profile.py

mallet是自然语言处理、机器学习领域的一个开源项目。

#! /usr/bin/env python
#
# Class for profiling python code. rev 1.0  6/2/94
#
# Based on prior profile module by Sjoerd Mullender...
#   which was hacked somewhat by: Guido van Rossum
#
# See profile.doc for more information

"""Class for profiling Python code."""

# Copyright 1994, by InfoSeek Corporation, all rights reserved.
# Written by James Roskind
#
# Permission to use, copy, modify, and distribute this Python software
# and its associated documentation for any purpose (subject to the
# restriction in the following sentence) without fee is hereby granted,
# provided that the above copyright notice appears in all copies, and
# that both that copyright notice and this permission notice appear in
# supporting documentation, and that the name of InfoSeek not be used in
# advertising or publicity pertaining to distribution of the software
# without specific, written prior permission.  This permission is
# explicitly restricted to the copying and modification of the software
# to remain in Python, compiled Python, or other languages (such as C)
# wherein the modified or derived code is exclusively imported into a
# Python module.
#
# INFOSEEK CORPORATION DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS
# SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
# FITNESS. IN NO EVENT SHALL INFOSEEK CORPORATION BE LIABLE FOR ANY
# SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER
# RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF
# CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
# CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.



import sys
import os
import time
import marshal

__all__ = ["run","help","Profile"]

# Sample timer for use with
#i_count = 0
#def integer_timer():
#       global i_count
#       i_count = i_count + 1
#       return i_count
#itimes = integer_timer # replace with C coded timer returning integers

#**************************************************************************
# The following are the static member functions for the profiler class
# Note that an instance of Profile() is *not* needed to call them.
#**************************************************************************

def run(statement, filename=None):
    """Run statement under profiler optionally saving results in filename

    This function takes a single argument that can be passed to the
    "exec" statement, and an optional file name.  In all cases this
    routine attempts to "exec" its first argument and gather profiling
    statistics from the execution. If no file name is present, then this
    function automatically prints a simple profiling report, sorted by the
    standard name string (file/line/function-name) that is presented in
    each line.
    """
    prof = Profile()
    try:
        prof = prof.run(statement)
    except SystemExit:
        pass
    if filename is not None:
        prof.dump_stats(filename)
    else:
        return prof.print_stats()

# print help
def help():
    for dirname in sys.path:
        fullname = os.path.join(dirname, 'profile.doc')
        if os.path.exists(fullname):
            sts = os.system('${PAGER-more} ' + fullname)
            if sts: print '*** Pager exit status:', sts
            break
    else:
        print 'Sorry, can\'t find the help file "profile.doc"',
        print 'along the Python search path.'


if os.name == "mac":
    import MacOS
    def _get_time_mac(timer=MacOS.GetTicks):
        return timer() / 60.0

if hasattr(os, "times"):
    def _get_time_times(timer=os.times):
        t = timer()
        return t[0] + t[1]


class Profile:
    """Profiler class.

    self.cur is always a tuple.  Each such tuple corresponds to a stack
    frame that is currently active (self.cur[-2]).  The following are the
    definitions of its members.  We use this external "parallel stack" to
    avoid contaminating the program that we are profiling. (old profiler
    used to write into the frames local dictionary!!) Derived classes
    can change the definition of some entries, as long as they leave
    [-2:] intact (frame and previous tuple).  In case an internal error is
    detected, the -3 element is used as the function name.

    [ 0] = Time that needs to be charged to the parent frame's function.
           It is used so that a function call will not have to access the
           timing data for the parent frame.
    [ 1] = Total time spent in this frame's function, excluding time in
           subfunctions (this latter is tallied in cur[2]).
    [ 2] = Total time spent in subfunctions, excluding time executing the
           frame's function (this latter is tallied in cur[1]).
    [-3] = Name of the function that corresponds to this frame.
    [-2] = Actual frame that we correspond to (used to sync exception handling).
    [-1] = Our parent 6-tuple (corresponds to frame.f_back).

    Timing data for each function is stored as a 5-tuple in the dictionary
    self.timings[].  The index is always the name stored in self.cur[-3].
    The following are the definitions of the members:

    [0] = The number of times this function was called, not counting direct
          or indirect recursion,
    [1] = Number of times this function appears on the stack, minus one
    [2] = Total time spent internal to this function
    [3] = Cumulative time that this function was present on the stack.  In
          non-recursive functions, this is the total execution time from start
          to finish of each invocation of a function, including time spent in
          all subfunctions.
    [4] = A dictionary indicating for each function name, the number of times
          it was called by us.
    """

    bias = 0  # calibration constant

    def __init__(self, timer=None, bias=None):
        self.timings = {}
        self.cur = None
        self.cmd = ""

        if bias is None:
            bias = self.bias
        self.bias = bias     # Materialize in local dict for lookup speed.

        if not timer:
            if os.name == 'mac':
                self.timer = MacOS.GetTicks
                self.dispatcher = self.trace_dispatch_mac
                self.get_time = _get_time_mac
            elif hasattr(time, 'clock'):
                self.timer = self.get_time = time.clock
                self.dispatcher = self.trace_dispatch_i
            elif hasattr(os, 'times'):
                self.timer = os.times
                self.dispatcher = self.trace_dispatch
                self.get_time = _get_time_times
            else:
                self.timer = self.get_time = time.time
                self.dispatcher = self.trace_dispatch_i
        else:
            self.timer = timer
            t = self.timer() # test out timer function
            try:
                length = len(t)
            except TypeError:
                self.get_time = timer
                self.dispatcher = self.trace_dispatch_i
            else:
                if length == 2:
                    self.dispatcher = self.trace_dispatch
                else:
                    self.dispatcher = self.trace_dispatch_l
                # This get_time() implementation needs to be defined
                # here to capture the passed-in timer in the parameter
                # list (for performance).  Note that we can't assume
                # the timer() result contains two values in all
                # cases.
                import operator
                def get_time_timer(timer=timer,
                                   reduce=reduce, reducer=operator.add):
                    return reduce(reducer, timer(), 0)
                self.get_time = get_time_timer
        self.t = self.get_time()
        self.simulate_call('profiler')

    # Heavily optimized dispatch routine for os.times() timer

    def trace_dispatch(self, frame, event, arg):
        timer = self.timer
        t = timer()
        t = t[0] + t[1] - self.t - self.bias

        if self.dispatch[event](self, frame,t):
            t = timer()
            self.t = t[0] + t[1]
        else:
            r = timer()
            self.t = r[0] + r[1] - t # put back unrecorded delta

    # Dispatch routine for best timer program (return = scalar, fastest if
    # an integer but float works too -- and time.clock() relies on that).

    def trace_dispatch_i(self, frame, event, arg):
        timer = self.timer
        t = timer() - self.t - self.bias
        if self.dispatch[event](self, frame,t):
            self.t = timer()
        else:
            self.t = timer() - t  # put back unrecorded delta

    # Dispatch routine for macintosh (timer returns time in ticks of
    # 1/60th second)

    def trace_dispatch_mac(self, frame, event, arg):
        timer = self.timer
        t = timer()/60.0 - self.t - self.bias
        if self.dispatch[event](self, frame, t):
            self.t = timer()/60.0
        else:
            self.t = timer()/60.0 - t  # put back unrecorded delta

    # SLOW generic dispatch routine for timer returning lists of numbers

    def trace_dispatch_l(self, frame, event, arg):
        get_time = self.get_time
        t = get_time() - self.t - self.bias

        if self.dispatch[event](self, frame, t):
            self.t = get_time()
        else:
            self.t = get_time() - t # put back unrecorded delta

    # In the event handlers, the first 3 elements of self.cur are unpacked
    # into vrbls w/ 3-letter names.  The last two characters are meant to be
    # mnemonic:
    #     _pt  self.cur[0] "parent time"   time to be charged to parent frame
    #     _it  self.cur[1] "internal time" time spent directly in the function
    #     _et  self.cur[2] "external time" time spent in subfunctions

    def trace_dispatch_exception(self, frame, t):
        rpt, rit, ret, rfn, rframe, rcur = self.cur
        if (rframe is not frame) and rcur:
            return self.trace_dispatch_return(rframe, t)
        self.cur = rpt, rit+t, ret, rfn, rframe, rcur
        return 1


    def trace_dispatch_call(self, frame, t):
        if self.cur and frame.f_back is not self.cur[-2]:
            rpt, rit, ret, rfn, rframe, rcur = self.cur
            if not isinstance(rframe, Profile.fake_frame):
                assert rframe.f_back is frame.f_back, ("Bad call", rfn,
                                                       rframe, rframe.f_back,
                                                       frame, frame.f_back)
                self.trace_dispatch_return(rframe, 0)
                assert (self.cur is None or \
                        frame.f_back is self.cur[-2]), ("Bad call",
                                                        self.cur[-3])
        fcode = frame.f_code
        fn = (fcode.co_filename, fcode.co_firstlineno, fcode.co_name)
        self.cur = (t, 0, 0, fn, frame, self.cur)
        timings = self.timings
        if timings.has_key(fn):
            cc, ns, tt, ct, callers = timings[fn]
            timings[fn] = cc, ns + 1, tt, ct, callers
        else:
            timings[fn] = 0, 0, 0, 0, {}
        return 1

    def trace_dispatch_return(self, frame, t):
        if frame is not self.cur[-2]: