The Python Profiler

To use the profiler in its simplest form:

	>>> import profile
	>>> profile.run(statement)

This will execute the statement and print statistics.  To get more
information out of the profiler, use:

	>>> import profile
	>>> profile.run(statement, dump_file)

where dump_file is a string naming a file to which the (binary)
profile statistics is to be dumped.  The binary format is a dump of a
dictionary.  The key is the function name in the format described
above; the value is a tuple consisting of, in order, number of calls,
total time spent in the function, total time spent in the function and
all functions called from it, a list of functions called by this
function, and a list of functions that called this function.  The dump
can be read back using the following code:

	>>> import marshal
	>>> f = open(dump_file, 'r')
	>>> dict = marshal.load(f)
	>>> f.close()

An easier way of doing this is by using the class `Stats' which is
also defined in profile:

	>>> import profile
	>>> s = profile.Stats().init(dump_file)

The following methods are defined for instances of `Stats':

	print_stats()	-- Print the statistics in a format similar to
			   the format profile.run() uses.
	print_callers()	-- For each function, print all functions
			   which it calls.
	print_callees()	-- For each function, print all functions from
			   which it is called.
	sort_stats(n)	-- Sort the statistics for subsequent
			   printing.  The argument determines on which
			   field the output should be sorted.
			   Possibilities are
				-1	function name
				0	number of calls
				1	total time spent in a function
				2	total time spent in a function
					plus all functions it called
	strip_dirs()	-- Strip the directory names off of the file
			   names which are part of the function names.
			   This undoes the effect of sort_stats(), but
			   a subsequent sort_stats() does work.

The methods sort_stats and strip_dirs may change in the future.

Output of profile.run(statement) and of the print_stats() method of
the `Stats' class consists of the following fields.

	Number of times the function was called.
	Total time spent in the function.
	Mean time per function call (second field divided by first).
	Total time spent in the function and all functions it called,
		recursively.
	Mean time time spent in the function and all functions it
		called (fourth field divided by first).
	Name of the function in the format
		<file name>:<line number>(<function name>)

The output of the print_callers and print_callees methods consists of
the name of the function and the names of all function it called or
was called from.  The latter names are followed by a parenthesised
number which is the number of calls for this function.