Simple debugging with meta programming
The 'watch' feature of modern debuggers allows a developer to see when a variable changes state. I use this feature a lot when doing small debugging sprints. Unfortunately, using it requires opening up a big IDE like PyCharm. I prefer to work primarily with the command line and vim.
So, I decided to write my own debugging 'utility' to handle very simple approach to this 'watching' debugging work flow.
- Easy to use
- Track several attributes at once
- Display when variable changes and what its' new value is
- Display stack trace/location where watched variable changes
The first step is to somehow track when any variable is set. Luckily,
Python provides a
solely for this purpose . The
is called each time an attribute is set (before the actual assignment
Watching specific attributes
Now we need a list of what attributes to watch. Then, we can provide a custom
that watches each attribute 'set' action and alert us when a specific attribute
def __setattr__(self, name, value): if name in ['myvar1', 'myvar2']: import traceback import sys # Print stack (without this __setattr__ call) traceback.print_stack(sys._getframe(1)) print '%s -> %s = %s' % (repr(self), name, value)
The above approach works, but it doesn't give us much flexibility.
What if I want to use this functionality in another class? This would require copying the whole method followed by a small change to the list of attributes. It seems like a shame to copy/paste all that code verbatim and then only change a single line. There must be a better way...
Steps to re-usability
We can be a bit more critical about the proposed solution and notice the following:
- The solution tightly couples the list of attributes and the real implementation details.
- We will always override the same method,
- This method must always exist inside of a class to work.
We need a way to pass a list of attribute names to a class method to watch all 'set' actions for these attributes. So, to be more precise and technical, we want override a built-in class method in a generic way and pass it a single argument, a list of attributes to watch.
This sounds like a great use of a
This is a perfect use case since the only information needed up front is the
attributes to watch. Then, we could add this behavior to any class
simply by using the
@ decorator syntax.
So, without further ado:
In a nutshell, we make our own
_setattr method and then replace the decorated
__setattr__ with our own. So, to watch attributes in a class simply
@watch_variables above your class with a list of attribute names. For
@watch_variables(['foo', 'bar']) class BuggyClass(object): def __init__(self, foo, bar): self.foo = foo self.bar = bar
Unfortunately, this solution has one major drawback. It cannot track changes
to mutable attributes. For example, this decorator cannot alert you if your
class has a
list attribute and it's modified by appending or some other
append() method and other related list modification mechanisms are
methods on the
list attribute, not our class. Thus, our custom version of
__setattr__ will never be called.
I'm sure this functionality could be added. However, it wasn't immediately obvious to me at the time, and the above solution worked for my most recent debugging sprint.
So, go show off your skills and fork my solution and 'fix' this deficiency or point out a better solution altogether!
 I briefly discussed connecting to Python's attribute setting mechanism during my dunder talk so check that out if this sort of functionality interests you.
Published: 02-10-2013 22:55:00