getenv name)
#+gcl (si:getenv name)
#+mkcl (mkcl:getenv name)
#+sbcl (sb-ext:posix-getenv name)
default)))
MY-GETENV
* (my-getenv "HOME")
"/home/edi"
* (my-getenv "HOM")
NIL
* (my-getenv "HOM" "huh?")
"huh?"
You should also note that some of these implementations also provide the ability to set these variables. These include ECL (si:setenv
) and AllegroCL, LispWorks, and CLISP where you can use the functions from above together with setf
. This feature might be important if you want to start subprocesses from your Lisp environment.
Also note that the
Osicat
library has the method (environment-variable "name")
, on POSIX-like
systems including Windows. It is also fset
-able.
Accessing command line arguments is implementation-specific but it
appears most implementations have a way of getting at
them. UIOP with uiop:command-line-arguments
or Roswell as well as external
libraries (see next section) make it portable.
SBCL stores the arguments list in the special variable sb-ext:*posix-argv*
$ sbcl my-command-line-arg
....
* sb-ext:*posix-argv*
("sbcl" "my-command-line-arg")
*
More on using this to write standalone Lisp scripts can be found in the SBCL Manual
LispWorks has system:*line-arguments-list*
* system:*line-arguments-list*
("/Users/cbrown/Projects/lisptty/tty-lispworks" "-init" "/Users/cbrown/Desktop/lisp/lispworks-init.lisp")
Here's a quick function to return the argument strings list across multiple implementations:
(defun my-command-line ()
(or
#+SBCL *posix-argv*
#+LISPWORKS system:*line-arguments-list*))
Now it would be handy to access them in a portable way and to parse them according to a schema definition.
We have a look at the Awesome CL list#scripting section and we'll show how to use unix-opts.
(ql:quickload "unix-opts")
We can now refer to it with its opts
nickname.
We first declare our arguments with opts:define-opts
, for example
(opts:define-opts
(:name :help
:description "print this help text"
:short #\h
:long "help")
(:name :level
:description "The level of something (integer)."
:short #\l
:long "level"
:arg-parser #'parse-integer))
Everything should be self-explanatory. Note that #'parse-integer
is
a built-in CL function.
Now we can parse and get them with opts:get-opts
, which returns two
values: the first is the list of valid options and the second the
remaining free arguments. We then must use multiple-value-bind to
catch everything:
(multiple-value-bind (options free-args)
;; There is no error handling yet (specially for options not having their argument).
(opts:get-opts)
We can explore this by giving a list of strings (as options) to
get-opts
:
(multiple-value-bind (options free-args)
(opts:get-opts '("hello" "-h" "-l" "1"))
(format t "Options: ~a~&" options)
(format t "free args: ~a~&" free-args))
Options: (HELP T LEVEL 1)
free args: (hello)
NIL
If we put an unknown option, we get into the debugger. We refer you to unix-opts' documentation and code sample to deal with erroneous options and other errors.
We can access the arguments stored in options
with getf
(it is a
property list), and we can exit (in a portable way) with
opts:exit
. So, for example:
(multiple-value-bind (options free-args)
;; No error handling.
(opts:get-opts)
(if (getf options :help)
(progn
(opts:describe
:prefix "My app. Usage:"
:args "[keywords]")
(exit))) ;; <= exit takes an optional return status.
...
And that's it for now, you know the essential. See the documentation for a complete example, and the Awesome CL list for useful packages to use in the terminal (ansi colors, printing tables and progress bars, interfaces to readline,…).
uiop has us covered, and is probably included in your Common Lisp implementation.
uiop:run-program
either takes a string as argument, denoting the
name of the executable to run, or a list of strings, for the program and its arguments:
(uiop:run-program "firefox")
or
(uiop:run-program (list "firefox" "http:url"))
This will process the program output as specified and return the processing results when the program and its output processing are complete.
Use :output t
to print to standard output.
This function has the following optional arguments:
run-program (command &rest keys &key
ignore-error-status
(force-shell nil force-shell-suppliedp)
input
(if-input-does-not-exist :error)
output
(if-output-exists :supersede)
error-output
(if-error-output-exists :supersede)
(element-type #-clozure *default-stream-element-type* #+clozure 'character)
(external-format *utf-8-external-format*)
&allow-other-keys)
It will always call a shell (rather than directly executing the command when possible)
if force-shell
is specified. Similarly, it will never call a shell if force-shell
is
specified to be nil
.
Signal a continuable subprocess-error
if the process wasn't successful (exit-code 0),
unless ignore-error-status
is specified.
If output
is a pathname, a string designating a pathname, or nil
(the default)
designating the null device, the file at that path is used as output.
If it's :interactive
, output is inherited from the current process;
beware that this may be different from your *standard-output*
,
and under slime
will be on your *inferior-lisp*
buffer.
If it's t
, output goes to your current *standard-output*
stream.
Otherwise, output
should be a value that is a suitable first argument to
slurp-input-stream
(qv.), or a list of such a value and keyword arguments.
In this case, run-program
will create a temporary stream for the program output;
the program output, in that stream, will be processed by a call to slurp-input-stream
,
using output
as the first argument (or the first element of output
, and the rest as keywords).
The primary value resulting from that call (or nil
if no call was needed)
will be the first value returned by run-program.
E.g., using :output :string
will have it return the entire output stream as a string.
And using :output '(:string :stripped t
) will have it return the same string
stripped of any ending newline.
if-output-exists
, which is only meaningful if output
is a string or a
pathname, can take the values :error
, :append
, and :supersede
(the
default). The meaning of these values and their effect on the case
where output
does not exist, is analogous to the if-exists
parameter
to open
with :direction
:output
.
error-output
is similar to output
, except that the resulting value is returned
as the second value of run-program
. t designates the *error-output*
.
Also :output
means redirecting the error output to the output stream,
in which case nil
is returned.
if-error-output-exists
is similar to if-output-exist
, except that it
affects error-output
rather than output
.
input
is similar to output
, except that vomit-output-stream
is used,
no value is returned, and T designates the *standard-input*
.
if-input-does-not-exist
, which is only meaningful if input
is a string
or a pathname, can take the values :create
and :error
(the
default). The meaning of these values is analogous to the
if-does-not-exist
parameter to open
with :direction :input
.
element-type
and external-format
are passed on
to your Lisp implementation, when applicable, for creation of the output stream.
One and only one of the stream slurping or vomiting may or may not happen in parallel in parallel with the subprocess, depending on options and implementation, and with priority being given to output processing. Other streams are completely produced or consumed before or after the subprocess is spawned, using temporary files.
run-program
returns 3 values:
- the result of the
output
slurping if any, ornil
- the result of the
error-output
slurping if any, ornil
- either 0 if the subprocess exited with success status, or an
indication of failure via the
exit-code
of the process
With uiop:launch-program
.
Its signature is the following:
launch-program (command &rest keys
&key
input
(if-input-does-not-exist :error)
output
(if-output-exists :supersede)
error-output
(if-error-output-exists :supersede)
(element-type #-clozure *default-stream-element-type*
#+clozure 'character)
(external-format *utf-8-external-format*)
directory
#+allegro separate-streams
&allow-other-keys)
Output (stdout) from the launched program is set using the output
keyword:
- If
output
is a pathname, a string designating a pathname, ornil
(the default) designating the null device, the file at that path is used as output. - If it's
:interactive
, output is inherited from the current process; beware that this may be different from your*standard-output*
, and under Slime will be on your*inferior-lisp*
buffer. - If it's
T
, output goes to your current*standard-output*
stream. - If it's
:stream
, a new stream will be made available that can be accessed viaprocess-info-output
and read from. - Otherwise,
output
should be a value that the underlying lisp implementation knows how to handle.
if-output-exists
, which is only meaningful if output
is a string or a
pathname, can take the values :error
, :append
, and :supersede
(the
default). The meaning of these values and their effect on the case
where output
does not exist, is analogous to the if-exists
parameter
to open
with :DIRECTION :output
.
error-output
is similar to output
. T designates the *error-output*
,
:output
means redirecting the error output to the output stream,
and :stream
causes a stream to be made available via
process-info-error-output
.
launch-program
returns a process-info
object, which look like the following (source):
(defclass process-info ()
(
;; The advantage of dealing with streams instead of PID is the
;; availability of functions like `sys:pipe-kill-process`.
(process :initform nil)
(input-stream :initform nil)
(output-stream :initform nil)
(bidir-stream :initform nil)
(error-output-stream :initform nil)
;; For backward-compatibility, to maintain the property (zerop
;; exit-code) <-> success, an exit in response to a signal is
;; encoded as 128+signum.
(exit-code :initform nil)
;; If the platform allows it, distinguish exiting with a code
;; >128 from exiting in response to a signal by setting this code
(signal-code :initform nil)))
See the docstrings.
uiop:process-alive-p
tests if a process is still alive, given a
process-info
object returned by launch-program
:
* (defparameter *shell* (uiop:launch-program "bash" :input :stream :output :stream))
;; inferior shell process now running
* (uiop:process-alive-p *shell*)
T
;; Close input and output streams
* (uiop:close-streams *shell*)
* (uiop:process-alive-p *shell*)
NIL
We can use uiop:wait-process
. If the process is finished, it returns
immediately, and returns the exit code. If not, it waits for the
process to terminate.
(uiop:process-alive-p *process*)
NIL
(uiop:wait-process *process*)
0
An exit code to 0 means success (use zerop
).
The exit code is also stored in the exit-code
slot of our
process-info
object. We see from the class definition above that it
has no accessor, so we'll use slot-value
. It has an initform
to
nil, so we don't have to check if the slot is bound. We can do:
(slot-value *my-process* 'uiop/launch-program::exit-code)
0
The trick is that we must run wait-process
beforehand, otherwise
the result will be nil
.
Since wait-process
is blocking, we can do it on a new thread:
(bt:make-thread
(lambda ()
(let ((exit-code (uiop:wait-process
(uiop:launch-program (list "of" "commands"))))
(if (zerop exit-code)
(print :success)
(print :failure)))))
:name "Waiting for <program>")
Note that run-program
returns the exit code as the third value.
If the input
keyword is set to :stream
, then a stream is created
and can be written to in the same way as a file. The stream can be
accessed using uiop:process-info-input
:
;; Start the inferior shell, with input and output streams
* (defparameter *shell* (uiop:launch-program "bash" :input :stream :output :stream))
;; Write a line to the shell
* (write-line "find . -name '*.md'" (uiop:process-info-input *shell*))
;; Flush stream
* (force-output (uiop:process-info-input *shell*))
where write-line writes the string to the given stream, adding a newline at the end. The force-output call attempts to flush the stream, but does not wait for completion.
Reading from the output stream is similar, with
uiop:process-info-output
returning the output stream:
* (read-line (uiop:process-info-output *shell*))
In some cases the amount of data to be read is known, or there are delimiters to determine when to stop reading. If this is not the case, then calls to read-line can hang while waiting for data. To avoid this, listen can be used to test if a character is available:
* (let ((stream (uiop:process-info-output *shell*)))
(loop while (listen stream) do
;; Characters are immediately available
(princ (read-line stream))
(terpri)))
There is also
read-char-no-hang which reads
a single character, or returns nil
if no character is available.
Note that due to issues like buffering, and the timing of when the
other process is executed, there is no guarantee that all data sent
will be received before listen
or read-char-no-hang
return nil
.
Here's an example to do the equivalent of ls | sort
. Note that "ls"
uses launch-program
(async) and outputs to a stream, where "sort",
the last command of the pipe, uses run-program
and outputs to a
string.
(uiop:run-program "sort"
:input
(uiop:process-info-output
(uiop:launch-program "ls"
:output :stream))
:output :string)
Implementations provide their own functions for this.
On SBCL:
(sb-posix:getpid)
It is possible portably with the osicat library:
(osicat-posix:getpid)
Here again, we could find it by using the apropos
function:
CL-USER> (apropos "pid")
OSICAT-POSIX:GETPID (fbound)
OSICAT-POSIX::PID
[…]
SB-IMPL::PID
SB-IMPL::WAITPID (fbound)
SB-POSIX:GETPID (fbound)
SB-POSIX:GETPPID (fbound)
SB-POSIX:LOG-PID (bound)
SB-POSIX::PID
SB-POSIX::PID-T
SB-POSIX:WAITPID (fbound)
[…]