nicecall 1.0.3

A library which provides a slightly more convinient way to launch processes, compared to Python's subprocess module.

While Python’s subprocess library is great, it may not be the easiest library to use. This is the reason I created nicecall: it allows to do simple tasks with processes very easily.

Note that nicecall is not a substitute to subprocess, because much of subprocess functionality doesn’t exist. For instance, one can’t use stdin or pipes with nicecall. The goal is not to replace subprocess, but only to provide an easy way to do the most common tasks.

How to use the library

Note: make sure you also check the tests which give a few examples of how to use the library. Most pieces of code below are in tests/smoke/test_docs.py.

Fluent interface

The library uses method chaining, which allows to add logic on the fly before actually launching the process. Methods such as on_stdout, ignore, etc. create a copy of the object, modify this copy, and return it to the caller. This makes it possible to reuse base objects in multiple locations in your code, reducing code duplication.

Exit code

Let’s start by executing a task:

result = nicecall.Process(["touch", "/tmp/hello"]).execute()

The result contains the exit code, which makes it possible to determine whether the process terminated successfully. Below, the value of result is expected to be zero. You may also make it fail:

result = nicecall.Process(["touch", "/tmp/a/b/c/d"]).execute()

The result should now be 1, assuming you don’t have /tmp/a/b/c directory.

stdout and stderr

One can also perform a bunch of actions on stdout and stderr. Let’s display stdout in terminal:

nicecall.Process(["echo", "a\nb\nc"]).on_stdout(print).execute()

The output should be:

a
b
c

If you’re absolutely sure that the process will be fast and produce a small amount of stdout or stderr, you can ask the library to buffer the contents in order to process them later:

stdout_buffer = nicecall.Buffer()
nicecall \
    .Process(["echo", "a\nb"]) \
    .on_stdout(stdout_buffer.store) \
    .execute()

Now you can access the content either as a list:

>>> print(stdout_buffer.lines)
['a', 'b']

or as a string with newlines:

>>> stdout_buffer.contents
a
b

Logging

A common thing, at least in my case, is to log stdout or stderr to syslog. With nicecall, it’s easy:

# Initialize logging.
log_handler = logging.handlers.SysLogHandler(address="/dev/log")
formatter = logging.Formatter("demo: [%(levelname)s] %(message)s")
log_handler.setFormatter(formatter)
log_handler.setLevel(logging.DEBUG)

demo_logger = logging.getLogger("demo")
demo_logger.setLevel(logging.DEBUG)
demo_logger.addHandler(log_handler)

...

# Log stdout.
logger = nicecall.StdoutLogger("test")
nicecall.Process(["echo", "a\nb"]).on_stdout(logger.log).execute()

Note that nicecall.StdoutLogger can be initialized with either the name of the logger, or the instance of the logger itself.

The library itself logs calls (INFO level) and call failures (WARNING level) through the logger named nicecall.process. For instance, executing touch /tmp/a/b/c/d will produce two log messages:

INFO:nicecall.process:Called “touch /tmp/a/b/c/d”.
WARNING:nicecall.process:“touch /tmp/a/b/c/d” failed with exit code 1.

Filtering

Sometimes, you don’t want to process specific content such as empty lines or whitespace. This is what filters are about:

nicecall \
    .Process(["echo", "a\n\nb"]) \
    .ignore(nicecall.filters.whitespace) \
    .on_stdout(print) \
    .execute()

Here, a and b will be displayed in terminal; however, the empty line will be ignored. The reverse is called keep. Both keep and ignore accept any function which takes a string as a parameter and returns a boolean. For instance, this will print only stdout content longer than fifteen characters:

nicecall \
    .Process(["echo", "Hello World!\nWhat a lovely day!"]) \
    .keep(lambda line: len(line) > 15) \
    .on_stdout(print) \
    .execute()

Multiple keep and ignore methods can be combined. The output will keep the lines which match all predicates from keep methods and none from ignore ones.

Filters apply to both stdout and stderr; there is no way to apply them to only one of the streams.

Classes

process.py

The class is the entry point of the library. It makes it possible to specify different options before actually starting the process.

• __init__: creates a new instance of the class.

  Parameters:

  args is an array which indicates the process to start, and its parameters. Example: ["touch", "/tmp/hello"].

• args property: the getter which returns the value initially passed to the constructor.

• execute: actually executes the process and blocks until the process finishes.

  Returns:

  Returns the exit code.

• keep: specifies a filter to apply to determine if the line of stdout or stderr should be processed by the actions specified through on_stdout and on_stderr.

  The method can be called multiple times and mixed with ignore to aggregate multiple filters.

  Parameters:

  predicate is a function which takes a string as a parameter and returns a boolean value: true if the line should be processed, or false otherwise.

  Returns:

  Returns a new instance of the Process class with the new filter.

• ignore: see keep. Here, the predicate is reverted.

• on_stdout: adds an action to perform when a line from stdout is received.

  The method can be called multiple times if multiple actions should be performed for every line of stdout.

  Parameters:

  action: a function which takes a string as a parameter and doesn’t return anything.

  Returns:

  Returns a new instance of the Process class with the new action.

• on_stderr: see on_stdout. Here, it deals with stderr instead.

filters.py

The file contains a bunch of filters which can be used in Process.keep and Process.ignore.

buffer.py

This class makes it possible to store in memory the output from stdout or stderr. It is expected to be used exclusively for short processes which output only a small amount of lines. In other cases, consider processing the output on the fly.

logger.py

This class is used to log output from stdout or stderr.

Compatibility

The library was written for Python 3 under Linux. I haven’t tested it neither with Python 2, nor under Windows.

Reliability

While I used Test Driven Development when creating this library and naturally have a 100% branch coverage, I don’t know neither Python, nor subprocess well enough to be sure that the library can be used reliably in production. Use at own risk.

Contributing

If you want to contribute, contact me at [e-mail]. You’ll be able to contribute to the project using the official SVN repository. If you find it more convinient to clone the source to GitHub, you can do that too.

The source code of the library and the corresponding documentation are covered by the MIT License.