Library version: | 7.2 |
---|---|
Library scope: | GLOBAL |
Robot Framework library for running processes.
This library utilizes Python's subprocess module and its Popen class.
The library has following main usages:
Both Run Process and Start Process accept the command to execute and all arguments passed to the command as separate arguments. This makes usage convenient and also allows these keywords to automatically escape possible spaces and other special characters in commands and arguments. Notice that if a command accepts options that themselves accept values, these options and their values must be given as separate arguments.
When running processes in shell, it is also possible to give the whole command to execute as a single string. The command can then contain multiple commands to be run together. When using this approach, the caller is responsible on escaping.
Examples:
Run Process | ${tools}${/}prog.py | argument | second arg with spaces | ||
Run Process | java | -jar | ${jars}${/}example.jar | --option | value |
Run Process | prog.py "one arg" && tool.sh | shell=yes | cwd=${tools} |
Possible non-string arguments are converted to strings automatically.
Run Process and Start Process keywords can be configured using optional **configuration
keyword arguments. Configuration arguments must be given after other arguments passed to these keywords and must use syntax like name=value
. Available configuration arguments are listed below and discussed further in sections afterward.
Name | Explanation |
---|---|
shell | Specifies whether to run the command in shell or not. |
cwd | Specifies the working directory. |
env | Specifies environment variables given to the process. |
env:<name> | Overrides the named environment variable(s) only. |
stdout | Path of a file where to write standard output. |
stderr | Path of a file where to write standard error. |
stdin | Configure process standard input. New in RF 4.1.2. |
output_encoding | Encoding to use when reading command outputs. |
alias | Alias given to the process. |
Note that because **configuration
is passed using name=value
syntax, possible equal signs in other arguments passed to Run Process and Start Process must be escaped with a backslash like name\=value
. See Run Process for an example.
The shell
argument specifies whether to run the process in a shell or not. By default, shell is not used, which means that shell specific commands, like copy
and dir
on Windows, are not available. You can, however, run shell scripts and batch files without using a shell.
Giving the shell
argument any non-false value, such as shell=True
, changes the program to be executed in a shell. It allows using the shell capabilities, but can also make the process invocation operating system dependent. Having a shell between the actually started process and this library can also interfere communication with the process such as stopping it and reading its outputs. Because of these problems, it is recommended to use the shell only when absolutely necessary.
When using a shell it is possible to give the whole command to execute as a single string. See Specifying command and arguments section for examples and more details in general.
By default, the child process will be executed in the same directory as the parent process, the process running Robot Framework, is executed. This can be changed by giving an alternative location using the cwd
argument. Forward slashes in the given path are automatically converted to backslashes on Windows.
Standard output and error streams, when redirected to files, are also relative to the current working directory possibly set using the cwd
argument.
Example:
Run Process | prog.exe | cwd=${ROOT}/directory | stdout=stdout.txt |
The child process will get a copy of the parent process's environment variables by default. The env
argument can be used to give the child a custom environment as a Python dictionary. If there is a need to specify only certain environment variable, it is possible to use the env:<name>=<value>
format to set or override only that named variables. It is also possible to use these two approaches together.
Examples:
Run Process | program | env=${environ} | |
Run Process | program | env:http_proxy=10.144.1.10:8080 | env:PATH=%{PATH}${:}${PROGDIR} |
Run Process | program | env=${environ} | env:EXTRA=value |
By default, processes are run so that their standard output and standard error streams are kept in the memory. This works fine normally, but if there is a lot of output, the output buffers may get full and the program can hang.
To avoid the above-mentioned problems, it is possible to use stdout
and stderr
arguments to specify files on the file system where to redirect the outputs. This can also be useful if other processes or other keywords need to read or manipulate the outputs somehow.
Given stdout
and stderr
paths are relative to the current working directory. Forward slashes in the given paths are automatically converted to backslashes on Windows.
As a special feature, it is possible to redirect the standard error to the standard output by using stderr=STDOUT
.
Regardless are outputs redirected to files or not, they are accessible through the result object returned when the process ends. Commands are expected to write outputs using the console encoding, but output encoding can be configured using the output_encoding
argument if needed.
If you are not interested in outputs at all, you can explicitly ignore them by using a special value DEVNULL
both with stdout
and stderr
. For example, stdout=DEVNULL
is the same as redirecting output on console with > /dev/null
on UNIX-like operating systems or > NUL
on Windows. This way the process will not hang even if there would be a lot of output, but naturally output is not available after execution either.
Examples:
${result} = | Run Process | program | stdout=${TEMPDIR}/stdout.txt | stderr=${TEMPDIR}/stderr.txt |
Log Many | stdout: ${result.stdout} | stderr: ${result.stderr} | ||
${result} = | Run Process | program | stderr=STDOUT | |
Log | all output: ${result.stdout} | |||
${result} = | Run Process | program | stdout=DEVNULL | stderr=DEVNULL |
Note that the created output files are not automatically removed after the test run. The user is responsible to remove them if needed.
The stdin
argument makes it possible to pass information to the standard input stream of the started process. How its value is interpreted is explained in the table below.
Value | Explanation |
---|---|
String NONE |
Inherit stdin from the parent process. This is the default. |
String PIPE |
Make stdin a pipe that can be written to. |
Path to a file | Open the specified file and use it as the stdin. |
Any other string | Create a temporary file with the text as its content and use it as the stdin. |
Any non-string value | Used as-is. Could be a file descriptor, stdout of another process, etc. |
Values PIPE
and NONE
are case-insensitive and internally mapped to subprocess.PIPE
and None
, respectively, when calling subprocess.Popen.
Examples:
Run Process | command | stdin=PIPE |
Run Process | command | stdin=${CURDIR}/stdin.txt |
Run Process | command | stdin=Stdin as text. |
The support to configure stdin
is new in Robot Framework 4.1.2. Its default value used to be PIPE
until Robot Framework 7.0.
Executed commands are, by default, expected to write outputs to the standard output and error streams using the encoding used by the system console. If the command uses some other encoding, that can be configured using the output_encoding
argument. This is especially useful on Windows where the console uses a different encoding than rest of the system, and many commands use the general system encoding instead of the console encoding.
The value used with the output_encoding
argument must be a valid encoding and must match the encoding actually used by the command. As a convenience, it is possible to use strings CONSOLE
and SYSTEM
to specify that the console or system encoding is used, respectively. If produced outputs use different encoding then configured, values got through the result object will be invalid.
Examples:
Start Process | program | output_encoding=UTF-8 | |
Run Process | program | stdout=${path} | output_encoding=SYSTEM |
A custom name given to the process that can be used when selecting the active process.
Examples:
Start Process | program | alias=example | ||
Run Process | python | -c | print('hello') | alias=hello |
The library keeps record which of the started processes is currently active. By default it is the latest process started with Start Process, but Switch Process can be used to activate a different process. Using Run Process does not affect the active process.
The keywords that operate on started processes will use the active process by default, but it is possible to explicitly select a different process using the handle
argument. The handle can be an alias
explicitly given to Start Process or the process object returned by it.
Run Process, Wait For Process and Terminate Process keywords return a result object that contains information about the process execution as its attributes. The same result object, or some of its attributes, can also be get using Get Process Result keyword. Attributes available in the object are documented in the table below.
Attribute | Explanation |
---|---|
rc | Return code of the process as an integer. |
stdout | Contents of the standard output stream. |
stderr | Contents of the standard error stream. |
stdout_path | Path where stdout was redirected or None if not redirected. |
stderr_path | Path where stderr was redirected or None if not redirected. |
Example:
${result} = | Run Process | program |
Should Be Equal As Integers | ${result.rc} | 0 |
Should Match | ${result.stdout} | Some t?xt* |
Should Be Empty | ${result.stderr} | |
${stdout} = | Get File | ${result.stdout_path} |
Should Be Equal | ${stdout} | ${result.stdout} |
File Should Be Empty | ${result.stderr_path} |
Some keywords accept arguments that are handled as Boolean values true or false. If such an argument is given as a string, it is considered false if it is an empty string or equal to FALSE
, NONE
, NO
, OFF
or 0
, case-insensitively. Other strings are considered true regardless their value, and other argument types are tested using the same rules as in Python.
True examples:
Terminate Process | kill=True | # Strings are generally true. |
Terminate Process | kill=yes | # Same as the above. |
Terminate Process | kill=${TRUE} | # Python True is true. |
Terminate Process | kill=${42} | # Numbers other than 0 are true. |
False examples:
Terminate Process | kill=False | # String false is false. |
Terminate Process | kill=no | # Also string no is false. |
Terminate Process | kill=${EMPTY} | # Empty string is false. |
Terminate Process | kill=${FALSE} | # Python False is false. |
*** Settings *** Library Process Suite Teardown Terminate All Processes kill=True *** Test Cases *** Example Start Process program arg1 arg2 alias=First ${handle} = Start Process command.sh arg | command2.sh shell=True cwd=/path ${result} = Run Process ${CURDIR}/script.py Should Not Contain ${result.stdout} FAIL Terminate Process ${handle} ${result} = Wait For Process First Should Be Equal As Integers ${result.rc} 0
Returns the process ID (pid) of the process as an integer.
If handle
is not given, uses the current active process.
Starting from Robot Framework 5.0, it is also possible to directly access the pid
attribute of the subprocess.Popen
object returned by Start Process like ${process.pid}
.
Return the underlying subprocess.Popen
object.
If handle
is not given, uses the current active process.
Starting from Robot Framework 5.0, Start Process returns the created subprocess.Popen
object, not a generic handle, making this keyword mostly redundant.
Returns the specified result object or some of its attributes.
The given handle
specifies the process whose results should be returned. If no handle
is given, results of the current active process are returned. In either case, the process must have been finishes before this keyword can be used. In practice this means that processes started with Start Process must be finished either with Wait For Process or Terminate Process before using this keyword.
If no other arguments than the optional handle
are given, a whole result object is returned. If one or more of the other arguments are given any true value, only the specified attributes of the result object are returned. These attributes are always returned in the same order as arguments are specified in the keyword signature. See Boolean arguments section for more details about true and false values.
Examples:
Run Process | python | -c | print('Hello, world!') | alias=myproc | |
# Get result object | |||||
${result} = | Get Process Result | myproc | |||
Should Be Equal | ${result.rc} | ${0} | |||
Should Be Equal | ${result.stdout} | Hello, world! | |||
Should Be Empty | ${result.stderr} | ||||
# Get one attribute | |||||
${stdout} = | Get Process Result | myproc | stdout=true | ||
Should Be Equal | ${stdout} | Hello, world! | |||
# Multiple attributes | |||||
${stdout} | ${stderr} = | Get Process Result | myproc | stdout=yes | stderr=yes |
Should Be Equal | ${stdout} | Hello, world! | |||
Should Be Empty | ${stderr} |
Although getting results of a previously executed process can be handy in general, the main use case for this keyword is returning results over the remote library interface. The remote interface does not support returning the whole result object, but individual attributes can be returned without problems.
Checks is the process running or not.
If handle
is not given, uses the current active process.
Returns True
if the process is still running and False
otherwise.
Joins arguments into one command line string.
In resulting command line string arguments are delimited with a space, arguments containing spaces are surrounded with quotes, and possible quotes are escaped with a backslash.
If this keyword is given only one argument and that is a list-like object, then the values of that list are joined instead.
Example:
${cmd} = | Join Command Line | --option | value with spaces |
Should Be Equal | ${cmd} | --option "value with spaces" |
Verifies that the process is running.
If handle
is not given, uses the current active process.
Fails if the process has stopped.
Verifies that the process is not running.
If handle
is not given, uses the current active process.
Fails if the process is still running.
Runs a process and waits for it to complete.
command
and *arguments
specify the command to execute and arguments passed to it. See Specifying command and arguments for more details.
**configuration
contains additional configuration related to starting processes and waiting for them to finish. See Process configuration for more details about configuration related to starting processes. Configuration related to waiting for processes consists of timeout
and on_timeout
arguments that have same semantics as with Wait For Process keyword. By default, there is no timeout, and if timeout is defined the default action on timeout is terminate
.
Process outputs are, by default, written into in-memory buffers. If there is a lot of output, these buffers may get full causing the process to hang. To avoid that, process outputs can be redirected using the stdout
and stderr
configuration parameters. For more information see the Standard output and error streams section.
Returns a result object containing information about the execution.
Note that possible equal signs in *arguments
must be escaped with a backslash (e.g. name\=value
) to avoid them to be passed in as **configuration
.
Examples:
${result} = | Run Process | python | -c | print('Hello, world!') |
Should Be Equal | ${result.stdout} | Hello, world! | ||
${result} = | Run Process | ${command} | stdout=${CURDIR}/stdout.txt | stderr=STDOUT |
${result} = | Run Process | ${command} | timeout=1min | on_timeout=continue |
${result} = | Run Process | java -Dname\=value Example | shell=True | cwd=${EXAMPLE} |
This keyword does not change the active process.
Sends the given signal
to the specified process.
If handle
is not given, uses the current active process.
Signal can be specified either as an integer as a signal name. In the latter case it is possible to give the name both with or without SIG
prefix, but names are case-sensitive. For example, all the examples below send signal INT (2)
:
Send Signal To Process | 2 | # Send to active process | |
Send Signal To Process | INT | ||
Send Signal To Process | SIGINT | myproc | # Send to named process |
This keyword is only supported on Unix-like machines, not on Windows. What signals are supported depends on the system. For a list of existing signals on your system, see the Unix man pages related to signal handling (typically man signal
or man 7 signal
).
By default sends the signal only to the parent process, not to possible child processes started by it. Notice that when running processes in shell, the shell is the parent process and it depends on the system does the shell propagate the signal to the actual started process.
To send the signal to the whole process group, group
argument can be set to any true value (see Boolean arguments).
Splits command line string into a list of arguments.
String is split from spaces, but argument surrounded in quotes may contain spaces in them.
If escaping
is given a true value, then backslash is treated as an escape character. It can escape unquoted spaces, quotes inside quotes, and so on, but it also requires using doubling backslashes in Windows paths and elsewhere.
Examples:
@{cmd} = | Split Command Line | --option "value with spaces" |
Should Be True | $cmd == ['--option', 'value with spaces'] |
Starts a new process on background.
See Specifying command and arguments and Process configuration for more information about the arguments, and Run Process keyword for related examples. This includes information about redirecting process outputs to avoid process handing due to output buffers getting full.
Makes the started process new active process. Returns the created subprocess.Popen object which can be used later to activate this process. Popen
attributes like pid
can also be accessed directly.
Processes are started so that they create a new process group. This allows terminating and sending signals to possible child processes.
Examples:
Start process and wait for it to end later using an alias:
Start Process | ${command} | alias=example |
# Other keywords | ||
${result} = | Wait For Process | example |
Use returned Popen
object:
${process} = | Start Process | ${command} |
Log | PID: ${process.pid} | |
# Other keywords | ||
${result} = | Terminate Process | ${process} |
Use started process in a pipeline with another process:
${process} = | Start Process | python | -c | print('Hello, world!') | |
${result} = | Run Process | python | -c | import sys; print(sys.stdin.read().upper().strip()) | stdin=${process.stdout} |
Wait For Process | ${process} | ||||
Should Be Equal | ${result.stdout} | HELLO, WORLD! |
Returning a subprocess.Popen
object is new in Robot Framework 5.0. Earlier versions returned a generic handle and getting the process object required using Get Process Object separately.
Makes the specified process the current active process.
The handle can be an identifier returned by Start Process or the alias
given to it explicitly.
Example:
Start Process | prog1 | alias=process1 |
Start Process | prog2 | alias=process2 |
# currently active process is process2 | ||
Switch Process | process1 | |
# now active process is process1 |
Terminates all still running processes started by this library.
This keyword can be used in suite teardown or elsewhere to make sure that all processes are stopped,
By default tries to terminate processes gracefully, but can be configured to forcefully kill them immediately. See Terminate Process that this keyword uses internally for more details.
Stops the process gracefully or forcefully.
If handle
is not given, uses the current active process.
By default first tries to stop the process gracefully. If the process does not stop in 30 seconds, or kill
argument is given a true value, (see Boolean arguments) kills the process forcefully. Stops also all the child processes of the originally started process.
Waits for the process to stop after terminating it. Returns a result object containing information about the execution similarly as Wait For Process.
On Unix-like machines graceful termination is done using TERM (15)
signal and killing using KILL (9)
. Use Send Signal To Process instead if you just want to send either of these signals without waiting for the process to stop.
On Windows graceful termination is done using CTRL_BREAK_EVENT
event and killing using Win32 API function TerminateProcess()
.
Examples:
${result} = | Terminate Process | ||
Should Be Equal As Integers | ${result.rc} | -15 | # On Unixes |
Terminate Process | myproc | kill=true |
Limitations:
Waits for the process to complete or to reach the given timeout.
The process to wait for must have been started earlier with Start Process. If handle
is not given, uses the current active process.
timeout
defines the maximum time to wait for the process. It can be given in various time formats supported by Robot Framework, for example, 42
, 42 s
, or 1 minute 30 seconds
. The timeout is ignored if it is Python None
(default), string NONE
(case-insensitively), zero, or negative.
on_timeout
defines what to do if the timeout occurs. Possible values and corresponding actions are explained in the table below. Notice that reaching the timeout never fails the test.
Value | Action |
---|---|
continue | The process is left running (default). |
terminate | The process is gracefully terminated. |
kill | The process is forcefully stopped. |
See Terminate Process keyword for more details how processes are terminated and killed.
If the process ends before the timeout or it is terminated or killed, this keyword returns a result object containing information about the execution. If the process is left running, Python None
is returned instead.
Examples:
# Process ends cleanly | |||
${result} = | Wait For Process | example | |
Process Should Be Stopped | example | ||
Should Be Equal As Integers | ${result.rc} | 0 | |
# Process does not end | |||
${result} = | Wait For Process | timeout=42 secs | |
Process Should Be Running | |||
Should Be Equal | ${result} | ${NONE} | |
# Kill non-ending process | |||
${result} = | Wait For Process | timeout=1min 30s | on_timeout=kill |
Process Should Be Stopped | |||
Should Be Equal As Integers | ${result.rc} | -9 |