profiler -- numeric overflow?
Tom Loredo
loredo at spacenet.tn.cornell.edu
Tue Jun 27 16:02:07 EDT 2000
Karsten-
This doesn't really address your question, but if you're tired of
seeing all those 0s in your profiler output, I've copied below
a slightly modified version of pstats.py (loaded by the profiler)
that modifies the output to give times in seconds (1.111),
milliseconds (1.111m) or microseconds (1.111u) as appropriate.
Note that if you capture the profiler output for some reason, this
might screw you up (e.g. if you try to read the strings in as numbers
to sort on).
The mods are the f9() function added at the end, and changes
in print_title() and print_line().
-Tom Loredo
~~~~~~~~~~ slightly hacked pstats.py ~~~~~~~~~~~~~
#
# Class for printing reports on profiled python code. rev 1.0 4/1/94
#
# Based on prior profile module by Sjoerd Mullender...
# which was hacked somewhat by: Guido van Rossum
#
# see jprofile.doc and jprofile.py for more info.
# 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 os
import time
import string
import marshal
import re
#**************************************************************************
# Class Stats documentation
#**************************************************************************
# This class is used for creating reports from data generated by the
# Profile class. It is a "friend" of that class, and imports data either
# by direct access to members of Profile class, or by reading in a dictionary
# that was emitted (via marshal) from the Profile class.
#
# The big change from the previous Profiler (in terms of raw functionality)
# is that an "add()" method has been provided to combine Stats from
# several distinct profile runs. Both the constructor and the add()
# method now take arbitrarilly many file names as arguments.
#
# All the print methods now take an argument that indicats how many lines
# to print. If the arg is a floating point number between 0 and 1.0, then
# it is taken as a decimal percentage of the availabel lines to be printed
# (e.g., .1 means print 10% of all available lines). If it is an integer,
# it is taken to mean the number of lines of data that you wish to have
# printed.
#
# The sort_stats() method now processes some additionaly options (i.e., in
# addition to the old -1, 0, 1, or 2). It takes an arbitrary number of quoted
# strings to select the sort order. For example sort_stats('time', 'name')
# sorts on the major key of "internal function time", and on the minor
# key of 'the name of the function'. Look at the two tables in sort_stats()
# and get_sort_arg_defs(self) for more examples.
#
# All methods now return "self", so you can string together commands like:
# Stats('foo', 'goo').strip_dirs().sort_stats('calls').\
# print_stats(5).print_callers(5)
#
#**************************************************************************
import fpformat
class Stats:
def __init__(self, *args):
if not len(args):
arg = None
else:
arg = args[0]
args = args[1:]
self.init(arg)
apply(self.add, args).ignore()
def init(self, arg):
self.all_callees = None # calc only if needed
self.files = []
self.fcn_list = None
self.total_tt = 0
self.total_calls = 0
self.prim_calls = 0
self.max_name_len = 0
self.top_level = {}
self.stats = {}
self.sort_arg_dict = {}
self.load_stats(arg)
trouble = 1
try:
self.get_top_level_stats()
trouble = 0
finally:
if trouble:
print "Invalid timing data",
if self.files: print self.files[-1],
print
def load_stats(self, arg):
if not arg: self.stats = {}
elif type(arg) == type(""):
f = open(arg, 'rb')
self.stats = marshal.load(f)
f.close()
try:
file_stats = os.stat(arg)
arg = time.ctime(file_stats[8]) + " " + arg
except: # in case this is not unix
pass
self.files = [ arg ]
elif hasattr(arg, 'create_stats'):
arg.create_stats()
self.stats = arg.stats
arg.stats = {}
if not self.stats:
raise TypeError, "Cannot create or construct a " \
+ `self.__class__` \
+ " object from '" + `arg` + "'"
return
def get_top_level_stats(self):
for func in self.stats.keys():
cc, nc, tt, ct, callers = self.stats[func]
self.total_calls = self.total_calls + nc
self.prim_calls = self.prim_calls + cc
self.total_tt = self.total_tt + tt
if callers.has_key(("jprofile", 0, "profiler")):
self.top_level[func] = None
if len(func_std_string(func)) > self.max_name_len:
self.max_name_len = len(func_std_string(func))
def add(self, *arg_list):
if not arg_list: return self
if len(arg_list) > 1: apply(self.add, arg_list[1:])
other = arg_list[0]
if type(self) != type(other) or \
self.__class__ != other.__class__:
other = Stats(other)
self.files = self.files + other.files
self.total_calls = self.total_calls + other.total_calls
self.prim_calls = self.prim_calls + other.prim_calls
self.total_tt = self.total_tt + other.total_tt
for func in other.top_level.keys():
self.top_level[func] = None
if self.max_name_len < other.max_name_len:
self.max_name_len = other.max_name_len
self.fcn_list = None
for func in other.stats.keys():
if self.stats.has_key(func):
old_func_stat = self.stats[func]
else:
old_func_stat = (0, 0, 0, 0, {},)
self.stats[func] = add_func_stats(old_func_stat, \
other.stats[func])
return self
# list the tuple indicies and directions for sorting,
# along with some printable description
sort_arg_dict_default = {\
"calls" : (((1,-1), ), "call count"),\
"cumulative": (((3,-1), ), "cumulative time"),\
"file" : (((4, 1), ), "file name"),\
"line" : (((5, 1), ), "line number"),\
"module" : (((4, 1), ), "file name"),\
"name" : (((6, 1), ), "function name"),\
"nfl" : (((6, 1),(4, 1),(5, 1),), "name/file/line"), \
"pcalls" : (((0,-1), ), "call count"),\
"stdname" : (((7, 1), ), "standard name"),\
"time" : (((2,-1), ), "internal time"),\
}
# Expand all abbreviations that are unique
def get_sort_arg_defs(self):
if not self.sort_arg_dict:
self.sort_arg_dict = dict = {}
std_list = dict.keys()
bad_list = {}
for word in self.sort_arg_dict_default.keys():
fragment = word
while fragment:
if not fragment:
break
if dict.has_key(fragment):
bad_list[fragment] = 0
break
dict[fragment] = self. \
sort_arg_dict_default[word]
fragment = fragment[:-1]
for word in bad_list.keys():
del dict[word]
return self.sort_arg_dict
def sort_stats(self, *field):
if not field:
self.fcn_list = 0
return self
if len(field) == 1 and type(field[0]) == type(1):
# Be compatible with old profiler
field = [ {-1: "stdname", \
0:"calls", \
1:"time", \
2: "cumulative" } [ field[0] ] ]
sort_arg_defs = self.get_sort_arg_defs()
sort_tuple = ()
self.sort_type = ""
connector = ""
for word in field:
sort_tuple = sort_tuple + sort_arg_defs[word][0]
self.sort_type = self.sort_type + connector + \
sort_arg_defs[word][1]
connector = ", "
stats_list = []
for func in self.stats.keys():
cc, nc, tt, ct, callers = self.stats[func]
stats_list.append((cc, nc, tt, ct) + func_split(func) \
+ (func_std_string(func), func,) )
stats_list.sort(TupleComp(sort_tuple).compare)
self.fcn_list = fcn_list = []
for tuple in stats_list:
fcn_list.append(tuple[-1])
return self
def reverse_order(self):
if self.fcn_list: self.fcn_list.reverse()
return self
def strip_dirs(self):
oldstats = self.stats
self.stats = newstats = {}
max_name_len = 0
for func in oldstats.keys():
cc, nc, tt, ct, callers = oldstats[func]
newfunc = func_strip_path(func)
if len(func_std_string(newfunc)) > max_name_len:
max_name_len = len(func_std_string(newfunc))
newcallers = {}
for func2 in callers.keys():
newcallers[func_strip_path(func2)] = \
callers[func2]
if newstats.has_key(newfunc):
newstats[newfunc] = add_func_stats( \
newstats[newfunc],\
(cc, nc, tt, ct, newcallers))
else:
newstats[newfunc] = (cc, nc, tt, ct, newcallers)
old_top = self.top_level
self.top_level = new_top = {}
for func in old_top.keys():
new_top[func_strip_path(func)] = None
self.max_name_len = max_name_len
self.fcn_list = None
self.all_callees = None
return self
def calc_callees(self):
if self.all_callees: return
self.all_callees = all_callees = {}
for func in self.stats.keys():
if not all_callees.has_key(func):
all_callees[func] = {}
cc, nc, tt, ct, callers = self.stats[func]
for func2 in callers.keys():
if not all_callees.has_key(func2):
all_callees[func2] = {}
all_callees[func2][func] = callers[func2]
return
#******************************************************************
# The following functions support actual printing of reports
#******************************************************************
# Optional "amount" is either a line count, or a percentage of lines.
def eval_print_amount(self, sel, list, msg):
new_list = list
if type(sel) == type(""):
new_list = []
for func in list:
if re.search(sel, func_std_string(func)):
new_list.append(func)
else:
count = len(list)
if type(sel) == type(1.0) and 0.0 <= sel < 1.0:
count = int (count * sel + .5)
new_list = list[:count]
elif type(sel) == type(1) and 0 <= sel < count:
count = sel
new_list = list[:count]
if len(list) != len(new_list):
msg = msg + " List reduced from " + `len(list)` \
+ " to " + `len(new_list)` + \
" due to restriction <" + `sel` + ">\n"
return new_list, msg
def get_print_list(self, sel_list):
width = self.max_name_len
if self.fcn_list:
list = self.fcn_list[:]
msg = " Ordered by: " + self.sort_type + '\n'
else:
list = self.stats.keys()
msg = " Random listing order was used\n"
for selection in sel_list:
list,msg = self.eval_print_amount(selection, list, msg)
count = len(list)
if not list:
return 0, list
print msg
if count < len(self.stats):
width = 0
for func in list:
if len(func_std_string(func)) > width:
width = len(func_std_string(func))
return width+2, list
def print_stats(self, *amount):
for filename in self.files:
print filename
if self.files: print
indent = " "
for func in self.top_level.keys():
print indent, func_get_function_name(func)
print indent, self.total_calls, "function calls",
if self.total_calls != self.prim_calls:
print "(" + `self.prim_calls`, "primitive calls)",
print "in", fpformat.fix(self.total_tt, 3), "CPU seconds"
print
width, list = self.get_print_list(amount)
if list:
self.print_title()
for func in list:
self.print_line(func)
print
print
return self
def print_callees(self, *amount):
width, list = self.get_print_list(amount)
if list:
self.calc_callees()
self.print_call_heading(width, "called...")
for func in list:
if self.all_callees.has_key(func):
self.print_call_line(width, \
func, self.all_callees[func])
else:
self.print_call_line(width, func, {})
print
print
return self
def print_callers(self, *amount):
width, list = self.get_print_list(amount)
if list:
self.print_call_heading(width, "was called by...")
for func in list:
cc, nc, tt, ct, callers = self.stats[func]
self.print_call_line(width, func, callers)
print
print
return self
def print_call_heading(self, name_size, column_title):
print string.ljust("Function ", name_size) + column_title
def print_call_line(self, name_size, source, call_dict):
print string.ljust(func_std_string(source), name_size),
if not call_dict:
print "--"
return
clist = call_dict.keys()
clist.sort()
name_size = name_size + 1
indent = ""
for func in clist:
name = func_std_string(func)
print indent*name_size + name + '(' \
+ `call_dict[func]`+')', \
f8(self.stats[func][3])
indent = " "
# Modified to note use of f9().
def print_title(self):
print ' ***** m = milliseconds, u = microseconds *****'
print string.rjust('ncalls', 9),
print string.rjust('tottime', 9),
print string.rjust('percall', 9),
print string.rjust('cumtime', 9),
print string.rjust('percall', 9),
print 'filename:lineno(function)'
# Modified to use f9().
def print_line(self, func): # hack : should print percentages
cc, nc, tt, ct, callers = self.stats[func]
c = `nc`
if nc != cc:
c = c + '/' + `cc`
print string.rjust(c, 9),
print f9(tt),
if nc == 0:
print ' '*9,
else:
print f9(tt/nc),
print f9(ct),
if cc == 0:
print ' '*9,
else:
print f9(ct/cc),
print func_std_string(func)
def ignore(self):
pass # has no return value, so use at end of line :-)
#**************************************************************************
# class TupleComp Documentation
#**************************************************************************
# This class provides a generic function for comparing any two tuples.
# Each instance records a list of tuple-indicies (from most significant
# to least significant), and sort direction (ascending or decending) for
# each tuple-index. The compare functions can then be used as the function
# argument to the system sort() function when a list of tuples need to be
# sorted in the instances order.
#**************************************************************************
class TupleComp:
def __init__(self, comp_select_list):
self.comp_select_list = comp_select_list
def compare (self, left, right):
for index, direction in self.comp_select_list:
l = left[index]
r = right[index]
if l < r:
return -direction
if l > r:
return direction
return 0
#**************************************************************************
def func_strip_path(func_name):
file, line, name = func_name
return os.path.basename(file), line, name
def func_get_function_name(func):
return func[2]
def func_std_string(func_name): # match what old profile produced
file, line, name = func_name
return file + ":" + `line` + "(" + name + ")"
def func_split(func_name):
return func_name
#**************************************************************************
# The following functions combine statists for pairs functions.
# The bulk of the processing involves correctly handling "call" lists,
# such as callers and callees.
#**************************************************************************
# Add together all the stats for two profile entries
def add_func_stats(target, source):
cc, nc, tt, ct, callers = source
t_cc, t_nc, t_tt, t_ct, t_callers = target
return (cc+t_cc, nc+t_nc, tt+t_tt, ct+t_ct, \
add_callers(t_callers, callers))
# Combine two caller lists in a single list.
def add_callers(target, source):
new_callers = {}
for func in target.keys():
new_callers[func] = target[func]
for func in source.keys():
if new_callers.has_key(func):
new_callers[func] = source[func] + new_callers[func]
else:
new_callers[func] = source[func]
return new_callers
# Sum the caller statistics to get total number of calls recieved
def count_calls(callers):
nc = 0
for func in callers.keys():
nc = nc + callers[func]
return nc
#**************************************************************************
# The following functions support printing of reports
#**************************************************************************
def f8(x):
return string.rjust(fpformat.fix(x, 3), 8)
# Added to handle short times.
def f9(x):
if (x >= 0.1):
return string.rjust(fpformat.fix(x, 3), 8) + ' '
elif (x >= 0.001):
return string.rjust(fpformat.fix(1.e3*x, 3), 8) + 'm'
else:
return string.rjust(fpformat.fix(1.e6*x, 3), 8) + 'u'
More information about the Python-list
mailing list