Struct mapped_command::Command

pub struct Command<Output, Error> where
    Output: 'static,
    Error: From<Error> + From<UnexpectedExitStatus> + 'static,  { /* fields omitted */ }

A alternative to std::process::Command see module level documentation.

Implementations

impl<Output, Error> Command<Output, Error> where
    Output: 'static,
    Error: From<Error> + From<UnexpectedExitStatus> + 'static, 

pub fn new(
    program: impl Into<OsString>,
    return_settings: impl OutputMapping<Output = Output, Error = Error>
) -> Self

Create a new command for given program and output mapping.

The output mapping will imply if stdout/stderr is captured and how the captured output is mapped to a Result<Self::Output, Self::Error>.

pub fn program(&self) -> &OsStr

Return the program the command will run.

pub fn arguments(&self) -> &[OsString]

Returns the arguments passed the the program when run.

pub fn with_arguments<T>(
    mut self: Self,
    args: impl IntoIterator<Item = T>
) -> Self where
    T: Into<OsString>, 

Returns this command with new arguments added to the end of the argument list

pub fn with_argument(mut self: Self, arg: impl Into<OsString>) -> Self

Returns this command with a new argument added to the end of the argument list

pub fn env_updates(&self) -> &HashMap<OsString, EnvChange>

Return a map of all env variables which will be set/overwritten in the subprocess.

Warning

The keys of env variables have not not been evaluated for syntactic validity. So the given keys can cause process spawning or calls to std::env::set_var to fail.

pub fn with_env_updates<K, V>(
    mut self: Self,
    map: impl IntoIterator<Item = (K, V)>
) -> Self where
    K: Into<OsString>,
    V: Into<EnvChange>, 

Returns this command with the map of env updates updated by given iterator of key value pairs.

If any key from the new map already did exist in the current updates it will replace the old key & value.

• Common supported values for keys include OsString, &OsStr, String, &str.
• Common supported values for values include EnvChange, OsString, &OsStr, String, &str

So you can pass in containers like Vec<(&str, &str)>, HashMap<&str, &str> or HashMap<OsString, EnvChange>, etc.

Warning

The keys of env variables will not be evaluated for syntactic validity. Setting a key invalid on given platform might cause the process spawning to fail (e.g. using a key lik "=" or ""). It also might also do other thinks like the env variable being passed in but being unaccessible or similar. It's completely dependent on the OS and the impl. of std::process::Command or whatever is used to execute the command.

pub fn with_env_update(
    mut self: Self,
    key: impl Into<OsString>,
    value: impl Into<EnvChange>
) -> Self

Returns this command with the map of env updates updated by one key value pair.

If the new key already did exist in the current updates it will replace that old key & value.

See Command::with_env_updates().

pub fn inherit_env(&self) -> bool

Returns true if the env of the current process is inherited.

Updates to then environment are applied after the inheritance:

• EnvChange::Set can be use to override inherited env vars, or add new ones if no variable with given key was inherited
• EnvChange::Remove can be used to remove an inherited (or previously added) env variable
• EnvChange::Inherit can be used to state a env variable should be inherited even if inherit_env is false. If inherit_env is true this will have no effect.

pub fn with_inherit_env(mut self: Self, do_inherit: bool) -> Self

Returns this command with a change to weather or the sub-process will inherit env variables.

See Command::inherit_env() for how this affects the sub-process env.

pub fn create_expected_env_iter(
    &self
) -> impl Iterator<Item = (Cow<'_, OsStr>, Cow<'_, OsStr>)>

Returns a map with all env variables the sub-process spawned by this command would have if the current processes env is not changed.

Site note about env::set_var() problems

Note that if you use std::env::set_var() in a multi-threaded setup depending on the operating system you run this on this can lead to all kind of problem, including unexpected race conditions in some situations (especially if inherit_env(true) is combined with EnvChange::Inherit and multiple variables are changed in another thread racing with this function and some but not all are covered by EnvChange::Inherit).

Given that std::env::set_var() should strictly be avoided in a multi-threaded context this is seen as an acceptable drawback.

Note that this function + std::env::set_var() is not unsafe it might just have a very unexpected result. Except if env::set_var() + reading env races are inherently unsafe on your system, in which case this has nothing to do with this function.

pub fn working_directory_override(&self) -> Option<&Path>

Return the working directory which will be used instead of the current working directory.

If None is returned it means no override is set and the working directory will be inherited from the spawning process.

pub fn with_working_directory_override(
    mut self: Self,
    wd_override: Option<impl Into<PathBuf>>
) -> Self

Replaces the working directory override.

Setting it to None will unset the override making the spawned process inherit the working directory from the spawning process.

pub fn expected_exit_status(&self) -> ExitStatus

Return which exit status is treated as success.

pub fn with_expected_exit_status(
    self,
    exit_status: impl Into<ExitStatus>
) -> Self

Set which exit status is treated as successful.

This enables exit status checking even if it was turned of before.

pub fn check_exit_status(&self) -> bool

Returns true if the exit status is checked before mapping the output(s).

pub fn with_check_exit_status(mut self: Self, val: bool) -> Self

Sets if the exit status is checked before mapping the output(s).

pub fn will_capture_stdout(&self) -> bool

Returns true if stdout will be captured.

Panics

If called in a exec_replacement_callback this will panic.

pub fn will_capture_stderr(&self) -> bool

Returns true if stderr will be captured.

Panics

If called in a exec_replacement_callback this will panic.

pub fn run(mut self: Self) -> Result<Output, Error>

Run the command, blocking until completion and then mapping the output.

This will:

1. run the program with the specified arguments and env variables
2. capture the necessary outputs as specified by the output mapping
3. if exit status checking was not disabled check the exit status and potentially fail.
4. if 3 doesn't fail now map captured outputs to a Result<Output, Error>

If Command::with_exec_replacement_callback() is used instead of running the program and capturing the output the given callback is called. The callback could mock the program execution. The exit status checking and output mapping are still done as normal.

Panics

This will panic if called in a exec_replacement_callback.

pub fn with_exec_replacement_callback(
    mut self: Self,
    callback: impl FnOnce(Self, &dyn OutputMapping<Output = Output, Error = Error>) -> Result<ExecResult, Error> + 'static
) -> Self

Sets a callback which is called instead of executing the command when running the command.

This is mainly meant to be used for mocking command execution during testing, but can be used for other thinks, too. E.g. the current implementation does have a default callback for normally executing the command this method was not called.

Implementing Mocks with an exec_replacement_callback

You must not call following methods in the callback:

• Command::run(), recursively calling run will not work.
• Command::will_capture_stdout(), use the passed in output mapping OutputMapping::capture_stdout() method instead.
• Command::will_capture_stderr(), use the passed in output mapping OutputMapping::capture_stderr() method instead.

An emulation of captured output and exit status is returned as ExecResult instance:

• Any exit code can be returned including a target specific one, the From<num> for ExitStatus implementations are useful here.
• If OutputMapping::capture_stdout() is true then ExecResult::stdout must be Some else it must be None. Failing to do so will panic on unwrap of debug assertions.
• If OutputMapping::capture_stdout() is true then ExecResult::stdout must be Some else it must be None. Failing to do so will panic on unwrap of debug assertions.

If used for mocking in tests you already know if stdout/stderr is assumed to (not) be captured so you do not need to access OutputMapping::capture_stdout()/OutputMapping::capture_stdout().

Settings like env updates and inheritance can be retrieved from the passed in Command instance.

Implement custom subprocess spawning

Be aware that if you execute the program in the callback you need to make sure the right program, arguments stdout/stderr capture setting and env variables are used. Especially note should be taken to how EnvChange::Inherit is handled.

The Command::create_expected_env_iter() method can be used to find what exact env variables are expected to be in the sub-process. Clearing the sub-process env and then setting all env vars using Command::create_expected_env_iter() is not the most efficient but most simple and robust to changes way to set the env. It's recommended to be used.