Welcome, guest | Sign In | My Account | Store | Cart

This package provides a decorator for tracing function and method calls in your applications. The tracing capabilities are managed through the logging package, and several mechanisms are provided for controlling the destination of the trace output.

It also provides functionality for adding decorators to existing classes or modules.

Python, 777 lines
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
#! /usr/bin/env python
######################################################################
#  Copyright (c) by Kevin L. Sitze on 2011-01-23.
#  This code may be used pursuant to the MIT License.
######################################################################

"""This package provides facilities to attach decorators to classes or
modules (possibly recursively).  A tracing decorator is provided for
tracing function and method calls in your applications.

from decorators import *

@trace
def function(...):
    '''Output goes to <TT>logging.getLogger(__module__)</TT>,
    where <TT>__module__</TT> is the name of the module in which
    the function is defined.  If the current module is '__main__'
    then the root logger will be used.
    '''
    pass

@trace(logger)
def function(...):
    '''Output goes to the specified logger instance.
    '''
    pass

@trace("Logger")
def function(...):
    '''Output goes to the named logger instance.
    '''

class ClassName(object):
    '''Output goes to the ClassName logger.  If ClassName is
    not in the '__main__' module then the name of its module
    will be prefixed to the classname (separated by a dot).
    All methods on the class will be traced.
    '''
    __logger__ = loggerOrLoggerName # optional
    __metaclass__ = decorators.TraceMetaClass
    def method(...):
        pass

class PostDecorate(object):
    def method(...):
        pass
# attach a decorator to an existing class.
attach(trace(logger), PostDecorate)

You can also attach a decorator to an existing module.
"""

import inspect
import logging
import sys
import thread
import types

from functools import wraps
from itertools import chain

FunctionTypes = tuple(set((
    types.BuiltinFunctionType,
    types.FunctionType
)))

MethodTypes = tuple(set((
    types.BuiltinMethodType,
    types.MethodType,
    types.UnboundMethodType
)))

class _C(object):
    @classmethod
    def classMethod(klass): pass
    classMethodType = type(classMethod)

    @staticmethod
    def staticMethod(): pass
    staticMethodType = type(staticMethod)

    @property
    def propertyMethod(self): pass

ClassMethodType = _C.classMethodType
StaticMethodType = _C.staticMethodType
PropertyType = type(_C.propertyMethod)

CallableTypes = tuple(set((
    types.BuiltinFunctionType,
    types.FunctionType,
    types.BuiltinMethodType,
    types.MethodType,
    types.UnboundMethodType,
    ClassMethodType
)))

__all__ = (
    'ThreadLocal',
    'TraceMetaClass',
    'attach',
    'getFormatter',
    'setFormatter',
    'getLoggerFactory',
    'setLoggerFactory',
    'trace'
)

######################################################################
#  Utility functions
######################################################################

MAX_SIZE = 320
def chop(value):
    s = repr(value)
    if len(s) > MAX_SIZE:
        return s[:MAX_SIZE] + '...' + s[-1]
    else:
        return s

def loggable(obj):
    """Return "True" if the obj implements the minimum Logger API
    required by the 'trace' decorator.
    """
    if isinstance(obj, logging.getLoggerClass()):
        return True
    else:
        return ( inspect.isclass(obj) and
                 inspect.ismethod(getattr(obj, 'debug', None)) and
                 inspect.ismethod(getattr(obj, 'isEnabledFor', None)) and
                 inspect.ismethod(getattr(obj, 'setLevel', None)) )

######################################################################
#  LoggerFactory Property
######################################################################

_logger_factory = logging
def getLoggerFactory():
    """Retrieve the current factory object for creating loggers.
    The default is to use the logging module.
    """
    global _logger_factory
    return _logger_factory

def setLoggerFactory(factory):
    """Set a factory object for creating loggers.  This object must
    publish a method or class named 'getLogger' that takes a string
    parameter naming the logger instance to retrieve.  Logger objects
    returned by this factory must, at a minimum, expose the methods
    'isEnabledFor' and 'debug'.
    """
    global _logger_factory
    _logger_factory = factory

######################################################################
#  class PrependLoggerFactory
######################################################################

class PrependLoggerFactory(object):
    """This is a convenience class for creating new loggers for the
    "trace" decorator.  All loggers created via this class will have a
    user specified prefix prepended to the name of the logger to
    instantiate.
    """
    def __init__(self, prefix = 'trace'):
        """Construct a new "PrependLoggerFactory" instance that
        prepends the value \var{prefix} to the name of each
        logger to be created by this class.
        """
        self.__prefix = prefix.strip('.')

    @property
    def prefix(self):
        """The value to prefix to each logger created by this factory.
        """
        return self.__prefix
    @prefix.setter
    def prefix(self, value):
        self.__prefix = value.strip('.')

    def getLogger(self, name):
        return logging.getLogger('.'.join((self.__prefix, name)))

######################################################################
#  class ThreadLocal
######################################################################

class ThreadLocal(object):
    """Instances of this class provide a thread-local variable.
    """
    def __init__(self):
        self.__lock = thread.allocate_lock()
        self.__vars = dict()
        self.__init = None

    @property
    def value(self):
        with self.__lock:
            try:
                return self.__vars[thread.get_ident()]
            except KeyError:
                return self.__init
    @value.setter
    def value(self, value):
        with self.__lock:
            self.__vars[thread.get_ident()] = value

    @property
    def initialValue(self):
        with self.__lock:
            return self.__init
    @initialValue.setter
    def initialValue(self, value):
        with self.__lock:
            self.__init = value

######################################################################
#  Formatter functions
######################################################################

def _formatter_self(name, value):
    """Format the "self" variable and value on instance methods.
    """
    __mname = value.__module__
    if __mname != '__main__':
        return '%s = <%s.%s object at 0x%x>' % (name, __mname, value.__class__.__name__, id(value))
    else:
        return '%s = <%s object at 0x%x>' % (name, value.__class__.__name__, id(value))

def _formatter_class(name, value):
    """Format the "klass" variable and value on class methods.
    """
    __mname = value.__module__
    if __mname != '__main__':
        return "%s = <type '%s.%s'>" % (name, __mname, value.__name__)
    else:
        return "%s = <type '%s'>" % (name, value.__name__)

def _formatter_named(name, value):
    """Format a named parameter and its value.
    """
    return '%s = %s' % (name, chop(value))

def _formatter_defaults(name, value):
    return '[%s = %s]' % (name, chop(value))

af_self = _formatter_self
af_class = _formatter_class
af_named = _formatter_named
af_default = _formatter_defaults
af_unnamed = chop
af_keyword = _formatter_named

def getFormatter(name):
    """Return the named formatter function.  See the function
    "setFormatter" for details.
    """
    if name in ( 'self', 'instance', 'this' ):
        return af_self
    elif name == 'class':
        return af_class
    elif name in ( 'named', 'param', 'parameter' ):
        return af_named
    elif name in ( 'default', 'optional' ):
        return af_default
    elif name in ( 'anonymous', 'arbitrary', 'unnamed' ):
        return af_anonymous
    elif name in ( 'keyword', 'pair', 'pairs' ):
        return af_keyword
    else:
        raise ValueError('unknown trace formatter %r' % name)

def setFormatter(name, func):
    """Replace the formatter function used by the trace decorator to
    handle formatting a specific kind of argument.  There are several
    kinds of arguments that trace discriminates between:

    * instance argument - the object bound to an instance method.
    * class argument - the class object bound to a class method.
    * positional arguments (named) - values bound to distinct names.
    * positional arguments (default) - named positional arguments with
      default values specified in the function declaration.
    * positional arguments (anonymous) - an arbitrary number of values
      that are all bound to the '*' variable.
    * keyword arguments - zero or more name-value pairs that are
      placed in a dictionary and bound to the double-star variable.

    \var{name} - specifies the name of the formatter to be modified.

        * instance argument - "self", "instance" or "this"
        * class argument - "class"
        * named argument - "named", "param" or "parameter"
        * default argument - "default", "optional"
        * anonymous argument - "anonymous", "arbitrary" or "unnamed"
        * keyword argument - "keyword", "pair" or "pairs"

    \var{func} - a function to format an argument.
    * For all but anonymous formatters this function must accept two
      arguments: the variable name and the value to which it is bound.
    * The anonymous formatter function is passed only one argument
      corresponding to an anonymous value.
    * if \var{func} is "None" then the default formatter will be used.
    """
    if name in ( 'self', 'instance', 'this' ):
        global af_self
        af_self = _formatter_self if func is None else func
    elif name == 'class':
        global af_class
        af_class = _formatter_class if func is None else func
    elif name in ( 'named', 'param', 'parameter' ):
        global af_named
        af_named = _formatter_named if func is None else func
    elif name in ( 'default', 'optional' ):
        global af_default
        af_default = _formatter_defaults if func is None else func
    elif name in ( 'anonymous', 'arbitrary', 'unnamed' ):
        global af_anonymous
        af_anonymous = chop if func is None else func
    elif name in ( 'keyword', 'pair', 'pairs' ):
        global af_keyword
        af_keyword = _formatter_named if func is None else func
    else:
        raise ValueError('unknown trace formatter %r' % name)

######################################################################
#  Decorator: trace
######################################################################

__builtins = (
    '__import__(name,globals={},locals={},fromlist=[],level=-1)',
    'abs(number)',
    'all(iterable)',
    'any(iterable)',
    'apply(object,args=[],kwargs={})',
    'bin(number)',
    'callable(object)',
    'chr(i)',
    'cmp(x,y)',
    'coerce(x,y)',
    'compile(source,filename,mode,flags=0,dont_inherit=0)',
    'delattr(object,name)',
    'dir()',
    'divmod(x,y)',
    'eval(source,globals={},locals={})',
    'execfile(filename,globals={},locals={})',
    'filter(function,sequence)',
    'format(value,format_spec="")',
    'getattr(object,name)',
    'globals()',
    'hasattr(object,name)',
    'hash(object)',
    'hex(number)',
    'id(object)',
    'input(prompt=None)',
    'intern(string)',
    'isinstance(object,klass)',
    'issubclass(C,B)',
    'iter(collection)',
    'len(object)',
    'locals()',
    'map(function,sequence)',
    'max(iterable,key=None)',
    'min(iterable,key=None)',
    'next(iterator)',
    'oct(number)',
    'open(name,mode=0666,buffering=True)',
    'ord(c)',
    'pow(x,y,z=None)',
    'range()',
    'raw_input(prompt=None)',
    'reduce(function,sequence)',
    'reload(module)',
    'repr(object)',
    'round(number,ndigits=0)',
    'setattr(object,name,value)',
    'sorted(iterable,cmp=None,key=None,reverse=False)',
    'sum(sequence,start=0)',
    'unichr(i)',
    'vars()',
    'zip(sequence)'
)

__builtin_defaults = {
    '""': "",
    '-1': -1,
    '0' : 0,
    '0666': 0666,
    'False': False,
    'None': None,
    'True': True,
    '[]': list(),
    '{}': dict()
}

__builtin_functions = None
def __lookup_builtin(name):
    """Lookup the parameter name and default parameter values for
    builtin functions.
    """
    global __builtin_functions
    if __builtin_functions is None:
        builtins = dict()
        for proto in __builtins:
            pos = proto.find('(')
            name, params, defaults = proto[:pos], list(), dict()
            for param in proto[pos+1:-1].split(','):
                pos = param.find('=')
                if not pos < 0:
                    param, value = param[:pos], param[pos+1:]
                    try:
                        defaults[param] = __builtin_defaults[value]
                    except KeyError:
                        raise ValueError( 'builtin function %s: parameter %s: unknown default %r' % ( name, param, value ) )
                params.append(param)
            builtins[name] = ( params, defaults )
        __builtin_functions = builtins

    try:
        params, defaults = __builtin_functions[name]
    except KeyError:
        params, defaults = tuple(), dict()
        __builtin_functions[name] = ( params, defaults )
        print >>sys.stderr, "Warning: builtin function %r is missing prototype" % name
    return ( len(params), params, defaults )

_ = ThreadLocal()
_.initialValue = False

def trace(_name):
    """Function decorator that logs function entry and exit details.

    \var{_name} a string, an instance of logging.Logger or a function.

    Construct a function or method proxy to generate call traces.
    """
    def decorator(_func):
        """This is the actual decorator function that wraps the
        \var{_func} function for detailed logging.
        """

        def positional(name, value):
            """Format one named positional argument.
            """
            if name is __self:
                return af_self(name, value)
            elif name is __klass:
                return af_class(name, value)
            else:
                return af_named(name, value)

        def wrapper(*__argv, **__kwds):
            if not logger.isEnabledFor(logging.DEBUG) or _.value:
                return _func(*__argv, **__kwds)

            try:
                _.value = True
                params = dict(co_defaults)
                params.update(__kwds)
                params.update(zip(co_varnames, __argv))

                position = [ positional(n, params.pop(n)) for n in co_varnames[:len(__argv)] ]
                defaults = [ af_default(n, params.pop(n)) for n in co_varnames[len(__argv):] ]
                nameless = ( af_unnamed(v) for v in __argv[co_argcount:] )
                keywords = ( af_keyword(n, params[n]) for n in sorted(params.keys()) )

                params = ', '.join(filter(None, chain(position, defaults, nameless, keywords)))
                # params = params.replace(', [', '[, ').replace('][, ', ', ')

                enter = [ pre_enter ]
                if params:
                    enter.append(' ')
                    enter.append(params)
                    enter.append(' ')
                enter.append(')')

                leave = [ pre_leave ]

                try:
                    logger.debug(''.join(enter))
                    try:
                        try:
                            _.value = False
                            result = _func(*__argv, **__kwds)
                        finally:
                            _.value = True
                    except:
                        type, value, traceback = sys.exc_info()
                        leave.append(' => exception thrown\n\traise ')
                        __mname = type.__module__
                        if __mname != '__main__':
                            leave.append(__mname)
                            leave.append('.')
                        leave.append(type.__name__)
                        if value.args:
                            leave.append('(')
                            leave.append(', '.join(chop(v) for v in value.args))
                            leave.append(')')
                        else:
                            leave.append('()')
                        raise
                    else:
                        if result is not None:
                            leave.append(' => ')
                            leave.append(chop(result))
                finally:
                    logger.debug(''.join(leave))
            finally:
                _.value = False

            return result

        ####
        #  decorator
        ####

        __self  = False
        __klass = False
        __rewrap = lambda x: x
        if type(_func) in FunctionTypes:
            # functions do not belong to a class.
            __cname = None
        elif type(_func) in MethodTypes:
            # im_self is None for unbound instance methods.
            # Assumption: trace is only called on unbound methods.
            if _func.im_self is not None:
                __rewrap = classmethod
                __cname = _func.im_self.__name__
                __klass = True
            else:
                __cname = _func.im_class.__name__
                __self  = True
            _func = _func.im_func
        else:
            # other callables are not supported yet.
            return _func
        __module = _func.__module__
        __fname  = _func.__name__

        # Do not wrap initialization and conversion methods.
        if __fname in ('__init__', '__new__', '__repr__', '__str__'):
            return __rewrap(_func)

        # Generate the Fully Qualified Function Name.

        __fqfn = list()
        if __module != '__main__':
            __fqfn.append(__module)
            __fqfn.append('.')
        if __cname is not None:
            __fqfn.append(__cname)
            __fqfn.append('.')
        __fqcn = ''.join(__fqfn)
        __fqfn.append(__fname)
        __fqfn = ''.join(__fqfn)

        if type(_name) in CallableTypes:
            logger = getLoggerFactory().getLogger(__fqfn)
        elif loggable(_name):
            logger = _name
        elif isinstance(_name, basestring):
            logger = getLoggerFactory().getLogger(_name)
        else:
            raise ValueError('invalid object %r: must be a function, a method, a string or an object that implements the Logger API' % _name)

        pre_enter = [ '>>> ' ]
        pre_enter.append(__fqfn)
        pre_enter.append('(')
        pre_enter = ''.join(pre_enter)

        pre_leave = [ '<<< ' ]
        pre_leave.append(__fqfn)
        pre_leave = ''.join(pre_leave)

        ####
        #  Here we are really mucking around in function internals.
        #  func_code is the low level 'code' instance that describes
        #  the function arguments, variable and other stuff.
        #
        #  func.func_code.co_argcount - number of function arguments.
        #  func.func_code.co_varnames - function variables names, the
        #      first co_argcount values are the argument names.
        #  func.func_defaults - contains default arguments

        try:
            code = _func.func_code
        except AttributeError:
            co_argcount , \
            co_varnames , \
            co_defaults = __lookup_builtin(_func.__name__)
        else:
            co_argcount = code.co_argcount
            co_varnames = code.co_varnames[:co_argcount]
            if _func.func_defaults:
                co_defaults = dict(zip(co_varnames[-len(_func.func_defaults):], _func.func_defaults))
            else:
                co_defaults = dict()
            if __klass:
                __klass = co_varnames[0]
            if __self:
                __self  = co_varnames[0]
        return __rewrap(wraps(_func)(wrapper))

    ####
    #  trace
    ####

    logging.basicConfig(level = logging.DEBUG)
    if type(_name) in CallableTypes:
        return decorator(_name)
    else:
        return decorator

######################################################################
#  attach: apply decorator to a class or module
######################################################################

def attachToProperty(decorator, klass, k, prop_attr, prop_decorator):
    if prop_attr is not None:
        setattr(klass, k, prop_attr)
        value = decorator(getattr(klass, k))
    else:
        value = None
    # Passing "None" to the property decorator causes the new property
    # to assume the original value of the associated attribute.
    return prop_decorator(value)

def attachToClass(decorator, klass, recursive = True):
    for k, v in klass.__dict__.iteritems():
        t = type(v)
        if t is types.FunctionType or t is ClassMethodType:
            setattr(klass, k, decorator(getattr(klass, k)))
        elif t is StaticMethodType:
            setattr(klass, k, staticmethod(decorator(getattr(klass, k))))
        elif t is PropertyType:
            value = getattr(klass, k)
            value = attachToProperty(decorator, klass, k, value.fget, value.getter)
            value = attachToProperty(decorator, klass, k, value.fset, value.setter)
            value = attachToProperty(decorator, klass, k, value.fdel, value.deleter)
            setattr(klass, k, value)
        elif recursive and inspect.isclass(v):
            attachToClass(decorator, v, recursive)

def attach(decorator, obj, recursive = True):
    """attach(decorator, class_or_module[, recursive = True])

    Utility to attach a \val{decorator} to the \val{obj} instance.

    If \val{obj} is a module, the decorator will be attached to every
    function and class in the module.

    If \val{obj} is a class, the decorator will be attached to every
    method and subclass of the class.

    if \val{recursive} is "True" then subclasses will be decorated.
    """
    if inspect.ismodule(obj):
        for name, fn in inspect.getmembers(obj, inspect.isfunction):
            setattr(obj, name, decorator(fn))
        for name, klass in inspect.getmembers(obj, inspect.isclass):
            attachToClass(decorator, klass, recursive)
    elif inspect.isclass(obj):
        attachToClass(decorator, obj, recursive)

######################################################################
#  class TraceMetaClass
######################################################################

class TraceMetaClass(type):
    """Metaclass to automatically attach the 'trace' decorator to all
    methods, static method and class methods of the class.
    """
    def __new__(meta, className, bases, classDict):
        klass = super(TraceMetaClass, meta).__new__(meta, className, bases, classDict)
        if classDict.has_key('__logger__'):
            hook = trace(classDict['__logger__'])
        else:
            hook = trace
        attachToClass(hook, klass, False)
        return klass

if __name__ == '__main__':
    logging.basicConfig(level = logging.DEBUG)
    logger = logging.root

    class Test(object):
        __logger__ = logger
        __metaclass__ = TraceMetaClass

        @classmethod
        def classMethod(klass):
            pass

        @staticmethod
        def staticMethod():
            pass

        __test = None
        @property
        def test(self):
            return self.__test
        @test.setter
        def test(self, value):
            self.__test = value

        def method(self):
            pass

    Test.classMethod()
    Test.staticMethod()
    test = Test()
    test.test = 1
    assert 1 == test.test
    test.method()

    class Test(object):
        @classmethod
        def classMethod(klass):
            pass

        @staticmethod
        def staticMethod():
            pass

        __test = None
        @property
        def test(self):
            return self.__test
        @test.setter
        def test(self, value):
            self.__test = value

        def method(self):
            pass

        def __str__(self):
            return 'Test(' + str(self.test) + ')'

    attach(trace(logger), Test)
    Test.classMethod()
    Test.staticMethod()
    test = Test()
    test.test = 1
    assert 1 == test.test
    test.method()
    print str(test)

    @trace(logger)
    def test(x, y, z = True):
        a = x + y
        b = x * y
        if z: return a
        else: return b

    test(5, 5)
    test(5, 5, False)

    setLoggerFactory(PrependLoggerFactory())

    @trace('main')
    def test(x, *argv, **kwds):
        """Simple test
        """
        return x + sum(argv)

    test(5)
    test(5, 5)
    test(5, 5, False)
    test(5, 5, False, name = 10)

    test( *xrange(50) )

    assert test.__doc__ == 'Simple test\n        '
    assert test.__name__ == 'test'

    myzip = trace('main')(zip)
    for i, j in myzip(xrange(5), xrange(5, 10)):
        print i, j

This package provides facilities to attach decorators to classes or modules (possibly recursively). A tracing decorator is provided for tracing function and method calls in your applications.

To use this module, save the code into a file named "decorators.py" in your python library path. Add one of the following import statements to your code.

from decorators import *
from decorators import trace

To trace a single function call add the "@trace" decorator to the function.

@trace
def function(...):
    pass

This trace decorator will send a log message to the debug log stream when function is called and when it returns (even if that return is a result of an exception). The logger itself is created using a logger factory (see decorators.getLoggerFactory in source code). The name of the logger is the concatenation of the module name and the function name, unless the module is '__main__' in which case only the function name is used.

You can also send the output of the trace decorator to a specific logger. This logger must be either a subclass of logging.Logger or it may be an instance of your own log class. To use your own class, you must expose the isEnabledFor and debug methods as published by the logging.Logger class.

@trace(logger)
def function(...):
    pass

You may optionally name the logger to use by passing the trace decorator a string instead of a logging object. This string will be passed unchanged to the getLogger method of the current LoggerFactory object (again, see decorators.getLoggerFactory).

@trace("Logger")
def function(...):
    pass

You can trace all methods on a class too. If the class is one your application owns, simply change its metaclass to decorators.TraceMetaClass. All methods on the class will automatically have the trace decorator applied.

class ClassName(object):
    __logger__ = loggerOrLoggerName # optional
    __metaclass__ = decorators.TraceMetaClass
    def method( self, ...):
        pass

Instance and class methods are handled specially by the trace decorator in that it understands the protocol that the first parameter is actually the instance or class object so that it can adjust the logging output slightly for these cases.

Note also the special __logger__ class variable that allows you to specify the logger name or object for this class.

If you need to trace a 3rd party class or module, you can attach the trace decorator to that class or module.

from 3rdPartyModule import AClass
attach(trace(my_logger_or_logger_name), AClass)

You can also attach a decorator to an existing module.

import 3rdPartyModule
attach(trace, 3rdPartyModule)

Notes and Bugs:

The package attempts to avoid infinite recursion scenarios by distinguishing between running application code and trace code. When trace code is executing and the trace decorator is entered the execution flow of the decorator will be short-circuited to the decorated function. To make this entry and exit detection thread safe I write a simple ThreadLocal class to track this information separately for different threads. ThreadLocal may be useful all on its own in your own threaded code (despite not automatically GC'ing values for threads that have exited).

While the __str__ and __repr__ methods on classes are never decorated by trace (even if explicitly decorated) infinite recursion through such methods can still occur. The __new__ and __init__ methods are likewise suppressed because attempting to log information about a class prior to its construction rarely leads to runnable code.

Tracing is inherently dangerous. If you need to trace 3rd party code the author of this module strongly suggests that you do so with a least-impact philosophy. Only decorate the specific function you need to observe. Don't decorate an entire class if a single method will do, don't decorate an entire module if a single class will do.

I've added code to "sort of handle" builtin (C-level) functions. This list is by no means comprehensive. If your application attempts to trace builtin functions found in Python packages you'll get a warning message for each function other than those already found in the __builtins__ module.

While I've made some effort at making sure that functions decorated with the trace decorator will remain thread safe if they were originally thread safe, the process of actually applying the trace decoration to functions is not an inherently thread safe process. What this means is that you should spawn threads in your application AFTER applying the trace decoration to those modules, classes and functions you wish to observe.

If you use logging.basicConfig to configure logging for your application, you are strongly encouraged to do this before using the trace decorator. This is because the decorator calls logging.basicConfig. This is likely a bug. It should delegate this to the factory (but the factory defaults to the logging module). In any case you can do this in your application code properly as follows:

# BEFORE importing any other module do this
import logging
logging.basicConfig(...)
import my_own_modules_here

If you want to send all trace output to a single logger, do this in your application

import logging
import decorators
trace = decorators.trace(logging.getLogger(my_logger_name))

@trace
def function(...): pass

By creating the trace decorator with a logger up front, it may make it easier for you to control your logging output. Another mechanism for centralizing control of the trace output is to construct a PrependLoggerFactory that adds a known value and a dot to the front of the name of each logger created by the trace descriptor. This allows your application to control all logging at a high level, as well as fine tuned control of logging over individual modules, classes or functions.

The name of this module is 'decorators' when 'observe' is probably a better name.

Changes:

This version has the formatting of various argument kinds broken out into separate functions. These functions may be changed under application control so that arguments in the trace output may be customized to increase or decrease the overall information level.

3 comments

Kevin L. Sitze (author) 13 years, 3 months ago  # | flag

The <TT> element is supposed to convert its contents to monospaced type. The Markdown syntax for doing this kind of thing is not obvious and I couldn't find it in a quick review of the documentation at http://daringfireball.net/projects/markdown/syntax. This sucks and I'm sorry.

Kevin L. Sitze (author) 13 years, 3 months ago  # | flag

This is the result of running the above code on my machine:

ksitze@linux:python[1426]$ python --version
Python 2.6.6
ksitze@linux:python[1427]$ ./decorators.py
DEBUG:root:>>> Test.classMethod( klass = <type 'Test'> )
DEBUG:root:<<< Test.classMethod
DEBUG:root:>>> staticMethod()
DEBUG:root:<<< staticMethod
DEBUG:root:>>> Test.test( self = <Test object at 0x774590>, value = 1 )
DEBUG:root:<<< Test.test
DEBUG:root:>>> Test.test( self = <Test object at 0x774590> )
DEBUG:root:<<< Test.test => 1
DEBUG:root:>>> Test.method( self = <Test object at 0x774590> )
DEBUG:root:<<< Test.method
DEBUG:root:>>> Test.classMethod( klass = <type 'Test'> )
DEBUG:root:<<< Test.classMethod
DEBUG:root:>>> staticMethod()
DEBUG:root:<<< staticMethod
DEBUG:root:>>> Test.test( self = <Test object at 0x774690>, value = 1 )
DEBUG:root:<<< Test.test
DEBUG:root:>>> Test.test( self = <Test object at 0x774690> )
DEBUG:root:<<< Test.test => 1
DEBUG:root:>>> Test.method( self = <Test object at 0x774690> )
DEBUG:root:<<< Test.method
DEBUG:root:>>> Test.test( self = <Test object at 0x774690> )
DEBUG:root:<<< Test.test => 1
Test(1)
DEBUG:root:>>> test( x = 5, y = 5, [z = True] )
DEBUG:root:<<< test => 10
DEBUG:root:>>> test( x = 5, y = 5, z = False )
DEBUG:root:<<< test => 25
DEBUG:trace.main:>>> test( x = 5 )
DEBUG:trace.main:<<< test => 5
DEBUG:trace.main:>>> test( x = 5, 5 )
DEBUG:trace.main:<<< test => 10
DEBUG:trace.main:>>> test( x = 5, 5, False )
DEBUG:trace.main:<<< test => 10
DEBUG:trace.main:>>> test( x = 5, 5, False, name = 10 )
DEBUG:trace.main:<<< test => 10
DEBUG:trace.main:>>> test( x = 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49 )
DEBUG:trace.main:<<< test => 1225
DEBUG:trace.main:>>> __builtin__.zip( sequence = xrange(5), xrange(5, 10) )
DEBUG:trace.main:<<< __builtin__.zip => [(0, 5), (1, 6), (2, 7), (3, 8), (4, 9)]
0 5
1 6
2 7
3 8
4 9
Alan Plum 13 years, 2 months ago  # | flag

Kevin, the markdown syntax for inline code is the backtick (`). It works like underscores/asterisks for emphasis (i.e. place one on each side). It's the same character you use for inline execution in shell commands or database/table/field names in MySQL.