Decorating your Python code

Patrick Boucher — Sat, 11 Oct 2008 03:16:25 +0000

I have touched on Python decorators in the past in this article (although that decorator was a bit convoluted). Decorators can be extremely useful in many situations and here are a few that I propose may help streamline script development.

Note: For those of you who would like a bit of background information on decorators you can check out this page.

Keeping things quiet

Let’s look at the code and we’ll go for a few explanations afterward.

def suspendXSILogging(func):
    def closure(*args, **kwargs):
        logPrefs = getXSILoggingPrefs()
        try:
            ret = func(*args, **kwargs)
        except Exception, e:
            setXSILoggingPrefs(logPrefs)
            raise
        setXSILoggingPrefs(logPrefs)
        return ret
    return closure

The first function suspendXSILogging can be used as a decorator and will disable all logging features of XSI. This has the advantage of making large scripts that may be command dependent a bit quicker.

You could always implement a plugin with a custom command but sometimes a script is simpler. Even in the command context this decorator could be useful if you wanted part of the command to log to the script editor while keeping other parts silent.

Another advantage of this decorator is that it puts your original function in a large trap that, if an error should occur, will make sure that logging is reset to the users original preferences.

Time for decorating

def timeExecution(func):
    def closure(*args, **kwargs):
        startTime = time.time()
        try:
            ret = func(*args, **kwargs)
        except Exception, e:
            delta = time.time() - startTime
            log.error('Failed in %f seconds' % delta)
            raise
        delta = time.time() - startTime
        log.info('Finished in %f seconds' % delta)
        return ret
    return closure

This second function, timeExecution, can decorate almost any function and will print to the script editor the amount of time the decorated function took to execute. This can be used as a debugging or performance tracking tool or as a way to provide feedback to your users. Again, a trap is implemented that will log execution time even if an error occurs.

Please not that this decorator uses a logging system described here. If you would rather use simple logging you should replace log.info( instances by log(, xsi.LogMessage( or Application.LogMessage( depending on your scripting habits.

Usage example

@timeExecution
@suspendXSILogging
def main():
    # Do some fun useful stuff - well... more useful than this!
    xsi.CreatePrim("Cube", "MeshSurface", "", "")

You can chain decorators without hesitation like in this example, one thing to note in this particular case is that you’ll want timeExecution higher in the decorator chain than suspendXSILogging otherwise the time information will never make it to the script editor. Duh!

Support code

Before I sign off on this latest article, here are two functions that, if useful on their own, are essential to the functioning of the suspendXSILogging decorator.

def getXSILoggingPrefs():
    prefs = xsi.Preferences
    vals = {}
    for n in ['scripting.cmdlog', 'scripting.msglog', 'scripting.msglogverbose']:
        vals[n] = prefs.GetPreferenceValue(n)
        prefs.SetPreferenceValue(n, False)
    return vals
 
def setXSILoggingPrefs(vals):
    for key, val in vals.iteritems():
        xsi.Preferences.SetPreferenceValue(key, val)

Cheers!

Logging and being proactive

Patrick Boucher — Sun, 05 Oct 2008 04:37:57 +0000

You write tools for artists under a deadline (yours and theirs). You live in a production oriented world. Unit testing, beta testing and anticipating can only go so far, and that is only if you have time to properly test. You have to accept that eventually, your code will break. How it breaks and how you react to such a break becomes as important as your capacity to create the tool in the first place.

You’ve probably seen the following idiom used many times on this website and in other places that do XSI scripting in Python:

1
2
3

from win32com.client import constants as c
xsi = Application
log = xsi.LogMessage

It is a shortcut that allows you to use log as if it was the native LogMessage call:

1	log('This is an info message.', c.siInfo)

Here is a way to extend this by using the standard logging module in a way that can help you be more proactive. I suggest you get familiar with the logging module from the standard Python docs, it’s bound to eventually be helpful.

Logging in XSI

Here is a construct I’ve started using lately that I am starting to enjoy, its advantages are:

Concise and self documenting
It has a trap to catch any unexpected error and log it
Logs both to the script editor and to file
Easily extensible to other logging mechanisms (email, event viewer)
You can pass non string messages and it will convert to string for you (Yay!)
Easily useable in scripts and other modules

from hookolo.xsi import *
 
def main():
	log.comment('comment')
	log.debug('debug')
	log.info('info')
	log.warning('warning')
	log.error('error')
	log.fatal('fatal')
	log.critical('critical')
	raise Exception('Totally unexpected exception!')
 
run('demoScript', main)

The code above would produce the following output in XSI’s script editor.

# comment
# VERBOSE : debug
# INFO : info
# WARNING : warning
# ERROR : error
# ERROR : fatal
# FATAL : critical
# ERROR : Trap reached in demoScript
# Traceback (most recent call last):
#   File "C:\hookolo\libs\hookolo\xsi\__init__.py", line 18, in run
#     func(*args, **kwargs)
#   File "

XSIBlog

Decorating your Python code

Logging and being proactive